Professional Documents
Culture Documents
Users Manual
Analysis Procedures
Release 9.4
TNO DIANA
BV
ii DIANA Finite Element Analysis Users Manual release 9.4 Analysis Procedures Edited by: Jonna Manie and Wijtze Pieter Kikstra Published by: TNO DIANA bv Schoemakerstraat 97, 2628 VK Delft, The Netherlands. Phone: +31 15 27 63 250 Fax: +31 15 27 63 019 E-mail: info@tnodiana.com Web page: tnodiana.com
Trademarks. Diana is a registered trademark of TNO DIANA bv. FemGV, FemGen, FemView are trademarks of Femsys Ltd. CADfix is a registered trademark of TranscenData Europe Limited. Windows is a registered trademark of Microsoft Corporation. PostScript, Acrobat and Acrobat Reader are registered trademarks of Adobe Systems, Inc. AutoCAD is a registered trademark of Autodesk Inc. DXF is a trademark of Autodesk Inc. ACIS is a registered trademark of Spatial Technology Inc. CADDS and Pro/ENGINEER are registered trademarks of Parametric Technology Corporation. CATIA is a registered trademark of Dassault Systemes S.A. IGES is a trademark of IGES Data Analysis, Inc. Parasolid is a registerd trademark of UGS Corporation. PATRAN is a registered trademark of MSC Software Corporation. The X Window System is a trademark of M.I.T. unix is a registered trademark of UNIX Systems Laboratories, Inc. Intel is a registered trademark of Intel Corporation. SUN and Solaris are trademarks or registered trademarks of Sun Microsystems, Inc. HP is a registered trademark of Hewlett-Packard Company. All other brand names, product names or trademarks belong to their respective holders.
First edition, December 15, 2009. Copyright 2009 by TNO DIANA bv, all rights reserved. No part of this publication may be reproduced in any form by print, photoprint, microlm or any other means, without the prior written permission of the publisher. The information in this document is subjected to change without notice and should not be construed as a commitment by TNO DIANA bv. TNO DIANA bv assumes no responsibility for any errors that may appear in this document. The Diana system is the sole property of TNO DIANA bv. Software materials made available are solely for use at a single site; they are not to be distributed to others without prior written permission of TNO DIANA bv.
A This document was prepared with the L TEX Document Preparation System.
Contents at a Glance
Preface Glossary of Symbols xix xxi
1
3 17 59
1 Input of Finite Element Model 2 Input for Structural Analysis 3 General Commands
II
87
89 111
III
Dynamic Analysis
115
117 127 149 175 193
6 Input for Dynamic Analysis 7 Linear Transient Analysis 8 Frequency Response Analysis 9 Spectral Response Analysis 10 Hybrid Frequency Time Domain Analysis
iv
IV
205
207 209
Stability Analysis
277
279 285
VI
299
301 303 311 335 357 365
15 Introduction to Potential Flow Analysis 16 Input for Potential Flow Analysis 17 Heat Flow Analysis 18 Groundwater Flow Analysis 19 Beam Cross-section Analysis 20 Reynolds Flow Lubrication Analysis
VII
371
373 377 383
VIII
Phased Analysis
399
403 421
IX
Solution Procedures
427
429
Background Theory
453
457 465 479 507 521 539 541 549 553 563
28 General Concepts of FEM 29 Solution Procedures of Sparse Linear Systems 30 Solution Procedures for Nonlinear Systems 31 Element Results and Nodal Forces 32 Dynamic Analysis 33 Hybrid Frequency Time Domain Analysis 34 Stability Analysis 35 Potential Flow Analysis 36 SoilPore Fluid Analysis 37 Fracture Mechanics Analysis
XI
Appendix
567
569 605 611
Bibliography Index
615 621
vi
Contents
Preface Glossary of Symbols xxi xxiii
1
3 3 5 5 9 9 11 13 14 14 15 17 17 18 18 19 20 21 24 27 30 33 36 38 39 40
1 Input of Finite Element Model 1.1 Units . . . . . . . . . . . . . . 1.1.1 Special Cases . . . . . 1.1.2 Syntax . . . . . . . . 1.2 Group Specication . . . . . . 1.2.1 Introduction . . . . . 1.2.2 Syntax . . . . . . . . 1.3 Node Coordinates . . . . . . . 1.4 Directions . . . . . . . . . . . . 1.5 Elements . . . . . . . . . . . . 1.6 Material Properties . . . . . .
2 Input for Structural Analysis 2.1 Rigid Supports . . . . . . . . . . . . . . 2.2 Linear Constraints . . . . . . . . . . . . 2.2.1 Nodes and Degrees of Freedom 2.2.2 Single- and Multi-point Tyings 2.2.3 General Input Syntax . . . . . 2.2.4 Equalities . . . . . . . . . . . . 2.2.5 Interconnection . . . . . . . . 2.2.6 Eccentricity . . . . . . . . . . . 2.2.7 General Connection . . . . . . 2.2.8 Automatic Tying . . . . . . . . 2.3 Loads . . . . . . . . . . . . . . . . . . . 2.3.1 Nodal Load . . . . . . . . . . . 2.3.2 Dead Weight . . . . . . . . . . 2.3.3 Centrifugal Load . . . . . . . .
Diana-9.4 Users Manual Analysis Procedures
viii 2.3.4 2.3.5 2.3.6 2.3.7 2.3.8 Fixed Displacements or Deformation Mobile Loads . . . . . . . . . . . . . Wind and Water Load . . . . . . . . Load Sets . . . . . . . . . . . . . . . Inuence Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
CONTENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 42 54 55 56 59 59 60 60 61 64 66 68 69 71 77 80 84 84
3 General Commands 3.1 Control Commands . . . . . . . . . . . . . . . . . 3.2 FILOS File Maintenance . . . . . . . . . . . . . . 3.2.1 Tutorial Introduction . . . . . . . . . . . 3.2.2 Command Syntax . . . . . . . . . . . . . 3.3 Reading or Remaking Input Data . . . . . . . . . 3.4 Element Evaluation . . . . . . . . . . . . . . . . . 3.5 Elements Assembly . . . . . . . . . . . . . . . . . 3.6 Batch Output . . . . . . . . . . . . . . . . . . . . 3.6.1 Model Selection . . . . . . . . . . . . . . 3.6.2 Output Selection . . . . . . . . . . . . . . 3.6.3 Tabular Output . . . . . . . . . . . . . . 3.6.4 Output for Postprocessing with iDIANA 3.6.5 Output for Postprocessing with FX+ . .
II
87
89 90 92 94 96 97 100 106 107 108 109
4 Regular Linear Static Analysis 4.1 Model Evaluation . . . . . . . . . . . . . 4.2 Output of Analysis Results . . . . . . . . 4.2.1 Loads and Extreme Values . . . 4.2.2 Displacements . . . . . . . . . . 4.2.3 Strains . . . . . . . . . . . . . . 4.2.4 Stresses . . . . . . . . . . . . . . 4.2.5 Nodal Forces . . . . . . . . . . . 4.2.6 Internal Nodal Element Forces . 4.2.7 Internal Element Forces . . . . . 4.2.8 Fracture Mechanics Parameters
5 Fatigue Failure Analysis 111 5.1 Number of Load Cycles to Failure . . . . . . . . . . . . . . . . . 111 5.1.1 Stress Amplitudes . . . . . . . . . . . . . . . . . . . . . 112 5.1.2 Output Selection . . . . . . . . . . . . . . . . . . . . . . 112
III
Dynamic Analysis
115
CONTENTS 6.2 6.3 6.4 Damping . . . . . . . . . . . . . . . Base Excitation . . . . . . . . . . . Transient Analysis . . . . . . . . . . 6.4.1 Initial Conditions . . . . . 6.4.2 TimeLoad Diagram . . . . 6.4.3 Ambient Time Dependency Frequency Domain Analysis . . . . 6.5.1 FrequencyLoad Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 118 119 119 121 123 123 123 127 129 130 130 131 131 132 132 134 135 137 138 141 145 146 147 148 149 151 151 152 153 156 157 159 160 161 163 164 165 166 169 172 173
ix
6.5
7 Linear Transient Analysis 7.1 Model Evaluation . . . . . . . . . . . . 7.2 Type of Transient Analysis . . . . . . . 7.3 Step Execution . . . . . . . . . . . . . . 7.3.1 Load Steps . . . . . . . . . . . 7.3.2 Time Steps . . . . . . . . . . . 7.3.3 No Equilibrium Iteration . . . 7.4 Output of Analysis results . . . . . . . 7.4.1 Displacements . . . . . . . . . 7.4.2 Velocities . . . . . . . . . . . . 7.4.3 Accelerations . . . . . . . . . . 7.4.4 Strains . . . . . . . . . . . . . 7.4.5 Stresses . . . . . . . . . . . . . 7.4.6 Nodal Forces and Moments . . 7.4.7 Internal Nodal Element Forces 7.4.8 Internal Element Forces . . . . 7.4.9 Dynamic Pressures . . . . . . .
8 Frequency Response Analysis 8.1 Modal Response Analysis . . . . . . . . . . . . 8.1.1 Model Evaluation . . . . . . . . . . . 8.1.2 Eigenmodes and Natural Frequencies 8.1.3 Response Analysis . . . . . . . . . . . 8.1.4 Example Commands . . . . . . . . . . 8.2 Direct Response Analysis . . . . . . . . . . . . 8.2.1 Model Evaluation . . . . . . . . . . . 8.2.2 Analysis Execution . . . . . . . . . . 8.3 Output of Analysis Results . . . . . . . . . . . 8.3.1 Displacements . . . . . . . . . . . . . 8.3.2 Velocities . . . . . . . . . . . . . . . . 8.3.3 Accelerations . . . . . . . . . . . . . . 8.3.4 Strains . . . . . . . . . . . . . . . . . 8.3.5 Stresses . . . . . . . . . . . . . . . . . 8.3.6 Nodal Forces and Moments . . . . . . 8.3.7 Dynamic Pressures . . . . . . . . . . .
x 9 Spectral Response Analysis 9.1 Model Evaluation . . . . . . . . . . . . 9.2 Eigenmodes and Natural Frequencies . 9.3 Response Analysis . . . . . . . . . . . . 9.3.1 Analysis Execution . . . . . . 9.4 Output of Analysis Results . . . . . . . 9.4.1 Displacements . . . . . . . . . 9.4.2 Velocities . . . . . . . . . . . . 9.4.3 Accelerations . . . . . . . . . . 9.4.4 Strains . . . . . . . . . . . . . 9.4.5 Stresses . . . . . . . . . . . . . 9.4.6 Nodal Forces . . . . . . . . . . 9.4.7 Internal Nodal Element Forces 9.4.8 Internal Element Forces . . . .
CONTENTS 175 176 177 177 178 178 180 181 181 182 185 190 191 192 193 194 194 196 197 198 199 201
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
10 Hybrid Frequency Time Domain Analysis 10.1 Model Evaluation . . . . . . . . . . . . . . . . 10.1.1 Element Matrices . . . . . . . . . . . 10.2 Eigenmodes and Natural Frequencies . . . . . 10.3 Response Analysis . . . . . . . . . . . . . . . . 10.3.1 Reduction of System of Equations . . 10.3.2 Analysis Execution . . . . . . . . . . 10.3.3 Output of Response Analysis Results
IV
205
11 Introduction to Nonlinear Structural Analysis 207 11.1 Input for Nonlinear Analysis . . . . . . . . . . . . . . . . . . . . 207 11.2 Element Choice for Nonlinear Analysis . . . . . . . . . . . . . . 208 12 Nonlinear Analysis 12.1 Model Evaluation . . . . . . . . . . . . . . . 12.2 Type of Nonlinear Analysis . . . . . . . . . . 12.2.1 Physical Nonlinear Analysis . . . . 12.2.2 Geometrical Nonlinear Analysis . . 12.2.3 Transient Nonlinear Analysis . . . . 12.3 Step Execution . . . . . . . . . . . . . . . . . 12.3.1 Initial State Evaluation . . . . . . . 12.3.2 Load Steps . . . . . . . . . . . . . . 12.3.3 Time Steps . . . . . . . . . . . . . . 12.3.4 Physical Nonlinear Analysis Options 12.3.5 Equilibrium Iteration . . . . . . . . 12.3.6 Stop Criteria for Step Execution . . 12.3.7 Logging . . . . . . . . . . . . . . . . 12.3.8 Initial State References . . . . . . .
December 15, 2009 First ed.
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
209 210 210 212 214 216 218 219 225 234 238 240 246 249 250
CONTENTS 12.3.9 12.4 Output 12.4.1 12.4.2 12.4.3 12.4.4 12.4.5 12.4.6 12.4.7 12.4.8 12.4.9 12.4.10 12.4.11 12.4.12 Save/Restore Steps . . . . . . . . . . . . . of Analysis Results . . . . . . . . . . . . . . Step Selection . . . . . . . . . . . . . . . . Strains . . . . . . . . . . . . . . . . . . . . Stresses . . . . . . . . . . . . . . . . . . . . Status . . . . . . . . . . . . . . . . . . . . . Pressure . . . . . . . . . . . . . . . . . . . Temperature, Concentration and Maturity External Forces, Reactions and Residuals . Internal Nodal Element Forces . . . . . . . Internal Element Forces . . . . . . . . . . . Model Parameters . . . . . . . . . . . . . . Fracture Mechanics Parameters . . . . . . Dynamic Pressures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251 252 254 255 260 265 269 270 271 272 273 274 275 276
xi
Stability Analysis
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
277
279 279 280 280 280 281 285 286 286 287 288 290 292 292 295
13 Introduction to Stability Analysis 13.1 Euler Stability Analysis . . . . . 13.2 Perturbation Analysis . . . . . . 13.3 Continuation Analysis . . . . . . 13.4 Input of Initial Imperfections . . 13.5 Applicability of Elements . . . .
14 Euler Stability Analysis 14.1 Model Evaluation . . . . . . . . . . . . 14.2 Buckling Modes . . . . . . . . . . . . . 14.2.1 Stability Analysis . . . . . . . 14.2.2 Imperfections . . . . . . . . . . 14.2.3 Output of Buckling Modes and 14.3 Postbuckling Analysis . . . . . . . . . . 14.3.1 Perturbation Analysis . . . . . 14.3.2 Continuation Analysis . . . . .
. . . . . . . . . . . . . . . . Values . . . . . . . . . . . .
VI
299
301 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303 303 303 304 305 306
15 Introduction to Potential Flow Analysis 16 Input for Potential Flow Analysis 16.1 Nodal Potentials . . . . . . . . . 16.1.1 Fixed Nodal Potentials 16.1.2 Equal Nodal Potentials 16.1.3 Initial Nodal Potentials 16.2 Boundary Conditions . . . . . .
Diana-9.4 Users Manual Analysis Procedures
. . . . .
. . . . .
. . . . .
. . . . .
CONTENTS Nodal Boundary Conditions . . . . . . . . . . . . . . . 307 Transient Boundary Conditions . . . . . . . . . . . . . 307 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311 312 312 312 312 313 313 313 314 316 317 317 318 322 322 324 326 329 331 331 331 332 333 333 334 335 335 335 335 336 336 336 337 337 338 338 338 338 338 339 340
17 Heat Flow Analysis 17.1 General Modeling Aspects . . . . . . . . . . . . . 17.1.1 Meshing . . . . . . . . . . . . . . . . . . . 17.1.2 Conductivity and Capacitance . . . . . . 17.1.3 Heat Sources and Fixed Temperatures . . 17.1.4 Convection and Radiation at Boundaries 17.1.5 Conduction and Convection Coecient . 17.1.6 Temperature and Flow Fields . . . . . . . 17.1.7 Special Features . . . . . . . . . . . . . . 17.2 Steady-state Analysis . . . . . . . . . . . . . . . . 17.2.1 Model Evaluation . . . . . . . . . . . . . 17.2.2 Linear Steadystate Analysis . . . . . . . 17.2.3 Nonlinear Steadystate Analysis . . . . . 17.3 Transient Analysis . . . . . . . . . . . . . . . . . . 17.3.1 Model Evaluation . . . . . . . . . . . . . 17.3.2 Initiate Transient Analysis . . . . . . . . 17.3.3 Time Steps Execution . . . . . . . . . . . 17.4 Output of Analysis Results . . . . . . . . . . . . . 17.4.1 Step Selection . . . . . . . . . . . . . . . 17.4.2 Temperatures . . . . . . . . . . . . . . . 17.4.3 Fluxes . . . . . . . . . . . . . . . . . . . . 17.4.4 Flows . . . . . . . . . . . . . . . . . . . . 17.4.5 Degrees of Reaction . . . . . . . . . . . . 17.4.6 Equivalent Age . . . . . . . . . . . . . . . 17.4.7 Internal Temperature of Cooling Pipe . . 18 Groundwater Flow Analysis 18.1 Detailed Groundwater Flow . . . . . . . . 18.1.1 Meshing . . . . . . . . . . . . . . 18.1.2 Conductivity and Storativity . . 18.1.3 Sources and Sinks . . . . . . . . 18.1.4 Seepage Faces . . . . . . . . . . 18.1.5 Hydraulic Head and Flow Fields 18.1.6 Special Features . . . . . . . . . 18.2 Regional Groundwater Flow . . . . . . . 18.2.1 Meshing . . . . . . . . . . . . . . 18.2.2 Conductivity and Storativity . . 18.2.3 Sources and Sinks . . . . . . . . 18.2.4 Hydraulic Head and Flow Fields 18.2.5 Special Features . . . . . . . . . 18.3 Steadystate Analysis . . . . . . . . . . . 18.3.1 Model Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
CONTENTS 18.3.2 Linear Steadystate Analysis . . 18.3.3 Nonlinear Steadystate Analysis 18.4 Transient Analysis . . . . . . . . . . . . . 18.4.1 Model Evaluation . . . . . . . . 18.4.2 Initiate Transient Analysis . . . 18.4.3 Time Steps Execution . . . . . . 18.5 Output of Analysis Results . . . . . . . . 18.5.1 Step Selection . . . . . . . . . . 18.5.2 Pressures . . . . . . . . . . . . . 18.5.3 Fluxes . . . . . . . . . . . . . . . 18.5.4 Flows . . . . . . . . . . . . . . . 19 Beam Cross-section Analysis 19.1 General Modeling Aspects . . . . . . . . 19.1.1 Meshing . . . . . . . . . . . . . . 19.1.2 Elastic Properties . . . . . . . . 19.2 Analysis Commands . . . . . . . . . . . . 19.2.1 Model Evaluation . . . . . . . . 19.2.2 Cross-section Analysis Execution 19.3 Output of Analysis Results . . . . . . . . 19.3.1 Cross-sectional Properties . . . . 19.3.2 Shear Stresses . . . . . . . . . . 20 Reynolds Flow Lubrication Analysis 20.1 General Modeling Aspects . . . . . . . . 20.1.1 Meshing . . . . . . . . . . . . . . 20.1.2 Viscosity . . . . . . . . . . . . . 20.1.3 Boundary Conditions . . . . . . 20.1.4 Velocity . . . . . . . . . . . . . . 20.2 Analysis Commands . . . . . . . . . . . . 20.2.1 Model Evaluation . . . . . . . . 20.2.2 Lubrication Analysis Execution 20.3 Output of Analysis Results . . . . . . . . 20.3.1 Pressures . . . . . . . . . . . . . 20.3.2 Fluxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 341 341 346 346 348 349 352 353 354 355 356 357 358 358 358 358 358 360 360 361 361 365 366 366 366 366 366 366 366 367 367 368 369
xiii
VII
371
21 Introduction to FlowStress Analysis 373 21.1 Simultaneous Solution with Pore Pressure . . . . . . . . . . . . . 373 21.2 Staggered Solution with Pore Pressure . . . . . . . . . . . . . . . 374 21.3 Staggered Solution with Temperature . . . . . . . . . . . . . . . 375
xiv 22 Staggered Analysis 22.1 How to Perform Staggered Analysis . . . . . . . . . 22.2 Model Denition . . . . . . . . . . . . . . . . . . . . 22.3 Potential Flow Analysis . . . . . . . . . . . . . . . . 22.4 Conversion to Structural Analysis . . . . . . . . . . 22.4.1 Heat and Concentration Flow to Structural 22.4.2 Groundwater Flow to Structural . . . . . . 22.4.3 Checking the Conversion . . . . . . . . . . 22.5 Structural Analysis . . . . . . . . . . . . . . . . . . 23 Mixture Analysis 23.1 Model Input . . . . . . . . . . . . . . . . 23.1.1 Mixture Elements . . . . . . . . 23.1.2 Input of Pore Pressure Potential 23.1.3 Input of Weight Load . . . . . . 23.2 Preliminary Linear Analysis . . . . . . . 23.3 Transient and Nonlinear Analysis . . . . 23.3.1 Initialization . . . . . . . . . . . 23.3.2 Initial Equilibrium . . . . . . . . 23.3.3 Load and Time Steps . . . . . . 23.4 Output of Analysis Results . . . . . . . . 23.4.1 Pressures . . . . . . . . . . . . . 23.4.2 Nodal Discharges . . . . . . . . 23.4.3 Saturation . . . . . . . . . . . . 23.4.4 Darcy Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
CONTENTS 377 377 378 380 380 381 381 382 382 383 383 383 386 389 389 390 390 391 392 394 395 397 397 398
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
VIII
Phased Analysis
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
399
403 404 404 404 405 407 408 410 410 410 411 411 411 414 417 418
24 Phased Structural Analysis 24.1 Input of the Finite Element Model . . . . . 24.1.1 Read First Model . . . . . . . . . 24.1.2 Adapt Model . . . . . . . . . . . . 24.1.3 Import Deformation . . . . . . . . 24.1.4 Initial Velocity . . . . . . . . . . . 24.2 Phase Initialization . . . . . . . . . . . . . 24.2.1 Reinforcement . . . . . . . . . . . 24.2.2 Displacement Constraints . . . . . 24.2.3 Loads . . . . . . . . . . . . . . . . 24.3 Output of Displacements . . . . . . . . . . 24.4 Incrementing Analysis Results . . . . . . . 24.4.1 Unchanged Model . . . . . . . . . 24.4.2 Initial Stress . . . . . . . . . . . . 24.4.3 Model Changes . . . . . . . . . . . 24.5 Fixed Displacements in Nonlinear Analysis
CONTENTS 25 Phased Potential Flow Analysis 25.1 Implementation Backgrounds . . . . . 25.1.1 Basic Equations . . . . . . . 25.1.2 Solution . . . . . . . . . . . . 25.2 Input of the Finite Element Model(s) 25.3 Phase Initialization . . . . . . . . . . 25.4 Transient Analysis . . . . . . . . . . . 25.4.1 Initialization . . . . . . . . . 25.4.2 Time Steps . . . . . . . . . . 25.4.3 Aging . . . . . . . . . . . . . 421 422 422 422 423 424 424 424 424 425
xv
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
IX
Solution Procedures
. . . . Solver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
427
429 430 430 431 431 433 435 439 439 440 442 444 445 446 447 447 447 448 448 448 448
26 Solve System of Equations 26.1 Automatic Solution Method . . . 26.2 PARDISO Parallel Direct Sparse 26.3 Out-of-Core Direct Solution . . . 26.4 Iterative Solution . . . . . . . . . 26.5 Substructuring . . . . . . . . . . . 27 Eigenvalue Analysis 27.1 Model Evaluation . . . . . . . . . 27.2 Type of Eigenvalue Problem . . . 27.2.1 Free Vibration . . . . . . 27.2.2 Standard Eigenproblem . 27.2.3 Linearized Buckling . . . 27.3 Execute Eigenvalue Analysis . . . 27.4 Output of Analysis Results . . . . 27.4.1 Eigenvalues . . . . . . . . 27.4.2 Relative Errors . . . . . . 27.4.3 Generalized Mass . . . . 27.4.4 Participation Factors . . 27.4.5 Eective Mass . . . . . . 27.4.6 Modal Damping Factors . 27.4.7 Eigenmodes . . . . . . . .
Background Theory
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
453
457 457 457 458 458 459
28 General Concepts of FEM 28.1 Global Formulation . . . . . . . . . . . . . 28.1.1 Displacements . . . . . . . . . . . 28.1.2 Strains and Stresses . . . . . . . . 28.1.3 Equilibrium . . . . . . . . . . . . . 28.1.4 Principle of Virtual Displacements
Diana-9.4 Users Manual Analysis Procedures
xvi 28.2 Discretization to Elements . . . . 28.2.1 Displacements . . . . . . 28.2.2 Strains and Stresses . . . 28.2.3 Element Assembly . . . . 28.2.4 Virtual Strain Energy . . 28.2.5 Element Stiness Matrix 28.3 Assembling the Load Vector . . . 28.4 Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
CONTENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460 460 460 461 461 463 464 464 465 466 466 467 467 468 468 468 471 472 473 473 473 473 474 474 475 475 476 479 479 480 486 487 488 491 499 500 502 502 505 505
29 Solution Procedures of Sparse Linear Systems 29.1 Direct Solution Methods . . . . . . . . . . . . . . 29.1.1 Factorization and Fill-in . . . . . . . . . 29.1.2 Sparse Cholesky Method . . . . . . . . . 29.1.3 PARDISO Parallel Direct Sparse Solver 29.2 Eigenvalue Procedures . . . . . . . . . . . . . . . . 29.2.1 Standard Eigenproblem . . . . . . . . . . 29.2.2 Generalized Eigenproblem . . . . . . . . 29.2.3 Shifting . . . . . . . . . . . . . . . . . . . 29.2.4 Eigenvalues and Frequencies . . . . . . . 29.2.5 Accuracy . . . . . . . . . . . . . . . . . . 29.2.6 Solution Techniques . . . . . . . . . . . . 29.3 Iterative Solution Methods . . . . . . . . . . . . . 29.3.1 Conjugate Gradient . . . . . . . . . . . . 29.3.2 Generalized Minimal Residual . . . . . . 29.3.3 Preconditioning . . . . . . . . . . . . . . 29.3.4 Termination Criterion . . . . . . . . . . . 29.3.5 Some Remarks on Practical Use . . . . . 29.4 Substructuring . . . . . . . . . . . . . . . . . . . . 30 Solution Procedures for Nonlinear Systems 30.1 Incremental-Iterative Solution . . . . . . . 30.1.1 Iterative Procedures . . . . . . . . 30.1.2 Continuation . . . . . . . . . . . . 30.1.3 Line Search . . . . . . . . . . . . . 30.1.4 Convergence Criteria . . . . . . . 30.1.5 Incremental Procedures . . . . . . 30.2 Geometric Nonlinearity . . . . . . . . . . . 30.2.1 Denitions . . . . . . . . . . . . . 30.2.2 Total Lagrange . . . . . . . . . . . 30.2.3 Updated Lagrange . . . . . . . . . 30.2.4 Nonconservative Loading . . . . . 30.2.5 Contact Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
CONTENTS 31 Element Results and Nodal Forces 31.1 Element Strains . . . . . . . . . . . . . . . . . . . . . . . . . 31.1.1 Equivalent Von Mises Strain . . . . . . . . . . . . . 31.1.2 Principal Strains . . . . . . . . . . . . . . . . . . . . 31.1.3 Volumetric Strain . . . . . . . . . . . . . . . . . . . 31.1.4 Distributed Seismic Moment . . . . . . . . . . . . . 31.2 Element Stresses . . . . . . . . . . . . . . . . . . . . . . . . . 31.2.1 Equivalent Von Mises Stress . . . . . . . . . . . . . 31.2.2 Principal Stresses . . . . . . . . . . . . . . . . . . . 31.2.3 Principal Shears . . . . . . . . . . . . . . . . . . . . 31.2.4 Plastic Yield . . . . . . . . . . . . . . . . . . . . . . 31.2.5 Pressure . . . . . . . . . . . . . . . . . . . . . . . . 31.2.6 Stress Invariants . . . . . . . . . . . . . . . . . . . . 31.2.7 Reinforcement Moments and Forces . . . . . . . . . 31.2.8 Stress Gradient in Reinforcement Bar . . . . . . . . 31.2.9 Shear Capacity and Hydrostatic Pressure Capacity . 31.3 Nodal Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.3.1 Internal Nodal Forces . . . . . . . . . . . . . . . . . 31.3.2 External Nodal Loads . . . . . . . . . . . . . . . . . 31.3.3 Reaction Forces and Residual Forces . . . . . . . . . 32 Dynamic Analysis 32.1 General Phenomena . . . . . . . . . . . . . . . . . 32.1.1 Mass . . . . . . . . . . . . . . . . . . . . 32.1.2 Damping . . . . . . . . . . . . . . . . . . 32.1.3 Loading . . . . . . . . . . . . . . . . . . . 32.1.4 Base Excitation . . . . . . . . . . . . . . 32.2 Frequency Response . . . . . . . . . . . . . . . . . 32.2.1 Mode Superposition . . . . . . . . . . . . 32.2.2 Direct Solution . . . . . . . . . . . . . . . 32.2.3 Analysis Results . . . . . . . . . . . . . . 32.3 Spectral Response . . . . . . . . . . . . . . . . . . 32.4 Transient Response . . . . . . . . . . . . . . . . . 32.4.1 Newmark . . . . . . . . . . . . . . . . . . 32.4.2 Euler Backward . . . . . . . . . . . . . . 32.4.3 HilberHughesTaylor . . . . . . . . . . . 32.4.4 Wilson- . . . . . . . . . . . . . . . . . . 32.4.5 RungeKutta . . . . . . . . . . . . . . . . 32.5 FluidStructure Interaction Analysis . . . . . . . . 32.5.1 Solid . . . . . . . . . . . . . . . . . . . . 32.5.2 Fluid . . . . . . . . . . . . . . . . . . . . 32.5.3 Discretized Coupled Equations . . . . . . 32.5.4 Solution of Coupled System . . . . . . . . 32.5.5 Simplication for Fixed Fluid Boundaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507 507 507 508 508 508 510 510 510 511 511 511 511 512 518 519 519 519 520 520 521 521 522 522 523 523 523 524 525 525 526 528 529 529 530 530 531 532 532 533 535 536 537
xvii
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
xviii
CONTENTS 33 Hybrid Frequency Time Domain Analysis 539 33.1 General Approach . . . . . . . . . . . . . . . . . . . . . . . . . . 539 34 Stability Analysis 541 34.1 Linear Buckling (Euler) . . . . . . . . . . . . . . . . . . . . . . . 542 34.2 Imperfections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545 34.3 Postbuckling Analysis . . . . . . . . . . . . . . . . . . . . . . . . 546 35 Potential Flow Analysis 35.1 Basic Equations . . . . . . . . . . . . . 35.1.1 ConvectionDiusion Equation 35.1.2 Boundary Conditions . . . . . 35.1.3 Finite Element Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549 549 549 549 550 553 553 554 554 555 555 555 556 557 557 558 559 560 561 563 563 564 564 565 565
36 SoilPore Fluid Analysis 36.1 Basic Equations and Assumptions . . . . . . . . . . . . 36.1.1 Porosity, Saturation and Apparent Density . . 36.1.2 Undrained Compressibility . . . . . . . . . . . 36.1.3 Stress Separation and Pore Pressure Potential 36.1.4 Momentum Conservation . . . . . . . . . . . . 36.1.5 Mass Conservation . . . . . . . . . . . . . . . . 36.1.6 Darcy Flow . . . . . . . . . . . . . . . . . . . . 36.1.7 Elastoplastic Stiness . . . . . . . . . . . . . . 36.2 Finite Element Equations . . . . . . . . . . . . . . . . . 36.2.1 Space Discretization . . . . . . . . . . . . . . . 36.2.2 Time Discretization . . . . . . . . . . . . . . . 36.2.3 Undrained Behavior . . . . . . . . . . . . . . . 36.2.4 Hydraulic Pore Pressure Load . . . . . . . . . 37 Fracture Mechanics Analysis 37.1 Stress Intensity Factor . . . . . . . . 37.1.1 Mesh Adaptation . . . . . 37.2 Energy Release Rate . . . . . . . . . 37.2.1 Two-dimensional Analysis . 37.2.2 Three-dimensional Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
XI
Appendix
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
567
569 570 570 572 573 574 575
A Using User-supplied Subroutines A.1 Supplying the Source Code . . . A.1.1 Direct Inclusion . . . . A.1.2 External File(s) . . . . A.2 Setting the User Environment . A.2.1 Environment Symbols . A.2.2 Files . . . . . . . . . . .
December 15, 2009 First ed.
CONTENTS A.3 A.4 Available Routines . . . . . . . Programmers Service Libraries A.4.1 Matrix Manipulation A.4.2 Vector Manipulation . A.4.3 Miscellaneous . . . . . A.4.4 Database I/O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 575 576 576 586 591 603 605 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 611 611 611 612 612 613
xix
B Available Element Types C Physical Properties Forms C.1 Mobile . . . . . . . . . . . . . C.1.1 Dutch VBB Code . . C.1.2 Dutch VOSB Code . C.1.3 European ENV Code C.1.4 General . . . . . . . .
Bibliography Index
615 621
xx
CONTENTS
Preface
Part I of this volume formally presents the general input for the element model, the additional input for structural analysis and the general analysis commands. Parts II to V deal with the execution of the various types of structural analysis. Part VI and Part VII respectively tell you how to perform potential ow and coupled owstress analysis. The next two parts describe features that are applicable for structural as well as for ow analysis: Part VIII tells you how to perform a phased analysis, Part IX outlines the application of various solution procedures. Part X gives background theory for many of the features described in the previous parts. This part also references publications, listed in the bibliography, for further reading. In the appendices of Part XI you will nd general information on Diana like notation conventions in the manuals, how to run a job and programmers service libraries. Related volumes. Novice users should read Volume Getting Started and browse through Volume Analysis Examples before going into details of this volume. Volume Element Library describes the elements that Diana oers you to create a nite element model, including the necessary input for these elements like geometric properties, loading and specic options to customize the behavior of the element.1 Volume Material Library is a reference manual for the available material models, it describes the input of these models and their basic background theory. Cautionary note Throughout this manual, it will be assumed that the reader has a basic understanding of computational mechanics and the Finite Element Method.2
Appendix B for a short overview of all available elements. informative introductions are the Guidelines to Finite Element Practice [42] and the book A Finite Element Primer [44], both published by NAFEMS.
2 Very
Diana-9.4 Users Manual Analysis Procedures December 15, 2009 First ed.
1 See
xxii
Preface
Glossary of Symbols
Scalars
3
H Hardening parameter. I Moment of inertia [m4 ]. J Creep function [m2 /N]. K Bulk or compression modulus [N/m2 ]. K Conduction coecient [W/(m2 K)]. K Spring stiness [N/m]. K Stress intensity factor (LEFM) []. K0 Lateral pressure ratio []. KD Drained compression modulus [N/m2 ]. Kf Fluid compression modulus [N/m2 ]. Ks Solid compression modulus [N/m2 ]. L Area coordinates. M Bending moment (Bernoulli) [N m]. M Concentrated mass [kg]. M Concentrated moment [N m]. M Maturity. M R Reaction moment [N m]. M r Residual moment [N m]. N Interpolation polynomial [].
A Area, cross-section [m2 ]. B Bandwidth of matrix []. B Boundary [m2 ]. C Concentration. C Damping coecient. C Hardening constant. Cc Compression index. D Interface stiness modulus [N/m3 ]. D Linear stiness modulus [N/m2 ]. DII Crack shear stiness [N/m2 ]. E Youngs modulus [N/m2 ]. Ehar Hardening modulus [N/m2 ]. Ep Hardening modulus [N/m2 ]. F Concentrated force [N]. F R.m.s. wavefront of matrix []. F R Reaction force [N]. F r Residual force [N]. G Shear modulus [N/m2 ]. G Energy release rate (LEFM). Gc Compressive fracture energy [N/m]. Gf Fracture energy [N/m2 ]. H Enthalpy [J/m3 ].
3 SI-units
in brackets.
December 15, 2009 First ed.
xxiv N Normal force (Bernoulli) [N]. N Number of nodes []. N Order of matrix [].
OCR
Glossary of Symbols Z Third global Cartesian coordinate [m], or axis. a a a b b c c c c cA ce cp co d e e e e f f f fc fcc fck fcm fk ft g h h hcr i j k Damping parameter. Mode amplitude. Crack length (LEFM) [m]. Damping parameter. Width [m]. Capacitance [J/(m3 K)]. Cohesion [N/m2 ]. Incompressibility penalty factor. Wave speed [m/s]. Arrhenius constant [K]. Elastic storativity. Phreatic storativity []. Reinforcement coverage [m]. Diameter [m]. Deviatoric strain []. Element number []. Emissivity []. Void ratio []. Distributed force [N/m, N/m2 ]. Natural frequency [1/s]. Volume source. Compressive strength [N/m2 ]. Cube compressive strength [N/m2 ]. Characteristic strength [N/m2 ]. Mean compressive strength [N/m2 ]. Characteristic strength [N/m2 ]. Tensile strength [N/m2 ]. Acceleration of gravity [m/s2 ]. Height [m]. Phreatic level [m]. Crack bandwidth [m]. Counter []. Counter []. Conductivity [W/(m K)].
Diana-9.4 Users Manual Analysis Procedures
P Perimeter [m]. P Potential. P Prole of matrix. Pex External potential (LEFM). Q Discharge [m3 /s]. Q Produced heat [J/m3 ]. Q Second Biot material parameter [N/m2 ]. Q Shear force (Bernoulli) [N]. RH Relative humidity []. Re Reynolds number []. S Boundary []. S Degree of saturation []. S Second PiolaKirchho stress [N/m2 ]. S Shear stress correction factor []. S Surface area [m2 ]. SA Spectral acceleration [m/s2 ]. SD Spectral displacement [m]. T Period of time [s]. T Temperature [K]. T Transmissivity. V Volume
[m3 ]
W Maximum wavefront of matrix []. W Moment of rigidity W Section modulus. W Strain energy function [J]. Win Elastic energy (LEFM) [J]. X First global Cartesian coordinate [m], or axis. Y Second global Cartesian coordinate [m], or axis.
December 15, 2009 First ed.
[m3 ]
xxv k Creep factor []. k Interface stiness modulus [N/m3 ]. k Permeability [m2 ]. k Modied permeability [m4 /(N s)]. l Length [m]. m Distributed moment [N m/m]. m Reinforcement moment. n Degree of nonlinear elasticity []. n Distributed in-plane force [N/m]. n Node number []. n Order of system matrix []. n Porosity []. n Traction axis, normal. p Isotropic strain invariant []. p Polynomial. p Pore pressure [N/m2 ]. p Pressure
[N/m2 ]
s t t t t teq u u u v v0 vc x y z t
Traction axis. Thickness [m]. Time [s]. Traction axis. Traction [N/m2 ]. Equivalent age of concrete [s]. Translational displacement [m]. Velocity [m/s]. Acceleration [m/s2 ]. Speed (wind, water) [m/s]. Specic volume []. Convection velocity [m/s2 ]. First local Cartesian coordinate [m], or axis. Second local Cartesian coordinate [m], or axis. Third local Cartesian coordinate [m], or axis. Increment of t [s]. Boundary (surface). Euler gamma function. Domain. Excitation frequency [Hz]. Rotation speed [rad/s]. Energy potential [J]. Generalized strain.
p Eective pressure [N/m2 ]. pc Preconsolidation stress [N/m2 ]. pe Excess pore pressure [N/m2 ]. phead Pressure head [m]. q Deviatoric strain invariant []. q Deviatoric stress [N/m2 ]. q Distributed shear force [N/m]. q Flux. q Specic discharge, distributed source. r Degree of reaction []. r Distance from crack tip (LEFM) [m]. r Maturity variable []. r Radius [m]. s Deviatoric stress [N/m2 ]. s Length along crack front (LEFM) [m].
Diana-9.4 Users Manual Analysis Procedures
First Biot material parameter []. General diusivity. Generalized modal displacement. Preconditioning parameter. Shape factor. Thermal expansion coecient [1/K]. Time integration parameter. Shear retention factor []. Compressibility parameter. Concentration expansion coecient.
December 15, 2009 First ed.
xxvi Decaying factor. Deviator strain []. GreenLagrange strain []. Iteration parameter. Shape factor. Shear deformation []. Structural damping factor. Modal participation factor. Volumetric weight [N/m3 ]. f Volumetric uid weight [N/m3 ]. p Deviatoric plastic strain. Extension, elongation [m]. Time integration parameter. ij Kronecker delta []. t Change in t [s]. Convergence criterion []. Logarithmic strain []. Tolerance. Engineering strain [].
C T
Glossary of Symbols f cr eq y head p Compression index. Eigenvalue. Loading parameter. Plastic multiplier []. Relaxation time [s]. Dynamic viscosity [N s/m2 ]. Friction coecient (Coulomb). Shifting factor []. Spring stiness [N/m]. Kinematic viscosity [m2 /s]. Poissons ratio []. Damping ratio []. First parametric coordinate []. Mass density [kg/m3 ]. Fluid density [kg/m3 ]. StefanBoltzmann constant [J/(m2 s K4 )]. Stress (Cauchy) [N/m2 ]. Eective stress [N/m2 ]. In-situ stress [N/m2 ]. Crack stress [N/m2 ]. Equivalent stress (Von Mises) [N/m2 ]. Yield stress [N/m2 ]. Dimensionless time []. Curvature (Bernoulli). Friction angle. Pore pressure potential [N/m2 ]. Potential. Rotational displacement. Hydraulic head [m]. Pressure component of hydraulic head. Phase angle. Dilatancy angle. Natural angular frequency. Reinforcement percentage. Rotational velocity (spin).
Diana-9.4 Users Manual Analysis Procedures
cr Crack strain []. c Creep strain []. e Elastic strain []. p Plastic strain []. sh Shrinkage strain []. Emission coecient []. Third parametric coordinate. Second parametric coordinate. Viscosity. Compression modulus. Curvature (Bernoulli) [1/m]. Equivalent plastic strain []. Hardening parameter. Swelling index. Buckling value.
December 15, 2009 First ed.
xxvii Vectors, Tensors a Pseudo-acceleration vector [m/s2 ]. f Distributed forces. f Equivalent element forces. f Load vector. fw Dead weight load. f Discharge vector [m3 /s]. fu Force vector [N]. g Body forces per unit volume. g Gravity acceleration vector [m/s2 ]. g Out-of-balance forces. i Unity vector. k Conductivity tensor. m Distributed moments. n Normal vector. n Generalized element forces. p Pressure vector. q Flux vector [m/s]. r Internal forces, reactions. r Residual vector. t Traction [N/m2 ]. u Displacement vector [m]. u Velocity vector [m/s]. u Acceleration vector [m/s2 ]. x Position vector [m]. y Result of forward substitution. E Strain tensor. Convective velocity eld. GreenLagrange strains. Strain vector. Stress vector [N/m2 ]. 2nd PiolaKirchho stresses. Buckling modes.
Diana-9.4 Users Manual Analysis Procedures
Eigenvector. Pressure potentials vector [N/m2 ]. Phase angles. Matrices B Interpolation matrix. B Straindisplacement relation. C Capacity matrix. C Compliance matrix. C Coupling matrix. C Damping matrix. C Right CauchyGreen stretch tensor. D Diagonal matrix. D Elasticity matrix. D Material stiness matrix. D Rigidity matrix. E GreenLagrange strain tensor. F Deformation gradient. H Modied elastic stiness matrix. I Identity (unity) matrix. J Jacobian matrix. K Conductivity matrix. K Permeability matrix. K Stiness matrix. L Dierential operator. L Lower triangular matrix. M Mass matrix. N Interpolation matrix. P Preconditioning matrix. Q Preconditioning matrix. R Rotation matrix. S Second PiolaKirchho stress tensor. S Stress matrix. S System directions.
December 15, 2009 First ed.
xxviii T Element directions. T Transformation matrix. U Upper triangular matrix. Eigenvalue diagonal matrix. Accents, sub- and superscripts u Average of . . . u Second time derivative of . . . u First time derivative of . . . u Amplitude of . . . Added . . . M KB . . . for the boundary. KF . . . for the uid domain. KI . . . for the interface. KS . . . for the structural domain. 123 Principal . . .
Glossary of Symbols uXY Z . . . in global orientation. uxyz . . . in local orientation. u(e) . . . in element orientation. u(s) . . . in system orientation. n Normal . . . t Tangential . . . Ke . . . for an element. Sn . . . for a node. fdy Dynamic . . . fext External . . . fint Internal . . . fst Static . . . K Eective . . . 0 Initial . . . . u Imaginary part of . . . u Real part of . . .
u
Phase angle of . . .
Part I
Chapter 1
1.1
Units
The input for Diana is free from units: you must choose a consistent system of units and enter all data in units from that system. Diana gives the output in the units that you chose for the input.
Because of its clear consistency we strongly advise to use the SI-unit system.1 However, other systems could also be used, for instance those shown in Table 1.1 on the following page. Because of its unnatural consistency for time, velocity and acceleration the mm-kg-N system should be used with great care in time-dependent analysis. Similarly, we advise against the use of customary inconsistent unit systems, like US and British Imperial in time-dependent analysis. Table 1.2 on the next page gives some values for constants in the various consistent unit systems. See also the books on unit conversion by Horvath [29] and Cardarelli [10].
1 SI:
Syst`me International e
Unit Length Mass Time Temperature Velocity Acceleration Force Stress Energy
derivation l m t T l t1 l t2 m l t2 m l1 t2 m l2 t2
SI m kg s K m/s m/s2 N Pa J
Unit Length Mass Time Temperature Velocity Acceleration Force Stress Energy
derivation l m t T l t1 l t2 m l t2 m l1 t2 m l2 t2
Unit system mm-kg-s mm-N-day mm kg s K mm/s mm/s2 mN mPa J mm 7.464961012 kg day C mm/day mm/day2 N MPa mJ
1.000106 1.000 1.000 31.78109 1.005106 10.05103 2.174103 2.174106 21.74109 3.62 2.5 220103 3.62103 79.103 220106 36.2106 7.9106 2.201012
1.1 Units
1.1.1
Special Cases
In some cases, outlined below, you must specify which system of units you used. You may specify the units of the input data via table UNITS [ 1.1.2]. If you dont, Diana assumes the SI-units system (m, kg, s, K). Material codes. In some cases of Diana input data, the material properties may be specied by the name of an international standard code, for instance MC1990 for the modeling of concrete creep according to the European CEB-FIP Model Code 90 as explained in Volume Material Library. Normally these codes require certain units to be used, for instance length in millimeters and time in hours. Diana translates the units of the input data to the units of the code. Unless you use SI-units for input data, you must specify your input units via table UNITS [ 1.1.2]. Steel proles. If you use the library of proles to indicate the cross-section of beam elements, it is important to realize that Diana converts the prole data from the units used in the system library les to the units used in the nite element model. It is good practice to specify table UNITS whenever you use the prole library. If you dont, Diana assumes that the model is input in SI-units, which may not be correct. See also Chapter Beam Elements in Volume Element Library. Staggered groundwater owstress analysis. In staggered groundwater owstress analysis Diana assumes an acceleration of gravity g = 9.81 m/s2 [ 22.4.2 p. 381]. Correct translation to the actual units of the model requires the applied units. Therefore you must specify these if other than SI.
1.1.2
Syntax
In some cases it is necessary to tell Diana via table UNITS in which units the input data is specied [ 1.1.1]. syntax UNITS
1 80
[ LENGTH ulen w lef r [le0 r ] [ MASS ] umas w maf r [ma0 r ] ] [ FORCE ufor w fof r [fo0 r ]
[M]
LENGTH species that the used unit of length is ulen : MM for millimeter, CM for centimeter, DM for decimeter, M for meter (the default), KM for kilometer, IN for inch, FT for foot, YD for yard, or MI for statute mile. The unit of length may be specied alternatively by factor lef and oset le0 for conversion to meters according to lm = lef linput + le0 . The default oset is zero. For example if the length is input in mils (in/1000) then le0 = 0 and lef 25.4106 . MASS species that the used unit of mass is umas : G for gram, KG for kilogram (the default), T for ton (1000 kg), OZ for ounce (UK), LB for pound (UK), KLB for kilopound (UK) or LTN for ton (UK) or long ton (US). The unit of mass may be specied alternatively by factor maf and oset ma0 for conversion to kilogram according to mkg = maf minput + ma0 . The default oset is zero. For example if the mass is input in stones (14 lb) then ma0 = 0 and maf 6.3503 . Instead of the unit of mass, you may specify the unit of force. Specication of the mass unit overrules specication of the force unit, i.e., if you specify both MASS and FORCE then Diana ignores the FORCE specication.
[le0 = 0]
[KG]
[ma0 = 0]
[N]
FORCE species that the used unit of force is ufor : N for newton (the default), KN for kilo-newton, DYN for dyne (105 N), KGF for kilogram-force, KP for kilo-pound (kgf), MT for metric-ton (MT), PDL for poundal (pdl), P for pound (gf), LBF for pound-force (lbf) or KLBF for kilo-pound (klbf). To derive the unit of mass from the specied unit of force, Diana applies an acceleration of gravity g = 9.80665 m/s2 . The unit of force may be specied alternatively by factor fof and oset fo0 for conversion to newton according to fN = fof finput + fo0 . The default oset is zero. TIME species that the used unit of time is utim : SEC for second (the default), MIN for minute, HOUR for hour, DAY for day, or YEAR for year (= 365 days). The ti0 The ti0 unit of time may be specied alternatively by factor tif and oset for conversion to seconds according to tsec = tif tinput + ti0 . default oset is zero. For example if the time is input in weeks then = 0 and tif = 7 24 3600 = 6.048105 .
[fo0 = 0] [SEC]
[ti0 = 0]
1.1 Units TEMPER species that the used unit of temperature is utem : CELSIU for degree Celsius (0 C = 273.15 K), KELVIN for kelvin (absolute temperature, the default) or FAHREN for degree Fahrenheit. The unit of temperature may be specied alternatively by factor tef and oset te0 for conversion to kelvin according to TK = tef Tinput + te0 . The default oset is zero. For example if the temperature is input in degrees Fahrenheit then te0 255.37 and tef 0.5556 . Default (SI)
UNITS LENGTH MASS TIME TEMPER M KG SEC KELVIN file .dat
[te0 = 0]
[KELVIN]
If you do not specify a particular unit or no table UNITS at all, Diana assumes the SI-units, see Table 1.1, which is equivalent to the example input above. mm-g-N
UNITS LENGTH MM FORCE N file .dat
In the example above the unit of mass is set to gram, leading to the mm-g-N unit system, see Table 1.1. m-t-kN
UNITS FORCE kN file .dat
In the example above leads to the m-t-kN unit system, see Table 1.1. The unit of mass is ton (1000 kg). CGS
UNITS LENGTH CM MASS G file .dat
In the example above leads to the CGS unit system, see Table 1.1. The unit of force is dyne.
8 mm-t-N
UNITS LENGTH MM FORCE N
In the example above leads to the mm-t-N unit system, see Table 1.1. The unit of mass is ton (1000 kg). mm-kg-s
UNITS FORCE kN file .dat
In the example above leads to the mm-kg-s unit system, see Table 1.1. The unit of force is kN . mm-N-day
UNITS TIME DAY LENGTH MM MASS 7.46496E+12 file .dat
For nonlinear analysis of creep and shrinkage of concrete models, Diana oers the CEB-FIP Model Code 1990, see for instance the rcbeam example in Volume Concrete and Masonry Analysis. The CEB-FIP regulations are commonly expressed in newton, millimeter, and day for force, length, and time respectively. These can be conceived as the extrinsic set of units. The intrinsic set of units in Diana are newton, meter, and second. The correction on the mass is then determined as follows. The extrinsic set of units for force is maf mass length time2 kg m The intrinsic set of units for force is s2 maf 1 kg 1 mm kg m =1 (1 day)2 s2 maf = 864002 = 7.464961012 103
You may input the mass factor maf as shown in the example above. However, it is more convenient to let the above calculation to Diana by specifying the unit of force directly:
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (I)
1.2
Group Specication
Specication of groups in a table of the input data le is an optional but powerful tool to reduce and simplify the input of element, node and reinforcement numbers in commands and input tables.
1.2.1
Introduction
A name and a number may be assigned to a group of elements, nodes or reinforcements in input table GROUPS. Input
GROUPS ELEMEN 1 ROOF file .dat
/ 3-10 /
This input data gives group number 1 the name ROOF and species that this group contains elements 3 up to and including 10. Nesting
GROUPS ELEMEN 1 ROOF1 2 ROOF2 3 ROOFS file .dat
Groups may be nested like in the example above. Nesting must be backward: groups ROOF1 and ROOF2 must be specied prior to group ROOFS. Automatic numbering
GROUPS ELEMEN 0 FLOORS 0 WALLS 0 ROOF file .dat
10
A group number zero indicates automatic numbering. The group number is incremented by one (highest group number + 1). In the example, FLOORS will be group one, WALLS group two and ROOF group three. Reference from commands
BEGIN SELECT BEGIN ELEMENTS 1 ROOF 17 20 END ELEMENTS END SELECT file .dcf
file .dcf BEGIN SELECT BEGIN ELEMENTS 1 END ELEMENTS END SELECT 3 17 20
Groups may be referenced in commands and input tables either by number or by name. Specication of the group name refers to all the elements in that group. Reference to a group number is done by placing the group number in braces, to distinguish it from an ordinary element number. Note: although formally correct, reference by number to automatically numbered groups is a bit obscure and therefore not recommended. Reference from input
ELEMEN CONNEC MATERI / 235-290(3) ROOF / 1 / 128 129 3-5 / 2 file .dat
11
Reference to groups by number and by name may be mixed with reference to elements, nodes or reinforcements. Elements to nodes
GROUPS ELEMEN 1 FLOORS file .dat
/ 1-120 200-250 /
Groups of elements may be embedded in sets of nodes. Group FLOORS, although specied with elements may be referred in a node selection command.
file .dcf BEGIN SELECT NODES FLOORS END SELECT
This command selects all the nodes attached to the elements in group FLOORS.
1.2.2
Syntax
syntax
GROUPS subtbl w
1 5 6 80
grnr n
subtbl is the subtable name: ELEMEN for groups of elements, NODES for groups of nodes, REINFO for groups of reinforcements, or COMPOS for groups of composed elements. Subtable names may occur in arbitrary order and even more than once.
12
Input of Finite Element Model grnr is the group number. If grnr > 0 then it is unique over all subtables. A grnr = 0 indicates an automatic number: Diana adds one to the current highest group number to get the number of the new group. Groups may be specied in arbitrary order and numbers may be skipped. grnam is the group name, unique over all subtables. The group name must start with a letter and may not have more than twenty characters. The group name END is not allowed. Actual contents of the group must be specied within slashes / /, with an arbitrary mixture of: nrs element, node or reinforcement numbers (depends on type of subtable). grnams group names, referring to previously specied groups in the same subtable. grnrs group numbers, specied within braces { }, also referring to previously specied groups in the same subtable. The braces must be surrounded by spaces.
Multiple specied numbers will be ltered out. Numbers may be specied in arbitrary order. No assumption shall be made of internal ordering of numbers in a group.
file .dat GROUPS ELEMEN 1 FLOOR1 4 "FLOOR 2" 5 FLOORS 8 ROOF NODES 2 UPPEREDGE 3 LOWEREDGE 9 EDGE1 10 EDGE2 11 EDGES REINFO 6 BAR1 7 BAR2 13 GRID1 14 GRID2
/ / / / / / / / / / / / /
Notation convention. In syntax descriptions, the possibility to refer to groups in a range of numbers is indicated with a subscript ng behind the variable:
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (I)
1.3 Node Coordinates syntax numbrs ng... An arbitrary mixture of numbers and group references may be specied for variable numbrs .
13
1.3
Node Coordinates
Nodal coordinates are input via table COORDI. The coordinates are expressed in global Cartesian XY Z axes. syntax COORDI [ DI=dimens n ]
1 5 6 80
node n
x r y r [z r ]
DI=dimens is an optional parameter which indicates the dimensionality of the nodal coordinate system. Specify DI=3 for a three-dimensional, or DI=2 for a two-dimensional system. The default is three-dimensional. node is the node number. Nodes may be input in arbitrary order, numbers may be skipped. Values x , y and z are the model XY Z coordinates. You must omit z for a two-dimensional system but specify it for a three-dimensional system. Two-dimensional coordinates
COORDI DI=2 1 1.0 0.5 3 2.0 0.0 file .dat
[DI=3]
[z = 0]
14
1.4
Directions
Input table DIRECT contains directions which are specied as vectors in the model coordinate system [ 1.3]. The directions in table DIRECT have no meaning in themselves. They are used for reference from other tables, for example in order to specify the direction of a support or a nodal loading. syntax DIRECT
1 5 6 80
dirnr n
xr yr zr
dirnr is the direction number. Directions may be input in arbitrary order, numbers may be skipped on the condition that they are not referred to from other tables. Values x , y and z are the model XY Z components of the direction vector. It is not necessary that the direction vectors are unit vectors, i.e., they do not have to be normalized. Model XY Z directions (default)
DIRECT 1 1.0 2 0.0 3 0.0 0.0 1.0 0.0 0.0 0.0 1.0 file .dat
Diana assumes three default directions: number 1 the model X axis, number 2 the model Y axis and number 3 the model Z axis. This is equivalent to the input shown above. You may overwrite the default directions by specifying other values in table DIRECT, like in the following example. Specied directions
DIRECT 1 0.0 1.0 2 0.0 2.0 4 1.0 1.0 5 1.0 0.3 6 0.0 -1.3 0.0 0.0 0.0 4.0 0.0 file .dat
In the above example 1 is the model Y axis (overwrites the default!), 2 is also the model Y axis, 4 is 45 with X and Y axis, 5 is an arbitrary direction in threedimensional space and 6 is the negative model Y axis. In this case direction 3 still represents the (default) Z axis.
1.5
Elements
15
1.6
Material Properties
16
Chapter 2
2.1
Rigid Supports
In table SUPPOR the rigid supports are given. Rigid supports have displacements equal to zero, unless a xed displacement is specied in subtable DEFORM of table LOADS [ 2.3.4 p. 41]. Each support is specied by a node number a type (translation or rotation) and a direction. syntax SUPPOR
1 5 6 80
node n
1
/ nodes ng... / type w dirnrs n... {type w dirnrs n... . . . } node is a single node number, nodes is a series of nodes with the same support type and direction. The series of nodes must be specied between slashes and may comprise numbers and/or groups. The support type is specied with type : TR for translation or RO for rotation (clamped). The support direction is specied with a number referring to table DIRECT [ 1.4]. The translation in, or the rotation
Diana-9.4 Users Manual Analysis Procedures (I) December 15, 2009 First ed.
18
Input for Structural Analysis around the specied direction is prevented. Direction numbers dirnrs may be input with a single number or with a series of numbers. The series means the same type of support for multiple directions. Supports input
SUPPORTS 12 TR 3 24 TR 2 RO 1 / 25-38 / TR 4 / 50-56 EDGE1 / TR 1 2 RO 3 file .dat
2.2
Linear Constraints
Linear constraints are user specied linear dependencies between degrees of freedom of the system of equations (displacements, rotations, temperatures etc.). These dependencies are specied in input table TYINGS, and in Diana terms called tyings. Some examples of the application of tyings are: hinges, sliding connections, symmetry, mesh renement, connection in case of incompatibility, eccentric connection, keeping edges or planes straight. Before you go into detail on these topics, keep in mind what NAFEMS [42] says on this point: Constraint equations or displacement transformations, as required by the analysis system, must be formulated with extreme care and geometric consistency to the full analysis accuracy. It is better to truncate the nodal geometry data to ensure transcription than to make errors in physically meaningless digits. If in any doubt, always seek expert advice! In general, you should always note when applying tyings whether they have a physical meaning or not. In particular, tyings depend on the system degrees of freedom that will arise in the nodes during the composition of the nite element model. If you apply tyings that are physically meaningless, this will often become clear from the absence of equilibrium and/or from inexplicable answers.
2.2.1
Generally speaking, a tying consists of a degree of freedom in a master node, in the gures of this section drawn as , and one or more degrees of freedom in slave nodes, drawn as . Figure 2.1 on the facing page shows the notation convention for degrees of freedom for tyings. As for rigid supports, degrees of freedom for tyings are specied by means of a type : TR for translation or RO
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (I)
19
type :
TR RO master
TR slave
RO
Figure 2.1: Tying degrees of freedom for rotation, and a direction number dirnr referring to table DIRECT [ 1.4 p. 14]. A slave degree of freedom cannot be a master in another tying, nor can it be supported. The solution of the system of equations yields the solution for the degrees of freedom of the master. Diana derives the solution for the slave degrees of freedom from the masters.
2.2.2
In single-point format you must explicitly specify all nodes for each tying separately. 2.2.2.2 Multi-point
In multi-point format the tying is specied as a connection between a master edge and a slave edge, where these edges consist of straight line sections [Fig. 2.2]. The principle is that you specify the corner nodes of these edges, the so-called vertices, indicated with double circles. You also specify a set of master and slave
master vertex master node slave vertex slave node master edge
Figure 2.2: Multi-point tying nodes in arbitrary order, the single circles. The master and slave nodes also comprise the vertices themselves, so you must specify the vertices twice. If the master and slave nodes are on the straight edge sections then Diana determines their sequence [Fig. 2.2]. Diana ignores specied nodes which are not a straight edge section.
Diana-9.4 Users Manual Analysis Procedures (I) December 15, 2009 First ed.
20
Input for Structural Analysis Conditions. For multi-point tyings, you must specify as many master vertices as slave vertices. For each edge section, the amount of (accepted) master and slave nodes depends on the type of tying.
2.2.3
Tyings must be input in one single table TYINGS, divided in separate subtables for the various types of tyings. These subtables, listed below, may appear repeatedly and in arbitrary order. syntax TYINGS EQUAL EQUMPC
1 5 6 80
Automatic tying . . . Equalities describe equality of degrees of freedom [ 2.2.4]. Subtable EQUAL is for single-point tyings, EQUMPC for multi-point tyings. Interconnections describe the interconnection of a node (and its degrees of freedom) to two other nodes [ 2.2.5]. Subtable BETWEE is for single-point tyings, BTWMPC for multi-point tyings. Eccentricities describe an eccentric connection of one or more nodes (and their degrees of freedom) to another node [ 2.2.6]. Subtable ECCENT is for singlepoint tyings, ECCMPC for multi-point tyings.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (I)
2.2 Linear Constraints General connections describe linear dependency between one degree of freedom and one or more other degrees of freedom [ 2.2.7]. Subtable FIX is for single-point tyings, FIXMPC for multi-point tyings. Automatic tying let Diana determine the correct tying of loose elements, for instance the connection of shell to solid elements [ 2.2.8.2]. Subtable ELEMEN species the master and slave elements for automatic tying. The input syntax for the various types of tyings is described in the referred sections which also comprise some instructive examples.
21
2.2.4
Equalities
The most common use of tyings is the equalization of degrees of freedom, or more specically: the constraint that one or more displacements are equal. Equality tyings may be input in single- or multi-point format. 2.2.4.1 Single-point Equalities
Equalities in single-point format must be specied in subtable EQUAL with one or more slave nodes, combined with a single master node. syntax TYINGS EQUAL type w dirnrs n... {type w dirnrs n... . . . }
1 5 6 80
snode n
1
mnode n
5 6 80
/ snodes ng... / mnode n type dirnrs describe one or more degrees of freedom [ 2.2.1 p. 18]. snode is the number of the slave node. snodes is a series of slave nodes. mnode is the number of the master node. General example. The following input is a general example of the singlepoint equality tyings as shown in Figure 2.3 on the next page.
file .dat DIRECTIONS 1 1. 0. 2 0. 1. 3 0. 0. 4 1. 1. 0. 0. 1. 0.
December 15, 2009 First ed.
22
19 4 u20 2 20 20 4 u21 2 21 21 4
u35 2 35
35 4
57
18 4
The rst tying [Fig. 2.3a] describes that the translation u1 in direction number 1 is the same for node 34 and 57. In the second tying [Fig. 2.3b], translations in direction 2 (u2 ) and rotations around direction 4 (4 ) of nodes 18 to 21 are the same as for node 35. Applications. Typical applications for equality tyings are straight edges or faces, hinges, and slides. The following input is an example of these applications for the tyings shown in Figure 2.4.
2 40 41 X Z 3 43 44 uX directions (a) straight edge (b) one-axial hinge (c) one-axial slide Z 1 42 34 57 73 29 uY
23
In these examples, only model XY Z directions are used, which by default correspond to direction numbers 1, 2, and 3 respectively. Therefore it is not necessary to specify table DIRECTIONS. The Straight edge [Fig. 2.4a] is specied such that the translation in direction number 1 for the edge nodes is equal to the same translation of the master node. For the One-axial hinge [Fig. 2.4b] the translations in three directions and the rotation around two axes are kept equal for the two nodes. Consequently, the rotation around the third axis is the only degree of freedom that may dier. The Slide [Fig. 2.4c] is specied likewise the hinge, but now the translation in one direction is left free for both nodes. Note that for the hinge as well as for the slide, the two nodes do coincide, i.e., share the same coordinates. 2.2.4.2 Multi-point Equalities
Equalities in multi-point format [ 2.2.2 p. 19] must be specied in subtable EQUMPC with series of slave vertices and nodes, combined with series of master vertices and nodes [Fig. 2.5]. For each edge section, the amount of master nodes
mvnods mnodes svnods snodes master edge
Figure 2.5: Multi-point equality n must be equal to the amount of slave nodes, i.e., each slave node is connected to one master node. syntax TYINGS EQUMPC type w dirnrs n... {type w dirnrs n... . . . }
1 5 6 80
type dirnrs specify one or more degrees of freedom [ 2.2.1 p. 18]. svnods is a series of nodes indicating the slave vertices. mvnods is a series of nodes indicating the master vertices. snodes is a series of slave nodes. mnodes is a series of master nodes.
file .dat TYINGS EQUMPC TR 1 2 / 16 26 / / 26 23 21 18 16
/ 1 11 / / 11 9 6 4 1 /
2.2.5
Interconnection
An interconnection of tying connects one node (the slave), to two others (the masters). Interconnection tyings may be input in single- or multi-point format. An interconnection tying indicates that a slave node lies in between two master nodes. 2.2.5.1 Single-point Interconnection
Interconnections in single-point format must be specied in subtable BETWEE with a single slave node and two master nodes [Fig. 2.6]. By default Diana
type dirnr mnode1 (.5) type dirnr snode type dirnr mnode2 (.5) type dirnr type dirnr type dirnr mnode2 type dirnr mnode1 f1 (1.) (a) in the middle (b) relative distance (c) absolute distance type type dirnr dirnr snode mnode2 f2
Figure 2.6: Single-point interconnection assumes that the slave node is just in the middle of the two masters [Fig. 2.6a]. You may overrule this assumption by specifying the position of the slave node explicitly [Fig. 2.6bc].
2.2 Linear Constraints syntax TYINGS BETWEE type w dirnrs n... {type w dirnrs n... . . . }
1 5 6 80
25
snode n
type dirnrs specify one or more degrees of freedom [ 2.2.1 p. 18]. snode is the number of the slave node. mnode1 , mnode2 are the two master nodes. The optional factors f1 and f2 specify the position of the slave node relative to the master nodes. A single factor f1 species the relative distance from the slave node to the rst master node [Fig. 2.6b]. Two factors f1 and f2 specify the absolute distances from the slave node to the masters [Fig. 2.6c]. If f1 + f2 is not equal to the distance between the master nodes, then snode is positioned in between them proportionally to f1 and f2 to determine the linear dependency. Mesh renement. A typical application of interconnection tyings is Mesh renement as shown in the input for the mesh of Figure 2.7.
2 uY Y X 1 (.5) directions (.5) (.4) (.6) 2.8 1.7 uX 15 16 17 18 19 20 21
[f1 = 0.5] (0 < f1 < 1)
Assuming that the nite element model consists of plane stress elements in the XY plane, only the uX and uY translations have to be tied.
Diana-9.4 Users Manual Analysis Procedures (I) December 15, 2009 First ed.
26
Input for Structural Analysis By default, the X and Y directions correspond to direction numbers 1 and 2 respectively. Therefore it is not necessary to specify table DIRECTIONS.
2.2.5.2
Multi-point Interconnection
Interconnections in multi-point format [ 2.2.2 p. 19] must be specied in subtable BTWMPC with series of slave vertices and nodes, combined with series of master vertices and nodes [Fig. 2.8]. If there are n master nodes between two
mvnods mnodes svnods snodes n = 2, k = 1 section 1 section 2 master edge
Figure 2.8: Multi-point interconnection vertices and k slave nodes between each pair of master nodes then there must be n + k (n + 1) nodes on each section of a slave edge. The values of n and k must be the same for each section. syntax TYINGS BTWMPC type w dirnrs n... {type w dirnrs n... . . . }
1 5 6 80
/ svnods n... / / mvnods n... / / snodes ng... / / mnodes ng... / type dirnrs specify one or more degrees of freedom [ 2.2.1 p. 18]. svnods is a series of nodes indicating the slave vertices. mvnods is a series of nodes indicating the master vertices. snodes is a series of slave nodes. mnodes is a series of master nodes.
file .dat TYINGS BTWMPC TR 1 2 / 13 9 / / 9 10 11 12 13 /
/ 26 28 / / 26 27 28 /
27
2.2.6
Eccentricity
An eccentricity tying indicates that one or more slave nodes are eccentrically connected to a master node. Eccentric connections may be useful if the real physical model comprises eccentricity, whereas in the nite element model the nodes do coincide. Eccentric connections must be applied with great care, they are a potential source of incorrect behavior of the nite element model! Tyings for eccentric connection may be input in single- or multi-point format. You may let Diana calculate the eccentricities from the locations of the nodes or you may explicitly specify the eccentricities yourself. 2.2.6.1 Calculated vs. User-specied Eccentricity
By default Diana calculates the eccentricity from the coordinate dierence of master and slave node: xe = Xmnode Xsnode , ye = Ymnode Ysnode , and ze = Zmnode Zsnode . By one or more of the parameters DX=xe , DY=ye , or DZ=ze (for the eccentricity in x, y, or z direction respectively) you may explicitly specify an eccentricity to overrule the default [Fig. 2.9]. If you specify one or two
type dirnr type dirnr type Y dirnr Z X snode xe
ze mnode ye
ze mnode ye ydirnr Y zdirnr Z type dirnr y xdirnr X (b) in local xyz axes z
xe
snode
Figure 2.9: User-specied eccentricity of these parameters, are zero. 2.2.6.2 then Diana assumes that the unspecied eccentricities
[xe = 0] [ye = 0] [ze = 0]
Axes of Eccentricity
By default user-specied eccentricities are expressed in the global model XY Z coordinate system [Fig. 2.9a]. However, you may specify eccentricities in a local xyz coordinate system with the slave node as origin [Fig. 2.9b]. Therefore you must use the AXES input option in the subtable heading, followed by three direction numbers xdirnr , ydirnr , and zdirnr for the direction of the local x, y and z axis respectively. These direction numbers refer to table DIRECT [ 1.4
Diana-9.4 Users Manual Analysis Procedures (I) December 15, 2009 First ed.
28
Input for Structural Analysis p. 14]. The specied xyz axes must form a right-handed Cartesian coordinate system. 2.2.6.3 Single-point Eccentricity
Eccentricities in single-point format must be specied in subtable ECCENT with one or more slave nodes, combined with a single master node. syntax TYINGS ECCENT {AXES xdirnr n ydirnr n zdirnr n } type w dirnrs n... . . .
1 5 6 80
snode n
5 6
80
/ snodes ng... / mnode n type dirnrs specify one or more degrees of freedom [ 2.2.1 p. 18]. With the AXES input option you may specify the local axes [ 2.2.6.2 p. 27]. snode is the number of the slave node, snodes is a series of slave nodes. mnode is the master node. If this master node is connected to a single slave node, then you may explicitly specify the eccentricity by one or more of the parameters DX=, DY=, or DZ= [ 2.2.6.1]. Two-dimensional example. The following example input is for a nite element model in two-dimensional space, for instance a plane frame with L6BEN elements. This example involves the displacement degrees of freedom uX , uY , and Z as shown in Figure 2.10 on the facing page.
file .dat DIRECTIONS 1 1. 0. 0. 2 0. 1. 0. 3 0. 0. 1. 4 1. 1. 0. 5 -1. 1. 0. TYINGS ECCENT TR 1 2 RO 3 : (a) eccentricity from node coordinates 24 25 : (b) artificial eccentricity in model axes
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (I)
29
In these examples, ve directions are used. Figure 2.10a shows the rst tying, an eccentric connection between two nodes which do not have the same location in the nite element model. Diana uses the distance between the two nodes to calculate the dependence of the slave node displacements on the master displacement. The tying of Figure 2.10b consists of two coinciding nodes. The eccentricity in the real physical model is applied articially: a distance Y in the direction of the model Y axis. For the tying of Figure 2.10c the articial eccentricity x is specied in a local xyz coordinate system. 2.2.6.4 Multi-point Eccentricity
Eccentricities in multi-point format [ 2.2.2 p. 19] must be specied in subtable ECCMPC with series of slave vertices and nodes, combined with series of master vertices and nodes [Fig. 2.11]. For each edge section, the amount of master
mvnods mnodes svnods snodes master edge
Figure 2.11: Multi-point eccentricity nodes n must be equal to the amount of slave nodes, i.e., each slave node is eccentrically connected to one master node syntax TYINGS
Diana-9.4 Users Manual Analysis Procedures (I) December 15, 2009 First ed.
/ svnods n... / / mvnods n... / / snodes ng... / / mnodes ng... / { DX=xe r DY=ye r DZ=ze r
type dirnrs specify one or more degrees of freedom [ 2.2.1 p. 18]. With the AXES input option you may specify the local axes [ 2.2.6.2 p. 27]. svnods is a series of nodes indicating the slave vertices. mvnods is a series of nodes indicating the master vertices. snodes is a series of slave nodes. mnodes is a series of master nodes. You may explicitly specify the eccentricity by one or more of the parameters DX=, DY=, or DZ= [ 2.2.6.1]. If you do so, Diana will apply this eccentricity for each pair of master and slave nodes.
file .dat TYINGS ECCMPC TR 1 2 3 RO 3 / 1 3 / / 9 11 / / 1 2 3 / / 9 10 11 /
2.2.7
General Connection
If none of the previous types of tying is applicable, you may specify a tying in its most general way: a linear relationship between one slave degree of freedom and one or more master degrees of freedom. These types of tyings, called general connections, may be input in single- or multi-point format. 2.2.7.1 Single-point General Connection
General connections in single-point format must be specied in subtable FIX with a single slave node connected to one or more master nodes.1
1 The REMAKE option of Module input [3.3 p.64] always delivers single-point general connection tyings regardless of the original specication.
31
snode n mnode n mtype w mdirnr n {fac r } . . . stype sdirnr specify one slave degree of freedom [ 2.2.1 p. 18]. snode is the number of the slave node. mnode is a master node which must be followed by the description of its degree of freedom [ 2.2.1 p. 18]. Factor fac is an optional multiplication factor: slave = fac master. You may specify multiple masters for each slave node, each one on a new line! Two-dimensional example. The input data below presents two simple examples of general connections in two-dimensional space as shown in Figure 2.12.
76 Y 2 X1 Z3 directions 50 (a) seesaw 25 52 53 75 25
[fac = 1]
50 (b) square
2 -2.
1 2.
32
For the Seesaw connection [Fig. 2.12a] the Y displacement for the slave node is twice as much as for the master and in the opposite direction (note the minus sign in factor 2). The Carpenters square [Fig. 2.12b] causes the slaves Y displacement to be twice as much as the masters X displacement. Multiple Masters Example. The input data below keeps the X distance between two pairs of nodes the same. In formula: u 1 - u 2 = u 3 - u 4 .
file .dat DIRECTIONS 1 1. 0. 2 0. 1. 3 0. 0. TYINGS FIX TR 1 1 2 TR 1 3 TR 1 4 TR 1 0. 0. 1.
1. 1. -1.
2.2.7.2
General connections in multi-point format [ 2.2.2 p. 19] must be specied in subtable FIXMPC with series of slave vertices and nodes, combined with series of master vertices and nodes [Fig. 2.13]. For each edge section, the amount of
mvnods mnodes svnods snodes master edge
Figure 2.13: Multi-point general connection master nodes n must be equal to the amount of slave nodes, i.e., each slave node is connected to one master node. syntax TYINGS FIXMPC stype w sdirnr n
1 5 6 80
2.2 Linear Constraints / snodes ng... / / mnodes ng... / mtype w mdirnr n {fac r } stype sdirnr specify one slave degree of freedom [ 2.2.1 p. 18]. svnods is a series of nodes indicating the slave vertices. mvnods is a series of nodes indicating the master vertices. snodes is a series of slave nodes. mnodes is a series of master nodes. Each series of master nodes must be followed by mtype and mdirnr , the description of a degree of freedom [ 2.2.1 p. 18]. Diana will apply this degree of freedom for all master nodes in the series. Factor fac is an optional multiplication factor: slave = fac master. This factor will also be applied for all master nodes in the series.
file .dat TYINGS FIXMPC TR 2 / 16 26 / / 16 18 21 23 26 /
33
[fac = 1]
/ 1 11 / / 1 4 6 9 11 /
TR 2 1.0
2.2.8
Automatic Tying
In Dianas library of nite elements for structural analysis we can distinguish two types of elements [Vol. Element Library]: elements with translational degrees of freedom only, such as truss, regular plane stress, plane strain and solid elements, and elements with both translational and rotational degrees of freedom such as beams and shells. If in a nite element model elements of the rst type are connected to elements of the second type, then additional tyings must be applied for reasons of compatibility: the rotational degree of freedom in the connection node must be tied to one or more translational degrees of freedom. Usually, these tyings are of the eccentric connection type and can be input in table TYINGS [ 2.2.6]. However, Diana can generate this type of tyings automatically for the connection of beams to plane stress and solid elements and for the connection of shell elements to solid elements. Restrictions. matic tyings: There are some important restrictions to connection by auto-
Currently the Automatic Tying option only works for class-III beams and curved shell elements [Vol. Element Library].
34
Input for Structural Analysis A beam element can only be connected fully to a plane stress element if both elements are in the model XY plane.
If multiple shell elements are connected to the same node and this node must be tied automatically to a solid element, then the thickness vectors of the shell elements at that node must coincide. This may be achieved by specication of a suitable SHAPE for these shell elements [Vol. Element Library]. Diana neglects the geometrical dimensions (thickness, cross-section) of beam and shell elements when searching for slave nodes of tyings.
Modes of operation. The Automatic Tying option can operate in two different modes. In the rst mode the elements to be tied have common node numbers in their connectivity, i.e., they would also be connected without tyings but not correctly, for instance hinged instead of clamped [ 2.2.8.1]. The second mode may be used if the elements to be tied dont have common node numbers, without tyings they would be loose, i.e., not connected at all [ 2.2.8.2]. For both cases, Diana will generate tyings automatically if you turn the automatic tying option on [ 3.5 p. 68]. 2.2.8.1 Connected Elements
Elements connected to a beam or shell element are candidates to deliver slave nodes for the tying. By default Diana examines all elements of the model. However, you may select parts of the model for which automatic tying must be activated [ 3.5 p. 68]. For automatic tying of connected elements, it is not necessary to specify input table TYINGS. Automatic tying of connected elements is possible for the connections as shown in Figure 2.14 and outlined below.
xbe ybe zbe ybe xbe tsh,1 tsh,3 tsh,2
2.2 Linear Constraints Beam to plane stress. [Fig. 2.14a] The nodes of the connected plane stress elements which are on the beams y axis will be tied automatically to the beams degrees of freedom. Beam to solid. [Fig. 2.14b] The nodes of the connected solid elements which are in the plane of the beams cross-section, i.e., the plane through the yz axes of the beam, will be tied automatically to the beams degrees of freedom. Shell to solid. [Fig. 2.14c] The nodes of the connected solid elements which are on the thickness vectors t of the shell will be tied automatically to the shells degrees of freedom. 2.2.8.2 Loose Elements
35
For loose elements Diana cannot determine the candidate slave elements for automatic tying, therefore you must specify for each tying the master element and its slave elements in subtable ELEMEN of table TYING. syntax TYINGS ELEMEN
1 5 6 80
/ selems ng... / melem n . . . / selems ng... / / melems ng... / . . . selems is a set of slave elements (plane stress or solid), specied by numbers and/or groups, which are candidates to deliver slave nodes for the tying. melem is a master elements (beam or shell) which must be tied automatically to the slave elements. melems is a set of master elements (beam or shell) which must be tied automatically to the slave elements. Elements of dierent types are skipped.
file .dat TYINGS ELEMEN / 1-1062 / 2113 / 1063-2097 / 2114
Automatic tying of loose elements is possible for the connections as shown in Figure 2.15 on the next page and outlined below.
36
ybe
zbe
Figure 2.15: Automatic tying of loose elements Beam to plane stress. [Fig. 2.15a] All nodes of the specied slave plane stress elements which are on the beams y axis will be tied automatically to the beams degrees of freedom. Beam to solid. [Fig. 2.15b] All nodes of the specied slave solid elements which are in the plane of the beams cross-section, i.e., the plane through the yz axes of the beam, will be tied automatically to the beams degrees of freedom. Shell to solid. [Fig. 2.15c] All nodes of the specied slave solid elements which are on the thickness vectors t of the shell will be tied automatically to the shells degrees of freedom. The nodes of shell and solid element may coincide, i.e., the distance e may be zero.
2.3
Loads
There are various classes of loads input, each of which is specied in a separate subtable of table LOADS. Regular loading input is divided into cases, which may be combined into load sets. Apart from these classes of regular loading, there is a special form of loading input for the determination of inuence elds [ 2.3.8 p. 56]. syntax LOADS CASE case n NODAL
1 5 6 80
2.3 Loads
37
5 6
12 13
80
loads sets CASE case is the number of a new load case. A case number specication line must follow immediately after the table heading. The case number may be respecied at the start of a new subtable. NODAL subtable for nodal load, i.e, concentrated on the nodes and directly resolved to system degrees of freedom [ 2.3.1]. ELEMEN subtable for element load, i.e, distributed over elements (faces, edges). Diana transforms element load to equivalent nodal loads. It also includes temperature loads on elements. See Volume Element Library for input syntax. WEIGHT subtable for dead weight load [ 2.3.2]. CENTRI subtable for centrifugal load [ 2.3.3]. Weight and centrifugal load will be superposed to nodal forces and resolved to system degrees of freedom.
38
Input for Structural Analysis DEFORM subtable for xed displacements (deformation) [ 2.3.4]. Fixed displacements will be translated to equivalent element forces and then resolved to system degrees of freedom. REINFO subtable for reinforcement load, for instance prestress in reinforcement bars. See Volume Element Library for input syntax. WIND subtable for wind load [ 2.3.6.1]. WATER subtable for water wave load [ 2.3.6.2]. Wind and water wave load will be evaluated as an average distributed load on certain element types. COMBIN subtable to dene the load sets as a combination of load cases [ 2.3.7].
file .dat LOADS CASE 1 NODAL ELEMEN CASE 2 WEIGHT CASE 3 DEFORM COMBIN
2.3.1
Nodal Load
Nodal loads are input in subtable NODAL. There are three forms of input syntax: (1) one nodal load per line, (2) nodes in a series of numbers or groups or both, with one load value valid for all the nodes in the series, (3) nodes in a series of numbers and/or groups, with a series of load values: one value for each node. syntax LOADS NODAL
1 5 6 80
node n
2.3 Loads
39
5 6
80
/ nodes ng... / type w dirnr n / values r... / node is a single node number. nodes is a series of nodes, it must be specied between slashes and may comprise numbers or groups or both. type is the load type: FORCE for force or MOMENT for moment. The direction number dirnr refers to table DIRECT [ 1.4 p. 14]: a force in, or a moment around the specied direction. value is the load value. For axisymmetric models, this value is the total circumferential load. values is a series of load values, one for each node in nodes , it must be specied between slashes.
file .dat LOADS CASE 2 NODAL 3 FORCE 2 5 MOMENT 1 / 4-12 WALL 31 / FORCE 5 / 20 22 25 27 35 FORCE 1
2.3.2
Dead Weight
Load due to dead weight is specied in subtable WEIGHT with direction and gravity acceleration. To evaluate dead weight loading, Diana retrieves the mass density for specic materials from the material properties in table MATERI [Vol. Material Library]. syntax LOADS WEIGHT
1 80
dirnr n gracce r dirnr is the direction number referring to table DIRECT [ 1.4 p. 14].
For mixture analysis you may additionally specify the origin of the position vector [ 23.1.3 p. 389].
file .dat LOADS CASE 5 WEIGHT 1 9.8 2 10.0
In this example, case 5 contains two weight loads in dierent directions and with dierent gravity acceleration!
2.3.3
Centrifugal Load
Centrifugal load, due to rotation of the nite element model, is specied in subtable CENTRI with turning speed and axis of rotation. To evaluate centrifugal loading, Diana retrieves the mass density for specic materials from the material properties in table MATERI [Vol. Material Library]. syntax LOADS CENTRI
1 80
omega r axipt1 r3 axipt2 r3 omega is the rotational velocity in rad/sec. axipt1 , axipt2 are the model XY Z coordinates of two points on the rotation axis. Only one single centrifugal load may be specied: one case, one speed, one rotation axis.
file .dat LOADS CASE 3 CENTRI 0.1 0.1 1.3 2.6
2.3 Loads
41
2.3.4
Fixed or prescribed displacements, so-called deformation load, are input in subtable DEFORM. There are three forms of input syntax: (1) one nodal displacement per line, (2) nodes in a series of numbers or groups or both, with one displacement value valid for all the nodes in the series, (3) nodes in a series of numbers and/or groups, with a series of displacement values: one value for each node. Fixed displacements in subtable DEFORM must be specied as supports in table SUPPOR [ 2.1 p. 17]. Consequently they will have a zero displacement for all cases, unless specied otherwise in subtable DEFORM. syntax LOADS DEFORM
1 5 6 80
node n
1
/ nodes ng... / type w dirnr n / values r... / node is a single node number, nodes is a series of nodes, it must be specied between slashes and may comprise numbers or groups or both. type is the displacement type: TR for translation or RO for rotation. The direction number dirnr refers to table DIRECT [ 1.4 p. 14]: a translation in, or a rotation around the specied direction. value is the displacement value, values is a series of displacement values, one for each node in nodes . You can get output of xed displacements via the PRESCR type specier of the DISPLA output item [ 4.2.2 p. 96].
file .dat LOADS CASE 3 DEFORM 3 TR 2 1.E-2 5 RO 3 0.25 / 4-12 EDGE / TR 5 .001 CASE 5 DEFORM 3 TR 2 2.E-2
Diana-9.4 Users Manual Analysis Procedures (I) December 15, 2009 First ed.
42
In this example the xed translation of node 3 in direction 2 diers in case 3 and case 5!
2.3.5
Mobile Loads
In bridge design, mobile (or moving) loads are often applied to get the extreme values of analysis results like strains, stresses, or support reactions. In the next sections we will present the main principles for mobile load analysis with Diana [ 2.3.5.1], followed by a description of the input syntax for mobile loads.2 2.3.5.1 Principles The application of mobile loads is subject to the following re-
Restrictions. strictions.
It is allowed to connect other elements to the beam elements on which a mobile load acts, for instance in a plate model of a bridge. When the stiness of the beam elements is negligible in comparison to the plate elements, the only function of the beam elements is to dene the mobile load pathway.
Mobile loads can only be applied in regular linear static analysis [Ch. 4]. Only one mobile load per load case may be dened. If a load case contains a mobile load, no other loadings are allowed in this load case.
Load cases. Usually, a specied mobile load will consist of a set of moving forces, which can represent a truck for example, and a static distributed load. Additionally, you must dene a set of unbroken beam elements which form the pathway of the load [Fig. 2.16]. Diana will automatically generate one load
distributed load nodes pathway truck position Fn truck F2 F1
Figure 2.16: Mobile load denition case for each position of the truck placed on a subsequent node of the pathway.
2 See
2.3 Loads Therefore, the number of load cases will be equal to the number of nodes along the pathway. In addition to the mobile load, every load case will also have the same static distributed load along the entire pathway. So the generated load cases only dier in the position of the truck. In the Results environment, iDiana can calculate envelopes of results for all these load cases [Vol. Pre- and Postprocessing]. Other load cases, like weight or prestress, can be taken into account as well, when calculating envelopes of results. Truck positions. Typically, a truck load comprises multiple force loads where each force represents the loading caused by one of the trucks axles, see the forces F1 to Fn in Figure 2.16 on the facing page. The position of the truck is dened by the position of its rst axle. Diana will spread the forces related to the other axles over the two adjacent nodes. When the truck is at the beginning of the pathway, i.e., with its rst axle on the rst node [Fig. 2.17a], only the rst axle will act on the model. The second axle load comes into action when the truck has moved over the distance between the two axles [Fig. 2.17b], and so on.
F1
43
(a)
F1
(b)
F1
(c) F1 (d)
F1
(e)
overhang
Figure 2.17: Ultimate truck positions When the truck is positioned at the end of the pathway, i.e., with its rst axle on the last node N , the other axles are still on the pathway and consequently will generate a load on the model [Fig. 2.17c]. However, when the truck leaves the pathway, i.e., when the rst axle arrives beyond the last node, Diana will not generate any truck load on the model [Fig. 2.17d]. Therefore, to account for the loading of the other axles, which are still on the model, you must dene an overhang, i.e., a part of the pathway beyond the end of the actual model [Fig. 2.17e]. The length of this overhang must be equal to, or greater than, the
Diana-9.4 Users Manual Analysis Procedures (I) December 15, 2009 First ed.
44
Input for Structural Analysis length of the truck. The number of nodes in the overhang denes the number of truck positions that Diana will take into account when the truck leaves the model. 2.3.5.2 Mobile Load Input
Input of a mobile load consists of the actual load specication in subtable MOBILE of input table LOADS and some geometrical data for the elements in table GEOMET. To apply more than one mobile load at the same time, you must specify all the MOBILE subtables in separate load case. syntax LOADS
1 5 6 12 13 80
You may superpose the results of mobile loads in the iDiana Results environment [Vol. Pre- and Postprocessing]. You may dene the mobile loads explicitly, or let them set up by Diana according to code regulations for bridges. syntax LOADS
1 5 6 12 13 80
CASE MOBILE
case n ELEMEN [ DIRECT CODE elems ng... dirnr n ] NONE VOSB VBB ENV
loading specs
CASE case is the number of the load case. Mobile loads may not be used in combination with load sets via subtable COMBIN. ELEMEN elems are the elements, specied by numbers and/or groups, which form the part of the model where the trac load must be applied. These
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (I)
2.3 Loads elements (at least two) must form a contiguous open chain of connected beam elements without branches. In this sense, two elements are considered to be connected if they share an end-node. Diana automatically determines the sequence of the elements in the chain, i.e., you may specify the elements or groups in arbitrary order. DIRECT dirnr species the direction of the mobile load, usually the gravity acceleration, with a direction number referring to table DIRECT [ 1.4 p. 14]. If you dont specify a direction, then Diana assumes Y for twodimensional models or Z for three-dimensional models. CODE indicates the code regulation to be applied: NONE for no code, i.e., explicit specication [ 2.3.5.3]. VOSB for the Dutch VOSB code [ 2.3.5.4]. VBB for the Dutch VBB code [ 2.3.5.4]. ENV for the European ENV 1991-3 code [ 2.3.5.5]. Loading specs is a code dependent specication of the mobile load. 2.3.5.3 Explicit Input
45
For explicit input of mobile loads you must specify at least the truck load and the distributed load in table LOADS and the eective width of the carriage way in table GEOMET. To simulate the inuence of the transverse position of these loads you may specify load factors in table GEOMET. Load LOADS
1 5 6 12 13 80
syntax
NONE f1 r [d2 r f2 r . . . dn r fn r ] qr
CODE NONE indicates that no code regulation must be applied, i.e., the mobile load will be specied explicitly. LORRY indicates the schematic truck load [Fig. 2.18]. Values f1 to fn are the axle forces Fi=1,n . Values d2 to dn are the distances di=2,n to the preceding force. DISTRI q is the value q of the distributed load [f orce/unitarea].
46
F1 d2 F2 d3 F3
Figure 2.18: Mobile truck and distributed load Width of carriage way GEOMET
1 5 6 12 13 80
syntax
BWIDTH
wr
BWIDTH w is the eective width w of the carriage way. Diana needs this width to determine the area at which the distributed load q is acting. Transverse position of load GEOMET
1 5 6 12 13 80
syntax
[ LODFAC [ DISFAC
ff r ] fq r ]
By default Diana neglects the inuence of the transverse position of an explicitly specied mobile truck load. To let Diana apply this transverse position you must specify multiplication factors in table GEOMET.
[fF =1 ]
LODFAC ff is a multiplication factor fF for the truck forces F . This factor simulates the inuence of the transverse position of the truck. DISFAC fq is a multiplication factor fq for the distributed load q. This factor simulates the inuence of the transverse position of the distributed load. By default this factor is equal to the eective width. Example of explicit input
ELEMEN GEOMET / 5 8 13 24 37 40 / 2 DIRECT 4 0. -1. 0. GEOMET 2 BWIDTH 4.75 LODFAC 0.83 file .dat
[fq = w]
2.3 Loads
DISFAC LOADS CASE 2 MOBILE ELEMEN DIRECT CODE LORRY DISTRI CASE 3 WEIGHT 4 9.8 3.72
47
2.3.5.4
There are two Dutch codes which dene mobile loads on bridges: the VOSB [46] for steel bridges and the VBB [45] for concrete bridges. According to these codes, the magnitude of the loads depend on the class as shown in Table 2.1. Table 2.1: Mobile loads according to the VOSB/VBB code
Class VBB VOSB 60 600 45 450 30 300 Distributed load 4 kN/m (max. 12 kN/m per lane) 3 kN/m2 (max. 9 kN/m per lane) 2 kN/m2 (max. 6 kN/m per lane)
2
To apply mobile loads according to the Dutch codes you must specify at least the loading class in table LOADS, and the width of the carriage way in table GEOMET. Additionally in table GEOMET, you may specify the transverse position of the truck, and the bump and load reduction factors. To apply the VOSB/VBB code, Diana assumes by default that you specied the model data in standard SI-units. If you used other units, then you must specify them in table UNITS [ 1.1 p. 3]. Code and class LOADS
1 5 6 12 13 80
syntax
CODE VOSB
Diana-9.4 Users Manual Analysis Procedures (I) December 15, 2009 First ed.
48 VBB class r ]
[ CLASS
CODE indicates the code regulations to be applied: VOSB for the Dutch VOSB code, or VBB for the Dutch VBB code.
(class > 0) [VBB: 60] [VOSB: 600]
CLASS class indicates the VOSB/VBB class which determines the size of the mobile load [Table 2.1 p. 47]. For class values other than those mentioned in the table, Diana applies a linear interpolation of the values with respect to the actual class number. Width of carriage way and lanes GEOMET
1 5 6 12 13 80
syntax
BWIDTH [ WLANE
wr lr ]
WLANE l is the width l of the lanes, applied for the VOSB code. According to both Dutch codes, Diana applies at most two lorries in transverse direction in a position which yields a maximum load contribution. Transverse position of truck. For the Dutch codes, Diana by default neglects the inuence of the transverse position of the truck load. To let Diana apply this transverse position you must specify either an inuence line or multiplication factors in table GEOMET. Inuence line GEOMET
1 5 6 12 13 80
syntax
[ TRVINF TWOBEA
] v1 r [v2 r . . . vn r ] br
2.3 Loads
w q v v1 v2 . . .
0
49
w q v b
1 0
Figure 2.20: Inuence line in transverse direction TRVINF species the inuence line for the truck position in transverse direction [Fig. 2.20a]. Values v1 to vn are the values of the inuence line vi=1,n . You may specify at most ten points. Diana assumes an equidistant distribution of the points. If you only specify a single point, then the inuence line is a constant. TWOBEA is a short-cut input for the inuence line if the nite element model simulates a bridge with two beams in transverse direction [Fig. 2.20b]. The value b is the distance b between the beams. In this case, Diana applies a straight inuence line. Diana applies the distributed load at the part where the inuence line is positive.
file .dat GEOMET 1 BWIDTH TRVINF 4. 1.5 -0.5
(n 10)
In this example the inuence line is a straight line. For a two-beam model the following input denes the same inuence line as the previous example.
file .dat GEOMET 1 BWIDTH TWOBEA 4. 2.
Multiplication factors
syntax
50 GEOMET
1 5 6 12 13
80
[ LODFAC [ DISFAC
ff r ] fq r ]
[fF =1 ]
LODFAC ff is a multiplication factor fF for the truck forces. This factor simulates the inuence of the transverse position of the truck. If you dont specify any inuence of the transverse position of truck forces Diana applies no multiplication factor. DISFAC fq is a multiplication factor fq for the distributed load q. This factor simulates the inuence of the transverse position of the distributed load. By default this factor is equal to the eective width. For the VOSB code Diana will correct it for the sum of the width of the lanes. Bumping and load reduction. Additional to the reduction factor for two lorries and the inuence in transverse direction, the mobile load has to be multiplied by a bump factor and a load reduction factor. These factors depend on a certain length l. For the VOSB code [46, A.2.10-11], Diana uses a bump factor S and a load reduction factor B dened as S =1+ 40 100 + l and B = 0.6 + 40 100 + l (2.1)
[fq = w]
For the VBB code the bump factor S is much more complicated because it depends strongly on geometrical properties [45, 4.3.1.2-3], therefore in Diana it must be specied explicitly, or you may indicate that bumping and load reduction must be applied according to the VOSB code. By default, Diana will apply neither a bump factor nor load reduction. VBB code GEOMET
1 5 6 12 13 80
syntax
[ BUMP [ LOARED
[S = 1]
sr ] br ]
BUMP s is the bump factor S which Diana will apply to the mobile load. LOARED b is the load reduction factor B which Diana will apply to the mobile load. VOSB code GEOMET
1 5 6 12 13 80
[B = 1]
syntax
[ LOALEN
lr ]
2.3 Loads LOALEN l is the length l which Diana will use to compute the bump factor and the load reduction factor according to the VOSB code even if the mobile load was input via the VBB code. 2.3.5.5 European ENV 1991-3 Code
51
The European ENV 1991-3 code [11] denes trac loads on bridges. According to this code, the magnitude of the loads depends on the width of the carriage way and the number of lanes [Fig. 2.21]. The total width w of the carriage way w r1 l
1
r2
l ...
r2
l n
r1
Figure 2.21: Lanes on a carriage way applied for ENV 1991-3 code depends on the number of lanes n, their width l, and the remaining width r: w =nl+r with r = 2 r1 + (n 1) r2 (2.2)
To apply mobile loads according to the ENV code you must at least specify the width of the carriage way. Diana will determine the number of lanes, their width, and the remaining width according to the ENV code [Table 2.2]. Table 2.2: Lanes and carriage way according to the ENV 1991-3 code
Carriage way width w [m] w < 5.4 5.4 w < 6 6w Lanes number n width l [m] 1 3 2 w/2 int(w/3) 3 Remaining width r [m] w3 0 wn3
The ENV 1991-3 code indicates various classes of loading by a number [Table 2.3]. Each class gives the distributed trac load qik , the distributed load on Table 2.3: Load classes according to the ENV 1991-3 code
Class 1 2 3 >3 Distributed load [kN/m2 ] qik qrk 9.0 2.5 2.5 2.5 2.5 2.5 2.5 2.5 Axle force [kN] Qik 300 200 100 0
52
Input for Structural Analysis the remaining width qrk and the axle force Qik of a truck [Table 2.3]. A mobile load is applied on each lane, consisting of a truck with two axle loads Qik at a xed distance of 1.2 m, plus the distributed force load qik over the full area of the lane [Fig. 2.22]. Additionally the distributed load qrk is applied as resting on the remaining area. By default, as prescribed by the ENV 1991-3 code, Diana will take the values for the loads Qik and qik from the load class of the lane number: class 1 for the rst lane, class 2 for the second and so on.
Qik 1.2 m qik Qik
Figure 2.22: Mobile truck and distributed load according to the ENV 1991-3 code
Input data. In batch input data format you must specify the loading according to the ENV 1991-3 code in tables LOADS and GEOMET as indicated in the following. To determine loading according to the ENV 1991-3 code, Diana assumes by default that you specied the model data in standard SIunits. If you used other units, then you must specify these in table UNITS [ 1.1 p. 3]. syntax LOADS
1 5 6 12 13 80
CODE [ CLASSE
ENV class n ]
CLASSE will load all lanes according to the indicated class [Table 2.3]. This overrules the ENV code! By default Diana will apply the loads according to the ENV code: class 1 on the rst lane, class 2 on the second and so on. syntax GEOMET
2.3 Loads
53
5 6
12 13
80
wr ff r ] fq r ]
BWIDTH w is the total width w of the carriage way. LODFAC ff is a multiplication factor for the truck axle loads. If you specify this factor then the axle loads become ff Qk instead of the calculated Qk [Eq. (2.4)]. This overrules the ENV code! DISFAC fq is a multiplication factor for the total distributed load. If you specify this factor then the distributed load becomes fq qtot instead of the calculated qtot [Eq. (2.3)]. This overrules the ENV code! Example
ELEMEN GEOMET / 1-5 / 1 LOADS CASE 1 MOBILE ELEMEN 1-5 CODE ENV GEOMET 1 BWIDTH 8.0 file .dat
[ff = 1.0]
[fq = 1.0]
For this example Diana determines the total distributed load and the total axle load according to the ENV 1991-3 code as follows:
According to Table 2.2 the number of lanes n is int(8/3) = 2; the width of the lanes l is equal to 3 m. According to Table 2.3 the remaining width r is equal to 8 2 3 = 2 m. As no class for the lanes has been specied in this example, class 1 will apply for the rst lane and class 2 for the second lane. This gives the following total distributed load per unit length:
(2.3)
According to Table 2.3 the total axle load of the truck is calculated as:
(2.4)
54
2.3.6
Wind and water load cause an average distributed load which can be used in static analysis.
In a static analysis, wind and water load may be specied both, even in the same load case. Diana evaluates wind or water load for a subset of the element types as indicated in a special section in Volume Element Library, which describes the input of necessary geometry parameters.
2.3.6.1
Wind Load
AVERAG species the average part of the horizontal wind speed: z0 is the roughness height z0 , d is the displacement height d and vstar is the friction wind speed v . All elements subjected to wind load must satisfy the condition that their average global Z coordinate ze is greater than d. DIRECT dirnr species the wind direction with a direction number referring to table DIRECT [ 1.4 p. 14]. DENSIT rho is the density of air .
[zg = 0]
GROUND zg is the global Z coordinate zg of the ground. No loads will be generated below zg . 2.3.6.2 Water Wave Load
AVERAG DIRECT
December 15, 2009 First ed.
z1 r z2 r . . . zn r v1 r v2 r . . . vn r dirnr n
Diana-9.4 Users Manual Analysis Procedures (I)
55
AVERAG species the average water prole: z1 to zn are the heights z, v1 to vn the corresponding average velocities v . Specify one v for each z. DIRECT dirnr species the water wave direction with a direction number referring to table DIRECT [ 1.4 p. 14]. DENSIT rho is the density of water . GROUND zg is the global Z coordinate zg of the ground. No loads will be generated below zg . LEVEL zw is the global Z coordinate zw of the still water level. No loads will be generated above zw .
[zg = 0]
(zw > zg )
2.3.7
Load Sets
In subtable COMBIN of table LOADS previously specied load cases are combined to load sets, which form the load vectors of the system of equations (right-handside). Solution of the system yields the displacements for these load sets.
Specication of load sets is optional. If you dont specify load sets, Diana assumes a one-to-one relation between load cases and load sets. Null-vectors will be generated for omitted load set numbers.
loset n
1
loset n
loset is the load set number. case is a case number, referring to previously specied cases [ 2.3 p. 36], cases is a series of case numbers in between slashes. factor is a multiplication factor for case or cases .
56
4 -0.25
2.3.8
Inuence Fields
An inuence value is the value of a stress or strain in a specic point of the model as a function of the location of a nodal load. In other words: the inuence value for node N represents the value of the stress or strain in point i of element E if a unit load acts in node N . The set of inuence values obtained if the load respectively acts in each node of the model is called the inuence eld. If you specify subtable INFLUE, with the element number and the type and direction of the nodal load, then Diana calculates the inuence elds for all the points in the element. Inuence eld determination cannot be combined with other loadings. No other subtables may be present in table LOADS if subtable INFLUE is specied. syntax LOADS INFLUE
1 5 6 80
elem n
type w dirnr n
elem is an element number. type is the load type: FORCE for force or MOMENT for moment. The direction number dirnr refers to table DIRECT [ 1.4 p. 14]: a force in, or a moment around the specied direction.
57
Diana determines the inuence elds with the standard commands for linear static analysis [Ch. 4] and for calculation of postanalysis results [ 4.2 p. 92]. Inuence elds may be output in tabular form, graphically or on a le for external postprocessing. The output of the inuence eld for a specic point may be selected with appropriate ELEMENT selection commands in the SELECT block of the OUTPUT commands. The usual STRAIN or STRESS commands specify the strain or stress for which the inuence eld must be determined. A node selection in the output commands indicates that the load only acts on parts of the model.
file .dcf analysis commands BEGIN OUTPUT FEMVIEW BEGIN SELECT BEGIN ELEMEN INTPNT 2 NODES 4 END ELEMEN NODES 35-80 END SELECT STRESS MOMENT XX END OUTPUT
The ELEMEN block gives the selection of points for which inuence elds must be plotted. Note that the element number is not specied because it was input in subtable INFLUE. Inuence elds are plotted for integration point 2 and for relative node number 4 of the element. The second NODES command indicates that the unit load acts on nodes 35 to 80. See also Volume Pre- and Postprocessing and examples bridge and skewpl [Vol. Analysis Examples].
58
Chapter 3
General Commands
See Volume Getting Started for a general description of Diana batch analysis commands.
3.1
Control Commands
Control commands control general aspects of a run. They are optionally specied in the command le before the rst module command. syntax [ ] LOG NOLOG [ ERRORS{.CPF} { MF=maxfat n MW=maxwar n module commands
}]
*END
LOG to enable log lines. See Volume Getting Started for general description of Diana job logging. NOLOG to disable log lines. ERRORS to ne-tune the error handling. See Volume Getting Started for general description of Dianas error message facility. ERRORS.C to print error codes, ERRORS.P to print parameters like severity and reference, ERRORS.F to print full error text. Parameter MF=maxfat is the maximum number of fatal errors allowed.
Diana-9.4 Users Manual Analysis Procedures (I)
[LOG]
[ERRORS.CPF]
[MF=10]
60
General Commands If the fatal error count exceeds this number, the run will be aborted.
[MW=50]
Parameter MW=maxwar is the maximum number of warnings. If the warning count exceeds this number, additional warnings will be suppressed. Example
NOLOG ERRORS.F MF=5 MW=20 *INPUT *END file .dcf
3.2
The central database for each analysis project with Diana is handled by filos, the FILe Organization System and hence called the filos le. The user must maintain a filos le during the whole lifetime of the analysis project for a particular nite element model. Basically there are two types of maintenance: external and internal. External maintenance by means of operating system commands. These commands perform such things as backup on magnetic tape, transfer in a computer network, copy, delete etc. Internal maintenance by means of Diana commands for Module filos. Most notably is initialization, furthermore there are commands to expand, compress or erase the filos le etc. The filos le is a binary le and cannot be ported between dierent brand computers, nor can it be printed, displayed on a terminal screen or edited by means of a text editor.
3.2.1
Tutorial Introduction
The possibilities of Module filos are manifold. Fortunately the commonly used commands are very simple as will be shown in the following examples. Initialize Filos le
*FILOS INITIALIZE file .dcf
61
Sometimes, after many runs, it may be useful to compress the filos le with the command as shown above. This causes unused space, due to deletion of items, to be collected and made available for reuse. Show disk space of Filos le
*FILOS INFORM SPACE file .dcf
This command tells you how many disk space the filos le currently uses, and how many free space it still contains. Command sequence. It must be emphasized that the commands for Module filos may be placed in between other module commands as shown in the following example: Backup the Filos le
*FILOS INITIA *INPUT *LINSTA *FILOS COPY FILE="Backup1" *NONLIN *FILOS COPY FILE="Backup2" *END file .dcf
These commands are for a complete linear and nonlinear analysis with two backup versions made: one just after the linear analysis and another one just after the nonlinear analysis.
3.2.2
Command Syntax
This section describes the commands for Module filos to perform internal maintenance of the filos le. The name of the filos le must be specied beforehand in an environment symbol or in the system command to start up the Diana run [Vol. Getting Started].
62
General Commands syntax *FILOS [ INITIA { DI=dirsiz n BF=bufsiz n5 MI=maxinc n CP=cmperc n [ UNLOCK ] [ CHECK ] [ BEGIN SAVE [ FILE=file s ] [ ITEMS=items w... ] END SAVE ] [ COPY FILE=file s ] [ BEGIN RESTOR [ FILE=file s ] [ ITEMS=items w... ] END RESTOR ] [ COMPRE ] [ BEGIN INFORM }] [ SPACE { DIRECT FILE [ ITEMS { }] VARIAB ARRAY SUBFIL END INFORM ] }]
*FILOS starts the module to maintain the filos le. There are no default commands for Module filos, if you only specify the *FILOS command nothing will happen. INITIA to initialize the filos le. Any le must once be initialized to become a legal filos le. The *FILOS INITIA command may only appear at the beginning of a command le, it will be ignored on other places! Various attributes may be specied by means of the following parameters: DI=dirsiz is the directory size, i.e., the maximum number of entries in the administration directory. BF=bufsiz species the buer sizes for the ve classes xs s m l xl.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (I)
3.2 FILOS File Maintenance MI=maxinc is the maximum number of element data directories to be held in core. CP=cmperc is the trigger percentage for automatic compression of element data space, where CP=0 means no compression. The default values for dirsiz , bufsiz , maxinc and cmperc are installation dependent and specied in le filoscap. Normally these defaults give good performance of analysis runs on your computer system, and these parameters may be omitted. However, there may be circumstances that make it useful to use other values, for instance a temporary lack of memory space, a sudden availability of faster disks or application in R&D environments. In these cases you can specify the values yourself to optimize the performance of the Diana jobs; contact your systems manager for this. The INITIA command causes all the data on the filos le to be lost; consequently the analysis of the nite element model must start from the beginning, with reading the input data. UNLOCK to unlock the filos le. A locking mechanism prevents that two Diana jobs simultaneously access the same filos le. The lock is removed when a job is legally nished. If by any chance the job is not legally nished, the filos le remains locked and must be unlocked at the beginning of the next job. CHECK to check the consistency of the filos le. After system problems (disk errors, break down) the consistency of the filos le may be damaged. This command checks if the filos le is still OK and usable for further analysis. If not, you must delete it and either initialize a new one and start all over again or restore a backup version that you kept somewhere. SAVE to save a selection of data items on another filos-like le. If you dont specify items then all data will be saved. By default Diana saves to the le diana save.ff . COPY to copy the filos le. There are two reasons to use this command instead of the operating system copy command: (1) it yields a compressed filos le or (2) it gives the opportunity to backup the filos le during the analysis process. Diana copies the filos le onto the le specied by FILE=file .1 RESTOR to restore a selection of data items from another filos-like le. If you dont specify items then all data will be restored. Diana copies the data items from the le specied by FILE=file .1 By default Diana restores from the le diana save.ff .
there is no extension in file (a period and some characters) Diana adds the extension .ff to it.
Diana-9.4 Users Manual Analysis Procedures (I)
63
1 If
64
General Commands COMPRE to compress the filos le. Diana collects wasted disk space, for instance due to deletion of data items, and puts it at the end of the filos le to make it available for reuse. INFORM to get information about the filos le: SPACE the amount of memory space used for the filos le: FILE the number of blocks currently used, DIRECT the number of directory entries. By default, if you specify SPACE without any option, Diana informs you about both blocks and directories. ITEMS information about items. This command is for Diana programmers only! 2 VARIAB gives information on values, ARRAY on arrays and SUBFIL on subles. By default, if you specify ITEMS without any option, Diana informs you about all available items.
3.3
Module input is primarily used to read the input le containing tables. Moreover input can printout (remake) tables that were previously read and it can build tables out of data on the filos le. All the other modules take their information from the filos le and need no input data, but only user commands. syntax *INPUT [ READ { option w } [ FILE=infil s ] [ TABLE tabnam w... ] ] ECHO APPEND [ DELETE TABLE tabnam w... ] [ REMAKE [ FILE=outfil s ] [ TABLE tabnam w... ] ]
[READ]
*INPUT starts Module input to read or write input data. If you dont specify any of the following commands then Diana reads all tables on the input le. READ to read input tables. All data from previously read tables will be overwritten! The ECHO option asks Diana to echo the input on the standard output le, while being read. Due to the APPEND option, the input data will be appended to previously read tables. By default, Diana takes the le name from the run command [Vol. Getting Started]. With parameter FILE= you may overrule this name by infil .
utility fu is available to get more information on items. See the Diana Programmers Manual section /bin/fu
2A
3.3 Reading or Remaking Input Data If there is no extension in this le name (a period and some characters) then Diana assumes the extension .dat. By default Diana will read all tables on the input data le. With the TABLE option you may select one or more specic tables tabnam to be read. DELETE TABLE to delete one or more specied tables tabnam from the filos le. These tables must have been read previously. REMAKE to write tables to a le. Diana will remake the tables, i.e., compose them out of evaluated and stored data from the filos le. Remade tables will contain the same data as those previously read, but possibly in an other lay out. By default, Diana writes the tables to the standard output le. If you specify parameter FILE= then the tables go to a le named outfil . If there is no extension in this le name (a period and some characters) then Diana adds the extension .dat to it. By default Diana will write all tables for which data is available on the filos le. With the TABLE option you may select one or more specic tables tabnam to be remade. Default
*INPUT file .dcf
65
By default all tables on the input le are read and not echoed if only the *INPUT command is specied, it has the same eect as:
file .dcf *INPUT READ
file .dcf
These commands cause the specied tables to be read and the input lines simultaneously written to the standard output le. Remake specied tables
*INPUT REMAKE FILE="myfile" TABLE LOADS SUPPOR file .dcf
These commands cause the specied tables to be written to the le myfile.dat even if they were never read with Module input.
Diana-9.4 Users Manual Analysis Procedures (I) December 15, 2009 First ed.
66
General Commands
3.4
Element Evaluation
Due to the EVALUA command(s) Diana will check and evaluate the geometric and material properties of elements. You may customize the evaluation process to check the quality of the mesh. The OFF option switches off the element evaluation which may save computing time if this task was performed previously. syntax BEGIN EVALUA [ OFF ] [ CHECK { }] SHAPE=eshape r RATIO=eratio r ]] [ REINFO INTERF [ ON OFF [ COMPOS ] END EVALUA CHECK performs an extended test on element shape and aspect ratio. By default Diana does not perform this test.
a
Figure 3.1: Element shapes The extended test checks whether the element shape deviates too much from the ideal shape and whether the aspect ratio is extremely high. Two parameters are appropriate for this test [Fig. 3.1]. Table 3.1 shows some typical values of the shape parameters for the slender and tapered quadrilaterals of Figure 3.1. See Roddeman [49] for background theory. You may overrule the default values by specifying parameters:
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (I)
67
[eshape =0.001]
SHAPE=eshape RATIO=eratio
the tolerance for the Shape test. the tolerance for the Aspect Ratio test.
[eratio =3.0]
If one of the shape parameters of an element is greater than the specied tolerance, Diana will produce a warning message. The extended check option checks linear and quadratic elements of the following element types: plane stress, plane strain, axisymmetric, plate bending, at shell, solid, and regular ow elements. Excluded are interfaces, beams trusses, boundary elements etc. Especially for solid elements, the extended test consumes a considerable amount of processing time. REINFO INTERF activates the generation of reinforcement parts in interface elements during the preprocessing of section input [Vol. Element Library]. This option is for structural analysis only, i.e., not for potential ow analysis. If you specify the OFF option, or if you dont specify the REINFO INTERF command at all, then Diana ignores the interface elements during preprocessing. COMPOS activates the automatic generation of composed beams and solids [Vol. Element Library].
file .dcf BEGIN EVALUA CHECK SHAPE=1.E-4 RATIO=2.5 REINFO INTERF END EVALUA
[ON]
68
General Commands
3.5
Elements Assembly
Due to the ASSEMB command(s) Diana will assemble the elements of the model and create the system degrees of freedom. The OFF option switches off the elements assembly which may save computing time if this task was performed previously. syntax BEGIN ASSEMB [ OFF ] [ TOLERA=eps r ] [ BEGIN AUTOTY [ OFF ] [ BEGIN SELECT [ ELEMEN elems ng... ] [ NODES nodes ng... ] END SELECT ] END AUTOTY ] END ASSEMB
[ = 106 ]
TOLERA=eps is the criterion for coincidence of a potential system degree of freedom and an existing one. AUTOTY turns on the automatic tying option. This option generates tyings for coupling of elements with rotational degrees of freedom to elements with translational degrees of freedom only, like beamplane stress, beamsolid and shellsolid. See 2.2.8 on page 33 for conditions. If you dont specify AUTOTY commands, then Diana does not perform automatic tying. OFF explicitly turns o the auto tying option. SELECT selects parts of the nite element model for which the automatic tying option is active. ELEMEN elems is a set of elements to be used by the automatic tying option. If you dont select elements, then all elements will be used. NODES nodes is a set of nodes to be used by the automatic tying option. If you dont select nodes, then all nodes will be used.
file .dcf BEGIN ASSEMB TOLERA=1.E-5 BEGIN AUTOTY SELECT ELEMEN 1-5
[OFF]
69
3.6
Batch Output
After analysis of the nite element model, Diana can output analysis results like displacements, strains and stresses. The actual data that can be output depends on the type of analysis and on the element type, as described in the other chapters in this part. The commands to get output of analysis results are grouped in the following command blocks. syntax [ BEGIN SELECT END SELECT ] [ BEGIN LAYOUT END LAYOUT ] BEGIN OUTPUT END OUTPUT SELECT selects parts of the model [ 3.6.1]. This selection is optional, if you dont specify it then Diana gives output for the entire model, i.e., all nodes, elements, and reinforcements. LAYOUT customizes the layout of tabular output [ 3.6.3.1 p. 82]. OUTPUT selects analysis results for output [ 3.6.2 p. 77]. Scope of command blocks. There may be multiple OUTPUT blocks. In this case, the selections and layout specications are active within their environment. For instance, if you put the SELECT block outside the OUTPUT blocks, then the model selection is valid for all subsequent OUTPUT blocks. If you put a SELECT block inside an OUTPUT block, then the model selection is only valid for that particular OUTPUT block and supersedes any selection that was specied previously outside the OUTPUT block.
70
General Commands
file .dcf *LINSTA BEGIN SELECT LOADS 4 6 NODES 10-80(5) ELEMENTS 25 38 END SELECT BEGIN OUTPUT DISPLA GLOBAL STRESS VONMIS END OUTPUT BEGIN OUTPUT STRESS PRINCI END OUTPUT BEGIN SELECT LOADS 8 END SELECT BEGIN OUTPUT STRESS PRINCI END OUTPUT
These commands select load sets 4 and 6 for output of linear static analysis results. You will get the displacements in the global coordinate system for nodes 10, 15, . . . , 80, and the Von Mises and principal stresses for elements 25 and 38. For load set 8 you will get the principal stresses for all elements. The same output would have resulted from the following commands.
file .dcf *LINSTA BEGIN OUTPUT BEGIN SELECT LOADS 4 6 NODES 10-80(5) ELEMENTS 25 38 END SELECT DISPLA GLOBAL STRESS VONMIS STRESS PRINCI END OUTPUT BEGIN OUTPUT BEGIN SELECT LOADS 8 END SELECT STRESS PRINCI END OUTPUT
In short format the commands for the last OUTPUT block could have been as follows.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (I)
71
3.6.1
Model Selection
The SELECT block comprises commands to select parts of the model for which Diana must produce output of analysis results. syntax BEGIN SELECT ] [ NODES nodes ng... ALL NONE [ BEGIN ELEMEN END ELEMEN ] [ BEGIN REINFO END REINFO ] [ BEGIN COMPOS END COMPOS ] [ field w [ ] [ special ] fldnrs n... ALL END SELECT NODES select nodes: nodes are numbers or groups of selected nodes, ALL indicates all nodes. NONE indicates no nodes at all. ELEMEN is a subcommand block for element selection [ 3.6.1.1]. REINFO is a subcommand block for reinforcement selection [ 3.6.1.3]. COMPOS is a subcommand block for selection of composed elements [ 3.6.1.4]. field select elds: fldnrs are eld numbers, ALL indicates all elds. The eld name field depends on the analysis type, for instance LOADS for load sets
Diana-9.4 Users Manual Analysis Procedures (I)
[ALL]
[ALL]
72
General Commands in linear static analysis, STEPS for nonlinear or dynamic analysis, MODES for eigenvalue or stability analysis. See the appropriate chapters in this part. Special selections may be available for a particular application, for instance in linear static analysis you may select the minimum and maximum values for (selected) load sets [ 4.2.1 p. 94].
file .dcf BEGIN SELECT NODES 1 3 10-80(5) 123 125 \ 200-300 LOADS 4 6 END SELECT
3.6.1.1
Element Selection
The ELEMEN command block is a sub-block in the SELECT block to select elements for output. syntax BEGIN ELEMEN [ elems ng... ALL NONE ] ]
[ NODES
rnonrs n... ALL [ INTPNT ] intnrs n... ALL [ STRPNT ] pntnrs n... ALL ] [ CROSSE crsnrs n... ALL [ REAXES parameters for stress and strain transformation ] END ELEMEN
[ALL]
ELEMEN element selection: elems is a series of elements specied by numbers or groups or both. ALL indicates all elements. NONE indicates no elements
Diana-9.4 Users Manual Analysis Procedures (I)
3.6 Batch Output at all which is useful to suppress output for elements, for instance to get reinforcement results only. NODES selects element nodes: rnonrs are relative node numbers, 1 for the rst node, 2 for the second etc., ALL indicates all element nodes. INTPNT selects element integration points: intnrs are integration point numbers, ALL indicates all element integration points. STRPNT selects user-specied stress points (for instance in beam elements): pntnrs are stress-point numbers, ALL indicates all stress points. CROSSE selects user-specied cross-sections (for instance in beam elements): crsnrs are cross-section numbers, ALL indicates all cross-sections. REAXES species some parameters that Diana applies to transform stresses and strains [ 3.6.1.2].
file .dcf BEGIN SELECT BEGIN ELEMENTS 5-100(5) NODES 1 2 END ELEMENTS END SELECT
[ALL]
73
[ALL]
[ALL]
[ALL]
This selection comprises the rst and second node of elements 5, 10, 15, . . . , 100.
file .dcf BEGIN SELECT BEGIN ELEMENTS ALL TYPE L6BEN CROSSE ALL STRPNT 3 END ELEMENTS END SELECT
This selection comprises stress point 3 in all cross-sections of all L6BEN elements in the model. 3.6.1.2 Stress and Strain Transformation
syntax
General Commands
REAXES in the ELEMEN command block species, for the selected elements, the axes for transformation of strains or stresses. In combination with the REAXES operation, these axes specify the normal axes of the planes on which the stress or strain vectors apply. In combination with the REINFO operation, these axes specify the two reinforcement axes, see 31.2.7 on page 512 for background theory. R1= species the rst axis. Two values x1 and y1 specify a vector in the element xy axes. Three values x1 , y1 and z1 specify a vector in the model XY Z axes. Default is the element x axis. R2= species the second axis. Two or three values are interpreted analogous to parameter R1. Default is the element y axis. For stress and strain vectors, there are no restrictions for the two axes. For reinforcement forces and moments, the model axes will be projected into the elements xy plane and therefore the specied axes should not be perpendicular to the xy plane of the element. Moreover, the two reinforcement axes should not coincide. If required, Diana will take the second axis opposite to the specied direction, in order to ensure that the third axis coincides with the element z axis. ZR= species the relative internal beam arms zr : zr1 for the rst reinforcement axis, zr2 for the second. If you only specify zr1 , then Diana applies the same internal beam arm for both reinforcement axes. CO= species the coverages co, i.e., the distance from the center of the reinforcement to the outer surface: co1 for the rst reinforcement axis, co2 for the second. If you only specify co1 , then Diana applies the same coverage for both reinforcement axes. Parameters ZR and CO only apply for reinforcement forces and moments but not for plane stress elements. 3.6.1.3 Reinforcement Selection
[R1=1. 0.]
[R2=0. 1.]
(zr > 0)
(co > 0)
The REINFO command block is a sub-block in the SELECT block to select reinforcements for output. syntax BEGIN REINFO [ reinfs ng... ALL
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (I)
3.6 Batch Output NONE ] intnrs n... ALL [ ELEMEN ] elems ng... ALL END REINFO [ INTPNT REINFO reinforcement selection: reinfs is a series of reinforcements specied by numbers or groups or both. ALL indicates all reinforcements. NONE indicates no reinforcements at all which is useful to suppress output for reinforcements, for instance to get element results only. INTPNT selects reinforcement integration points: intnrs are integration point numbers, ALL indicates all reinforcement integration points. ELEMEN selects elements for reinforcement output: elems is a series of elements specied by numbers or groups or both. Results will only be output if the reinforcement is embedded in one of the selected elements. ALL indicates all elements.
file .dcf BEGIN SELECT BEGIN REINFO 1 4 5 ELEMEN 30-40 END REINFO END SELECT
75
[ALL]
[ALL]
[ALL]
This selection comprises those parts of reinforcements 1, 4 and 5 which are embedded in one of the elements 30 to 40.
file .dcf BEGIN SELECT ELEMEN NONE REINFO 1 4 5 END SELECT
This selection comprises output for reinforcements 1, 4 and 5 only. No output will be given for elements! 3.6.1.4 Composed Element Selection
The COMPOS command block is a sub-block in the SELECT block to select composed elements for output. The command syntax depends on the type of the composed elements: beams or solids.
Diana-9.4 Users Manual Analysis Procedures (I) December 15, 2009 First ed.
[ NODES
rnonrs n... ALL [ CROSSE ] crsnrs n... ALL END COMPOS COMPOS composed beam selection: elemcs is a series of composition numbers referring to table COMPOS [Vol. Element Library]. ALL indicates all compositions (the default). NONE suppresses output for compositions, for instance to get results for individual elements only. NODES selects nodes of a beam composition: rnonrs are relative node numbers, 1 for the rst node, 2 for the second etc., ALL indicates all nodes of the composition (the default). CROSSE selects user-specied cross-sections of the beam composition: crsnrs are cross-section numbers. ALL indicates all cross-sections (the default).
file .dcf BEGIN SELECT BEGIN COMPOS 3 5 CROSSE 2 END COMPOS END SELECT BEGIN OUTPUT STRESS MOMENT END OUTPUT
[ALL]
[ALL]
[ALL]
These commands ask to output the bending moments of composed beams 3 and 5 in their second cross-section. See also the example cbeam in Volume Analysis Examples. Composed solids COMPOS [ elemcs n... ALL NONE ] syntax
3.6 Batch Output COMPOS composed solid selection: elemcs is a series of composition numbers referring to table COMPOS [Vol. Element Library]. ALL indicates all compositions (the default). NONE suppresses output for compositions, for instance to get results for individual elements only.
file .dcf BEGIN SELECT COMPOS ALL END SELECT BEGIN OUTPUT STRESS DISMOM END OUTPUT
77
[ALL]
These commands ask to output the distributed bending moments of all composed solids.
3.6.2
Output Selection
The OUTPUT block comprises commands to specify for which analysis results Diana must produce output. The output device (tabular, postprocessing etc.), and some options and parameters may be specied in the starting line of this command block. syntax BEGIN OUTPUT [ device w ] { outopt w } { params } TABULA BINARY FILE=file s FEMVIE ASCII FXPLUS APPEND [ OFF ] item w [ type w ] [ form w ] [ oper w ] { comp w } { loca w } { opti w } . . . TOTAL TRANSL LOCAL NODES AXES INCREM ROTATI GLOBAL INTPNT NOAXES INITIA FORCE PRINCI STRPNT COOR PHASE DISFOR VONMIS CROSSE NOCOOR REACTI MOMENT INVARI CENTER AVERAG RESIDU DISMOM REINFO SMOOTH PRESCR CAUCHY VOLUME ERROR ELEMEN PIOLAK REAXES NOBOND REINFO GREEN ... ... TRACTI PRESSU TEMPER END OUTPUT
78
General Commands OUTPUT command block to specify desired output of analysis results. You may put the following additional keywords in the starting line:
[FEMVIE]
device species the way of output: TABULA for tabular output [ 3.6.3 p. 80]. FEMVIE for interactive postprocessing of analysis results with iDiana (the default) [ 3.6.4 p. 84]. FXPLUS for interactive postprocessing of analysis results with FX+ [ 3.6.5 p. 84]. outopt Are additional output options, availability depend on the output device. BINARY to get output in binary format. ASCII to get output in ascii text format. APPEND to append the output to an existing output le. params are additional parameters, depending on device and analysis type. FILE=file is a general parameter which species the name of the output le. This name overrules the le name specied in the diana run command [Vol. Getting Started]. Diana adds an appropriate extension to the specied le name for instance .tb for tabular output [ 3.6.3 p. 80]. See also Volume Getting Started for les in a Diana job. OFF suppresses all output of analysis results. item is the name of the result item to be output. Names depend on analysis type for instance DISPLA for displacements, STRESS for stresses. Most result items have additional speciers for type, location etc. Item names and speciers depend on the item itself, the analysis type and the element type, they are described in the appropriate volumes of the Users Manual. type is the type of the result item. TOTAL for total. INCREM for incremental. INITIA for initial. PHASE for the increment of a phase in phased analysis. REACTI for reaction. RESIDU for residual. PRESCR for prescribed. more types for specic analysis types. form is the theoretical formulation of the result item. TRANSL for translation. ROTATI for rotation.
3.6 Batch Output FORCE for force. DISFOR for distributed force. MOMENT for moment. DISMOM for distributed moment. CAUCHY for Cauchy stress. PIOLAK for PiolaKirchho stress. GREEN for GreenLagrange strain. TRACTI for tractions of interface elements. PRESSU for pressure. TEMPER for temperature. See also the description of particular result items. oper is an operation on the result item to be performed prior to its output. General operations are listed below. For analysis and/or result specic operations see the descriptions of the result items. LOCAL for transformation to local xyz directions, GLOBAL for transformation to global XY Z directions, PRINCI for transformation to principal directions. VONMIS for transformation to equivalent Von Mises strains [ 31.1.1 p. 507], or Von Mises stresses [ 31.2.1 p. 510]. INVARI for stress invariants [ 31.2.6 p. 511]. REINFO for reinforcement moments and forces [ 31.2.7 p. 512]. VOLUME for volumetric strain [ 31.1.3 p. 508]. REAXES for strain vectors. You may give special commands in the SELECT block if you ask output of strain vectors [ 3.6.1.2 p. 73]. more oper ations for specic analysis types. comp are the components of the result item to be output, for instance XX for xx or XY for xy . Components depend on the particular result item, see the descriptions. If you dont select components then Diana gives output for all components of the result item. Components only apply for tabular output; for output to the iDiana Results environment all components will be written. loca species the location, see Volume Element Library for denitions. NODES for nodes. INTPNT for element integration points. STRPNT for stress points. CROSSE for cross-sections. CENTER for averaged results in the element center point. opti are additional options for tabular output. If appropriate, options are extensively explained in the chapters for a particular type of analysis or in Volume Element Library.
Diana-9.4 Users Manual Analysis Procedures (I)
79
80
General Commands AXES gives output of the relevant axes directions, for example: directions of the degree of freedom for nodal local results directions of the element axes for element local results directions of the principal stresses or strains NOAXES suppresses this output. COOR gives output of the coordinates of the location of the result item, for instance the integration points, NOCOOR suppresses this output. SMOOTH to smooth the element data values like strains and stresses, after extrapolation to the nodes. Smoothing yields the average value of the element contributions at the node. This option can only be applied for results in global orientation. Smoothing is particularly useful for graphic output, it avoids overlapped plotting of strain and stress rosettes. ERROR gives the maximum value of the error (deviation) of the smoothed value and the extrapolated value. NOBOND includes the contribution of prestress in posttensioned reinforcements to element forces and moments [Vol. Element Library]. . . . other options may be special for a particular type of analysis, for instance AMPLIT and COMPLE for a frequency response analysis [ 8.3 p. 162].
file .dcf BEGIN OUTPUT TABULA NONLIN FILE="myfile" DISPLA TOTAL GLOBAL Y COOR STRESS TOTAL CAUCHY LOCAL XX YY XY INTPNT AXES END OUTPUT
[NOAXES]
[NOCOOR]
These commands produce tabulated output of nonlinear analysis on a le named myfile.tb. The output comprises displacements and stresses: the total displacements in global Y direction uY including the original coordinates of the nodes and the total Cauchy stresses in local element direction, xx yy xy , located in the integration points. The directions of the local xy axes are also output.
3.6.3
Tabular Output
Diana produces tabular output if you specify the result items (displacement, strain, stress etc.) in an OUTPUT command block with the device name TABULA. By default Diana chooses a page and table layout which suits most cases. To customize the layout of the tabular output you may specify a LAYOUT command block.
3.6 Batch Output syntax BEGIN OUTPUT TABULA { outopt w } {params } APPEND [ TEXT text s ] [ BEGIN LAYOUT END LAYOUT ] item w END OUTPUT APPEND causes the output to be appended to an existing le. TEXT text is a string of text which Diana will put left justied in the header line. By default, Diana applies the base name of the tabular le as header text. LAYOUT customizes the layout of the tabular output [ 3.6.3.1 p. 82]. item species the output item(s) [ 3.6.2 p. 77].
file .dcf BEGIN OUTPUT TABULA STRESS CAUCHY PRINCI STRESS CAUCHY VONMIS END OUTPUT
81
As shown below, these commands produce two tables: one for the principal stresses and another one for the equivalent Von Mises stress.
file .tb
Analysis type Load case nr. Result Tensor invariant Location of results Elmnr Nodnr 1 1 2 6 5 2 2 3 7 6 LINSTA 1 STRESS TOTAL PRINCI NODES CAUCHY
S1 S2 3.814E+05 -3.332E+05 5.041E+05 -2.856E+05 1.503E+06 1.995E+05 1.174E+06 2.651E+05 2.651E+06 5.335E+05 2.691E+06 4.167E+05 2.166E+06 -1.657E+05 2.172E+06 4.495E+04
Analysis type Load case nr. Result Tensor invariant Location of results
CAUCHY
Elmnr Nodnr 1 1 2 6 5
82
General Commands
2 3 7 6
This output is from a linear static analysis of a model with plane stress elements. See Volume Element Library for description of the available stresses for these elements. See also 4.2.4 on page 100 for description of output labels like S1, S2, S3 and Seq.
file .dcf BEGIN OUTPUT TABULA LINSTA APPEND FILE="myfile" TEXT "Example of displacement output" DISPLA COOR END OUTPUT
These commands append tabular output of the displacements of linear static analysis, including the coordinates of the nodes, to the le myfile.tb. The header text will be Example of displacement output. 3.6.3.1 Layout Specication syntax BEGIN LAYOUT [ LINPAG nlines n ] [ COLLIN ncols n ] }] [ DIGITS { RESULT ndgres n COORD ndgcoo n AXES ndgaxe n [ COMBIN ] END LAYOUT LINPAG nlines species the number of lines per page. The default is 66 lines per page, including the header and the blank line. COLLIN ncols species the number of columns per line. The default is 130 columns. DIGITS species the amount of digits (number symbols) in the format of oating point values. Diana always puts a single digit in front of the decimal point. For instance, if you specify three digits then the format looks like 2.38E+01.
(1 ndg... 10)
[nlines = 66]
[ncols = 130]
Diana never formats more than ten digits. As Diana runs on machines with double precision accuracy, yielding 15 decimals, the maximum number of digits to be formatted is within the accuracy.
Diana-9.4 Users Manual Analysis Procedures (I)
3.6 Batch Output RESULT ndgres species the number of digits for the analysis results. The default format for result values is four digits which looks like 2.385E+01. COORD ndgcoo species the number of digits for point (node) coordinates. The default format for coordinates is four digits which looks like 2.385E+01. AXES ndgaxe species the number of digits for the direction of local axes. The default format for axes directions is three digits which looks like 0.38. Diana normalizes the axes directions which yields values between 1 and 1 therefore an exponent is omitted. COMBIN causes the output items to be assembled in one table. In this case, the location specier loca [ 3.6.2 p. 77] must be the same for all items to be combined in one table. Therefore Diana assumes the location of the rst item to be valid for all items in the set. If you specify dierent locations for the items in the set, then Diana gives a warning message. This option may only be applied with explicitly specied output items, i.e., it does not work with default output. Diana will combine at most fteen items in one table.
file .dcf BEGIN OUTPUT TABULA LAYOUT COMBIN STRESS CAUCHY PRINCI STRESS CAUCHY VONMIS END OUTPUT
83
[ndgres = 4]
[ndgcoo = 4]
[ndgaxe = 3]
As shown below, these commands produce one table containing the principal stresses as well as the equivalent Von Mises stress.
file .tb
Analysis type Load case nr. Result Location of results Elmnr Nodnr 1 1 2 6 5 2 2 3 7 6 LINSTA 1 Combination NODES S3 0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.000E+00 Seq 6.193E+05 6.925E+05 1.414E+06 1.066E+06 2.428E+06 2.509E+06 2.254E+06 2.150E+06
S1 S2 3.814E+05 -3.332E+05 5.041E+05 -2.856E+05 1.503E+06 1.995E+05 1.174E+06 2.651E+05 2.651E+06 5.335E+05 2.691E+06 4.167E+05 2.166E+06 -1.657E+05 2.172E+06 4.495E+04
This output is from a linear static analysis of a model with plane stress elements. See Volume Element Library for description of the available stresses for these elements. See also 4.2.4 on page 100 for description of output labels S1, S2, S3 and Seq.
84
General Commands
3.6.4
To get output for interactive postprocessing with iDiana you must specify the FEMVIE device option. Volume Pre- and Postprocessing describes the postprocessing procedures with iDiana. See also 3.6.2 on page 77 for general syntax description of OUTPUT commands. syntax BEGIN OUTPUT FEMVIE { outopt w } { params } BINARY ASCII APPEND [ MODEL OFF ] item w END OUTPUT
[BINARY]
BINARY writes output to an iDiana database (the default). ASCII writes an iDiana neutral le in ascii text format. When choosing between database or neutral le, you should realize that neutral les are portable between dierent brands of computers. Databases (binary) are usually smaller than neutral les but generally they are not portable. APPEND appends the output to an existing iDiana neutral le or database. MODEL OFF suppresses the output of the data for the nite element model, like node coordinates and element connection. In this case Diana only outputs the analysis results. For an iDiana database, which must contain the model data, you must also specify the APPEND option.
3.6.5
To get output for interactive postprocessing with FX+ you must specify the FXPLUS device option. Volume FX+ for DIANA describes the postprocessing procedures with FX+. See also 3.6.2 on page 77 for general syntax description of OUTPUT commands. syntax BEGIN OUTPUT FXPLUS { outopt w } { params } BINARY item w END OUTPUT
85
BINARY writes output to Post-Neutral les for FX+ (the default). These neutral les are portable between dierent brands of computers. item species the output item(s) [ 3.6.2 p. 77]. Volume FX+ for DIANA includes a list of items which are available for output to FX+.
86
General Commands
Part II
Chapter 4
syntax
90
Regular Linear Static Analysis An example of the minimum set of commands to initialize a new filos le, read the input le, perform a linear static analysis, and get output of displacements and stresses is:
file .dcf *FILOS INITIA *INPUT *LINSTA BEGIN OUTPUT DISPLA STRESS END OUTPUT *END
For syntax description of the *FILOS commands see 3.2 on page 60 and for the *INPUT commands 3.3 on page 64. Default
*LINSTA *END file .dcf
If you only give a single *LINSTA command, like in the above example, then Diana evaluates the model and solves the system of equations. You will not get output of analysis results in this case. The default behavior is as if you had given the following commands.
file .dcf *LINSTA MODEL SOLVE *END
4.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual analysis. syntax BEGIN MODEL [ OFF ] [ EVALUA [ OFF ] ] [ ASSEMB [ OFF ] ]
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (II)
4.1 Model Evaluation [ MATRIX [ OFF ] ] [ LOADS [ OFF ] ] END MODEL EVALUA to check and evaluate geometric and material properties for elements [ 3.4 p. 66]. ASSEMB to assemble the elements, i.e., creating appropriate system degrees of freedom [ 3.5 p. 68]. MATRIX to setup the element stiness matrices. LOADS to setup the load vectors. You can get output of load vectors via the INTERN and EXTERN type speciers of the FORCE output item [ 4.2.5 p. 106]. Provided that the filos le of a previous analysis is still available, some tasks may be skipped which may save a considerable amount of computing time, especially for large nite element models. We show a few examples. Change loading
*INPUT READ TABLE LOADS *LINSTA BEGIN MODEL EVALUA OFF ASSEMB OFF MATRIX OFF LOADS END MODEL SOLVE BEGIN OUTPUT END OUTPUT *END file .dcf
91
A typical example of customized commands is the change of loading and the calculation of new analysis results. Module linsta only has to setup the load vectors and solve the system of equations to get the analysis results. We could also have left out the LOADS and SOLVE commands because linsta performs a main task unless it is explicitly switched o. The minimum set of commands is as follows.
file .dcf *INPUT READ TABLE LOADS *LINSTA MODEL OFF
92
BEGIN OUTPUT END OUTPUT *END
Change supports
*INPUT READ TABLE SUPPOR *LINSTA BEGIN MODEL EVALUA OFF END MODEL BEGIN OUTPUT END OUTPUT *END
file .dcf
If only some supports change in the nite element model, the evaluation of the material and geometric properties of elements may be skipped.
4.2
You can obtain output of regular linear static analysis results via an obligatory OUTPUT command block which selects the analysis results to be output. For the general, analysis type independent options for output of analysis results see 3.6 on page 69. syntax BEGIN OUTPUT [ device w ] [ outopt w... ] [ params ] [ OFF ] [ SELECT ] [ LAYOUT ] item w DISPLA STRAIN STRESS FORCE NODFOR ELMFOR FRACTU END OUTPUT
4.2 Output of Analysis Results SELECT optional commands to customize the batch output: model selection [ 3.6.1 p. 71], selection of load sets and extreme values [ 4.2.1 p. 94], stressand strain transformation [ 3.6.1.2 p. 73]. LAYOUT optional commands to customize the layout and style of tabular output [ 3.6.3.1 p. 82]. item is the name of the analysis result to be output. See 3.6.2 on page 77 for complete syntax of this command. DISPLA for displacements [ 4.2.2]. STRAIN for strains [ 4.2.3]. STRESS for stresses [ 4.2.4]. FORCE for nodal forces [ 4.2.5]. NODFOR for element nodal forces [ 4.2.6]. ELMFOR for internal element forces [ 4.2.7]. FRACTU for Linear Elastic Fracture Mechanics analysis parameters of crack tip elements [ 4.2.8]. Default
*LINSTA [ commands ] OUTPUT *END file .dcf
93
If you only give a single OUTPUT command, like in the above example, or if you dont give the OUTPUT command at all, then Diana gives a default output as if you had given the following commands.
file .dcf *LINSTA [ commands ] BEGIN OUTPUT DISPLA TOTAL TRANSL GLOBAL FORCE REACTI GLOBAL FORCE EXTERN GLOBAL STRESS TOTAL CAUCHY GLOBAL STRESS TOTAL FORCE LOCAL STRESS TOTAL MOMENT LOCAL STRESS TOTAL DISFOR LOCAL STRESS TOTAL DISMOM LOCAL STRAIN TOTAL GREEN GLOBAL END OUTPUT *END
94 Example
BEGIN OUTPUT BEGIN SELECT BEGIN ELEMEN 4-30 REAXES R1=1.0 1.0 END ELEMEN END SELECT STRESS TOTAL DISFOR STRESS TOTAL DISMOM END OUTPUT BEGIN OUTPUT BEGIN SELECT BEGIN ELEMEN 31-40 REAXES R1=1.0 0.5 END ELEMEN END SELECT STRESS TOTAL CAUCHY STRESS TOTAL DISFOR END OUTPUT
REINFO REINFO
/ R2=1.0 0.2
REAXES REAXES
Output only
*LINSTA MODEL OFF BEGIN OUTPUT STRESS TOTAL DISFOR REINFO [ more output ] END OUTPUT
file .dcf
4.2.1
For linear static analysis, you may select load sets for which Diana must output analysis results. If you dont select load sets, then Diana will output result values for all load sets. Moreover, you may ask Diana to output only the extreme result values for the (selected) load sets. syntax BEGIN SELECT LOADS [ { }] losets n... MIN ALL MAX END SELECT
4.2 Output of Analysis Results LOADS species a load selection: losets are load set numbers referring to input table LOADS [ 2.3.7 p. 55]. ALL indicates all load sets, which also is the default. MIN selects the minimum value of the (selected) load sets to be output, MAX selects the maximum value.
file .dcf BEGIN OUTPUT TABULA BEGIN SELECT LOADS 2 4 6 MAX BEGIN ELEMEN 415 533 NODES 1 END ELEMEN END SELECT STRESS CAUCHY LOCAL XX NODES END OUTPUT
95
[ALL]
CAUCHY
Element type L2TRU Extreme values Sxx 3.198E+02 Elmnr Nodnr 415 315 533 387 Sxx 3.198E+02 ( 2) 1.763E+02 ( 6)
Loadcasenr 2
Elmnr 415
Nodnr 315
This tabular output shows a maximum value for stress component xx at node 315 in element 415 to be 319.8 for load set 2. The maximum value at node 387 in element 533 is 176.3 for load set 6. Notice that Diana writes the maximum value for all selected output points just below the header paragraph. Combined items. If you ask for a combination of various output items in tabular output [ 3.6.3.1 p. 83], then Diana determines the extreme value(s) for the output item that you specied rst. The values for the other items for the same load set will be printed on the same line of the output table. Mobile loads. If you ask for tabular output for an analysis in which a mobile load is applied, you get a tabular le like:
96
DISFOR
Element type CT30S Extreme values Qxz -2.337E+01 Elmnr Nodnr 11 1 11 21 22 23 12 15 23 13 3 2 1 12 -3.742E+00 -1.709E+00 3.247E-01 -7.695E+00 -2.337E+01 -9.492E+00 -2.170E+01 6.646E-01 1.003E+01 1.095E+01 1.186E+01 1.353E+00 Qxz (12 ) (12 ) (12 ) ( 1-3) ( 1-3) ( 1-3) ( 1-2) ( 5-2) (12 ) (12 ) (12 ) ( 5-2)
Loadcasenr 1
Truckpos 3
Elmnr 11
Nodnr 23
In this example in the load cases 1 and 5 a mobile load is dened whereas load case 12 is a regular loading. The minimum total distributed shear force Qxz has a value -2.337E+01 for the truck position number 3 in load case number 1 at the location of the node with number 23 in the element with number 11. The numbers between brackets indicate the load case number, and in case of a mobile load, also the truck position after the hyphen.
4.2.2
Displacements
syntax
DISPLA [ type w ] [ form w ] [ oper w ] { comp w } { opti w } TOTAL TRANSL LOCAL PHASE ROTATI GLOBAL PRESCR DISPLA species displacements of the nodes as output item.
[TOTAL]
type
species the displacement type. TOTAL for the total displacements of a structure, i.e., the deformed geometry after a linear static analysis. PHASE for the phased displacements, i.e., the change in the deformed geometry between two consecutive phases in a phased structural analysis [ 24 p. 403]. PRESCR for the prescribed displacements, i.e., the deformation load specied in subtable DEFORM of table LOADS [ 2.3.4 p. 41]. Prescribed displacements can be output as soon as the elements have been assembled, it is not necessary that the system of equations has been solved.
4.2 Output of Analysis Results form species the displacement formulation [ 3.6.2 p. 78].
[TRANSL] [GLOBAL]
97
oper species an operation (transformation) to be performed on the displacements [ 3.6.2 p. 79]. comp selects displacement components for output. Default is all available components. opti are additional options [ 3.6.2 p. 79].
Total displacements item type form oper DISPLA TOTAL TRANSL LOCAL DISPLA TOTAL DISPLA TOTAL DISPLA TOTAL comp X Y Z
Dtx Dty Dtz
[all]
uy uz
DtY DtZ
uY uZ
Dry Drz
y z
DrY DrZ
Y Z
Phased displacements item type form oper DISPLA PHASE TRANSL LOCAL DISPLA PHASE DISPLA PHASE DISPLA PHASE
comp X Y Z
PDtx PDty PDtz
uy uY y Y
uz uZ z Z
Prescribed displacements item type form oper DISPLA PRESCR TRANSL LOCAL
comp X Y Z
pDtx pDty pDtz
ux DISPLA PRESCR TRANSL GLOBAL pDtX uX DISPLA PRESCR ROTATI LOCAL pDrx x DISPLA PRESCR ROTATI GLOBAL pDrX X
uy uY y Y
uz uZ z Z
4.2.3
Strains
syntax
STRAIN [ type w ] [ form w ] [ oper w ] { comp w } { loca w } { opti w } TOTAL GREEN LOCAL INTPNT SMOOTH
Diana-9.4 Users Manual Analysis Procedures (II) December 15, 2009 First ed.
98 FORCE MOMENT DISFOR DISMOM TRACTI DISSEI GLOBAL PRINCI VONMIS REAXES VOLUME
STRAIN species strains as output item. Table 4.1 outlines the availability and applicability of the various strain output options for each of the element families. Table 4.1: Availability of strain output
am pl .s t pl res .s s t ax rai is n y pl mm at e . . be sh nd . cu ell .s h so ell lid st r. in sp ter ri n fa p. g ce m a e. ss re in fo .
item type
form
oper
loca
opti
STRAIN TOTAL GREEN FORCE MOMENT DISFOR DISMOM TRACTI DISSEI LOCAL GLOBAL PRINCI VONMIS REAXES VOLUME INTPNT NODES STRPNT CROSSE CENTER SMOOTH ERROR
us tr
a a a a a a a a a a a a
be
a a c c a a a a a a a a a a a a
a a a a a a a a a a a a a
a a a a a a a a a a a a
a a a a a a a a a a a a
a a a a a a a a a a a a -
a a a a a a a a a a a a a -
a a a a a a a a a a a a a
a a a a a a a a a a a a a
a a d a a a a -
a a a a e e -
a a f g f a f h f -
(a) All elements. (b) Not for elements with orthotropic geometry. (c) For all beam elements, classII and classIII only in combination with local coordinate system. (d) Only for plane structural interface elements. (e) Only for base springs. (f) Only for bondslip reinforcements. (g) Only for bondslip reinforcements modeled by beam elements. (h) Not for bondslip reinforcements. (-) Not available or not suitable.
[TOTAL] [GREEN]
type form
species the strain type [ 3.6.2 p. 78]. species the strain formulation. GREEN for GreenLagrange strains [ 4.2.3.1].
4.2 Output of Analysis Results FORCE for deformations due to normal and shear forces [ 4.2.3.2]. DISFOR for generalized strains [ 4.2.3.2]. MOMENT for curvatures due to a concentrated bending moment [ 4.2.3.3]. DISMOM for curvatures due to a distributed bending moment [ 4.2.3.3]. TRACTI for tractions in structural interface elements [ 4.2.3.4]. DISSEI for distributed seismic moments in plane structural interface elements [ 4.2.3.5]. oper species an operation (transformation) to be performed on the primary strains [ 3.6.2 p. 79].
[GLOBAL]
99
comp selects strain components for output. Default is all available components. loca species the location for the strains to be output [ 3.6.2 p. 79]. opti are additional options [ 3.6.2 p. 79]. 4.2.3.1 GreenLagrange Strains
comp XX YY ZZ
Exx Eyy
[all]
Primary strains [ 31.1] item type form oper STRAIN TOTAL GREEN LOCAL STRAIN TOTAL STRAIN TOTAL STRAIN TOTAL
XY
YZ
Gyz
ZX
Gzx
Ezz Gxy
xx yy zz xy yz zx GREEN GLOBAL EXX EYY EZZ GXY GYZ GZX XX Y Y ZZ XY Y Z ZX GREEN PRINCI E1 1 GREEN REAXES E1RA a 1
E2
E3
2
E2RA
a 2
Von Mises strain [ 31.1.1] item type form oper STRAIN TOTAL GREEN VONMIS Eeq eq Volumetric strain [ 31.1.3] item type form oper STRAIN TOTAL GREEN VOLUME Evol vol
4.2.3.2
Deformations
comp X Y
Px Py
Force deformations item type form oper STRAIN TOTAL FORCE LOCAL STRAIN TOTAL FORCE
Z
Pz
ux uy uz GLOBAL PX PY PZ uX uY uZ
December 15, 2009 First ed.
100
Generalized strains itm type form oper STRAIN TOTAL DISFOR LOCAL STRAIN TOTAL DISFOR REAXES comp XX YY ZZ
XY
YZ
ZX
xx yy zz xy yz zx
P1RA P2RA
a 1
a 2
4.2.3.3
Curvatures
comp X Y Z
Kx Ky Kz
Concentrated curvatures item type form oper STRAIN TOTAL MOMENT LOCAL
x y z STRAIN TOTAL MOMENT GLOBAL KX KY KZ X Y Z Distributed curvatures item type form oper STRAIN TOTAL DISMOM LOCAL STRAIN TOTAL DISMOM REAXES comp XX YY ZZ XY YZ ZX 1
Kxx Kyy Kzz Kxy Kyz Kzx
xx yy zz xy yz zx
K1RA K2RA
a 1
a 2
4.2.3.4
Tractions
comp X Y
PTx PTy
Traction deformations item type form oper STRAIN TOTAL TRACTI LOCAL
Z
PTz
4.2.3.5
Distributed seismic moment [ 31.1.4] item type form STRAIN TOTAL DISSEI PSTOT PSNEG PSPOS PS PS PS+
4.2.4
Stresses
[ oper w ] { comp w } { loca w } { opti w } LOCAL INTPNT NOBOND GLOBAL NODES SMOOTH PRINCI STRPNT ERROR VONMIS CROSSE
syntax
4.2 Output of Analysis Results DISMOM TRACTI GRADIE USRRBE USRRSH INVARI REINFO REAXES BIAXFE CAPACI CENTER
101
am
item type
form
oper
loca
opti
STRESS TOTAL INITIA CAUCHY FORCE MOMENT DISFOR DISMOM TRACTI GRADIE USRRBE USRRSH LOCAL GLOBAL PRINCI VONMIS INVARI REINFO REAXES BIAXFE CAPACI INTPNT NODES STRPNT CROSSE CENTER NOBOND SMOOTH ERROR
us
be
a a a a a a a a a a a a a a
a a a a a a a a a a a a a a a a a a a
pl
tr
a a a a a a a a a a a a a a a a a a
.s
a a a e e a a a a a a a a a a a a
a a a c c a a a a a a a a a a a
a a b a a a a a a a a a a a a -
a a b a a a a a a a a a a a a -
a a a a a a a a a a a a a a a a a a a
a a a d d a a a a a a a a a a a a a a
a a a a a a a -
a a a a a g g -
a a a a i h f a h j h -
(a) All elements. (b) Not for elements with orthotropic geometry. (c) For shells of revolution only. (d) For composed solids only [Vol. Element Library]. (e) For innite shells only. (f) Only for bar reinforcements. (g) Only for base springs. (h) Only for bondslip reinforcements. (i) Only for bondslip reinforcements modeled by beam elements. (j) Not for bondslip reinforcements. (-) Not available or not suitable.
STRESS species stresses as output item. Table 4.2 outlines the availability and applicability of the various stress output options for each of the element families.
Diana-9.4 Users Manual Analysis Procedures (II) December 15, 2009 First ed.
102
[TOTAL] [CAUCHY]
Regular Linear Static Analysis type species the stress type [ 3.6.2 p. 78]. form species the stress formulation. CAUCHY for Cauchy stresses [ 4.2.4.1]. FORCE for concentrated forces and tractions [ 4.2.4.2]. MOMENT for concentrated bending moments [ 4.2.4.3]. DISFOR for distributed forces [ 4.2.4.2]. DISMOM for distributed bending moments [ 4.2.4.3]. TRACTI for tractions in structural interface elements [ 4.2.4.4]. GRADIE for gradients of stresses in reinforcement bars [ 4.2.4.5]. USRRBE for userdened derived results for beam elements [ 4.2.4.6]. To determine the userdened derived results Diana needs additional material properties and the usersuplied subroutine USRBEA, see Volume Material Library. USRRSH for userdened derived results for plate and shell elements [ 4.2.4.7]. To determine the userdened derived results Diana needs additional material properties and the usersuplied subroutine USRSHL, see Volume Material Library.
[GLOBAL]
oper
species an operation (transformation) to be performed on the primary stresses [ 3.6.2 p. 79]. BIAXFE gives the following Safety Factors for concrete under static and dynamic loading conditions with reference to a biaxial failure envelope:
F Sstatic usual = Rstatic usual /r F Sstatic unusual = Rstatic unusual /r F Sdynamic unusual = Rdynamic unusual /r F Sdynamic extreme = Rdynamic extreme /r
Where r is the distance from the origin to the actual stress point (1 , 2 ): r = (1 2 + 2 2 ) (4.1) To determine the Safety Factors Diana needs additional material properties. For the required parameters and additional information on the Safety Factors, see Volume Material Library. Note that all Safety Factors are limited to 100. CAPACI gives the shear capacity and hydrostatic pressure capacity of stress against MohrCoulomb failure criterion [ 31.2.9 p. 519]. To determine the shear capacity and hydrostatic pressure capacity Diana needs additional material properties, see Volume Material Library.
[all]
comp selects stress components for output. Default is all available components.
Diana-9.4 Users Manual Analysis Procedures (II)
4.2 Output of Analysis Results loca species the location for the stresses to be output [ 3.6.2 p. 79]. opti are additional options [ 3.6.2 p. 79]. 4.2.4.1 Cauchy Stresses
comp XX YY
Sxx Syy
103
Primary stresses [ 31.2] item type form oper STRESS TOTAL CAUCHY LOCAL STRESS TOTAL STRESS TOTAL STRESS TOTAL STRESS INITIA STRESS INITIA STRESS INITIA STRESS INITIA
ZZ
Szz
XY
Sxy
YZ
Syz
ZX
Szx
xx yy zz xy yz zx CAUCHY GLOBAL SXX SYY SZZ SXY SYZ SZX XX Y Y ZZ XY Y Z ZX CAUCHY PRINCI S1 1 CAUCHY REAXES S1RA a 1 CAUCHY LOCAL S0xx S0yy S0zz S0xy S0yz S0zx 0 0 0 0 0 0 xx yy zz xy yz zx CAUCHY GLOBAL S0XX S0YY S0ZZ S0XY S0YZ S0ZX 0 0 0 0 0 0 XX Y Y ZZ XY Y Z ZX CAUCHY PRINCI S01 0 1 CAUCHY REAXES S01RA 0a 1
S2
S3
2
S2RA
a 2
S02
0 2
S03
0 3
S02RA
0a 2
Von Mises stress [ 31.2.1] item type form oper STRESS TOTAL CAUCHY VONMIS Seq eq STRESS INITIA CAUCHY VONMIS S0eq 0 eq Stress invariants [ 31.2.6] comp item type form oper P Q LODE STRESS TOTAL CAUCHY INVARI P Q Lode p q Biaxial failure envelope comp item type form oper STUS STUN DYUN DYEX STRESS TOTAL CAUCHY BIAXFE FSstus FSstun FSdyun FSdyex F Sstus F Sstun F Sdyun F Sdyex Shear capacity [ 31.2.9] comp SHRCAP PPCAP item type form oper STRESS TOTAL CAUCHY CAPACI SHRCAP PPCAP
Concentrated forces item type form oper STRESS TOTAL FORCE LOCAL STRESS TOTAL FORCE
X
Qy
Y
Nz
Qy
NY
Qz Qx Qy Nz
NZ
NY
Q0y
NZ
Q0z Q0x Q0y N0z
0 Q0 Q0 Q0 Nz z x y
Q0 y
N0Y
0 NY
N0Z
0 NZ
Distributed forces item type form oper STRESS TOTAL DISFOR LOCAL STRESS TOTAL DISFOR REAXES
comp XX YY ZZ
Nxx Nyy Nzz
XY
Nxy
YZ
Qyz
ZX
Qxz
qxz
N1RA N2RA
na 2
Reinforcement forces [ 31.2.7] comp item type form oper 1R 2R 1RC 2RC QT STRESS TOTAL DISFOR REINFO N1R N2R N1RC N2RC QT n1 n2 n1c n2c q
4.2.4.3
Bending moments
comp X Y Z
Mx My MY Mz MZ
Concentrated moments item type form oper STRESS TOTAL MOMENT LOCAL STRESS TOTAL
Mx My Mz MOMENT GLOBAL MX MX STRESS INITIA MOMENT LOCAL M0x 0 Mx STRESS INITIA MOMENT GLOBAL M0X 0 MX MY MZ
M0y M0z
0 0 My Mz
M0Y M0Z
0 0 MY MZ
105
comp XX YY ZZ XY
Mxx Myy Mxy
YZ ZX 1
mxx myy
mxy
M1RA M2RA
ma 2
Reinforcement moments [ 31.2.7] comp item type form oper 1R 2R STRESS TOTAL DISMOM REINFO M1R M2R m1 m2
4.2.4.4
Tractions
form oper TRACTI LOCAL comp X Y Z
STx STy STz
tx TRACTI GLOBAL STX tX STRESS INITIA TRACTI LOCAL ST0x t0 x STRESS INITIA TRACTI LOCAL ST0X t0 X
ty
STY
tz
STZ
tY t0 y t0 Y
tZ t0 z t0 Z
4.2.4.5
Stress gradients [ 31.2.8] item type form oper STRESS TOTAL GRADIE LOCAL SG xx
4.2.4.6
4.2.4.7
106
4.2.5
Nodal Forces
syntax
FORCE [ type w ] [ form w ] [ oper w ] { comp w } { opti w } REACTI TRANSL LOCAL RESIDU ROTATI GLOBAL EXTERN INTERN FORCE species forces and moments in the nodes as output item [ 31.3 p. 519].
[REACTI]
type
species the type of the nodal forces. REACTI for the reaction forces in all supported nodes. RESIDU for the residual forces (by default Diana only gives the extreme residual forces). EXTERN for the external load vectors. INTERN for the right-hand-side vectors (internal load). Load vectors can be output as soon as these have been setup [ 4.1 p. 91], it is not necessary that the system of equations has been solved.
[TRANSL] [GLOBAL]
species the formulation [ 3.6.2 p. 78] species an operation (transformation) to be performed on the forces and moments [ 3.6.2 p. 79]. selects force or moment components for output. Default is all available components.
[all]
Z
FBz
b Fz
FBZ
b FZ
MBz
b Mz
MBZ
b MZ
107
FORCE RESIDU TRANSL GLOBAL FRX r FX FORCE RESIDU ROTATI LOCAL MRx r Mx FORCE RESIDU ROTATI GLOBAL MRX r MX Load vectors [ 31.3.2] item type form oper FORCE EXTERN TRANSL LOCAL
MRY MRZ
r r MY MZ
comp X Y
FEx FEy FEY MEy MEY FIy FIY MIy MIY
Z
FEz FEZ MEz MEZ FIz FIZ MIz MIZ
ext ext Fx Fy Fz
FORCE EXTERN TRANSL GLOBAL FEX ext FX FORCE EXTERN ROTATI LOCAL MExt ext Mx FORCE EXTERN ROTATI GLOBAL MEX ext MX FORCE INTERN TRANSL LOCAL FIx int Fx FORCE INTERN TRANSL GLOBAL FIX int FX FORCE INTERN ROTATI LOCAL MIxt int Mx FORCE INTERN ROTATI GLOBAL MIX int MX
4.2.6
syntax
NODFOR [ type w ] [ form w ] [ oper w ] { comp w } { opti w } ELEMEN TRANSL LOCAL REINFO ROTATI GLOBAL TOTAL NODFOR species internal nodal element forces and moments in the nodes as output item. This command gives the contributions of the element or reinforcement internal nodal forces (or both), to a certain node. A selection of elements which form a section of the model gives the total internal forces that act on that section. type species the forces type. ELEMEN for contribution of elements only.
Diana-9.4 Users Manual Analysis Procedures (II) December 15, 2009 First ed.
[TOTAL]
108
Regular Linear Static Analysis REINFO for contribution of embedded reinforcements only. TOTAL for contribution of both elements and embedded reinforcements.
[TRANSL] [GLOBAL]
species the formulation [ 3.6.2 p. 78]. species an operation (transformation) to be performed on the forces and moments [ 3.6.2 p. 79]. selects force or moment components for output. Default is all available components.
[all]
Z
FNz
n Fz
FNZ
n FZ
MNz
n Mz
MNZ
n MZ
4.2.7
syntax
ELMFOR [ type w ] [ form w ] [ oper w ] { comp w } { opti w } ELEMEN TRANSL GLOBAL REINFO ROTATI TOTAL ELMFOR species internal element forces and moments in the nodes of an element as output item. This command gives the contributions of the element or reinforcement internal forces (or both), to a certain node of an element.
[TOTAL]
type
species the forces type. ELEMEN for contribution of elements only. REINFO for contribution of embedded reinforcements only. TOTAL for contribution of both elements and embedded reinforcements.
[TRANSL] [GLOBAL]
species the formulation [ 3.6.2 p. 78]. species an operation (transformation) to be performed on the forces and moments [ 3.6.2 p. 79]. selects force or moment components for output. Default is all available components.
Diana-9.4 Users Manual Analysis Procedures (II)
[all]
4.2 Output of Analysis Results opti are additional options [ 3.6.2 p. 79].
Internal element forces comp item type form oper X Y ELMFOR type TRANSL GLOBAL FEX FEY e e FX F Y ELMFOR type ROTATI GLOBAL MEX MEY e e MX MY
109
Z
FEZ
e FZ
MEZ
e MZ
4.2.8
For crack tip elements [Vol. Element Library] Diana can calculate and output the parameters for Linear Elastic Fracture Mechanics analysis (LEFM). See Chapter 37 for background theory. See also example crackpl in Volume Analysis Examples. syntax FRACTU FRACTU species the LEFM parameters as output item. No further options apply for this output. The LEFM parameters are always output as three components.
LEFM parameters item FRACTU K G KI G I
KI is the Mode-I stress intensity factor [Eq. (37.2) p. 564]. The stress intensity factor can only be calculated if the material parameters and the thickness (for plane stress elements) of all the elements in the crack front are identical. GI is the Mode-I energy release rate [Eq. (37.8) p. 565].
110
Chapter 5
5.1
If the stress amplitude is calculated in a specic material point of the model, then the number of load cycles to fatigue failure can be easily determined for this point using the Whler diagram. For each selected point, Diana automatically o derives the stress amplitude. Therefore, assuming that in one load cycle all selected load sets are passed through, the dierence of the extreme stress values of these load sets yield the stress amplitude of this point [Fig. 5.1].
S
Sampl
1 2 3 4 5 6 7 8
load set
112
5.1.1
Stress Amplitudes
Diana can calculate the stress amplitude from the Von Mises stresses or from the principal stresses. For the Von Mises stresses, the amplitude is the dierence between the smallest and largest stress. For the principal stresses, the amplitude is the dierence between the largest absolute value of the maximum and minimum stresses. Examples of principal stress amplitude. situation of extreme principal stresses S1.max = 1.0 S1.min = 0.5 The stress amplitude is calculated by Sampl = S1.max S1.min = 1.0 0.5 = 0.5 (5.2) Consider the following stress
(5.1)
Another example is the following stress situation of extreme principal stresses S1.max = 1.0 S1.min = 0.0 the stress amplitude yields Sampl = S1.max S2.min = 1.0 0.5 = 1.5 (5.4) S2.max = 0.5 S2.min = 0.5 (5.3)
5.1.2
Output Selection
syntax
BEGIN OUTPUT [device w ] [outopt w... ] [params ] [ OFF ] [ BEGIN SELECT [ ] LOADS [ ] losets n... ALL END SELECT ] NFATIG { oper w } { loca w } { opti w } VONMIS NODES SMOOTH PRINCI INTPNT ERROR END OUTPUT
5.1 Number of Load Cycles to Failure for model selection see 3.6.1 on page 71. A load cycle is dened by a series of two or more load sets which subsequently occur in one cycle. By default Diana takes all load sets specied in table LOADS [ 2.3.7 p. 55], you may select load sets with the following command. LOADS losets is a series of load sets, referring to input table LOADS, which dene the load cycle for fatigue failure analysis. OUTPUT command block to select analysis results for output. See 3.6.2 on page 77 for options. You must specify the output of the number of load cycles to fatigue failure in separate OUTPUT blocks, because the interpretation of selected loads is not like for other output items. It is not allowed to put the item NFATIG together with other items in one OUTPUT block. NFATIG species the number of load cycles to fatigue failure as output item. For each output point, Diana calculates the stress amplitude from the extreme stresses. Using the specied Whler diagram [Vol. Mateo rial Library], Diana translates the stress amplitude into the number of load cycles to fatigue failure. oper species the formulation of the stress amplitude used. VONMIS for dierence between extreme Von Mises stresses.1 PRINCI for dierence between extreme principal stresses. loca species the location for the strains to be output [ 3.6.2 p. 79]. opti are additional options [ 3.6.2 p. 79].
item oper NFATIG VONMIS NF Nf NFATIG PRINCI NF Nf
[VONMIS]
113
[ALL]
1 Von
114
Part III
Dynamic Analysis
Chapter 6
6.1
Mass
In dynamic analysis Diana requires the input of mass parameters. There are two forms of mass input: mass density for structural elements or concentrated mass via point elements. See Volume Material Library for description of input syntax. Mass density. In dynamic analysis you must specify the mass density for all structural elements. In static analysis, the mass density is required to determine dead weight load. Concentrated mass. Concentrated mass in dynamic analysis can be modeled with point mass/damping elements. Generally speaking, these elements dont inuence the static behavior of the model, i.e., they dont have stiness, strain or stress. In static analysis, the concentrated mass acts as concentrated loading for dead weight.
6.2
Damping
There are various forms of damping input: viscous damping and structural damping for solid and structural elements, continuous damping (dashpots) via discrete spring/dashpot elements or point mass/damping elements. See Volume
Diana-9.4 Users Manual Analysis Procedures (III) December 15, 2009 First ed.
118
Input for Dynamic Analysis Material Library for description of input syntax. See also 32.1.2 on page 522 for background theory of damping. Viscous damping. Viscous damping is a form of damping which is proportional to the velocity. For structural elements you may specify Raleigh damping coecients to simulate viscous damping. Structural damping. Structural damping, also called hysteretic damping, is a form of damping which is independent of the frequency and proportional to the displacement. It may be specied for structural elements in the model and is only valid in a frequency response analysis. Continuous damping. Continuous damping may be specied via discrete spring/dashpot elements or one-node point mass/damping elements.
6.3
Base Excitation
A rigid base motion may be specied with accelerations in various directions. A typical application for base excitation is a well known earthquake spectrum. See 32.1.4 on page 523 for background theory. syntax LOADS CASE case n BASE
1 80
dirnr n accel r . . .
CASE starts a new subtable, case is the load case number, used for reference from the analysis commands. BASE indicates that the load case consists of a base motion: number dirnr refers to a direction in table DIRECT, value accel is the uniform translational acceleration applied to the rigid base. For all translational supports, Diana will resolve the acceleration in the supported direction.
file .dat DIRECT 1 1. 2 0. 3 0. 4 1. LOADS
December 15, 2009 First ed.
0. 1. 0. 1.
0. 0. 1. 0.
119
In this example the rigid base involves a uniform translational acceleration of 6 length/time2 in the model Y direction and of 10 length/time2 in a direction of 45 with the model X and Y axes.
6.4
Transient Analysis
This section describes the input for transient structural and uidstructure interaction analysis which is additional to the general input for the nite element model [Ch. 1] and for structural analysis [Ch. 2]. See also Chapter 7 for commands for transient analysis.
6.4.1
Initial Conditions
You may apply various types of initial conditions for a transient analysis.
Specied initial displacements or velocities. You may specify initial displacements or velocities in input table INIVAR [ 6.4.1.1]. To let Diana actually use these displacements or velocities as an initial condition for a transient analysis, you must supply the START command with special options [ 7.3 p. 130] [ 12.3.1 p. 219]. Calculated initial displacements. You may ask Diana to apply previously calculated displacements as an initial condition for a transient dynamic analysis. Calculated initial displacements require no further input data, you only must supply the START command with special options depending on the origin of the displacement eld. Specied initial stresses. The initial stresses as specied via an element prestress load in input table LOADS [Vol. Element Library] may be applied as initial stresses for the transient analysis. To let Diana actually apply these stresses as an initial condition you must supply the START command with special options. Calculated initial stresses. The stresses as calculated for a load set of a linear static analysis may be applied as initial stresses for a transient analysis. Calculated initial stresses require no further input data, you only must supply the START command with special options.
6.4.1.1
Initial displacement or velocity elds for transient analysis are input via table INIVAR. These elds may be used as initial (strain free) condition for a transient analysis via special start option [ 12.3.1.1 p. 222].
Diana-9.4 Users Manual Analysis Procedures (III) December 15, 2009 First ed.
120
node n
1
node n
1
DISPLA is the subtable heading for initial displacements and VELOCI for initial velocities. Number field is the eld number for reference in the start option command. node is a single node number. nodes is a series of nodes, it must be specied between slashes and may comprise numbers or groups or both. type is the type of variable: TR for translation or RO for rotation. dirnr species the direction, referring to table DIRECT [ 1.4 p. 14]. displa is the initial displacement. displs is a series of initial displacements, one for each node in nodes , it must be specied between slashes. veloci is the initial velocity. velocs is a series of initial velocities, one for each node in nodes , it must be specied between slashes.
file .dat INIVAR DISPLA 3 1 TR 3 0.001 / 10-12 / TR 2 / 0.002(3) / 4 RO 5 0.00005
121
This example species initial displacement eld number 3. The initial translation of node 1 is 0.001 in direction 3. The initial translation of nodes 10 to 12 is 0.002 in direction 2. The initial rotation of node 4 is 0.00005 around direction 5. All other translations and rotations are initially equal to zero.
6.4.2
TimeLoad Diagram
You may specify a timeload diagram completely, i.e., times and load factors, in table TIMELO. Alternatively, you may specify only the times in table TIMELO and import the load factors from an external le. Complete table TIMELO
1 80
syntax
LOAD tlo n TIMES times r... / FACTOR lf r... / TIMELO is the table heading for the timeload diagram input. LOAD tlo species the transient load which is active during the following times. The load number tlo refers to a load set in Table LOADS [ 2.3.7 p. 55]. TIMES times are the times t for the corresponding load factors. Times must be specied in increasing order. FACTOR lf are the load factors for the corresponding times. You may specify multiple sets of times and load factors, optionally with a different load number for each set. Within one set, the number of factors lf must always match the number of times in times as shown in the following example.
file .dat TIMELO LOAD 2 TIMES 0.00 0.10 / FACTOR 2.5 3.6 / LOAD 1 TIMES 0.13 0.16 0.16 0.45 / FACTOR 6.4 4.2 2.5 0.0 / LOAD 3 TIMES 0.45:0.501(0.01) / FACTOR 4.8 5.3 7.9 7.8 6.2 1.4 /
(ti+1 ti )
LOAD nlo n TIMES times r... / FACTOR IMPORT file s [ SKIP nlin n ] [ ] SCALE sfac r PEAK pval r TIMELO is the table heading for the timeload diagram input. LOAD nlo species the load which is active during the following times. The load number nlo refers to a load set in Table LOADS [ 2.3.7 p. 55].
(ti+1 ti )
TIMES times are the times t for the corresponding load factors. Times must be specied in increasing order. FACTOR IMPORT asks Diana to read the load factors from an external le named file . Factors on this le must be separated by spaces, commas, tabs or newlines. There are two additional options.
[nlin = 0]
[sfac = 1]
(pval > 0)
SKIP causes Diana to skip the rst nlin lines before starting to read load factors. SCALE species a scale factor sfac . Diana will multiply each load factor from the external le by this factor before applying the load. PEAK asks Diana to scale the values from the external le such that the greatest absolute vale of the load factors is equal to pval . You may specify multiple sets of times and le, optionally with a dierent load number for each set. Within one set, Diana reads as many factors from file as there are times in times . For repeated import from file reading continues from the last read factor of the previous import.
file .dat TIMELO LOAD 4 TIMES 0.00 0.10 0.13 0.16 0.16 0.45 / FACTOR IMPORT "load.log" SKIP 2 SCALE 3.
In this example Diana will read six load factors from external le load.log. This le could be like shown below.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (III)
123
The rst two lines will be skipped. The actual load factors for transient load 4 will be: lf1 = 3 1 = 3, lf2 = 3 3 = 9, etc. Note that in this example the load factor suddenly drops at time t = 0.16 from 3 5.2 = 15.6 to 3 3.2 = 9.6 . Value formats. All regular Fortran formats are allowed for the factors on the external le. Integers (without decimal point) are interpreted as reals. Exponents of ten may be specied via E or D format. For instance, 45, 45.0, 45.0E0, and 45.0D0 all represent the real value of 45.0 for the scale factor.
6.4.3
Time dependent temperature, concentration, and pressure is taken into account as a loading in a transient analysis. See input tables TEMPER, CONCEN, and PRESSU in Volume Material Library. This transient loading is analogous to specication via element loads TEMPER, CONCEN, and PRESSU in subtable ELEMEN of table LOADS [Vol. Element Library], in combination with a timeload diagram in table TIMELO [ 6.4.2].
6.5
This section describes the input for frequency domain structural and uid structure interaction analysis which is additional to the general input for the nite element model [Ch. 1] and for structural analysis [Ch. 2]. See also Chapter 8 for commands for frequency response analysis and Chapter 9 for commands for spectral response analysis.
6.5.1
FrequencyLoad Diagram
You may specify a frequencyload diagram completely, i.e., frequencies and load factors, in table FREQLO. Alternatively, you may specify only the frequencies in table FREQLO and import the load factors from an external le. Complete table FREQLO
1 80
syntax
124
Input for Dynamic Analysis FREQLO is the table heading for the frequencyload diagram input. LOAD tlo species the transient load which is active during the following frequencies. The load number tlo refers to a load set in Table LOADS [ 2.3.7 p. 55].
(fi+1 fi )
FREQUE frequencies are the frequencies f for the corresponding load factors. frequencies must be specied in increasing order. FACTOR lf are the load factors for the corresponding frequencies. You may specify multiple sets of frequencies and load factors, optionally with a dierent load number for each set. Within one set, the number of factors lf must always match the number of frequencies in times as shown in the following example.
file .dat FREQLO LOAD 2 FREQUE 0.00 0.10 / FACTOR 2.5 3.6 / LOAD 1 FREQUE 0.13 0.16 0.16 0.45 / FACTOR 6.4 4.2 2.5 0.0 / LOAD 3 FREQUE 0.45:0.501(0.01) / FACTOR 4.8 5.3 7.9 7.8 6.2 1.4 /
In this example no load is active from f = 0.10 to f = 0.13. Factors imported FREQLO
1 5 6 80
syntax
LOAD nlo n FREQUE frequencies r... / FACTOR IMPORT file s [ SKIP nlin n ] ] [ SCALE sfac r PEAK pval r FREQLO is the table heading for the frequencyload diagram input. LOAD nlo species the load which is active during the following frequencies. The load number nlo refers to a load set in Table LOADS [ 2.3.7 p. 55].
6.5 Frequency Domain Analysis FREQUE frequencies are the frequencies f for the corresponding load factors. Frequencies must be specied in increasing order. FACTOR IMPORT asks Diana to read the load factors from an external le named file . Factors on this le must be separated by spaces, commas, tabs or newlines. There are two additional options. SKIP causes Diana to skip the rst nlin lines before starting to read load factors. SCALE species a scale factor sfac . Diana will multiply each load factor from the external le by this factor before applying the load. PEAK asks Diana to scale the values from the external le such that the greatest absolute vale of the load factors is equal to pval . You may specify multiple sets of frequencies and le, optionally with a dierent load number for each set. Within one set, Diana reads as many factors from file as there are frequencies in frequencies . For repeated import from file reading continues from the last read factor of the previous import.
file .dat FREQLO LOAD 4 FREQUE 0.00 0.10 0.13 0.16 0.16 0.45 / FACTOR IMPORT "load.log" SKIP 2 SCALE 3.
[nlin = 0]
125
(fi+1 fi )
[sfac = 1]
(pval > 0)
In this example Diana will read six load factors from external le load.log. This le could be like shown below.
load.log Load logging for tower footing Thu Mar 19 16:33:55 MET 1998 1 3 4 5.2 3.2 1.6
The rst two lines will be skipped. The actual load factors for transient load 4 will be: lf1 = 3 1 = 3, lf2 = 3 3 = 9, etc. Note that in this example the load factor suddenly drops at frequency f = 0.16 from 3 5.2 = 15.6 to 3 3.2 = 9.6 . Value formats. All regular Fortran formats are allowed for the factors on the external le. Integers (without decimal point) are interpreted as reals. Exponents of ten may be specied via E or D format. For instance, 45, 45.0, 45.0E0, and 45.0D0 all represent the real value of 45.0 for the scale factor.
126
Chapter 7
[Ch. 8].
Diana-9.4 Users Manual Analysis Procedures (III) December 15, 2009 First ed.
128 TYPE species the type of the analysis [ 7.2]. EXECUT executes steps [ 7.3].
OUTPUT selects analysis results for output [ 7.4 p. 132]. This output selection, at the *NONLIN command level, is valid for all EXECUT blocks of the current analysis. However, an output selection within any EXECUTE block [ 7.3 p. 131], overrules the output selection at the *NONLIN level. Example. The commands for linear transient analysis are a sub-set of those for nonlinear transient analysis as described in Chapter 12. The following is an example of commands for linear transient analysis.
file .dcf *FILOS INITIA *INPUT *NONLIN BEGIN TYPE BEGIN TRANSI METHOD NEWMAR GAMMA=0.5 BETA=0.25 DYNAMI END TRANSI END TYPE BEGIN EXECUTE BEGIN TIME BEGIN STEPS step sizes END STEPS END TIME BEGIN ITERAT MAXITE=0 no iteration! END ITERAT END EXECUTE *END
The *FILOS command with INITIA invokes Module filos to initialize the filos le which is the central database for each analysis with Diana [ 3.2 p. 60]. The *INPUT command invokes Module input to read the complete input data le. This le must contain the general input data for the nite element mesh [Ch. 1], additional data for structural analysis [Ch. 2], the data for mass and damping [Vol. Material Library], and additional data for transient analysis [ 6.4 p. 119] like the timeload diagram as shown below.
file .dat TIMELO LOAD 2 TIMES 0.0 0.00001 1.0 / FACTOR 0.0 2.85 2.85 /
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (III)
129
Preliminary static analysis. If you dene an initial state with calculated displacements or stresses, then the displacements must be determined via an EXECUT START command block [ 12.3.1 p. 219]. In all other cases the commands described previously are formally sucient for a linear transient analysis. However, you are urged to perform a linear static analysis and some postprocessing before continuing with the transient dynamic analysis [Ch. 4]. Fluidstructure interaction analysis. In case of a model containing uid, structural elements and uid-structure interface elements, automatically a uid structure linear transient analysis will be performed. Diana recognizes this element combination and will perform the required preparation to perform a linear transient uidstructure analysis. The added mass eect of the uid [Eq. (32.92) p. 537] is taken into account to calculate the linear transient response. Results will be available for the structural part of the model. Lumped element matrices may not be used in a uidstructure response analysis.
7.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual analysis. syntax BEGIN MODEL [ OFF ] [ EVALUA [ OFF ] ] [ ASSEMB [ OFF ] ] [ MATRIX [ OFF ] ] [ LOADS [ OFF ] ] END MODEL EVALUA to check and evaluate geometric and material properties for elements and reinforcements [ 3.4 p. 66]. ASSEMB to assemble the elements, i.e., creating appropriate system degrees of freedom [ 3.5 p. 68]. MATRIX to setup the element stiness matrices. LOADS to setup the the load vectors.
Diana-9.4 Users Manual Analysis Procedures (III) December 15, 2009 First ed.
130
7.2
BEGIN TYPE [ OFF ] PHYSIC OFF TRANSI END TYPE PHYSIC OFF may switch o physical nonlinear analysis, which by default is on for Module nonlin. TRANSI sets options and parameters for transient analysis. Commands are analogous to those for nonlinear transient analysis [ 12.2.3].
7.3
Step Execution
With the EXECUT commands you ask Diana to execute load or time steps. A command le may contain one or more EXECUT blocks. Details of various commands are given in the referred sections. syntax BEGIN EXECUT [ OFF ] [ type w ] START LOAD TIME [ ITERAT ] [ SOLVE ] [ STOP ] [ OUTPUT ] END EXECUT
[LOAD]
type indicates the type of step(s) to be executed. START to evaluate the initial state [ 12.3.1 p. 219]. LOAD to execute load steps [ 7.3.1]. TIME to execute time steps [ 7.3.2]. ITERAT switches o the process of equilibrium iteration [ 7.3.3 p. 132].
7.3 Step Execution SOLVE customizes the settings for the solution method [Ch. 26 p. 429]. STOP species a stop criterion for step execution [ 12.3.6 p. 246]. OUTPUT selects output for steps [ 7.4 p. 132]. This selection overrules any output selection specied at the *NONLIN command level [Ch. 7 p. 128].
131
7.3.1
Load Steps
syntax
BEGIN LOAD [ OFF ] [ LOADNR=loset n ] [ RESTOR step n ] [ BEGIN STEPS [ method w ] EXPLIC ENERGY AUTOMA END STEPS ] [ SAVE ] END LOAD LOADNR loset refers to a load set from table LOADS [ 2.3.7 p. 55]. Default is the lowest available load set number. RESTOR restores data for step step which must have been saved previously [ 12.3.9 p. 251]. STEPS species how to apply step sizes. Via method you may specify the step sizes explicitly, or you may let Diana determine them automatically. EXPLIC explicitly specied step sizes [ 12.3.2.1 p. 227]. ENERGY energy based automatic step size control [ 12.3.2.3 p. 229]. AUTOMA automatic step size control [ 12.3.2.5 p. 233]. Note that Arc-length methods are not protable in linear analysis. SAVE saves data of specied steps for future restart [ 12.3.9 p. 251].
[EXPLIC]
7.3.2
Time Steps
132
Linear Transient Analysis syntax BEGIN TIME [ OFF ] [ RESTOR step n ] [ BEGIN STEPS [ method w ] EXPLIC EXPONE AUTOMA END STEPS ] [ SAVE ] END TIME RESTOR restores data for step step which must have been saved previously.
[EXPLIC]
STEPS species how to apply step sizes. Via method you may specify the step sizes explicitly, or you may let Diana determine them automatically. EXPLIC explicitly specied step sizes [ 12.3.3.1 p. 235]. EXPONE applies exponential time increments [ 12.3.3.3 p. 236]. AUTOMA automatic step size control [ 12.3.2.5 p. 233]. SAVE saves data of specied steps for future restart [ 12.3.9 p. 251].
7.3.3
No Equilibrium Iteration
Linear transient analysis does not require equilibrium iteration. Therefore you may switch o the iteration process via the ITERAT commands. syntax BEGIN ITERAT MAXITE=0 END ITERAT MAXITE=0 sets the maximum number of iterations for each time or load step to zero.
7.4
You may indicate the analysis results to be output via the OUTPUT commands. For the general, analysis type independent options for output of analysis results see 3.6 on page 69.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (III)
7.4 Output of Analysis results syntax BEGIN OUTPUT [device w ] [outopt w... ] [params ] [ OFF ] [ BEGIN SELECT [ ] [ STEPS ] END SELECT ] [ LAYOUT ] item w DISPLA VELOCI ACCELE STRAIN STRESS FORCE NODFOR ELMFOR FSPRES END OUTPUT SELECT command block to customize the batch output. for model selection see 3.6.1 on page 71, for stress- and strain transformation 3.6.1.2 on page 73. STEPS selects steps for output [ 12.4.1 p. 254]. LAYOUT optional commands to specify the layout and style of tabular output [ 3.6.3 p. 80]. item is the name of the analysis result to be output. See 3.6.2 on page 77 for complete syntax of this command. DISPLA for displacements [ 7.4.1]. VELOCI for velocities [ 7.4.2]. ACCELE for accelerations [ 7.4.3]. STRAIN for strains [ 7.4.4]. STRESS for stresses [ 7.4.5]. FORCE for nodal forces and moments [ 7.4.6]. NODFOR for element nodal forces [ 7.4.7]. ELMFOR for internal element forces [ 7.4.8]. FSPRES for dynamic pressures of uidstructure interface elements [ 7.4.9].
133
134
7.4.1
Displacements
syntax
DISPLA [ type w ] [ form w ] [ oper w ] { comp w } { opti w } TOTAL TRANSL LOCAL RELATI INCREM ROTATI GLOBAL PHASE DISPLA species displacements of the nodes as output item.
[TOTAL]
type
species the displacement type. TOTAL for the total displacements of a structure, i.e., the deformed geometry at a certain stage. INCREM for the incremental displacements, i.e., the change in the deformed geometry between two consecutive steps. PHASE for the phased displacements, i.e., the change in the deformed geometry between two consecutive phases in a phased analysis.
[TRANSL] [GLOBAL]
form
oper species an operation (transformation) to be performed on the displacements [ 3.6.2 p. 79]. comp selects displacement components for output. Default is all available components. opti are additional options. The RELATI option is special for analyses with base excitation [ 6.3 p. 118], it causes the resulting displacements to be output relative to the base displacements. See 3.6.2 on page 79 for other options.
Total displacements item type form oper DISPLA TOTAL TRANSL LOCAL DISPLA TOTAL DISPLA TOTAL DISPLA TOTAL DISPLA TOTAL DISPLA TOTAL DISPLA TOTAL DISPLA TOTAL TRANSL LOCAL TRANSL GLOBAL opti comp X Y
TDtx TDty
[all]
Z
TDtz
ux
uy
uz
uX TRANSL GLOBAL RELATI TrDtX uX r ROTATI LOCAL TDrx x ROTATI LOCAL RELATI TrDrx xr ROTATI GLOBAL TDrX X ROTATI GLOBAL RELATI TrDrX Xr
uY uYr
TDry
uZ u Zr
TDrz
TrDtY TrDtZ
y yr
TDrY
z zr
TDrZ
TrDry TrDrz
Y Yr
Z Zr
TrDrY TrDrZ
135
Z
IDtz
ux
uy uz
uX uY uZ DISPLA INCREM TRANSL GLOBAL RELATI IrDtX uXr DISPLA INCREM ROTATI LOCAL IDrx x DISPLA INCREM ROTATI LOCAL RELATI IrDrx xr DISPLA INCREM ROTATI GLOBAL IDrX X DISPLA INCREM ROTATI GLOBAL RELATI IrDrX Xr Phased displacements item type form oper DISPLA PHASE TRANSL LOCAL DISPLA PHASE DISPLA PHASE DISPLA PHASE DISPLA PHASE DISPLA PHASE DISPLA PHASE DISPLA PHASE TRANSL LOCAL TRANSL GLOBAL
IrDtY IrDtZ
uYr uZr
IDry IDrz
y z
IrDry IrDrz
yr zr
IDrY IDrZ
Y Z
IrDrY IrDrZ
Yr Zr
opti
comp X Y
PDtx PDty
Z
PDtz
ux
uy
uz
uX TRANSL GLOBAL RELATI PrDtX uX r ROTATI LOCAL PDrx x ROTATI LOCAL RELATI PrDrx xr ROTATI GLOBAL PDrX X ROTATI GLOBAL RELATI PrDrX Xr
uY uYr
PDry
uZ u Zr
PDrz
PrDtY PrDtZ
y yr
PDrY
z zr
PDrZ
PrDry PrDrz
Y Yr
Z Zr
PrDrY PrDrZ
7.4.2
Velocities
syntax
VELOCI [ type w ] [ form w ] [ oper w ] { comp w } { opti w } TOTAL TRANSL LOCAL RELATI PHASE ROTATI GLOBAL VELOCI species the velocity of the nodes as output item. type species the velocity type [ 3.6.2 p. 78].
[TOTAL]
136
Linear Transient Analysis TOTAL for the total velocities of a structure, i.e., the velocity at a certain time. PHASE for the phased velocities, i.e., the change in the velocity of the nodes between two consecutive phases in a phased analysis.
[TRANSL] [GLOBAL]
form
oper species an operation (transformation) to be performed on the velocities [ 3.6.2 p. 79]. comp selects velocity components for output. Default is all available components.
[all]
opti are additional options. The RELATI option is special for analyses with base excitation [ 6.3 p. 118], it causes the resulting velocities to be output relative to the base velocities. See 3.6.2 on page 79 for other options.
Total velocities item type form oper VELOCI TOTAL TRANSL LOCAL VELOCI TOTAL VELOCI TOTAL VELOCI TOTAL VELOCI TOTAL VELOCI TOTAL VELOCI TOTAL VELOCI TOTAL TRANSL LOCAL TRANSL GLOBAL comp X Y
TVtx TVty
opti
Z
TVtz
ux
uy
uz
uX TRANSL GLOBAL RELATI TrVtX uX r ROTATI LOCAL TVrx x ROTATI LOCAL ROTATI GLOBAL
uY uYr
TVry
uZ u Zr
TVrz
TrVtY TrVtZ
137
opti
Z
PVtz
ux
uy
uz
uX TRANSL GLOBAL RELATI PrVtX uX r ROTATI LOCAL PVrx x ROTATI LOCAL ROTATI GLOBAL
uY uYr
PVry
uZ u Zr
PVrz
PrVtY PrVtZ
7.4.3
Accelerations
syntax
ACCELE [ type w ] [ form w ] [ oper w ] { comp w } { opti w } TOTAL TRANSL LOCAL PHASE ROTATI GLOBAL ACCELE species the acceleration to be output in the nodes. type species the acceleration type [ 3.6.2 p. 78]. TOTAL for the total accelerations of a structure, i.e., the acceleration at a certain time. PHASE for the phased accelerations, i.e., the change in the acceleration of the nodes between two consecutive phases in a phased analysis. form species the acceleration formulation [ 3.6.2 p. 78].
[TRANSL] [GLOBAL] [TOTAL]
oper species an operation (transformation) to be performed on the accelerations [ 3.6.2 p. 79]. comp selects acceleration components for output. Default is all available components. opti are additional options [ 3.6.2 p. 79].
[all]
138
Total accelerations item type form oper ACCELE TOTAL TRANSL LOCAL ACCELE TOTAL ACCELE TOTAL ACCELE TOTAL comp X Y Z
TAtx TAty TAtz
uy uY
uz uZ
Phased accelerations item type form oper ACCELE PHASE TRANSL LOCAL ACCELE PHASE ACCELE PHASE ACCELE PHASE
uy uY
uz uZ
7.4.4
Strains
[ oper w ] { comp w } { loca w } { opti w } LOCAL INTPNT GLOBAL NODES PRINCI STRPNT VONMIS CROSSE REAXES CENTER VOLUME
syntax
STRAIN [ type w ] [ form w ] TOTAL GREEN FORCE MOMENT DISFOR DISMOM TRACTI
STRAIN species strains as output item. Table 7.1 on the facing page outlines the availability and applicability of the various strain output options for each of the element families.
[TOTAL] [GREEN]
type form
species the strain type [ 3.6.2 p. 78]. species the strain formulation. GREEN for GreenLagrange strains [ 7.4.4.1]. FORCE for deformations due to normal and shear forces [ 7.4.4.2]. DISFOR for generalized strains [ 7.4.4.2]. MOMENT for curvatures due to concentrated bending moments [ 7.4.4.3].
139
item type
form
oper
loca
STRAIN TOTAL GREEN FORCE MOMENT DISFOR DISMOM TRACTI LOCAL GLOBAL PRINCI VONMIS REAXES VOLUME INTPNT NODES STRPNT CROSSE CENTER
tr
a a a a a a a a a a
a a b b a a a a a a a a a a
a a a a a a a a a a a
a a a a a a a a a a
a a a a a a a a a a
a a a a a a a a a a a a
a a a a a a a a a a a a a
a a a a a a a a a a a
a a a a a a a a a a a
a a a a a a
a a a a -
a a c d c a c e c -
(a) All elements. (b) For all beam elements, classII and classIII only in combination with local coordinate system. (c) Only for bondslip reinforcements. (d) Only for bondslip reinforcements modeled by beam elements. (e) Not for bondslip reinforcements. (-) Not available or not suitable.
DISMOM for curvatures due to distributed bending moments [ 7.4.4.3]. TRACTI for tractions in structural interface elements [ 7.4.4.4]. oper species an operation (transformation) to be performed on the primary strains [ 3.6.2 p. 79]. comp selects strain components for output. Default is all available components. loca species the location for the strains to be output [ 3.6.2 p. 79]. opti are additional options [ 3.6.2 p. 79].
[LOCAL] [all]
Primary strains [ 31.1] item type form oper STRAIN TOTAL GREEN LOCAL STRAIN TOTAL STRAIN TOTAL STRAIN TOTAL GREEN GREEN GREEN
XY
YZ
Gyz
ZX
Gzx
Ezz Gxy
xx
yy zz xy
yz zx
GLOBAL EXX EYY EZZ GXY GYZ GZX XX Y Y ZZ XY Y Z ZX PRINCI E1 1 REAXES E1RA a 1
E2
E3
2
E2RA
a 2
Von Mises strain [ 31.1.1] item type form oper STRAIN TOTAL GREEN VONMIS Eeq eq Volumetric strain [ 31.1.3] item type form oper STRAIN TOTAL GREEN VOLUME Evol vol
7.4.4.2
Deformations
oper LOCAL comp X Y
Px Py
Force deformations item type form STRAIN TOTAL FORCE STRAIN TOTAL FORCE
Z
Pz
ux uy uz GLOBAL PX PY PZ uX uY uZ comp XX YY ZZ
Generalized strains item type form oper STRAIN TOTAL DISFOR LOCAL
XY
YZ
ZX
xx yy zz xy yz zx
7.4.4.3
Curvatures
comp X Y Z
Kx Ky Kz
Concentrated curvatures item type form oper STRAIN TOTAL MOMENT LOCAL STRAIN TOTAL
x y z MOMENT GLOBAL KX KY KZ X Y Z
141
7.4.4.4
Tractions
comp X Y
PTx PTy
Traction deformations item type form oper STRAIN TOTAL TRACTI LOCAL
Z
PTz
7.4.5
Stresses
[ oper w ] { comp w } { loca w } { opti w } LOCAL INTPNT GLOBAL NODES PRINCI STRPNT VONMIS CROSSE INVARI CENTER REINFO REAXES BIAXFE CAPACI
syntax
STRESS [ type w ] [ form w ] TOTAL CAUCHY FORCE MOMENT DISFOR DISMOM TRACTI
STRESS species stresses as output item. Table 7.2 on the next page outlines the availability and applicability of the various stress output options for each of the element families. type species the stress type [ 3.6.2 p. 78]. form species the stress formulation. CAUCHY for Cauchy stresses [ 7.4.5.1]. FORCE for concentrated forces [ 7.4.5.2]. MOMENT for concentrated bending moments [ 7.4.5.3]. DISFOR for distributed forces [ 7.4.5.2]. DISMOM for distributed moments [ 7.4.5.3]. TRACTI for tractions in structural interface elements [ 7.4.5.4]. oper species an operation (transformation) to be performed on the primary stresses [ 3.6.2 p. 79].
[LOCAL] [TOTAL] [CAUCHY]
142
item type
form
oper
loca
STRESS TOTAL CAUCHY FORCE MOMENT DISFOR DISMOM TRACTI LOCAL GLOBAL PRINCI VONMIS INVARI REINFO REAXES BIAXFE CAPACI INTPNT NODES STRPNT CROSSE CENTER
tr
a a a a a a a a a a a
a a a a a a a a a a a a a a
a a a a a a a a a a a a a a a
a a a a a a a a a a a a
a a a a a a a a a a a
a a a a a a a a a a a a a
a a a a a a a a a a a a a
a a a a a a a a a a a a a a
a a b b a a a a a a a a a a a a
a a a a a a
a a a a -
a a a d c a c e c -
(a) All elements. (b) For composed solids only [Vol. Element Library]. (c) Only for bondslip reinforcements. (d) Only for bondslip reinforcements modeled by beam elements. (e) Not for bondslip reinforcements. (-) Not available or not suitable.
BIAXFE gives the following Safety Factors for concrete under static and dynamic loading conditions with reference to a biaxial failure envelope:
F Sstatic usual = Rstatic usual /r F Sstatic unusual = Rstatic unusual /r F Sdynamic unusual = Rdynamic unusual /r F Sdynamic extreme = Rdynamic extreme /r
Where r is the distance from the origin to the actual stress point (1 , 2 ): r = (1 2 + 2 2 ) (7.1) To determine the Safety Factors Diana needs additional material properties. For the required parameters and additional information on the Safety Factors, see Volume Material Library. Note that all Safety Factors are limited to 100.
7.4 Output of Analysis results CAPACI gives the shear capacity and hydrostatic pressure capacity of stress against MohrCoulomb failure criterion [ 31.2.9 p. 519]. To determine the shear capacity and hydrostatic pressure capacity Diana needs additional material properties, see Volume Material Library. comp selects strain components for output. Default is all available components. loca species the location for the strains to be output [ 3.6.2 p. 79]. opti are additional options [ 3.6.2 p. 79]. 7.4.5.1 Cauchy Stresses
comp XX YY
Sxx Syy
[all]
143
Primary stresses [ 31.2] item type form oper STRESS TOTAL CAUCHY LOCAL STRESS TOTAL STRESS TOTAL STRESS TOTAL
ZZ
Szz
XY
Sxy
YZ
Syz
ZX
Szx
xx yy zz xy yz zx CAUCHY GLOBAL SXX SYY SZZ SXY SYZ SZX XX Y Y ZZ XY Y Z ZX CAUCHY PRINCI S1 1 CAUCHY REAXES S1RA a 1
S2
S3
2
S2RA
a 2
Von Mises stress [ 31.2.1] item type form oper STRESS TOTAL CAUCHY VONMIS Seq eq Stress invariants [ 31.2.6] comp item type form oper P Q LODE STRESS TOTAL CAUCHY INVARI P Q Lode p q Biaxial failure envelope comp item type form oper STUS STUN DYUN DYEX STRESS TOTAL CAUCHY BIAXFE FSstus FSstun FSdyun FSdyex F Sstus F Sstun F Sdyun F Sdyex Shear capacity [ 31.2.9] comp item type form oper SHRCAP PPCAP STRESS TOTAL CAUCHY CAPACI SHRCAP PPCAP
Concentrated forces item type form oper STRESS TOTAL FORCE LOCAL STRESS TOTAL FORCE
Qx Qy Nz
Nx Qy Qz Qx Qy Nz GLOBAL NX NY NZ NX NY NZ comp XX YY ZZ XY
Distributed forces item type form oper STRESS TOTAL DISFOR LOCAL
YZ ZX
nxx nyy nzz nxy qyz qxz Reinforcement forces [ 31.2.7] comp item type form oper 1R 2R 1RC 2RC STRESS TOTAL DISFOR REINFO N1R N2R N1RC N2RC n1 n2 n1c n2c
7.4.5.3
Bending Moments
comp X Y Z
Mx My Mz
Concentrated moments item type form oper STRESS TOTAL MOMENT LOCAL STRESS TOTAL
Distributed moments item type form oper STRESS TOTAL DISMOM LOCAL
YZ ZX
mxx myy
mxy
Reinforcement moments [ 31.2.7] comp item type form oper 1R 2R STRESS TOTAL DISMOM REINFO M1R M2R m1 m2
7.4.5.4
Tractions
form oper TRACTI LOCAL comp X Y Z
STx STy STz
145
7.4.6
syntax
FORCE [ type w ] [ form w ] [ oper w ] { comp w } { opti w } TOTAL TRANSL LOCAL RESIDU ROTATI GLOBAL REACTI FORCE species forces (and moments) in the nodes as output item [ 31.3 p. 519]. type species the type [ 3.6.2 p. 78]. TOTAL for the total forces and moments. RESIDU for the residual forces (by default Diana only gives the extreme residual forces). REACTI for the reaction forces in all supported nodes. form oper comp species the formulation [ 3.6.2 p. 78] species an operation (transformation) to be performed on the forces and moments [ 3.6.2 p. 79]. selects force or moment components for output. Default is all available components.
[TRANSL] [GLOBAL] [TOTAL]
[all]
Z
FTz
Fz
FTZ
FZ
MTz
Mz
MTZ
MZ
comp X Y Z
FRx FRx FRx
R R R Fx Fy Fz
FORCE RESIDU TRANSL GLOBAL FRX R FX FORCE RESIDU ROTATI LOCAL MRx R Mx FORCE RESIDU ROTATI GLOBAL MRX R MX
FRX FRX
R R FY FZ
MRy MRz
R R My Mz
MRY MRZ
R R MY MZ
146
Support reactions [ 31.3.3] comp item type form oper X Y FORCE REACTI TRANSL LOCAL FBx FBy B B Fx F y FORCE REACTI TRANSL GLOBAL FBX FBY B B FX F Y FORCE REACTI ROTATI LOCAL MBx MBy B B Mx My FORCE REACTI ROTATI GLOBAL MBX MBY B B MX MY
Z
FBz
B Fz
FBZ
B FZ
MBz
B Mz
MBZ
B MZ
7.4.7
syntax
NODFOR [ type w ] [ form w ] [ oper w ] { comp w } { opti w } ELEMEN TRANSL LOCAL REINFO ROTATI GLOBAL TOTAL NODFOR species internal nodal element forces and moments in the nodes as output item. This command gives the contributions of the element or reinforcement internal nodal forces (or both), to a certain node. A selection of elements which form a section of the model gives the total internal forces that act on that section.
[TOTAL]
type
species the forces type. ELEMEN for contribution of elements only. REINFO for contribution of embedded reinforcements only. TOTAL for contribution of both elements and embedded reinforcements.
[TRANSL] [GLOBAL]
species the formulation [ 3.6.2 p. 78]. species an operation (transformation) to be performed on the forces and moments [ 3.6.2 p. 79]. selects force or moment components for output. Default is all available components.
[all]
147
Z
FNz
n Fz
FNZ
n FZ
MNz
n Mz
MNZ
n MZ
7.4.8
syntax
ELMFOR [ type w ] [ form w ] [ oper w ] { comp w } { opti w } ELEMEN TRANSL GLOBAL REINFO ROTATI TOTAL ELMFOR species internal element forces and moments in the nodes of an element as output item. This command gives the contributions of the element or reinforcement internal forces (or both), to a certain node of an element. type species the forces type. ELEMEN for contribution of elements only. REINFO for contribution of embedded reinforcements only. TOTAL for contribution of both elements and embedded reinforcements. form oper comp species the formulation [ 3.6.2 p. 78]. species an operation (transformation) to be performed on the forces and moments [ 3.6.2 p. 79]. selects force or moment components for output. Default is all available components.
[TRANSL] [GLOBAL] [TOTAL]
[all]
Z
FEZ
e FZ
MEZ
e MZ
148
7.4.9
Dynamic Pressures
For uidstructure interface elements [Vol. Element Library] Diana can calculate and output the dynamic pressures. See [ 32.5 p. 532] for background theory. syntax FSPRES [ type w ] [ loca w ] { opti w } TOTAL NODES FSPRES species the dynamic pressures of the uidstructure interface elements as output item.
[TOTAL]
type
species the dynamic pressure type. TOTAL for the total dynamic pressures of the uidstructure interfaces.
[NODES]
loca
species the location for the dynamic pressures to be output [ 3.6.2 p. 79].
Chapter 8
The steady-state harmonic analysis may be applied when the loading is a deterministic frequency content and the system is subjected to continuous harmonic excitations. The response may be determined via a mode-superposition technique or via a direct steady-state method (direct dynamic stiness method). The modesuperposition method requires a preliminary eigenfrequency analysis. By means of a selected number of eigenpairs Diana performs a transformation to modal coordinates. See 8.1 on page 151 for analysis commands. Damping. In case of damping, only modal damping can be used in a mode superposition technique. For local damping or for heavily damped systems, a mode superposition technique is no longer appropriate. In these cases a direct method must be used. Loading. The system can be subjected to two types of loading excitation. The rst is a mechanical load in terms of externally applied concentrated forces or distributed forces like pressures. The second type of excitation is a base motion which must be prescribed via a support motion.
Diana-9.4 Users Manual Analysis Procedures (III) December 15, 2009 First ed.
150
Frequency Response Analysis A mechanical load can be applied via nodal or element loads. For a base or support motion, Diana assumes that all supports have the same prescribed motion. You must specify the motion in terms of accelerations, translational accelerations only. Diana converts these accelerations to equivalent inertia loads. For a base motion, the response is relative to the base. For a direct response analysis the system can also be subjected to a prescribed deformation load excitation. All other types of loading will be ignored. Multiple excitation loads may be used in a modal or direct frequency response analysis. For each excitation frequency the responses of the excitation loads are superposed. Dierent loading types may be combined in the same frequency response analysis. In case a deformation excitation load is applied in combination with a base acceleration loading the deformation is relative to the base. Multiplication coecient. For each load set the frequency dependent multiplication coecient is piecewise linearly interpolated from the corresponding diagram specied in table FREQLO [ 6.5.1 p. 123]. If table FREQLO is not specied a default multiplication coecient equal to 1.0 will be applied for each excitation frequency on the rst existing load set. Example
FREQLO LOAD 1 FREQUE 3.0 5.0 9.0 12.0 / FACTOR 1.0 6.0 4.0 9.0 / LOAD 3 FREQUE 0.0 20.0 / FACTOR 2.0 2.0 / file .dat
In this example the diagram is specied by four points (fi , i ) = (3, 1), (5, 6), (9, 4), (12, 9) [Fig. 8.1]. The multiplication factor for load set 3 is equal to 2.0 for the frequency range 0.0 to 20.0.
factor n 2 ... 1 0 specied interpolated
f1
f2
f...
fn
frequency
8.1 Modal Response Analysis Fluidstructure interaction analysis. In case of a model containing uid, structural elements and uidstructure interface elements, automatically a uidstructure frequency response analysis will be performed. Diana recognizes this element combination and will perform the required preparation to perform a frequency response uidstructure analysis. In a direct response analysis compression of the uid as well as free surface waves are admitted and a radiation boundary may be incorporated. To include these eects specic material data is required, see Fluidstructure interaction analysis in Volume Material Library. In a modal response analysis the added mass eect of the uid [Eq. (32.92) p. 537] is taken into account. Results will be available for the structural part of the model. Lumped element matrices may not be used in a uidstructure response analysis.
151
8.1
The primary tasks for Module modal are invoked via the following command sequence. The OFF options suppress the execution of the specied task, which may save computing time if it was performed previously. syntax *MODAL [ MODEL ] [ EIGEN ] [ RESPON ] *END MODEL evaluates and assembles the nite element model [ 8.1.1]. EIGEN to solve the eigenvalue problem for a modal response analysis [ 8.1.2 p. 152]. RESPON to perform the actual modal response analysis [ 8.1.3 p. 153].
8.1.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual modal response analysis. syntax BEGIN MODEL [ OFF ] [ EVALUA ]
Diana-9.4 Users Manual Analysis Procedures (III) December 15, 2009 First ed.
EVALUA to check and evaluate geometric and material properties for elements [ 3.4 p. 66]. ASSEMB to assemble the elements, i.e., creating appropriate system degrees of freedom [ 3.5 p. 68].
8.1.2
To solve the free vibration eigenvalue problem in a modal response analysis, i.e., to determine the eigenmodes and natural frequencies, you must give commands in the EIGEN block which are analogous to those for Module eigen [ 27.2 p. 439]. The OFF options suppress the execution of the specied task. This may be useful to save computing time if previously determined results still reside on the filos le. syntax BEGIN EIGEN [ OFF ] [ FREEVI [ OFF ] ] [ EXECUT [ OFF ] ] [ OUTPUT ] END EIGEN FREEVI sets up the free vibration eigenmode analysis [ 27.2.1 p. 440]. EXECUT species how to solve the eigenvalue problem. In particular this involves the type of the solution procedure [ 27.3 p. 445]. OUTPUT to specify the eigenmodes to be output [ 27.4.7 p. 448]. Default output
*MODAL BEGIN EIGEN OUTPUT END EIGEN [ commands ] *END file .dcf
If you only give a single OUTPUT command, like in the above example, or if you omit the OUTPUT command, then Diana gives a default output as if you had given the following commands.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (III)
153
Example
*MODAL BEGIN EIGEN FREEVI MASS BEGIN EXECUT NMODES=20 END EXECUT OUTPUT DISPLA END EIGEN *END
file .dcf
This command sequence is a typical example for a preliminary eigenvalue analysis. In the free vibration problem, indicated by FREEVIB, the system mass matrix has to be used which is invoked by the MASS command option. By default Diana assembles the consistent element mass matrices. The EXECUT command block calculates the natural frequencies and corresponding mode shapes needed for modal analysis. Parameter NMODES=20 asks for an arbitrarily chosen number of twenty frequencies. Finally, printed output of results like natural frequencies and mode shapes is obtained through the OUTPUT command block.
8.1.3
Response Analysis
To perform the actual modal response analysis, i.e., to determine the response of a construction at a certain load frequency, you must give commands in the RESPON block. The OFF options suppress the specied tasks. This may be useful to save computing time if previously determined results still reside on the filos le. syntax BEGIN RESPON [ OFF ] [ REDUCE [ OFF ] ] [ EXECUT [ OFF ] ] [ OUTPUT ] END RESPON
154
Frequency Response Analysis REDUCE reduces the system of equations [ 8.1.3.1]. EXECUT executes the actual modal response analysis [ 8.1.3.2]. OUTPUT species the modal response analysis results to be output [ 8.3]. Default output
*MODAL [ commands ] BEGIN RESPON OUTPUT END RESPON *END file .dcf
If you only give a single OUTPUT command, like in the above example, or if you omit the OUTPUT command, then Diana gives a default output as if you had given the following commands.
file .dcf *MODAL [ commands ] BEGIN RESPON BEGIN OUTPUT DISPLA TOTAL TRANSL GLOBAL COMPLE VELOCI TOTAL TRANSL GLOBAL COMPLE ACCELE TOTAL TRANSL GLOBAL COMPLE END OUTPUT END RESPON *END
8.1.3.1
To reduce the full set of equations, you must give commands in the REDUCE block within the RESPON block. syntax BEGIN REDUCE [ OFF ] [ BEGIN SELECT MODES modes n... ALL END SELECT ] END REDUCE
8.1 Modal Response Analysis OFF suppresses the reduction of the system of equations. This may save computing time when the reduced system of equations is still available from a previous modal response analysis. SELECT MODES selects eigenmodes for the modal response analysis. The modes must have been determined previously [ 8.1.2 p. 152]. You may explicitly specify a set of modes , or you may require a modal response analysis for ALL determined modes. All modes is also the default if you dont select modes. 8.1.3.2 Analysis Execution
[ALL]
155
The EXECUT command block actually performs the modal response analysis by applying the excitation frequencies. syntax BEGIN EXECUT [ OFF ] [ ACCELE ] ]] [ MODES [ OFF NUMBER=i n... [ EXPLIC [ ]] OFF FREQUE=om r... [ DAMPIN c1 r [ . . . cn r ] ] END EXECUT OFF suppresses the execution of the modal response analysis. ACCELE employs a static correction to the displacement response according to a mode acceleration method. MODES indicates an excitation with natural frequencies. Parameter NUMBER species a series of natural frequency numbers i indicating the natural frequencies fi . If you dont specify any frequency numbers then Diana assumes excitation with the rst natural frequency f1 . No frequency number may exceed the highest natural frequency number calculated [ 8.1.2]. EXPLIC indicates an excitation with explicitly specied frequencies i . Parameter FREQUE species a series of frequencies om . If you dont specify any frequencies then Diana assumes excitation with a discrete frequency of 1 by default. DAMPIN species modal damping. Values c1 to cn are a series of damping coecients in parts of the critical damping factor ccrit [Eq. (32.5) p. 522]; a
Diana-9.4 Users Manual Analysis Procedures (III) December 15, 2009 First ed.
[EXPLIC]
[om =1.0]
156
Frequency Response Analysis value of 0.01 indicates 1 %. If only one factor c1 is specied, then the damping is the same for all natural frequencies of the system. Alternatively, the damping coecients must be specied for all natural frequencies of the system, which may be reduced [ 8.1.3.1]. Modal damping must be applied if excitations coincide with the natural frequencies. If you specify neither natural nor discrete frequencies, then Diana assumes an excitation with a discrete frequency of 1. Default
*MODAL [ commands ] BEGIN RESPON EXECUT END RESPON [ commands ] file .dcf
[c1 =0.01]
[1 = 1.0]
If you only give a single EXECUT command, then Diana will execute a frequency response analysis according to the following command sequence.
file .dcf *MODAL BEGIN RESPON BEGIN EXECUT ACCELE EXPLIC FREQUE=1.0 END EXECUT END RESPON [ commands ]
8.1.4
8.1.4.1
Example Commands
Response to Forced Periodic Excitation
The following commands are an example for steady-state response to forced periodic excitations. In this case, Module modal gives amplitudes of displacement and velocity. The nite element model is subjected to an harmonically varying forcing function.
file .dcf *MODAL BEGIN RESPON BEGIN EXECUT
157
A modal analysis is performed by the EXECUT command block and a static correction is employed to the displacement response by the ACCELE command. The EXPLIC command requests the analysis of response for three explicit excitation frequencies of 1, 7, and 12 Hz, if the unit of time is second [s]. See 8.3 on page 161 for output commands. 8.1.4.2 Frequency Response
This example shows how to setup a frequency response analysis. In this analysis a combination of response amplitude, like displacement and phase angle versus frequency ratio, are determined for excitations covering a specied range with some resonance1 frequencies when the system is subjected to a certain input. In order to avoid an innite response at resonance, eects of damping must be included.
file .dcf *MODAL BEGIN RESPON REDUCE OFF BEGIN EXECUTE MODES NUMBER=1-5(2) DAMPIN 0.02 END EXECUTE END RESPON [ commands ]
REDUCE OFF species that no reduction will be used. Modal damping is specied in the DAMPIN command, where 0.02 is a coecient in parts of the critical damping and which is, in this example, the same for all natural frequencies. The excitation frequencies at which the response has to be calculated are specied in the EXECUT command block. The MODES command indicates a sequence of excitations in the rst, third, and fth natural frequency (f1 , f3 , f5 ).
8.2
The primary tasks for Module freque are invoked via the following command sequence.
1 resonance
= the condition when the excitation frequency equals the natural frequency.
158
Frequency Response Analysis syntax *FREQUE [ MODEL ] [ EXECUT ] [ OUTPUT ] *END MODEL evaluates and assembles the nite element model [ 8.2.1]. EXECUT executes the actual direct response analysis [ 8.2.2]. OUTPUT species the direct response analysis results to be output [ 8.3]. Default output
*FREQUE [ commands ] OUTPUT *END file .dcf
If you only give a single OUTPUT command, like in the above example, or if you omit the OUTPUT command, then Diana gives a default output as if you had given the following commands.
file .dcf *FREQUE [ commands ] BEGIN OUTPUT DISPLA TOTAL TRANSL GLOBAL COMPLE VELOCI TOTAL TRANSL GLOBAL COMPLE ACCELE TOTAL TRANSL GLOBAL COMPLE END OUTPUT *END
Example
*FREQUE BEGIN EXECUT EXPLIC FREQUE=1. 7. 12. END EXECUT [ output commands ]
file .dcf
This command sequence is an example for direct response to forced periodic excitations. In this case, Module freque gives amplitudes of displacement and
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (III)
8.2 Direct Response Analysis velocity. The nite element model is subjected to an harmonically varying forcing function. A direct response analysis is performed by the EXECUT command block. The EXPLIC command requests the analysis of response for three explicit excitation frequencies of 1, 7, and 12 Hz, if the unit of time is second [s]. See 8.3 on page 161 for output commands.
159
8.2.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual direct response analysis. syntax BEGIN MODEL [ OFF ] [ EVALUA ] [ ASSEMB ] [ MATRIX ] END MODEL EVALUA to check and evaluate geometric and material properties for elements [ 3.4 p. 66]. ASSEMB to assemble the elements, i.e., creating appropriate system degrees of freedom [ 3.5 p. 68]. MATRIX to setup the element matrices [ 8.2.1.1]. 8.2.1.1 Element Matrices
The MATRIX commands specify the type of element matrices to be applied in the direct response analysis. syntax BEGIN MATRIX [ OFF ] [ MASS [ OFF ] [
] [ ROTATI [ OFF ] ] ] CONSIS LUMPED ]] [ DAMPIN [ OFF ] [ CONSIS LUMPED ] ] [ STRESS [ OFF ] [ INPUT { } LOAD=loset n
Diana-9.4 Users Manual Analysis Procedures (III) December 15, 2009 First ed.
MASS species the kind of mass matrices to be applied: CONSIS for consistent matrices (the default). LUMPED for lumped matrices. ROTATI indicates that rotational terms must be included in the mass matrices. The OFF option suppresses the inclusion of rotational terms. If you dont specify the ROTATI option explicitly, then Diana will include rotational terms by default. OFF suppresses the setup of mass matrices.
[CONSIS]
DAMPIN species the kind of damping matrices to be applied: CONSIS for consistent matrices (the default). LUMPED for lumped matrices. OFF suppresses the setup of damping matrices.
[CALCUL]
STRESS adds the geometric stress-stiness matrix KG to the linear elastic stiness matrix KL0 . INPUT indicates a geometric stress-stiness matrix, with stresses specied via a prestress load for load set loset , in subtable ELEMEN and/or REINFO of table LOADS [Vol. Element Library]. CALCUL indicates a geometric stress-stiness matrix, with stresses which Diana will calculate automatically in a linear static analysis for load set loset (the default). PHASE indicates a geometric stress-stiness matrix, with stresses present from a previous phase. To setup the geometric stress-stiness matrix KG from a stress eld you may specify a load set number loset via the LOAD parameter. This load set number corresponds to a load set in input table LOADS [ 2.3.7 p. 55]. Default is the lowest available load set number. The optional parameter FACTOR species a multiplication factor fac .
[fac =1.0]
8.2.2
Analysis Execution
The EXECUT command block actually performs the direct response analysis by applying the excitation frequencies.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (III)
8.3 Output of Analysis Results syntax BEGIN EXECUT [ OFF ] [ EXPLIC [ FREQUE=om r... ] ] END EXECUT OFF suppresses the execution of the direct response analysis. EXPLIC indicates an excitation with explicitly specied frequencies i . Parameter FREQUE species a series of frequencies om . If you dont specify any frequencies then Diana assumes excitation with a discrete frequency of 1 by default. Default
*FREQUE [ commands ] EXECUT [ commands ] file .dcf
161
[om =1.0]
If you only give a single EXECUT command, then Diana will execute a direct response analysis according to the following command sequence.
file .dcf *FREQUE BEGIN EXECUT EXPLIC FREQUE=1.0 END EXECUT [ commands ]
8.3
With command in the OUTPUT block you can get output of results from a frequency response analysis. For the general, analysis type independent options for output of analysis results see 3.6 on page 69. syntax BEGIN OUTPUT [device w ] [outopt w... ] [ params ] [ OFF ] [ BEGIN SELECT [ ] [ FREQUE [ ]]
Diana-9.4 Users Manual Analysis Procedures (III) December 15, 2009 First ed.
162
Frequency Response Analysis frqnrs n... ALL END SELECT ] [ LAYOUT ] item w { type w } { form w } { oper w } { comp w } { loca w } { opti w } DISPLA AMPLIT VELOCI COMPLE ACCELE FORCE STRAIN STRESS FSPRES END OUTPUT SELECT command block to customize the batch output: for model selection see 3.6.1 on page 71.
[ALL]
FREQUE selects specic frequencies for output: frqnrs is a series of frequency numbers, ALL selects all frequencies. LAYOUT commands to customize the layout and style of tabular output [ 3.6.3 p. 80]. item is the name of the analysis result to be output. See 3.6.2 on page 77 for complete syntax of this command. DISPLA for displacements [ 8.3.1]. VELOCI for velocities [ 8.3.2]. ACCELE for accelerations [ 8.3.3]. FORCE for nodal forces and moments [ 8.3.6]. STRAIN for strains [ 8.3.4]. STRESS for stresses [ 8.3.5]. FSPRES for dynamic pressures of uidstructure interface elements [ 8.3.7].
[COMPLE]
opti are options applied to the output item [ 3.6.2 p. 79]. Special options for a frequency response analysis indicate how the result must be expressed. AMPLIT to get the result of item expressed in amplitude and phase angle. This option is only for tabular output. COMPLE to get the complex representation of the result item , i.e., the real and imaginary part. This option is for tabular output and for output to the iDiana Results environment.
163
8.3.1
Displacements
syntax
DISPLA [ type w ] [ form w ] [ oper w ] { comp w } { opti w } TOTAL TRANSL LOCAL AMPLIT ROTATI GLOBAL COMPLE DISPLA species displacements of the nodes as output item. type species the displacement type. TOTAL for the total displacements of a structure, i.e., the deformed geometry at a certain stage. form species the displacement formulation [ 3.6.2 p. 78].
[TRANSL] [GLOBAL] [TOTAL]
oper species an operation (transformation) to be performed on the displacements [ 3.6.2 p. 79]. comp selects displacement components for output. Default is all available components. opti are additional options [ 8.3 p. 162].
Translation displacements comp item type form oper opti X Y DISPLA TOTAL TRANSL LOCAL AMPLIT DCtxA DCtyA ux uy DISPLA TOTAL TRANSL LOCAL COMPLE DCtxR DCtyR ux uy DISPLA TOTAL TRANSL GLOBAL AMPLIT DCtXA DCtYA uX uY DISPLA TOTAL TRANSL GLOBAL COMPLE DCtXA DCtYA uX uY Rotation displacements item type form oper DISPLA TOTAL ROTATI LOCAL comp opti X Y AMPLIT DCrxA DCryA x y DISPLA TOTAL ROTATI LOCAL COMPLE DCrxR DCryR x y DISPLA TOTAL ROTATI GLOBAL AMPLIT DCrXA DCrYA X Y
[all]
Z uz uz uZ uZ
X
ux
Y
uy
Z
uz
ux
uX
uy
uY
uz
uZ
uX
uY
uZ
Z z z
X
x
Y
y
Z
z
Z X Y Z DISPLA TOTAL ROTATI GLOBAL COMPLE DCrXR DCrYR DCrZR DCrXP DCrYP DCrZP X Y Z X Y Z
164
8.3.2
Velocities
syntax
VELOCI [ type w ] [ form w ] [ oper w ] { comp w } { opti w } TOTAL TRANSL LOCAL AMPLIT ROTATI GLOBAL COMPLE VELOCI species velocities of the nodes as output item.
[TOTAL]
type
species the velocity type. TOTAL for the total velocities of a structure.
[TRANSL] [GLOBAL]
form
oper species an operation (transformation) to be performed on the velocities [ 3.6.2 p. 79]. comp selects velocity components for output. Default is all available components.
[all]
Z uz uz uZ uZ
X ux ux
Y uy uy
Z uz uz uZ uZ
VtyA VtzA VtxP VtyP VtzP VtyR VtzR VtxI VtyI VtzI VtYA VtZA VtXP VtYP VtZP
uX uY uX uY
Z z z Z
X
x
Y
y
Z
z
z
Z
VELOCI TOTAL ROTATI GLOBAL COMPLE VrXR VrYR VrZR VrXI VrYI VrZI X Y Z X Y Z
165
8.3.3
Accelerations
syntax
ACCELE [ type w ] [ form w ] [ oper w ] { comp w } { opti w } TOTAL TRANSL LOCAL AMPLIT ROTATI GLOBAL COMPLE ACCELE species accelerations of the nodes as output item. type species the acceleration type. TOTAL for the total accelerations of a structure, i.e., the acceleration at a certain stage. form species the acceleration formulation [ 3.6.2 p. 78].
[TRANSL] [GLOBAL] [TOTAL]
oper species an operation (transformation) to be performed on the accelerations [ 3.6.2 p. 79]. comp selects acceleration components for output. Default is all available components. opti are additional options [ 8.3].
Translation accelerations item type form oper ACCELE TOTAL TRANSL LOCAL comp opti X AMPLIT AtxA ux ACCELE TOTAL TRANSL LOCAL COMPLE AtxR ux ACCELE TOTAL TRANSL GLOBAL AMPLIT AtXA uX Y uy uy
[all]
Z uz uz
X ux ux
Y uy uy
Z uz uz
AtyA AtzA AtxP AtyP AtzP AtyR AtzR AtxI AtyI AtzI AtYA AtZA AtXP AtYP AtZP
uY uZ u X u Y uZ ACCELE TOTAL TRANSL GLOBAL COMPLE AtXR AtYR AtZR AtXI AtYI AtZI uX uY uZ uX uY uZ Rotation accelerations item type form oper ACCELE TOTAL ROTATI LOCAL comp opti X AMPLIT ArxA x ACCELE TOTAL ROTATI LOCAL COMPLE ArxR x ACCELE TOTAL ROTATI GLOBAL AMPLIT ArXA X ACCELE TOTAL ROTATI GLOBAL COMPLE ArXR X Y y y Y Y
Z z z Z Z
X
x
Y
y
Z
z
z
Z
166
8.3.4
Strains
[ oper w ] { comp w } { loca w } LOCAL INTPNT GLOBAL NODES VONMIS STRPNT CROSSE CENTER { opti w } AMPLIT COMPLE
syntax
STRAIN [ type w ] [ form w ] TOTAL GREEN FORCE MOMENT DISFOR DISMOM TRACTI
STRAIN species strains as output item. Table 8.1 outlines the availability and applicability of the various strain output options for each of the element families. Table 8.1: Availability of strain output
am pl .s t pl res .s s t ax rai is n y pl mm at e . . b e sh nd . cu ell .s h so ell lid st r. i sp nte ri rfa n p. g ce m a e. ss re in fo .
item type
form
oper
loca
opti
STRAIN TOTAL GREEN FORCE MOMENT DISFOR DISMOM TRACTI LOCAL GLOBAL VONMIS INTPNT NODES STRPNT CROSSE CENTER AMPLIT COMPLE
tr
a a a a a a a a a
be
us s
a a b b a a a a a a a a a
a a a a a a a a a a
a a a a a a a a a a
a a a a a a a a a a
a a a a a a a a a a a a
a a a a a a a a a a a a
a a a a a a a a a a
a a a a a a a a a a
a a a a a a a
a a a a a
a a c d c a c e c a a
(a) All elements. (b) For all beam elements, classII and classIII only in combination with local coordinate system. (c) Only for bondslip reinforcements. (d) Only for bondslip reinforcements modeled by beam elements. (e) Not for bondslip reinforcements. (-) Not available or not suitable.
[TOTAL] [GREEN]
type form
species the strain type [ 3.6.2 p. 78]. species the strain formulation. GREEN for GreenLagrange strains [ 8.3.4.1]. FORCE for deformations due to normal and shear forces [ 8.3.4.2].
8.3 Output of Analysis Results DISFOR MOMENT DISMOM TRACTI oper for for for for generalized strains [ 8.3.4.2]. curvatures due to concentrated bending moments [ 8.3.4.3]. curvatures due to distributed bending moments [ 8.3.4.3]. tractions in structural interface elements [ 8.3.4.4].
[GLOBAL]
167
Primary strains [ 31.1] item type form oper STRAIN TOTAL GREEN LOCAL
ZZ
EzzA
XY
GxyA
YZ
GyzA
ZX
GzxA
zz
EzzP
zz
xy
GxyP
xy
yz
GyzP
yz
zx
GzxP
zx
EyyP
yy
COMPLE ExxR xx
ExxI
EyyR
EzzR
GxyR
GyzR
GzxR
yy
EyyI
zz
EzzI
xy
GxyI
yz
GyzI
zx
GzxI
yy
EYYA
zz
EZZA
xy
GXYA
yz
GYZA
zx
GZXA
Y Y
EYYP
ZZ
EZZP
XY
GXYP
Y Z
GYZP
ZX
GZXP
XX Y Y
ZZ XY
Y Z ZX
STRAIN TOTAL GREEN GLOBAL COMPLE EXXR EYYR EZZR GXYR GYZR GZXR XX Y Y ZZ XY Y Z ZX
EXXI EYYI EZZI GXYI GYZI GZXI
XX
Y Y
ZZ
XY
Y Z
ZX
Von Mises strains [ 31.1.1] item type form oper opti STRAIN TOTAL GREEN VONMIS AMPLIT EeqA EeqP eq eq STRAIN TOTAL GREEN VONMIS COMPLE EeqR EeqI eq eq
Force deformations item type form oper STRAIN TOTAL FORCE LOCAL STRAIN TOTAL FORCE STRAIN TOTAL FORCE STRAIN TOTAL FORCE
comp opti X Y AMPLIT PxA PyA x y u u LOCAL COMPLE PxR PyR ux uy GLOBAL AMPLIT PXA PYA X Y u u GLOBAL COMPLE PXR PYR uX uY comp opti XX AMPLIT PxxA xx
PxxP
Z
PzA
X
PxP PxI PXP PXI
Y
PyP PyI PYP PYI
Z
PzP PzI PZP PZI
z u
PzR PZA
ux uy uz
uz ux uy uz Z u
PZR
uX uY uZ
uZ uX uY uZ
Generalized strains item type form oper STRAIN TOTAL DISFOR LOCAL
YY
PyyA
ZZ
PzzA
XY
PxyA
YZ
PyzA
ZX
PzxA
yy
PyyP
zz
PzzP
xy
PxyP
yz
PyzP
zx
PzxP
xx yy zz xy yz zx
xx
yy
zz
xy
yz
zx
8.3.4.3
Curvatures
Y
KyA
Concentrated curvatures item type form oper STRAIN TOTAL MOMENT LOCAL
comp opti X AMPLIT KxA x STRAIN TOTAL MOMENT LOCAL COMPLE KxR x STRAIN TOTAL MOMENT GLOBAL AMPLIT KXA X STRAIN TOTAL MOMENT GLOBAL COMPLE KXR X Distributed curvatures item type form oper STRAIN TOTAL DISMOM LOCAL comp opti XX AMPLIT KxxA xx
Z
KzA
X
KxP
x
Y
KyP
y
Z
KzP
z
y
KyR
z
KzR
KxI
KyI
KzI
y
KYA
z
KZA
x
KXP KXI
y
KYP KYI
z
KZP
Z
Y
KYR
Z
KZR
X Y
KZI
Y YY yy
ZZ zz
XY xy
YZ yz
ZX zx
xx
yy
zz
xy
yz
zx
169
Traction deformations item type form oper STRAIN TOTAL TRACTI LOCAL
comp opti X Y AMPLIT PTxA PTyA x y u u STRAIN TOTAL TRACTI LOCAL COMPLE PTxR PTyR ux uy STRAIN TOTAL TRACTI GLOBAL AMPLIT PTXA PTYA X Y u u STRAIN TOTAL TRACTI GLOBAL COMPLE PTXR PTYR uX uY
X
PTxP
ux
Y
PTyP
uy
Z
PTzP
uz
z u
PTzR
PTxI
PTyI
PTzI
uz
PTZA
ux
PTXP PTXI
uy
PTYP PTYI
uz
PTZP
uZ
Z u
PTZR
uX uY
PTZI
uZ
uX
uY
uZ
8.3.5
Stresses
syntax [ oper w ] { comp w } { loca w } { opti w } LOCAL INTPNT AMPLIT GLOBAL NODES COMPLE VONMIS STRPNT CROSSE CENTER
STRESS [ type w ] [ form w ] TOTAL CAUCHY FORCE MOMENT DISFOR DISMOM TRACTI
STRESS species stresses as output item. Table 8.2 on the next page outlines the availability and applicability of the various stress output options for each of the element families. type form species the stress type [ 3.6.2 p. 78] species the stress formulation. CAUCHY for Cauchy stresses [ 8.3.5.1]. FORCE for concentrated forces [ 8.3.5.2]. DISFOR for distributed forces [ 8.3.5.2]. MOMENT for concentrated bending moments [ 8.3.5.3]. DISMOM for distributed bending moments [ 8.3.5.3]. TRACTI for tractions in structural interface elements [ 8.3.5.4]. oper species an operation (transformation) to be performed on the primary stresses [ 3.6.2 p. 79].
[GLOBAL] [TOTAL] [CAUCHY]
170
item type
form
oper
loca
opti
STRESS TOTAL CAUCHY FORCE MOMENT DISFOR DISMOM TRACTI LOCAL GLOBAL VONMIS INTPNT NODES STRPNT CROSSE CENTER AMPLIT COMPLE
tr
a a a a a a a a a a a
a a a a a a a a a a a a a a
a a a a a a a a a a a
a a a a a a a a a a
a a a a a a a a a a
a a a a a a a a a a a a
a a a a a a a a a a a a
a a a a a a a a a a a a
a a b b a a a a a a a a
a a a a a a a a
a a a a a a
a a a d c a c e c a a
(a) All elements. (b) For composed solids only [Vol. Element Library]. (c) Only for bondslip reinforcements. (d) Only for bondslip reinforcements modeled by beam elements. (e) Not for bondslip reinforcements. (-) Not available or not suitable.
8.3.5.1
Cauchy Stresses
comp opti XX YY AMPLIT SxxA SyyA xx yy
SxxP
xx
Primary stresses item type form oper STRESS TOTAL CAUCHY LOCAL
ZZ
SzzA
XY
SxyA
YZ
SyzA
ZX
SzxA
zz
SzzP
zz
xy
SxyP
xy
yz
SyzP
yz
zx
SzxP
zx
SyyP
yy
COMPLE SxxR xx
SxxI
SyyR
SzzR
SxyR
SyzR
SzxR
yy
SyyI
zz
SzzI
xy
SxyI
yz
SyzI
zx
SzxI
yy
SYYA
zz
SZZA
xy
SXYA
yz
SYZA
zx
SZXA
Y Y
SYYP
ZZ
SZZP
XY
SXYP
Y Z
SYZP
ZX
SZXP
XX Y Y
ZZ XY
Y Z ZX
STRESS TOTAL CAUCHY GLOBAL COMPLE SXXR SYYR SZZR SXYR SYZR SZXR XX Y Y ZZ XY Y Z ZX
SXXI SYYI SZZI SXYI SYZI SZXI
XX
Y Y
ZZ
XY
Y Z
ZX
171
8.3.5.2
Forces
comp opti X AMPLIT NxA Nx Y
QyA
Concentrated forces item type form oper STRESS TOTAL FORCE LOCAL STRESS TOTAL FORCE STRESS TOTAL FORCE STRESS TOTAL FORCE LOCAL
Z
QzA
X
NxP
Nx
Y
QyP
Qy
Z
QzP
Qz
Qy
NyR
Qz
NzR
COMPLE NxR Nx GLOBAL AMPLIT NXA NX GLOBAL COMPLE NXR NX comp opti XX AMPLIT NxxA nxx
NxI
NyI
NzI
Qy
NYA
Qz
NZA
Nx
NXP NXI
Qy
NYP NYI
Qz
NZP
NZ
NY
NYR
NZ
NZR
NX NY
NZI
NY YY nyy
NZ
NX
NY
NZ
Distributed forces item type form oper STRESS TOTAL DISFOR LOCAL
ZZ nzz
XY nxy
YZ qyz
ZX qxz
nxx nyy nzz nxy qyz qxz COMPLE NxxR NyyR NzzR NxyR QyzR QxzR nxx nyy nzz nxy qyz qxz
nxx
nyy
nzz
nxy
qyz
qxz
8.3.5.3
Bending moments
comp opti X AMPLIT MxA Mx Y
MyA
Concentrated moments item type form oper STRESS TOTAL MOMENT LOCAL STRESS TOTAL MOMENT LOCAL
Z
MzA
X
MxP
Mx
Y
MyP
My
Z
MzP
Mz
My
MyR
Mz
MzR
COMPLE MxR Mx STRESS TOTAL MOMENT GLOBAL AMPLIT MXA MX STRESS TOTAL MOMENT GLOBAL COMPLE MXR MX
MxI
MyI
MzI
My
MYA
Mz
MZA
Mx
MXP MXI
My
MYP MYI
Mz
MZP
MZ
MY
MYR
MZ
MZR
MX MY
MZI
MY
MZ
MX
MY
MZ
172
Distributed moments item type form oper STRESS TOTAL DISMOM LOCAL STRESS TOTAL DISMOM LOCAL
XY
MxyA
XX
MxxP MxxI
YY
MyyP MyyI
XY
MxyP MxyI
mxy
MxyR
mxy
mxx
myy
mxy
8.3.5.4
Tractions
form oper TRACTI LOCAL comp opti X Y AMPLIT STxA STyA tx ty TRACTI LOCAL COMPLE STxR STyR tx ty TRACTI GLOBAL AMPLIT STXA STYA tX tY TRACTI GLOBAL COMPLE STXR STYR tX tY Z tz tz tZ tZ X
tx
Tractions item type STRESS TOTAL STRESS TOTAL STRESS TOTAL STRESS TOTAL
Y
ty
Z
tz
tx
tX
ty
tY
tz
tZ
tX
tY
tZ
8.3.6
syntax
FORCE [ type w ] [ form w ] [ oper w ] { comp w } { opti w } REACTI TRANSL LOCAL AMPLIT ROTATI GLOBAL COMPLE FORCE species forces (and moments) in the nodes as output item [ 31.3 p. 519].
[REACTI]
type
species the type [ 3.6.2 p. 78]. REACTI for the reaction forces in all supported nodes.
[TRANSL] [GLOBAL]
species the formulation [ 3.6.2 p. 78]. species an operation (transformation) to be performed on the forces and moments [ 3.6.2 p. 79]. selects force or moment components for output. Default is all available components.
[all]
173
Z b Fz
b Fz
X
b Fx b Fx
Y
b Fy b Fy
Z
b Fz b Fz b FZ b FZ
FByA FBzA FBxP FByP FBzP FByR FBzR FBxI FByI FBzI FBYA FBZA FBXP FBYP FBZP
COMPLE FBxR b Fx FORCE REACTI TRANSL GLOBAL AMPLIT FBXA b FX FORCE REACTI TRANSL GLOBAL COMPLE FBXR b FX Moment reactions item type form oper FORCE REACTI ROTATI LOCAL comp opti X AMPLIT MBxA b Mx FORCE REACTI ROTATI LOCAL COMPLE MBxR b Mx FORCE REACTI ROTATI GLOBAL AMPLIT MBXA b MX FORCE REACTI ROTATI GLOBAL COMPLE MBXR b MX
b FY
b FY
b FZ
b FZ
b b FX FY b FX b FY
Y
MByA
Z
MBzA
X
MBxP
b Mx
Y
MByP
b My
Z
MBzP
b Mz
b My
MByR
b My
b Mz
MBzR
b Mz
MBxI
b Mx
MByI
b My
MBzI
b Mz
MBYA
MBZA
MBXP MBXI
b MX
MBYP MBYI
b MY
MBZP
b MZ
b MY
MBYR
b MY
b MZ
MBZR
b MZ
b b MX MY
MBZI
b MZ
8.3.7
Dynamic Pressures
For uidstructure interface elements [Vol. Element Library] Diana can calculate and output the dynamic pressures. See [ 32.5 p. 532] for background theory. syntax FSPRES [ type w ] [ loca w ] { opti w } TOTAL NODES AMPLIT COMPLE FSPRES species the dynamic pressures of the uidstructure interface elements as output item. type species the dynamic pressure type. TOTAL for the total dynamic pressures of the uidstructure interfaces. loca opti species the location for the strains to be output [ 3.6.2 p. 79]. are additional options [ 8.3 p. 162].
[NODES] [COMPLE] [TOTAL]
174
Dynamic pressures item type loca opti FSPRES TOTAL NODES AMPLIT PRfsA PRfsP pf s pf s FSPRES TOTAL NODES COMPLE PRfsR PRfsI pf s pf s
Chapter 9
In this example the diagram for load set 1 is specied by four points (fi , i ) = (3, 1), (5, 6), (9, 4), (12, 9) [Fig. 9.1]. The multiplication factor for load set 3 is equal to 2.0 for the frequency range 0.0 to 20.0.
Diana-9.4 Users Manual Analysis Procedures (III) December 15, 2009 First ed.
176
spectral acceleration SAn SA2 SAi SA1 0 Sa specied interpolated
f1
f2
fi
fn
frequency
Figure 9.1: Loading multiplication specication Fluidstructure interaction analysis. In case of a model containing uid, structural elements and uidstructure interface elements, automatically a uidstructure spectral response analysis will be performed. Diana recognizes this element combination and will perform the required preparation to perform a uidstructure spectral response analysis. In a spectral response analysis the added mass eect of the uid [Eq. (32.92) p. 537] is taken into account. Results will be available for the structural part of the model. Lumped element matrices may not be used in a uidstructure response analysis. Command sequence. lows. The command sequence for Module spectr is as folsyntax *SPECTR [ MODEL ] [ EIGEN ] [ RESPON ] *END MODEL evaluates the nite element model [ 9.1]. EIGEN solves the eigenvalue problem for a spectral response analysis [ 9.2]. RESPON performs the actual spectral response analysis [ 9.3].
9.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual analysis.
9.2 Eigenmodes and Natural Frequencies syntax BEGIN MODEL [ OFF ] [ EVALUA [ OFF ] ] [ ASSEMB [ OFF ] ] END MODEL EVALUA checks and evaluates geometric and material properties for elements and reinforcements [ 3.4 p. 66]. ASSEMB assembles the elements, i.e., creating appropriate system degrees of freedom [ 3.5 p. 68].
177
9.2
To solve the free vibration eigenvalue problem in a spectral response analysis, i.e., to determine the eigenmodes and natural frequencies, you must give commands in the EIGEN block which are analogous to those for Module eigen [ 27.2 p. 439]. The OFF options suppress the execution of the specied task. This may be useful to save computing time if previously determined results still reside on the filos le. syntax BEGIN EIGEN [ OFF ] [ FREEVI [ OFF ] ] [ EXECUT [ OFF ] ] [ OUTPUT ] END EIGEN FREEVI sets up the free vibration eigenmode analysis [ 27.2.1 p. 440]. EXECUT species how to solve the eigenvalue problem. In particular this involves the type of the solution procedure [ 27.3 p. 445]. OUTPUT species the eigenmodes to be output [ 27.4.7 p. 448].
9.3
Response Analysis
To perform the actual spectral response analysis, i.e., to determine the response of a construction to a certain excitation spectrum, you must give commands in the RESPON block. The OFF options suppress the specied tasks. This may be useful to save computing time if previously determined results still reside on the filos le.
Diana-9.4 Users Manual Analysis Procedures (III) December 15, 2009 First ed.
178
Spectral Response Analysis syntax BEGIN RESPON [ EXECUT [ OFF ] ] [ OUTPUT [ OFF ] ] END RESPON EXECUT executes the actual spectral response analysis [ 9.3.1]. OUTPUT species the spectral response analysis results to be output [ 9.4].
9.3.1
Analysis Execution
The EXECUT command block actually performs the spectral response analysis by applying the excitation spectrum specied by a base excitation load and a diagram with frequency dependent load multiplication coecients. See 32.3 on page 526 for background theory. syntax BEGIN EXECUT [ OFF ] [ DAMPIN c1 r [ . . . cn r ] ] END EXECUT OFF suppresses the execution of the spectral response analysis. DAMPIN species modal damping. Values c1 to cn are a series of damping coecients in parts of the critical damping factor ccrit [Eq. (32.5) p. 522]; a value of 0.01 indicates 1 %. If only one factor c1 is specied, then the damping is the same for all natural frequencies of the system. Alternatively, the damping coecients must be specied for all natural frequencies of the system. Modal damping coecients will be used to superpose the results of the modes according to the CQCrule. If no damping coecients are specied zero damping will be assumed.
9.4
With command in the OUTPUT block you can get output of results from a spectral response analysis. For the general, analysis type independent options for output of analysis results see 3.6 on page 69.
9.4 Output of Analysis Results syntax BEGIN OUTPUT [device w ] [outopt w... ] [ params ] [ OFF ] [ BEGIN SELECT [ NODES [ ]] nodes ng... ALL [ MODES [ ]{ }] modes n... ABS ALL SRSS CQC END SELECT ] [ LAYOUT ] item w DISPLA VELOCI ACCELE STRAIN STRESS FORCE NODFOR ELMFOR END OUTPUT SELECT command block to customize the batch output: NODES selects nodes for output of results. You specify a series of nodes by numbers and/or groups, or you may select ALL nodes. MODES species a mode selection: modes are the mode numbers referring to the calculated modes in the eigenvalue analysis. ALL indicates all calculated modes, which is also the default. ABS superposes the results of the (selected) modes according to the ABSrule. SRSS superposes the results of the (selected) modes according to the SRSS rule. CQC superposes the results of the (selected) modes according to the CQCrule. LAYOUT commands to customize the layout and style of tabular output [ 3.6.3 p. 80]. item is the name of the analysis result to be output. See 3.6.2 on page 77 for complete syntax of this command. DISPLA for displacements [ 9.4.1].
Diana-9.4 Users Manual Analysis Procedures (III)
179
[ALL]
[ALL]
180
Spectral Response Analysis VELOCI for displacements [ 9.4.2]. ACCELE for displacements [ 9.4.3]. STRAIN for strains [ 9.4.4]. STRESS for stresses [ 9.4.5]. FORCE for nodal forces [ 9.4.6]. NODFOR for element nodal forces [ 9.4.7]. ELMFOR for internal element forces [ 9.4.8].
9.4.1
Displacements
syntax
DISPLA [ type w ] [ form w ] [ oper w ] { comp w } { opti w } TOTAL TRANSL LOCAL ROTATI GLOBAL DISPLA species displacement amplitudes of the nodes as output item.
[TOTAL]
type
species the displacement type. TOTAL for the total displacements of a structure, i.e., the deformed geometry after a spectral response analysis.
[TRANSL] [GLOBAL]
form
oper species an operation (transformation) to be performed on the displacements [ 3.6.2 p. 79]. comp selects displacement components for output. Default is all available components. opti are additional options [ 3.6.2 p. 79].
Total displacements item type form oper DISPLA TOTAL TRANSL LOCAL DISPLA TOTAL DISPLA TOTAL DISPLA TOTAL comp X Y Z
DtxA DtyA DtzA
[all]
uy uY y Y
uz uZ z Z
181
9.4.2
Velocities
syntax
VELOCI [ type w ] [ form w ] [ oper w ] { comp w } { opti w } TOTAL TRANSL LOCAL ROTATI GLOBAL VELOCI species velocity amplitudes of the nodes as output item. type species the velocity type. TOTAL for the total velocities of a structure. form species the velocity formulation [ 3.6.2 p. 78].
[TRANSL] [GLOBAL] [TOTAL]
oper species an operation (transformation) to be performed on the velocities [ 3.6.2 p. 79]. comp selects velocity components for output. Default is all available components.
[all]
uy u Y y Y
uz uZ z Z
9.4.3
Accelerations
syntax
ACCELE [ type w ] [ form w ] [ oper w ] { comp w } { opti w } TOTAL TRANSL LOCAL ROTATI GLOBAL ACCELE species acceleration amplitudes of the nodes as output item. type species the acceleration type. TOTAL for the total accelerations of a structure. form species the acceleration formulation [ 3.6.2 p. 78].
[TRANSL] [TOTAL]
182
Spectral Response Analysis oper species an operation (transformation) to be performed on the accelerations [ 3.6.2 p. 79].
[all] [GLOBAL]
comp selects acceleration components for output. Default is all available components. opti are additional options [ 3.6.2 p. 79].
Total accelerations item type form oper ACCELE TOTAL TRANSL LOCAL ACCELE TOTAL ACCELE TOTAL ACCELE TOTAL comp X Y Z
AtxA AtyA AtzA
uy u Y y Y
uz uZ z Z
9.4.4
Strains
[ oper w ] { comp w } { loca w } { opti w } LOCAL INTPNT SMOOTH GLOBAL NODES ERROR PRINCI STRPNT VONMIS CROSSE REAXES CENTER VOLUME
syntax
STRAIN [ type w ] [ form w ] TOTAL GREEN FORCE MOMENT DISFOR DISMOM TRACTI DISSEI
STRAIN species strain amplitudes as output item. Table 9.1 on the facing page outlines the availability and applicability of the various strain output options for each of the element families.
[TOTAL] [GREEN]
type form
species the strain type [ 3.6.2 p. 78]. species the strain formulation. GREEN for GreenLagrange strains [ 9.4.4.1]. FORCE for deformations due to normal and shear forces [ 9.4.4.2]. DISFOR for generalized strains [ 9.4.4.2]. MOMENT for curvatures due to a concentrated bending moment [ 9.4.4.3]. DISMOM for curvatures due to a distributed bending moment [ 9.4.4.3]. TRACTI for tractions in structural interface elements [ 9.4.4.4].
183
item type
form
oper
loca
opti
STRAIN TOTAL GREEN FORCE MOMENT DISFOR DISMOM TRACTI DISSEI LOCAL GLOBAL PRINCI VONMIS REAXES VOLUME INTPNT NODES STRPNT CROSSE CENTER SMOOTH ERROR
tr
a a a a a a a a a a a a
a a c c a a a a a a a a a a a a
a a a a a a a a a a a a a
a a a a a a a a a a a a
a a a a a a a a a a a a
a a a a a a a a a a a a -
a a a a a a a a a a a a a -
a a a a a a a a a a a a a
a a a a a a a a a a a a a
a a d a a a a -
a a a a e e -
a a f g f a f h f -
(a) All elements. (b) Not for elements with orthotropic geometry. (c) For all beam elements, classII and classIII only in combination with local coordinate system. (d) Only for plane structural interface elements. (e) Only for base springs. (f) Only for bondslip reinforcements. (g) Only for bondslip reinforcements modeled by beam elements. (h) Not for bondslip reinforcements. (-) Not available or not suitable.
DISSEI for distributed seismic moments in plane structural interface elements [ 9.4.4.5].
[GLOBAL]
oper
comp selects strain components for output. Default is all available components. loca species the location for the strains to be output [ 3.6.2 p. 79]. opti are additional options [ 3.6.2 p. 79].
Primary strains [ 31.1] item type form oper STRAIN TOTAL GREEN LOCAL STRAIN TOTAL STRAIN TOTAL STRAIN TOTAL
YZ
ZX
GyzA GzxA
xx yy zz xy yz zx GREEN GLOBAL EXXA EYYA EZZA GXYA GYZA GZXA XX Y Y ZZ XY Y Z ZX GREEN PRINCI E1A 1 GREEN REAXES E1RAA a 1
E2A
E3A
2
E2RAA
a 2
Von Mises strain [ 31.1.1] item type form oper STRAIN TOTAL GREEN VONMIS EeqA eq Volumetric strain [ 31.1.3] item type form oper STRAIN TOTAL GREEN VOLUME EvolA vol
9.4.4.2
Deformations
comp X Y
PxA PyA
Force deformations item type form oper STRAIN TOTAL FORCE LOCAL STRAIN TOTAL FORCE
Z
PzA
Generalized strains itm type form oper STRAIN TOTAL DISFOR LOCAL STRAIN TOTAL DISFOR REAXES
XY
YZ
ZX
xx yy zz xy yz zx
P1RAA P2RAA
a 1
a 2
9.4.4.3
Curvatures
comp X Y Z
KxA KyA KzA
Concentrated curvatures item type form oper STRAIN TOTAL MOMENT LOCAL
185
XY
YZ
ZX
xx yy zz xy yz zx
K1RAA K2RAA
a 1
a 2
9.4.4.4
Tractions
comp X Y Z
Traction deformations item type form oper STRAIN TOTAL TRACTI LOCAL
9.4.4.5
Distributed seismic moment [ 31.1.4] item type form STRAIN TOTAL DISSEI PSTOTA PSNEGA PSPOSA PS PS PS+
9.4.5
Stresses
[ oper w ] { comp w } { loca w } { opti w } LOCAL INTPNT NOBOND GLOBAL NODES SMOOTH PRINCI STRPNT ERROR VONMIS CROSSE INVARI CENTER REINFO REAXES BIAXFE CAPACI
syntax
STRESS [ type w ] [ form w ] TOTAL CAUCHY INITIA FORCE MOMENT DISFOR DISMOM TRACTI GRADIE
STRESS species stress amplitudes as output item. Table 9.2 on the next page outlines the availability and applicability of the various stress output options for each of the element families. type species the stress type [ 3.6.2 p. 78]. form species the stress formulation. CAUCHY for Cauchy stresses [ 9.4.5.1].
Diana-9.4 Users Manual Analysis Procedures (III) December 15, 2009 First ed.
[TOTAL] [CAUCHY]
186
item type
form
oper
loca
opti
STRESS TOTAL INITIA CAUCHY FORCE MOMENT DISFOR DISMOM TRACTI GRADIE LOCAL GLOBAL PRINCI VONMIS INVARI REINFO REAXES BIAXFE CAPACI INTPNT NODES STRPNT CROSSE CENTER NOBOND SMOOTH ERROR
tr
a a a a a a a a a a a a a a
a a a a a a a a a a a a a a a a a a
a a a a a a a a a a a a a a a a a a
a a a e e a a a a a a a a a a a a
a a a c c a a a a a a a a a a a
a a b a a a a a a a a a a a -
a a b a a a a a a a a a a a -
a a a a a a a a a a a a a a a a a a
a a a d d a a a a a a a a a a a a a a
a a a a a a a -
a a a a a g g -
a a a a i h f a h j h -
(a) All elements. (b) Not for elements with orthotropic geometry. (c) For shells of revolution only. (d) For composed solids only [Vol. Element Library]. (e) For innite shells only. (f) Only for bar reinforcements. (g) Only for base springs. (h) Only for bondslip reinforcements. (i) Only for bondslip reinforcements modeled by beam elements. (j) Not for bondslip reinforcements. (-) Not available or not suitable.
FORCE for concentrated forces and tractions [ 9.4.5.2]. MOMENT for concentrated bending moments [ 9.4.5.3]. DISFOR for distributed forces [ 9.4.5.2]. DISMOM for distributed bending moments [ 9.4.5.3]. TRACTI for tractions in structural interface elements [ 9.4.5.4]. GRADIE for gradients of stresses in reinforcement bars [ 9.4.5.5].
[GLOBAL]
oper
species an operation (transformation) to be performed on the primary stresses [ 3.6.2 p. 79]. BIAXFE gives the following Safety Factors for concrete under static and
9.4 Output of Analysis Results dynamic loading conditions with reference to a biaxial failure envelope:
187
F Sstatic usual = Rstatic usual /r F Sstatic unusual = Rstatic unusual /r F Sdynamic unusual = Rdynamic unusual /r F Sdynamic extreme = Rdynamic extreme /r
Where r is the distance from the origin to the actual stress point (1 , 2 ): r = (1 2 + 2 2 ) (9.1) To determine the Safety Factors Diana needs additional material properties. For the required parameters and additional information on the Safety Factors, see Volume Material Library. Note that all Safety Factors are limited to 100. CAPACI gives the shear capacity and hydrostatic pressure capacity of stress against MohrCoulomb failure criterion [ 31.2.9 p. 519]. To determine the shear capacity and hydrostatic pressure capacity Diana needs additional material properties, see Volume Material Library. comp selects stress components for output. Default is all available components. loca species the location for the stresses to be output [ 3.6.2 p. 79]. opti are additional options [ 3.6.2 p. 79]. 9.4.5.1 Cauchy Stresses
comp XX YY
SxxA SyyA
[all]
Primary stresses [ 31.2] item type form oper STRESS TOTAL CAUCHY LOCAL STRESS TOTAL STRESS TOTAL STRESS TOTAL STRESS INITIA STRESS INITIA STRESS INITIA STRESS INITIA
ZZ
SzzA
XY
SxyA
YZ
SyzA
ZX
SzxA
xx yy zz xy yz zx CAUCHY GLOBAL SXXA SYYA SZZA SXYA SYZA SZXA XX Y Y ZZ XY Y Z ZX CAUCHY PRINCI S1A 1 CAUCHY REAXES S1RAA a 1 CAUCHY LOCAL S0xxA S0yyA S0zzA S0xyA S0yzA S0zxA 0 0 0 0 0 0 xx yy zz xy yz zx CAUCHY GLOBAL S0XXA S0YYA S0ZZA S0XYA S0YZA S0ZXA 0 0 0 0 0 0 XX Y Y ZZ XY Y Z ZX CAUCHY PRINCI S01A 0 1 CAUCHY REAXES S01RAA 0a 1
S2A
S3A
2
S2RAA
a 2
S02A
0 2
S03A
0 3
S02RAA
0a 2
188
Von Mises stress [ 31.2.1] item type form oper STRESS TOTAL CAUCHY VONMIS SeqA eq STRESS INITIA CAUCHY VONMIS S0eqA 0 eq Stress invariants [ 31.2.6] comp item type form oper P Q LODE STRESS TOTAL CAUCHY INVARI PA QA LodeA p q
Biaxial failure envelope comp item type form oper STUS STUN DYUN DYEX STRESS TOTAL CAUCHY BIAXFE FSstus FSstun FSdyun FSdyex F Sstus F Sstun F Sdyun F Sdyex Shear capacity [ 31.2.9] comp item type form oper SHRCAP PPCAP STRESS TOTAL CAUCHY CAPACI SHRCAP PPCAP
9.4.5.2
Forces
comp X Y
NxA
Concentrated forces item type form oper STRESS TOTAL FORCE LOCAL STRESS TOTAL FORCE
Z Qz
NZA
X
QyA
Y
NzA
Z Nz
Qy
NYA
Qx
Qy
NY
Q0yA
NZ
Q0zA Q0xA Q0yA N0zA
Q0 y
N0YA
0 NY
Q0 z
N0ZA
0 NZ
Q0 x
Q0 y
0 Nz
Distributed forces item type form oper STRESS TOTAL DISFOR LOCAL STRESS TOTAL DISFOR REAXES
comp XX YY
NxxA NyyA
ZZ
NzzA
XY
NxyA
YZ
QyzA
ZX
QxzA
nxx
nyy
nzz
nxy
qyz
qxz
N1RAA N2RAA
na 2
n0 xx
n0 yy
n0 zz
n0 xy
0 qyz
0 qxz
Reinforcement forces [ 31.2.7] comp item type form oper 1R 2R 1RC 2RC QT STRESS TOTAL DISFOR REINFO N1RA N2RA N1RCA N2RCA QTA n1 n2 n1c n2c q
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (III)
189
9.4.5.3
Bending moments
comp X Y Z
MxA MyA MYA MzA MZA
Concentrated moments item type form oper STRESS TOTAL MOMENT LOCAL STRESS TOTAL
Mx My Mz MOMENT GLOBAL MXA MX STRESS INITIA MOMENT LOCAL M0xA 0 Mx STRESS INITIA MOMENT GLOBAL M0XA 0 MX Distributed moments item type form oper STRESS TOTAL DISMOM LOCAL STRESS TOTAL DISMOM REAXES
M0xxA M0yyA M0xyA
MY MZ
M0yA M0zA
0 0 My Mz
M0YA M0ZA
0 0 MY MZ
comp XX YY
MxxA MyyA
ZZ XY
MxyA
YZ ZX 1
mxx
myy
mxy
M1RAA M2RAA
ma 2
Reinforcement moments [ 31.2.7] comp item type form oper 1R 2R STRESS TOTAL DISMOM REINFO M1RA M2RA m1 m2
9.4.5.4
Tractions
form oper TRACTI LOCAL comp X Y
STxA STyA
Z
STzA
tx TRACTI GLOBAL STXA tX STRESS INITIA TRACTI LOCAL ST0xA t0 x STRESS INITIA TRACTI LOCAL ST0XA t0 X
ty
STYA
tz
STZA
tY t0 y t0 Y
tZ t0 z t0 Z
190 9.4.5.5
Stress gradients [ 31.2.8] item type form oper STRESS TOTAL GRADIE LOCAL SGA xx
9.4.6
Nodal Forces
syntax
FORCE [ type w ] [ form w ] [ oper w ] { comp w } { opti w } REACTI TRANSL LOCAL RESIDU ROTATI GLOBAL FORCE species force and moment amplitudes in the nodes as output item [ 31.3 p. 519].
[REACTI]
type
species the type of the nodal forces. REACTI for the reaction forces in all supported nodes. RESIDU for the residual forces (by default Diana only gives the extreme residual forces).
[TRANSL] [GLOBAL]
species the formulation [ 3.6.2 p. 78] species an operation (transformation) to be performed on the forces and moments [ 3.6.2 p. 79]. selects force or moment components for output. Default is all available components.
[all]
Z
FBzA
b Fz
FBZA
b FZ
MBzA
b Mz
MBZA
b MZ
191
FORCE RESIDU TRANSL GLOBAL FRXA r FX FORCE RESIDU ROTATI LOCAL MRxA r Mx FORCE RESIDU ROTATI GLOBAL MRXA r MX
MRYA MRZA
r r MY MZ
9.4.7
syntax
NODFOR [ type w ] [ form w ] [ oper w ] { comp w } { opti w } ELEMEN TRANSL LOCAL REINFO ROTATI GLOBAL TOTAL NODFOR species internal nodal element force and moment amplitudes in the nodes as output item. This command gives the contributions of the element or reinforcement internal nodal forces (or both), to a certain node. A selection of elements which form a section of the model gives the total internal forces that act on that section. type species the forces type. ELEMEN for contribution of elements only. REINFO for contribution of embedded reinforcements only. TOTAL for contribution of both elements and embedded reinforcements. form oper comp species the formulation [ 3.6.2 p. 78]. species an operation (transformation) to be performed on the forces and moments [ 3.6.2 p. 79]. selects force or moment components for output. Default is all available components.
[TRANSL] [GLOBAL] [TOTAL]
[all]
192
Nodal element forces [ 31.3.1] comp item type form oper X Y NODFOR type TRANSL LOCAL FNxA FNyA n n Fx Fy NODFOR type TRANSL GLOBAL FNXA FNYA n n FX FY NODFOR type ROTATI LOCAL MNxA MNyA n n Mx My NODFOR type ROTATI GLOBAL MNXA MNYA n n MX MY
Z
FNzA
n Fz
FNZA
n FZ
MNzA
n Mz
MNZA
n MZ
9.4.8
syntax
ELMFOR [ type w ] [ form w ] [ oper w ] { comp w } { opti w } ELEMEN TRANSL GLOBAL REINFO ROTATI TOTAL ELMFOR species internal element forces and moments in the nodes of an element as output item. This command gives the contributions of the element or reinforcement internal forces (or both), to a certain node of an element.
[TOTAL]
type
species the forces type. ELEMEN for contribution of elements only. REINFO for contribution of embedded reinforcements only. TOTAL for contribution of both elements and embedded reinforcements.
[TRANSL] [GLOBAL]
species the formulation [ 3.6.2 p. 78]. species an operation (transformation) to be performed on the forces and moments [ 3.6.2 p. 79]. selects force or moment components for output. Default is all available components.
[all]
Z
FEZ
e FZ
MEZ
e MZ
Chapter 10
MODEL evaluates the nite element model [ 10.1]. EIGEN solves the eigenvalue problem for a hybrid frequency time domain analysis [ 10.2]. RESPON performs the actual hybrid frequency time domain analysis [ 10.3].
10.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual analysis. syntax BEGIN MODEL [ OFF ] [ EVALUA [ OFF ] ] [ ASSEMB [ OFF ] ] [ MATRIX [ OFF ] ] [ LOADS [ OFF ] ] END MODEL EVALUA checks and evaluates geometric and material properties for elements and reinforcements [ 3.4 p. 66]. ASSEMB assembles the elements, i.e., creating appropriate system degrees of freedom [ 3.5 p. 68]. MATRIX to setup the element matrices [ 10.1.1]. LOADS to setup the load vectors.
10.1.1
Element Matrices
The MATRIX commands specify the type of element matrices to be applied in the direct response analysis. syntax BEGIN MATRIX [ OFF ] [ MASS [ OFF ] [
] [ ROTATI [ OFF ] ] ]
10.1 Model Evaluation CONSIS LUMPED [ DAMPIN [ OFF ] [ ]] CONSIS LUMPED [ STRESS [ OFF ] [ ] INPUT { LOAD=loset n FACTOR=fac r CALCUL { LOAD=loset n FACTOR=fac r PHASE END MATRIX OFF suppresses the setup of element matrices. MASS species the kind of mass matrices to be applied: CONSIS for consistent matrices (the default). LUMPED for lumped matrices. ROTATI indicates that rotational terms must be included in the mass matrices. The OFF option suppresses the inclusion of rotational terms. If you dont specify the ROTATI option explicitly, then Diana will include rotational terms by default. OFF suppresses the setup of mass matrices. DAMPIN species the kind of damping matrices to be applied: CONSIS for consistent matrices (the default). LUMPED for lumped matrices. OFF suppresses the setup of damping matrices. STRESS adds the geometric stress-stiness matrix KG to the linear elastic stiness matrix KL0 . INPUT indicates a geometric stress-stiness matrix, with stresses specied via a prestress load for load set loset , in subtable ELEMEN and/or REINFO of table LOADS [Vol. Element Library]. CALCUL indicates a geometric stress-stiness matrix, with stresses which Diana will calculate automatically in a linear static analysis for load set loset (the default). PHASE indicates a geometric stress-stiness matrix, with stresses present from a previous phase.
Diana-9.4 Users Manual Analysis Procedures (III)
195
] }
[CONSIS]
[CONSIS]
[CALCUL]
196
Hybrid Frequency Time Domain Analysis To setup the geometric stress-stiness matrix KG from a stress eld you may specify a load set number loset via the LOAD parameter. This load set number corresponds to a load set in input table LOADS [ 2.3.7 p. 55]. Default is the lowest available load set number. The optional parameter FACTOR species a multiplication factor fac .
[fac =1.0]
10.2
To solve the free vibration eigenvalue problem in a hybrid frequency time domain analysis, i.e., to determine the eigenmodes and natural frequencies, you must give commands in the EIGEN block which are analogous to those for a free vibration eigenmode analysis of Module eigen [ 27.2 p. 439]. The OFF options suppress the execution of the specied task. This may be useful to save computing time if previously determined results still reside on the filos le. syntax BEGIN EIGEN [ OFF ] [ EXECUT [ OFF ] ] [ OUTPUT ] END EIGEN EXECUT species how to solve the free vibration eigenvalue problem. In particular this involves the type of the solution procedure [ 27.3 p. 445]. OUTPUT to specify the eigenmodes to be output [ 27.4.7 p. 448]. Default output
*HFTD BEGIN EIGEN OUTPUT END EIGEN [ commands ] *END file .dcf
If you only give a single OUTPUT command, like in the above example, or if you omit the OUTPUT command, then Diana gives a default output as if you had given the following commands.
file .dcf *HFTD BEGIN EIGEN BEGIN OUTPUT DISPLA TOTAL TRANSL GLOBAL
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (III)
197
Example
*HFTD BEGIN EIGEN EXECUT NMODES=30 OUTPUT DISPLA END EIGEN *END
file .dcf
This command sequence is a typical example for a preliminary eigenvalue analysis. The EXECUT command block calculates the natural frequencies and corresponding mode shapes needed for the hybrid frequency time domain analysis. Parameter NMODES=30 asks for an arbitrarily chosen number of thirty frequencies. Finally, printed output of results like natural frequencies and mode shapes is obtained through the OUTPUT command block.
10.3
Response Analysis
To perform the actual hybrid frequency time domain analysis, i.e., to determine the response of a construction due to base excitation loading at a certain time, you must give commands in the RESPON block. The OFF options suppress the specied tasks. This may be useful to save computing time if previously determined results still reside on the filos le. syntax BEGIN RESPON [ OFF ] [ REDUCE [ OFF ] ] [ EXECUT [ OFF ] ] [ OUTPUT [ OFF ] ] END RESPON REDUCE reduces the system of equations [ 10.3.1]. EXECUT executes the actual hybrid frequency time domain analysis [ 10.3.2]. OUTPUT species the hybrid frequency time domain analysis results to be output [ 10.3.3].
If you only give a single OUTPUT command, like in the above example, or if you omit the OUTPUT command, then Diana gives a default output as if you had given the following commands.
file .dcf *HFTD [ commands ] BEGIN RESPON BEGIN OUTPUT DISPLA TOTAL FORCE REACTI FORCE EXTERN STRESS TOTAL STRESS TOTAL STRESS TOTAL STRESS TOTAL STRESS TOTAL STRESS TOTAL STRAIN TOTAL STRAIN TOTAL END OUTPUT END RESPON *END
TRANSL GLOBAL
10.3.1
To reduce the full set of equations, you must give commands in the REDUCE block within the RESPON block. syntax BEGIN REDUCE [ OFF ] [ BEGIN SELECT MODES modes n... ALL
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (III)
10.3 Response Analysis END SELECT ] END REDUCE OFF suppresses the reduction of the system of equations. This may save computing time when the reduced system of equations is still available from a previous modal response analysis. SELECT MODES selects eigenmodes for the hybrid frequency time domain analysis. The modes must have been determined previously [ 10.2 p. 196]. You may explicitly specify a set of modes , or you may require a hybrid frequency time domain analysis for ALL determined modes. All modes is also the default if you dont select modes.
[ALL]
199
10.3.2
Analysis Execution
The EXECUT command block actually performs the hybrid frequency time domain analysis by calculating the response due to the base excitation loading. syntax BEGIN EXECUT [ OFF ] [ TIMESE ] [ ITERAT ] [ MINFRE minfre r ] [ MAXFRE maxfre r ] END EXECUT TIMESE denes the time segments [ 10.3.2.1]. ITERAT species the time segment based convergence criterion to be applied [ 10.3.2.2]. MINFRE denes the minimum excitation frequency minfre . The unit is depending on the chosen time unit. MINFRE can be used as a high-pass lter. MAXFRE denes the maximum excitation frequency maxfre . The unit is depending on the chosen time unit. MAXFRE can be used as a low-pass lter.
[maxfre = 50] [minfre = 0]
10.3.2.1
Time Segment
The TIMESE commands specify the settings for the time segments.
200
Hybrid Frequency Time Domain Analysis syntax BEGIN TIMESE [ BEGIN STEPS [ EXPLIC NUMBER number i... ] END STEPS ] END TIMESE STEPS sets the number of time steps contained by the time segments. EXPLIC explicitly specied number of time steps contained by the time segments. NUMBER= number denes the number of time steps contained by the time segments. 10.3.2.2 Equilibrium Iteration
The CONVER commands specify the convergence criterium for the equilibrium iteration process. Diana terminates the iteration process if the specied criterium is satised. syntax BEGIN ITERAT [ MAXITE=mi n ] [ BEGIN CONVER } FORCE [ noconv w ] { TERMIN TOLCON=tc r CONTIN TOLABT=ta r END CONVER ] END ITERAT
[mi = 10]
MAXITE= mi is the maximum number of iterations for each time segment. CONVER species the convergence criterion for the iteration process. FORCE species an convergence criterion based on the pseudo-forces. noconv indicates what to do if no convergence occurs within the maximum number of iterations. With option TERMIN Diana stops the analysis run, no further steps will be executed (the default). However, further steps may be executed in subsequent runs. With option CONTIN the analysis run will be continued, you should check the relevance of the analysis results carefully in this case.
[TERMIN]
10.3 Response Analysis Two parameters specify the tolerance values. The default values depend on the used criterion as indicated in the examples below. TOLCON=tc is the tolerance for convergence. When the norm has become less than tc the reference norm, Diana assumes sucient accuracy, stops the iteration process and continues with the next step. TOLABT=ta is the tolerance for divergence. When the norm exceeds a value of ta the reference norm, Diana assumes divergence and aborts the nonlinear analysis.
201
[tc = 102 ]
[ta = 104 ]
10.3.3
You may indicate the analysis results to be output via the OUTPUT command block. For the general, analysis type independent options for output of analysis results see 3.6 on page 69. syntax BEGIN OUTPUT [device w ] [outopt w... ] [params ] [ OFF ] [ BEGIN SELECT [ ] [ STEPS ] END SELECT ] [ LAYOUT ] item w . . . DISPLA VELOCI ACCELE STRAIN STRESS STATUS PRESSU TEMPER CONCEN MATURI FORCE NODFOR ELMFOR PARAME FRACTU END OUTPUT SELECT command block to customize the batch output.
Diana-9.4 Users Manual Analysis Procedures (III) December 15, 2009 First ed.
202
Hybrid Frequency Time Domain Analysis for model selection see 3.6.1 on page 71, for stress- and strain transformation 3.6.1.2 on page 73. STEPS selects steps for output [ 10.3.3.1 p. 203]. LAYOUT optional commands to customize the layout and style of tabular output [ 3.6.3 p. 80]. item is the name of the analysis result to be output. See 3.6.2 on page 77 for complete syntax of this command. DISPLA for displacements, syntax and output is analogous to linear transient analysis [ 7.4.1 p. 134]. VELOCI for velocities, syntax and output is analogous to linear transient analysis [ 7.4.2 p. 135]. ACCELE for accelerations, syntax and output is analogous to linear transient analysis [ 7.4.3 p. 137]. STRAIN for strains, syntax and output is analogous to nonlinear transient analysis [ 12.4.2 p. 255]. STRESS for stresses, syntax and output is analogous to nonlinear transient analysis [ 12.4.3 p. 260]. STATUS for status, syntax and output is analogous to nonlinear transient analysis [ 12.4.4 p. 265]. PRESSU for pressure, syntax and output is analogous to nonlinear transient analysis [ 12.4.5 p. 269]. TEMPER for temperature, syntax and output is analogous to nonlinear transient analysis [ 12.4.6 p. 270]. CONCEN for concentration, syntax and output is analogous to nonlinear transient analysis [ 12.4.6 p. 270]. MATURI for maturity, syntax and output is analogous to nonlinear transient analysis [ 12.4.6 p. 270]. FORCE for nodal forces and moments, syntax and output is analogous to nonlinear transient analysis [ 12.4.7 p. 271]. NODFOR for element nodal forces, syntax and output is analogous to nonlinear transient analysis [ 12.4.8 p. 272]. ELMFOR for internal element forces, syntax and output is analogous to nonlinear transient analysis [ 12.4.9]. PARAME for model parameters, syntax and output is analogous to nonlinear transient analysis [ 12.4.10 p. 274]. FRACTU for Linear Elastic Fracture Mechanics analysis parameters of crack tip elements, syntax and output is analogous to nonlinear transient analysis [ 12.4.11].
203
The STEPS command selects steps for output of response analysis results. syntax STEPS [ steps n... ALL LAST ]{ MIN MAX }
steps are numbers of selected steps. ALL will produce output for all steps (the default). LAST selects the nal step. Diana will output the extreme values that occurred up to and including the selected steps if you specify one of the options: MIN for the minimum values, MAX for the maximum values.
[ALL]
204
Part IV
Chapter 11
11.1
In nonlinear structural analysis you may specify data additional to the input data for linear static analysis as described in Chapter 1 and Chapter 2. Material properties. The additional data mainly comprises the nonlinear phenomena like material behavior [Vol. Material Library]. Contact analysis. For contact analysis you must specify contact zones with contact elements [Vol. Element Library]. Transient analysis. For nonlinear transient (dynamic) analysis you may specify additional data analogous to the data for linear transient analysis as described in Chapter 6 and 6.4. This data comprises the specication of mass and damping, the initial displacements or velocities, the timeload diagram, and the base excitation. Moreover the time and/or ambient dependency of some material properties may be specied [Vol. Material Library].
208
11.2
In models for nonlinear analysis you are advised not to apply linearly interpolated isoparametric elements, like for instance Q8MEM, Q8OME, Q8EPS, Q8AXI, or HX24L. These elements have intrinsic shortcomings, like parasitic shear and volumetric locking, which cannot be easily dealt with in nonlinear analysis. In linear elastic analysis so-called assumed strain concepts can be applied to improve the element behavior [Vol. Element Library]. Unfortunately, in nonlinear analysis the assumed strain concepts harm the stability of the solution procedure and may cause spurious kinematic modes. Furthermore, the assumed strain concepts are not compatible with elements for mixture analysis and for Total and Updated Lagrange geometric nonlinearity. It is therefore recommended to apply higher order elements, for instance the quadratically interpolated CQ16M, CQ16O, CQ16E, CQ16A, or CHX60. The line interfaces to shell elements (L16IF, L20IF, CL24I, CL32I) cannot be applied in geometrical nonlinear analysis.
Chapter 12
Nonlinear Analysis
In order to perform a nonlinear structural analysis with Diana you must take the following actions: 1. Invoke Module filos to initialize an analysis database [ 3.2 p. 60]. 2. Invoke Module input to read the nite element model into the database [ 3.3 p. 64]. 3. Invoke Module nonlin to perform a nonlinear analysis. 4. You may get plots of the analysis results in the Results working environment of iDiana [Vol. Pre- and Postprocessing]. The command sequence for Module nonlin is as follows. syntax *NONLIN [ MODEL ] [ TYPE ] [ EXECUT ] . . . [ OUTPUT ] . . . *END MODEL evaluates the nite element model [ 12.1]. TYPE species the type of the nonlinear analysis [ 12.2]. EXECUT executes steps [ 12.3 p. 218]. OUTPUT selects analysis results for output [ 12.4 p. 252]. This output selection, at the *NONLIN command level, is valid for all EXECUT blocks of the current analysis. However, an output selection within any EXECUTE block overrules the output selection at the *NONLIN level [ 12.3 p. 219].
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
210
Nonlinear Analysis
12.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual analysis. syntax BEGIN MODEL [ OFF ] [ EVALUA [ OFF ] ] [ ASSEMB [ OFF ] ] [ MATRIX [ OFF ] ] [ LOADS [ OFF ] ] END MODEL EVALUA to check and evaluate geometric and material properties for elements and reinforcements [ 3.4 p. 66]. ASSEMB to assemble the elements, i.e., creating appropriate system degrees of freedom [ 3.5 p. 68]. MATRIX to setup the element stiness matrices. LOADS to setup the load vectors.
12.2
BEGIN TYPE [ OFF ] [ PHYSIC ] [ GEOMET ] [ TRANSI ] [ LINSTR [ OFF ] ] END TYPE PHYSIC sets options and parameters for physical nonlinear analysis [ 12.2.1]. GEOMET sets options and parameters for geometrical nonlinear analysis [ 12.2.2]. TRANSI sets options and parameters for transient nonlinear analysis [ 12.2.3].
12.2 Type of Nonlinear Analysis LINSTR activates linear stress/strain determination for linear elements. By default this option is not active. If you activate it then Diana will handle any element without physical and geometrical nonlinearities in a special way:
Strains and stresses will be determined only if you request these for output. Requested strains and stress will be determined directly from the total displacement eld, like for linear static analysis. No incremental update will be performed. Recalculation of the tangential matrix will be skipped, because this matrix remains constant.
211
[OFF]
For models with relatively large areas of linearly behaving elements this option may yield a considerable saving of computing time and data storage, but please note the following: The LINSTR option yields incorrect results for modied elasticity [Vol. Material Library], or if you change the reinforcement in a phased nonlinear analysis [Ch. 24 p. 403], or if you suppress the superposition of displacements and strains via the SUPPRE STRAIN command [ 12.3.4 p. 239], or if you suppress the superposition of stresses via the SUPPRE STRESS command [ 12.3.4 p. 239]. Default
*NONLIN TYPE [ commands ] *END file .dcf
These commands initiate a physical nonlinear analysis with application of all nonlinear phenomena as specied in the input data le. The same would occur if you had given the following commands:
file .dcf *NONLIN BEGIN TYPE PHYSIC END TYPE [ commands ] *END
212
Nonlinear Analysis
12.2.1
The use of the phenomena specied in the input data can be overruled for a limited number of options via the PHYSIC commands in the TYPE block. Phenomena and parameters are overruled for all the subsequent EXECUT blocks. syntax BEGIN PHYSIC [ OFF ] [ PLASTI [ OFF ] [
]{ }] FIRST MITERA=ni n SECOND TOLERA=ty r [ CREEP [ OFF ] [ ]{ }] ZERO MITERA=ni n FIRST TOLERA=tc r [ CORROS [ OFF ] ] [ TEMPER [ OFF ] ] [ CONCEN [ OFF ] ] [ CRACKI [ OFF ] [ ]{ }] SECANT TOLDIR=atd r CONSIS TOLTCO=tco r ]] [ TOTCRK [ OFF ] [ SECANT CONSIS [ ELASTI [ OFF ] ] [ VISCOE [ OFF ] ] [ VISCOP [ OFF ] ] [ MATURI [ OFF ] ] [ HYPER [ OFF ] ] [ INTERF [ OFF ] ] [ SHRINK [ OFF ] ] [ SOIL [ OFF ] ] END PHYSIC PLASTI for use of plasticity. An option indicates the tangent stiness to be used.
[FIRST]
FIRST for plasticity with classical (rst order) tangent stiness. SECOND for plasticity with consistent (second order) tangent stiness. Reduces the number of necessary iterations when used in combination with Regular NewtonRaphson iteration.
12.2 Type of Nonlinear Analysis Via two parameters you may customize the internal plasticity iteration process: MITERA=ni is the maximum number of iterations and TOLERA=ty the tolerance on satisfaction of the yield function. CREEP for use of creep. An option indicates the creep order. ZERO for creep with zero order approximation. FIRST for creep with properly linearized tangent stiness (rst order creep). Requires less computing time if used in combination with Regular NewtonRaphson iteration (only for transient creep). Via two parameters you may customize the internal creep iteration process: MITERA=ni is the maximum number of iterations and TOLERA=tc is the tolerance on stress accuracy. CORROS for inuence of corrosion. TEMPER takes the inuence of temperature on the strains into account (thermal strains). CONCEN takes the inuence of concentration on the strains into account (concentration strains). CRACKI for use of cracking. An option indicates the crack normal stiness to be used. See also the background theory on the multidirectional xed crack model in Volume Material Library. SECANT for cracking with secant crack normal stiness, i.e., for all cracks (either on softening branch or not) the positive secant unloading modulus is used as tangent stiness. This is the default cracking model. CONSIS for cracking with consistent crack normal stiness, i.e., for cracks on softening branch the true negative softening modulus is used as tangent stiness. This option may require less computing time, especially when used in combination with Regular NewtonRaphson iteration. It opens up the possibility to analyze snap-back behavior in combination with Arc-length control. Be careful when the crack pattern is diuse (for instance with reinforced concrete), then the method easily fails because of bifurcations. In such cases, continue with the secant crack normal stiness (SECANT). Two parameters may be used to aect the internal cracking iteration process: TOLDIR=atd the threshold angle TD between multidirectional cracks (in degrees) and TOLTCO=tco the tolerance on satisfaction of the tension-cut-o criterion. TOTCRK for use of crack models based on total strain. An option indicates the crack normal stiness to be used. See also the background theory on total strain crack models in Volume Material Library.
Diana-9.4 Users Manual Analysis Procedures (IV)
213
[ni =10]
[ty =0.0001]
[ZERO]
[SECANT]
214
Nonlinear Analysis SECANT for cracking with secant crack normal stiness, i.e., for all cracks (either on softening branch or not) the positive secant unloading modulus is used as tangent stiness. This is the default cracking model. CONSIS for cracking with consistent crack normal stiness, i.e., for cracks on the softening branch the true negative softening modulus is used as tangent stiness. ELASTI for use of elastic material behavior. VISCOE for use of viscoelastic material behavior. VISCOP for use of viscoplastic material behavior. MATURI for use of maturity dependent material properties in stress analysis of young, hardening concrete in combination with Power Law viscoelasticity. HYPER for use of hyperelastic (rubbery) material properties. INTERF for use of interface nonlinearity. The type of nonlinearity depends on input, for instance discrete cracking, crack dilatancy, bond-slip or friction. SHRINK for use of material shrinkage. SOIL for use of simple soil material behavior.
[SECANT]
12.2.2
The GEOMET commands initiate a geometrical nonlinear analysis. See 30.2 on page 499 for background theory. syntax BEGIN GEOMET [ OFF ] [ formul w ] TOTAL UPDATE [ BEGIN NCLOAD [ OFF ] [ LOADS losets n... ] END NCLOAD ] END GEOMET formul species the formulation to be applied.
12.2 Type of Nonlinear Analysis TOTAL applies the Total Lagrange formulation [ 30.2.2 p. 502] (the default). UPDATE applies the Updated Lagrange formulation [ 30.2.3 p. 502]. Updated Lagrange analysis is only available for solids, plane strain elements (except innite shells), and axisymmetric elements (except for shells of revolution). For other elements Diana automatically applies Total Lagrange. Plasticity, creep, and viscoelasticity are the only physical nonlinearities that can be applied in combination with Updated Lagrange analysis. Anisotropic eects of cracking, material anisotropy and embedded reinforcements are not handled correctly in Updated Lagrange analysis, because the rotations are not accounted for in the current implementation. Therefore, cracking, material anisotropy and embedded reinforcements can not be used in combination with Updated Lagrange analysis. NCLOAD species whether or not nonconservative pressure load must be applied. OFF switches o the application of nonconservative pressure load. LOADS losets are load set numbers referring to table LOADS [ 2.3.7 p. 55]. Diana will update the forces according to the load denition in the input le: if the direction was dened in a global direction then only the magnitude will be updated, if the direction was dened in an element direction then both the magnitude and the direction will be updated. Important conditions:
[OFF]
215
Only element loads can be nonconservative, other loads may not be part of the chosen load set number. Nonconservative load cannot be combined with Arc-length control methods [ 12.3.2.4 p. 230].
Default
*NONLIN TYPE GEOMET [ commands ] *END file .dcf
These commands initiate a geometrical nonlinear analysis according to the Total Lagrange formulation. The same would occur if you had given the following commands:
file .dcf *NONLIN BEGIN TYPE GEOMET TOTAL END TYPE
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
216
[ commands ] *END
Nonlinear Analysis
12.2.3
The TRANSI commands initiate a transient nonlinear analysis, i.e., applying time integration. As an option, you may specify the integration method to be used and the type of matrices to be applied in dynamic transient analysis. syntax BEGIN TRANSI [ OFF ] [ BEGIN METHOD [ method w ] BACKWA NEWMAR { } GAMMA=gamma r BETA=beta r HHT [ ALPHA=alpha r ] WILSON [ THETA=theta r ] SDIRK2 END METHOD ] [ BEGIN DYNAMI [ OFF ] [ MASS [ OFF ] [ CONSIS LUMPED [ DAMPIN [ OFF ] [ CONSIS LUMPED END DYNAMI ] [ TIMEDE [ OFF ] ] END TRANSI METHOD method species the time integration method.
[BACKWA] [ =
1 2,
] [ ROTATI [ OFF ] ] ]
] [ RAYLEI ] ]
BACKWA for Euler Backward time integration [ 32.4.2 p. 529]. NEWMAR for time integration according to Newmarks method [ 32.4.1 p. 529]. You may customize the integration algorithm with two optional parameters: GAMMA=gamma species factor , and BETA=beta
Diana-9.4 Users Manual Analysis Procedures (IV)
1 4]
1 2)
217
1 1 4(2
species factor . The conditions for and are to guarantee the stability of the algorithm. HHT for the HilberHughesTaylor time integration method [ 32.4.3 p. 530]. The optional parameter ALPHA=alpha species the factor . WILSON for the Wilson- time integration method [ 32.4.4 p. 530]. The optional parameter THETA=theta species the factor . SDIRK2 for two stage RungeKutta time integration [ 32.4.5 p. 531]. DYNAMI will apply a transient dynamic analysis.1 Please note that for transient dynamic analysis the creation of mass and damping matrices may require additional input data [Ch. 6 p. 117]. MASS species the kind of mass matrices to be applied for inclusion of inertia forces. CONSIS for consistent mass matrices (the default). LUMPED for lumped mass matrices. ROTATI to include rotational terms in the mass matrices. The OFF options suppresses the inclusion of rotational terms. If you dont specify the ROTATI option explicitly, then Diana will include rotational terms by default. DAMPIN species the kind of damping matrices to be applied for inclusion of damping forces. CONSIS uses consistent damping matrices. LUMPED uses lumped damping matrices. RAYLEI uses Rayleigh damping based on the current stiness matrices. With this option Diana will update the damping matrix within each time increment. This option requires the specication of the Rayleigh damping parameters as a material property, see Viscous Damping in Volume Material Library. TIMEDE will take the time derivative terms u and in to account in soilpore uid (mixture) analysis [Eq. (36.25) p. 558]. Default
*NONLIN TYPE TRANSI DYNAMI [ commands ] *END file .dcf
[OFF]
[MASS] [CONSIS]
[ON]
[CONSIS]
These commands initiate a transient dynamic nonlinear analysis with Euler Backward time integration. Consistent mass matrices will be applied, including rotational terms. The same would occur if you had given the following commands:
1 Dynamic
[Ch. 8].
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
218
Nonlinear Analysis
file .dcf *NONLIN BEGIN TYPE BEGIN TRANSI BACKWA DYNAMI MASS CONSIS ROTATI ON END TRANSI END TYPE [ commands ] *END
12.3
Step Execution
With the EXECUT commands you ask Diana to execute load or time steps. A command le may contain one or more EXECUT blocks. Details of various commands are given in the referred sections. syntax BEGIN EXECUT [ OFF ] [ type w ] START LOAD TIME [ PHYSIC ] [ ITERAT ] [ SOLVE ] [ STOP ] [ LOGGIN ] [ OUTPUT ] . . . [ REFERE ] END EXECUT
[LOAD]
type indicates the type of step(s) to be executed. START to evaluate the initial state [ 12.3.1]. LOAD to execute load steps [ 12.3.2 p. 225]. TIME to execute time steps [ 12.3.3 p. 234]. PHYSIC sets options for physical nonlinear analysis [ 12.3.4 p. 238].
12.3 Step Execution ITERAT species the process of equilibrium iteration to be applied [ 12.3.5 p. 240]. SOLVE customizes the settings for the solution method [Ch. 26 p. 429]. STOP species a stop criterion for step execution [ 12.3.6 p. 246]. LOGGIN customizes the logging information that will be output during execution of the steps [ 12.3.7 p. 249]. OUTPUT selects output for steps [ 12.4 p. 252]. This selection overrules any output selection specied at the *NONLIN command level [Ch. 12 p. 209]. REFERE sets initial state references for geotechnical analysis [ 12.3.8 p. 250]. Default
*NONLIN [ commands ] EXECUT *END file .dcf
219
Due to these commands Diana will execute a load step with a default factor, applying the default integration scheme and solution procedure, but no stop criteria. Diana will give the default logging information and output of analysis results. The same would occur if you had given the following commands:
file .dcf *NONLIN [ commands ] BEGIN EXECUT LOAD PHYSIC OFF ITERAT SOLVE LOGGIN OUTPUT END EXECUT *END
Note that the PHYSIC OFF command only switches o the setting of special physical options for step execution [ 12.3.4 p. 238]. The physical phenomena as specied on the input le and/or overruled via the TYPE PHYSIC command will be applied anyhow [ 12.2.1 p. 212].
12.3.1
The START commands evaluate the initial state of the model, i.e, prior to the execution of the rst load or time step. Consequently, the START commands may only appear in the rst EXECUT block of a nonlinear analysis.
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
220
Nonlinear Analysis syntax BEGIN START [ OFF ] [ TIME=sttime r ] [ INITIA ] [ LOAD [ ]] LOADNR=loset n OFF [ STEPS ] END START
[sttime =0]
TIME sttime is a user specied starting time. INITIA species initial conditions [ 12.3.1.1].
0 LOAD applies an initial external load fext [Eq. (12.1) p. 224]. The parameter LOADNR=loset indicates a load set number. Default is the lowest available load set number. The OFF option applies no initial external load, which is equivalent to the omission of the LOAD command.
STEPS executes steps for incremental adaptation to the initial conditions [ 12.3.1.2 p. 224]. Default
*NONLIN [ commands ] EXECUT START *END file .dcf
These commands will evaluate the initial state with start time t = 0 and with the default initial stresses [ 12.3.1.1]. Diana will adapt the status of the model to any nonlinearities caused by the initial conditions, for instance plastic ow. The same would occur if you had given the following commands:
file .dcf *NONLIN [ commands ] BEGIN EXECUT BEGIN START TIME=0. INITIA STRESS LOAD OFF STEPS END START END EXECUT *END
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IV)
221
12.3.1.1
Initial Conditions
To start with static equilibrium, you must let Diana rst perform one or more steps [ 12.3.1.2 p. 224]. Generally speaking there are three kinds of initial conditions: displacements, velocities, and stresses which may also be applied simultaneously (see also 6.4.1 on page 119). You may either specify initial displacements, velocities and stresses explicitly, or apply displacements and stresses calculated in a preliminary analysis as an initial condition for the nonlinear analysis. syntax BEGIN INITIA [ OFF ] [ DISPLA [ OFF ] [ INPUT
CALCUL
[ SOLVE ] END INITIA For initial elds the optional parameter FACTOR species a multiplication factor fac . For most initial elds you may either specify a field number via the
[fac =1.0]
222
Nonlinear Analysis FIELD parameter, or a load set number loset via the LOAD parameter. Default is the lowest available eld or load set number.
[INPUT]
DISPLA species a start with initial displacements. INPUT indicates a start with specied initial displacements. The field number refers to a eld in subtable DISPLA of table INIVAR [ 6.4.1.1 p. 119]. CALCUL indicates a start with initial displacements which Diana will calculate automatically in a linear static analysis for load set loset . POSTBU indicates a start with a postbuckling displacement eld from the most recent continuation stability analysis with Module euler [ 14.3.2 p. 295]. In combination with a Continuation iteration scheme for the rst step [ 12.3.2.2 p. 228] [ 12.3.2.3 p. 229], Diana will apply the postbuckling displacement eld as a rst predictor of the nonlinear analysis. The load value calculated by Module euler overrules the load step size that you specify for the rst step. This option is only useful in geometrically nonlinear analysis. VELOCI species a start with initial velocities. This option only applies for transient dynamic analysis. INPUT indicates a start with specied velocities, which is the only and default option. The field number refers to a eld in subtable VELOCI of table INIVAR [ 6.4.1.1 p. 119],
[INPUT]
STRESS species a start with initial stresses. This option also sets the initial state references. You may reset these references during the analysis by means of the REFERE command [ 12.3.8 p. 250]. In a nonlinear phased analysis the STRESS INPUT and STRESS CALCUL commands initialize the stresses in elements that become active for the rst time in the current phase. Stresses in all other active elements originate from a previous phase. INPUT indicates a start with stresses, specied via a prestress load for load set loset , in subtable ELEMEN and/or REINFO of table LOADS [Vol. Element Library]. The EQUILI option indicates a start with the non-prestress loads found in load set loset, applied as a load to balance the prestress. This option is particularly useful when you apply loads for which you can give an initial estimate for the resulting stresses as prestress. CALCUL indicates a start with the stresses which Diana will calculate automatically in a linear static analysis for load set loset . The EQUILI option indicates a start with an evaluated equilibrium state: it not only applies the stresses but also the external load as an initial condition. Actually, it is equivalent to the single CALCUL
12.3 Step Execution option, i.e., without performing START STEPS [ 12.3.1.2], followed by execution of a load step with load set loset as nonlinear load. The EQUILI option is particularly useful to initialize in situ (overconsolidated) stresses for geotechnical analysis. Usually this option yields an initial equilibrium state, i.e., no deformation. However, the initial equilibrium may be distorted in two cases: when you apply a K0 procedure to adapt the initial stress (see initial stress ratio in Volume Material Library), or when physical nonlinearity occurs. PHASE initiates a new phase in a nonlinear phased analysis. To reconstruct the internal force vector for the active elements Diana will integrate the stresses present from a previous phase. SOLVE customizes the settings for the solution method to be applied for the initial linear analysis [Ch. 26 p. 429]. Diana does not accept the combination of explicitly specied and calculated elds of the same type. Table 12.1 indicates the possibility of various combinations of options which you may apply via multiple START commands. Table 12.1: Combinations of initial condition options
UL BU PU LC ST PU LC UL T T
223
CA
IN
IN CI SS RE ST
LA
PO
SP
SP
SP
LO
DI
VE
Default
*NONLIN [ commands ] BEGIN EXECUT START INITIA [ commands ] END EXECUT *END
ST
DI
DI
These commands will cause the evaluation of the initial state with start time t = 0 and with initial prestresses from load set 1. Diana will adapt the status of the model to any nonlinearities caused by the initial conditions, for instance plastic ow. The same would occur if you had given the following commands:
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
RE
SS
LA
LA
CA
224
Nonlinear Analysis
file .dcf *NONLIN [ commands ] BEGIN EXECUT BEGIN START TIME=0.0 INITIA STRESS INPUT LOAD=lowest FACTOR=1.0 END START [ commands ] END EXECUT *END
12.3.1.2
The START STEPS commands apply incremental adaptation to the initial conditions, for instance initial stresses. You should use this option only if the internal 0 forces following from initial stresses fint are not in equilibrium with the initial 0 external load fext . Due to this option, Diana rst temporarily adds the initial out-of-balance force to the initial external load. Next, Diana uses the load steps that you specify to remove this addition. The external loading in step nstep reads
nstep 0 fext = fint + i=1 0 0 size i fext fint
(12.1)
The sum of explicitly specied step sizes must be equal to 1. The initial external load is dened as the nonlinear load set loset [ 12.3.1 p. 219]. If you dont specify START STEPS commands, then Diana will apply the internal forces, following from the initial stresses, at once during the rst step of the next EXECUT block. Note that Diana internally creates an additional load set for the load vector 0 0 equal to fext fint . This load set is assigned a load set number which is one higher than the highest userdened load set number. This additional load set number will appear in logging and output. syntax BEGIN STEPS [ OFF ] [ RESTOR step n ] [ method w ] EXPLIC ITERAT ENERGY AUTOMA
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IV)
12.3 Step Execution [ SAVE ] END STEPS RESTOR restores data for step step which must have been saved previously [ 12.3.9 p. 251]. method indicates how the step sizes are to be chosen. Options and syntax are analogous to load step control. EXPLIC explicitly specied step sizes [ 12.3.2.1 p. 227]. ITERAT iteration based step size control [ 12.3.2.2 p. 227]. ENERGY energy based step size control [ 12.3.2.3 p. 229]. AUTOMA automatic step size control [ 12.3.2.5 p. 233]. SAVE saves data of specied steps for future restart [ 12.3.9 p. 251]. Default
*NONLIN [ commands ] BEGIN EXECUT START STEPS [ commands ] END EXECUT *END file .dcf
[EXPLIC]
225
Due to these commands Diana will incrementally adapt the initial conditions in one step of size 1, which is equivalent to the following commands:
file .dcf *NONLIN [ commands ] BEGIN EXECUT BEGIN START STEPS EXPLIC SIZES 1.0 END START [ commands ] END EXECUT *END
12.3.2
Load Steps
226
Nonlinear Analysis syntax BEGIN LOAD [ OFF ] [ LOADNR=loset n ] [ BEGIN STEPS [ RESTOR step n ] [ method w ] EXPLIC ITERAT ENERGY AUTOMA [ SAVE ] END STEPS ] END LOAD LOADNR loset refers to a load set from table LOADS [ 2.3.7 p. 55]. Default is the lowest available load set number.
[STEPS]
STEPS species how to apply step sizes. Via method you may specify the step sizes explicitly, or you may let Diana determine them automatically. RESTOR restores data for step step which must have been saved previously [ 12.3.9 p. 251].
[EXPLIC]
method indicates how the step sizes are to be chosen. EXPLIC ITERAT ENERGY AUTOMA explicitly specied step sizes [ 12.3.2.1 p. 227]. iteration based automatic step size control [ 12.3.2.2 p. 227]. energy based automatic step size control [ 12.3.2.3 p. 229]. automatic step size control [ 12.3.2.5 p. 233].
SAVE saves data of specied steps for future restart [ 12.3.9 p. 251]. In case time dependent material or geometrical properties are used, the current time is taken into account. The current time is zero by default [ 12.3.1 p. 219]. Default
*NONLIN [ commands ] EXECUT LOAD *END file .dcf
Due to these commands Diana will execute a load step with load set 1 and a factor of 1, applying the default integration scheme and solution procedure, but no stop criteria. The same would occur if you had given the following commands:
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IV)
227
12.3.2.1
Via the EXPLIC commands you may specify load step sizes explicitly. syntax BEGIN EXPLIC [ OFF ] SIZES sizes r... [ ARCLEN ] END EXPLIC SIZES sizes are explicitly specied load step sizes. The number of values species the number of steps to be executed. ARCLEN applies Arc-length control [ 12.3.2.4 p. 230]. 12.3.2.2 Iteration Based Adaptive Loading
The ITERAT commands cause automatic adaptive load increments, based on the number of iterations. syntax BEGIN ITERAT [ OFF ] [ INISIZ=is r ] [ CONTIN ] [ NITERA=nitr n ] [ MAXSIZ=maxs r ] [ MINSIZ=mins r ]
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
Nonlinear Analysis
INISIZ=is is the initial size for the rst step. CONTIN asks Diana to ignore the initial step size is and to continue with the step size of the previous EXECUTE block.
NITERA=ni is the number of iterations (should be considered as optimal). MAXSIZ=maxs is the upper limit of the step size. MINSIZ=mins is the lower limit of the step size. Diana applies the specied limits in the zero-iteration of each step to determine a rst estimation of the step size.
NSTEPS=ns is the number of steps [ 30.1.5.3 p. 495]. GAMMA=gam species the exponent from Equation (30.33) on page 496 which we can now write as
t+t
0 =
l uT u0 0
ni tN
(12.2)
with t N the number of iterations from the previous step, this results in t+t 0 the size increment for the current step. lodunl species a loadingunloading method. In the Adaptive Arc-length methods, loadingunloading can be determined with various methods. See 30.1.5.3 on page 496 for background theory. SIGN for automatic loadingunloading for the incremental step sizes similar to Spherical Path Arc-length control. This method may only be used in combination with Arc-length control. The Updated Normal Plane method is recommended in this case [ 12.3.2.4]. PIVOTS for loadingunloading depending on existence of negative pivots in the triangulated stiness matrix. This method may only be used in combination with a NewtonRaphson iteration process [ 12.3.5.1 p. 241]. This method cannot be used in combination with an iterative solution procedure [ 26.4 p. 431]. ARCLEN applies Arc-length control [ 12.3.2.4].
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IV)
[SIGN]
229
Due to the ENERGY commands Diana will apply automatic adaptive load increments, based on energy. syntax BEGIN ENERGY [ OFF ] [ INISIZ=is r ] [ CONTIN ] [ MAXSIZ=maxs r ] [ MINSIZ=mins r ] [ NSTEPS=ns n ] [ lodunl w ] SIGN PIVOTS [ ARCLEN ] END ENERGY INISIZ=is is the initial size for the rst step. CONTIN asks Diana to ignore the initial step size is and to continue with the step size of the previous EXECUTE block. MAXSIZ=maxs is the upper limit of the step size. MINSIZ=mins is the lower limit of the step size. Diana applies the specied limits in the zero-iteration of each step to determine a rst estimation of the step size. NSTEPS=ns is the number of steps [ 30.1.5.3 p. 495]. lodunl species a loadingunloading method. In the Adaptive Arc-length methods, loadingunloading can be determined with various methods, see 30.1.5.3 on page 496 for background theory. SIGN for automatic loadingunloading for the incremental step sizes similar to Spherical Path Arc-length control. This method may only be used in combination with Arc-length control. The Updated Normal Plane method is recommended in this case [ 12.3.2.4 p. 230]. PIVOTS for loadingunloading depending on existence of negative pivots in the triangulated stiness matrix. This method may only be used in combination with a NewtonRaphson iteration process [ 12.3.5.1 p. 241]. This method cannot be used in combination with an iterative solution procedure [ 26.4 p. 431]. ARCLEN applies Arc-length control [ 12.3.2.4 p. 230].
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
[is =1.]
[ns =1]
[SIGN]
Nonlinear Analysis
The optional ARCLEN commands specify a selection of displacements to be accounted for in Arc-length iteration methods. This option is called Indirect Displacement control and may be useful in case of local snap-through or snapback behavior. Arc-length methods may only be used in combination with nodal or element loads, not with prescribed non-zero displacements or with time steps. Arc-length methods fail if the loading is dominantly nonconservative. Arc-length methods adapt the loading during iterations in one load step. For adaptive loading in consecutive load steps use the adaptive loading options of the STEPS command [ 12.3.2 p. 225]. See 30.1.5.2 on page 494 for background theory. syntax BEGIN ARCLEN [ OFF ] [ method w ] SPHERI UPDATE [ control w ] REGULA CMOD END ARCLEN
[UPDATE]
method indicates the Arc-length method to be applied: SPHERI for the Spherical Path Arc-length method [ 30.1.5.2 p. 493], UPDATE for the Updated Normal Plane Arc-length method [ 30.1.5.2 p. 494]. control selects a set of degrees of freedom to be applied in Arc-length control: REGULA for regular indirect displacement control, CMOD for CMOD control. Indirect Displacement Control BEGIN REGULA BEGIN SET NODES nodes ng... TYPE type w DIRECT dir n ALPHA alpha r END SET END REGULA In regular indirect displacement Arc-length control the SET block denes a set of degrees of freedom [Eq. (30.31) p. 494]. You may specify multiple SET blocks. syntax
[REGULA]
12.3 Step Execution NODES nodes is a series of nodes, specied by numbers and/or groups. TYPE type species a degree of freedom type: TRANSL for translation or ROTATI for rotation. DIRECT dirnr is a direction number referring to table DIRECT. ALPHA alpha and the direction determine the weight factors .
file .dcf *NONLIN [ commands ] BEGIN EXECUTE BEGIN LOAD LOADNR=2 BEGIN STEPS BEGIN EXPLIC SIZES 0.5 0.3 0.1 BEGIN ARCLEN UPDATE BEGIN REGULA BEGIN SET NODES 1 3 5 7-13 TYPE TRANSL DIRECT 5 ALPHA 1.0 END SET END REGULA END ARCLEN END EXPLIC END STEPS END EXECUTE *END
231
CMOD control BEGIN CMOD BEGIN SET BEGIN SIDE1 NODES nodes1 ng... TYPE type1 w DIRECT dir1 n ALPHA alpha1 r END SIDE1 BEGIN SIDE2
Diana-9.4 Users Manual Analysis Procedures (IV)
syntax
232 NODES nodes2 ng... TYPE type2 w DIRECT dir2 n ALPHA alpha2 r END SIDE2 END SET [ BEGIN SET END SET . . . ] END CMOD
Nonlinear Analysis
In CMOD Arc-length control the SET block denes related pairs of degrees of freedom [Eq. (30.32) p. 495]. You may specify multiple SET blocks, one for each crack mouth. SIDE1 denes the degrees of freedom along the rst side of the crack mouth. NODES nodes1 is a series of node numbers. TYPE type1 species a degree of freedom type: TRANSL for translation or ROTATI for rotation. DIRECT dir1 is a direction number referring to table DIRECT. ALPHA alpha1 and the direction determine the weight factors 1 for each of the degrees of freedom. SIDE2 denes the degrees of freedom along the second side of the crack mouth. NODES nodes2 is a series of node numbers. There must be just as much nodes along both sides of the crack mouth. TYPE type2 species a degree of freedom type: TRANSL for translation or ROTATI for rotation. DIRECT dir2 is a direction number referring to table DIRECT. ALPHA alpha2 and the direction determine the weight factors 2 for each of the degrees of freedom.
file .dcf *NONLIN [ commands ] BEGIN EXECUTE BEGIN LOAD LOADNR=2 BEGIN STEPS BEGIN EXPLIC SIZES 0.5 0.3 0.1
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IV)
233
In the above example cmod(1) is 1 the translation in direction 1 of node 7 minus 1 the translation in direction 1 of node 74, cmod(2) is 0.8 the translation in direction 2 of node 9 minus 0.9 the translation in direction 1 of node 75. 12.3.2.5 Automatic Adaptive Loading
The AUTOMA command causes automatic adaptive load increments. See 30.1.5.4 on page 497 for background theory.
234
Nonlinear Analysis syntax BEGIN AUTOMA [ OFF ] [ SIZE=size r ] [ MINSIZ=mins r ] [ MAXSIZ=maxs r ] [ CUTBCK=cutb r ] END AUTOMA
[size =1]
SIZE=size is the total load increment. MINSIZ=mins is the lower limit of the step size, relative to size . MAXSIZ=maxs is the upper limit of the step size, relative to size . CUTBCK=maxs is the factor by which the step size is scaled down in case of non-convergence.
[mins =10
[maxs =1]
[cutb =0.25]
12.3.3
Time Steps
syntax
BEGIN TIME [ OFF ] [ BEGIN STEPS [ RESTOR step n ] [ method w ] EXPLIC ITERAT EXPONE AUTOMA [ SAVE ] END STEPS ] END TIME
[STEPS]
STEPS species how to apply step sizes. RESTOR restores data for step step which must have been saved previously [ 12.3.9 p. 251].
[EXPLIC]
method indicates how the step sizes are to be chosen. EXPLIC explicitly specied step sizes [ 12.3.3.1 p. 235]. ITERAT iteration based step size control [ 12.3.3.2 p. 235].
12.3 Step Execution EXPONE applies exponential time increments [ 12.3.3.3 p. 236]. AUTOMA applies automatic time increments [ 12.3.3.4 p. 237]. SAVE saves data of specied steps for future restart [ 12.3.9 p. 251]. Default
*NONLIN [ commands ] EXECUT TIME *END file .dcf
235
Due to these commands Diana will execute a time step with a factor of 1, applying the default integration scheme and solution procedure, but no stop criteria. The same would occur if you had given the following commands:
file .dcf *NONLIN [ commands ] BEGIN EXECUT BEGIN TIME BEGIN STEPS EXPLIC SIZES 1.0 END STEPS END TIME END EXECUT *END
12.3.3.1
Via the EXPLIC commands you may specify step sizes explicitly. syntax BEGIN EXPLIC [ OFF ] [ SIZES sizes r... ] END EXPLIC SIZES sizes are explicitly specied time step sizes, i.e., time increments. The number of values species the number of steps to be executed. 12.3.3.2 Iteration Based Adaptive Time Increments
[SIZES 1.0]
The ITERAT commands cause automatic adaptive time increments, based on the number of iterations.
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
236
Nonlinear Analysis syntax BEGIN ITERAT [ OFF ] [ INISIZ=is r ] [ CONTIN ] [ NITERA=nitr n ] [ MAXSIZ=maxs r ] [ MINSIZ=mins r ] [ NSTEPS=ns n ] [ GAMMA=gam r ] END ITERAT
[is =1.]
INISIZ=is is the initial size for the rst step. CONTIN asks Diana to ignore the initial step size is and to continue with the step size of the previous EXECUTE block.
NITERA=ni is the number of iterations (should be considered as optimal). MAXSIZ=maxs is the upper limit of the step size. MINSIZ=mins is the lower limit of the step size. Diana applies the specied limits in the zero-iteration of each step to determine a rst estimation of the step size.
NSTEPS=ns is the number of steps [ 30.1.5.3 p. 495]. GAMMA=gam species the exponent from Equation (30.33) on page 496 which we can now write as
t+t
0 =
l uT u0 0
ni tN
ga
(12.3)
with t N the number of iterations from the previous step, this results in t+t 0 the size increment for the current step. 12.3.3.3 Exponential Time Increments
The EXPONE commands invoke an exponential time increment algorithm. With this option Diana determines the time increment ti for time step i from ti = f i t0 (12.4) where t0 is the initial time increment and f a multiplication factor. The initial time increment t0 and the factor f are related to the total time increment tinc by the power series t0 1 + f + f 2 + f 3 + + f n1 = t0
December 15, 2009 First ed.
fn 1 = tinc f 1
(12.5)
12.3 Step Execution with n the number of time steps. To determine the time increment ti for step i from (12.4) and (12.5) Diana uses some input parameters, of which the rst three are optional. syntax BEGIN EXPONE [ OFF ] [ ] DELTAT=dt0 r NSTEPS=ns n [ FACTOR=fac r ] [ TOTALT=tt r ] END EXPONE DELTAT=dt0 is the initial time increment t0 . If you specify this parameter, then Diana determines an initial time increment close to t0 that yields an integer number of steps n. NSTEPS=ns is the number of time steps n in the current EXECUT command block. FACTOR=fac is the multiplication factor f which Diana applies between two subsequent time steps. TOTALT=ti is the total time increment tinc for the current EXECUT command block. 12.3.3.4 Automatic Time Increments
237
The AUTOMA commands invoke an automatic time incremental algorithm [ 30.1.5.4 p. 498]. If the SDIRK2 time integration method is used, it is possible to obtain time accurate solutions, else it is equivalent to automatic load stepping [ 30.1.5.4 p. 497]. syntax BEGIN AUTOMA [ OFF ] [ SIZE=size r ] [ MINSIZ=mins r ] [ MAXSIZ=maxs r ] [ CUTBCK=cutb r ] }] [ SDIRK2 { TOLREL=epsrel r TOLABS=epsabs r
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
Nonlinear Analysis
SIZE=size is the total time increment. MINSIZ=mins is the lower limit of the step size, relative to size . MAXSIZ=maxs is the upper limit of the step size, relative to size . CUTBCK=maxs is the factor by which the step size is scaled down in case of non-convergence. SDIRK2 applies time integration error control according to the SDIRK2 method [ 30.1.5.4 p. 498]. This method only makes sense if the solution depends on the time step size (dynamics, viscoelastic and viscoplastic behavior). Two optional parameters customize the error criterion [Eq. (30.40) p. 498]:
[maxs =1]
[cutb =0.25]
[ [
rel abs
= 103 ] = 10
5
TOLREL=epsrel species the relative error criterion TOLABS=epsabs species the absolute error criterion
rel . abs .
12.3.4
The PHYSIC commands overrule the use of physical nonlinear phenomena as specied in the input data le. If the PHYSIC commands are absent then Diana uses all phenomena from the input data le. syntax BEGIN PHYSIC [ OFF ] [ BOND [ ]] REINFO reinfs ng... OFF [ SLIP [ ]] REINFO reinfs ng... OFF [ SUPPRE { }] DISPLA STRAIN STRESS OFF [ DRAINE [ OFF ] ] [ LIQUEF [ OFF ] ] END PHYSIC BOND use the bond option even for bond-less reinforcement. REINFO reinfs species a series of reinforcement numbers or groups or both for which the bond option must be switched on. This command overrules the NOBOND input option [Vol. Material Library].
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IV)
12.3 Step Execution OFF switches o the bond option for bond-less reinforcement, i.e., bonding depends on the presence of the NOBOND input option. SLIP use the slip option for bond-less reinforcement. REINFO reinfs species a series of reinforcement numbers or groups or both for which the slip option must be applied. The inuence of loading, creep and/or shrinkage may cause changes in reinforcement strains and stresses in the time period between prestressing and grouting. The slip option provides a rather coarse method to model these changes. Diana will calculate the mean value of the change in reinforcement strain for the whole reinforcement. This mean change of strain is then uniformly applied to the whole reinforcement. Eects of friction between the reinforcement and its mother elements are not taken into account. This immediately shows the limits to the applicability of this method. The slip option may be used to model a partly bonded reinforcement. Therefore you should identify the bonded and not bonded part of the reinforcement. Then you should replace the reinforcement with a series of bonded reinforcements and reinforcements without bonding, but with the slip option applied. OFF switches o the slip option for bond-less reinforcement. SUPPRE suppress the superposition of current step results. DISPLA for all degrees of freedom (translations, rotations, pressures, etc.). STRAIN for strain related degrees of freedom (translations, rotations, special spline element degrees of freedom) and the total strain itself. This option is particularly useful for mixture analysis, to initialize pressures whilst keeping displacements and strains at zero level. STRESS for stresses only. This option is particular useful in combination with the DEFORM IMPORT [ 24.1.3 p. 405] loading in a phased geotechnical analysis for suppressing stresses when a new soil layer is placed. DRAINE use drained behavior even for elements with basically undrained behavior. This command overrules the UNDRAI input option [Vol. Material Library]. OFF switches drained behavior o, i.e., applies undrained behavior. This is the default if you specify input item UNDRAI in table MATERI [Vol. Material Library]. Undrained behavior is particularly used during the loading of clay and peat. First the dead weight is applied in combination with drained behavior, followed by the additional load in combination with undrained behavior. LIQUEF OFF deactivates all liquefaction models during the execution of the steps in the current EXECUT block [Vol. Material Library]. Diana will use other constitutive models like (Modied) MohrCoulomb instead, provided that you have supplied the input for such a model in table MATERI.
Diana-9.4 Users Manual Analysis Procedures (IV)
239
240
Nonlinear Analysis
12.3.5
Equilibrium Iteration
The ITERAT commands dene the equilibrium iteration process to be used for steps. See 30.1.1 on page 480 for background theory. syntax BEGIN ITERAT [ MAXITE=mi n ] [ CONTIN [ OFF ] ] [ METHOD ] [ LINESE ] [ CONVER ] END ITERAT
[mi =10]
MAXITE= mi is the maximum number of iterations for each time or load step. CONTIN for iteration with the Continuation method [ 30.1.2 p. 486]. The OFF option switches o this method, which also is the default behavior. METHOD species the iteration method [ 12.3.5.1]. LINESE invokes a Line Search algorithm to scale the incremental displacements [ 12.3.5.2]. CONVER species the convergence criterion for the iteration process [ 12.3.5.3]. Default iteration process. If you dont give any ITERAT commands, then the default is a Regular NewtonRaphson method with a maximum of ve iterations. As convergence criteria, the norms of the force and displacement elds will be applied. This is equivalent to the following commands:
file .dcf *NONLIN [ commands ] BEGIN EXECUT [ commands ] BEGIN ITERAT MAXITE=5 METHOD NEWTON REGULA BEGIN CONVER FORCE NEWREF TERMIN TOLCON=1.E-2 TOLABT=1.E+4 DISPLA NEWREF TERMIN TOLCON=1.E-2 TOLABT=1.E+4 END CONVER END ITERAT [ commands ] END EXECUT *END
[OFF]
241
With the METHOD commands you may choose one out of various iteration methods: Constant or Linear Stiness, NewtonRaphson, or Quasi-Newton (Secant). syntax BEGIN METHOD [ ] CONSTA LINEAR NEWTON [ [ ]] REGULA TANGEN MODIFI LINEAR PREVIO SECANT [ [ ]] BROYDE TANGEN BFGS LINEAR CRISFI PREVIO END METHOD CONSTA invokes the Constant Stiness method [ 30.1.1.3 p. 486]. This command may also be used to freeze the tangent stiness matrix. LINEAR invokes the Linear Stiness method [ 30.1.1.3 p. 486]. This command initiates an explicit run to the linear elastic stiness, which is particularly attractive in case of sudden unloading. NEWTON invokes a NewtonRaphson iteration method [ 30.1.1.1 p. 482]. The default NewtonRaphson method is Regular. REGULA for Regular NewtonRaphson Diana sets up the tangential stiness before each iteration. MODIFI for Modied NewtonRaphson, Diana sets up the tangential stiness before the rst iteration and subsequently keeps it constant during the step. By default, the NewtonRaphson methods set up the tangential stiness before each iteration. An option species an alternative for the rst iteration: TANGEN sets up a new tangential stiness. LINEAR uses the linear stiness. PREVIO uses the stiness of the last iteration of the previous step. SECANT invokes a Quasi-Newton (Secant) iteration method [ 30.1.1.2 p. 484].
Diana-9.4 Users Manual Analysis Procedures (IV)
[REGULA]
[BFGS]
242 BROYDE Broydens method. BFGS The BFGS method (the default). CRISFI Criselds method.
[TANGEN]
Nonlinear Analysis
By default, the Quasi-Newton methods set up the tangential stiness at the start of each step. An option species an alternative start: TANGEN sets up a new tangential stiness. LINEAR uses the linear stiness. PREVIO uses the stiness of the last iteration of the previous step. 12.3.5.2 Line Search
Diana oers a Line Search algorithm to scale the incremental displacements in the iteration process automatically. As outlined in 30.1.3 on page 487, this may stabilize the convergence behavior or increase the convergence speed. Line Search cannot be applied in contact analysis [ 30.2.5 p. 505]. You may invoke a Line Search algorithm via the LINESE commands. The algorithm may be guided by one ore more optional parameters. Usually the default values of these parameters are satisfactory. syntax BEGIN LINESE [ OFF ] [ ETAMAX=maxsc r ] [ ETAMIN=minsc r ] [ PSI=psi r ] [ DETA=deta r ] [ MAXLS=mls n ] END LINESE ETAMAX=maxsc is the upper bound for the Line Search scaling factor max . The default is equivalent to no extrapolation. ETAMIN=minsc is the lower bound for the Line Search scaling factor min . PSI=psi is the stop criterion . Line Search stops if the energy change (residue displacement variation) is less than the original value. DETA=deta is the stop criterion . Line Search stops if the interval in a Regula-Falsi process is less than . MAXLS=mls is the maximum number of line searches in each global iteration.
Diana-9.4 Users Manual Analysis Procedures (IV)
[ = 0.1]
[mls =5]
243
Due to these commands Diana will apply a Line Search algorithm with default values for all parameters which is equivalent to the following commands:
file .dcf *NONLIN [ commands ] BEGIN EXECUT [ commands ] BEGIN ITERAT LINESE ETAMAX=1.0 ETAMIN=0.1 PSI=0.8 DETA=0.1 MAXLS=5 END ITERAT END EXECUT *END
12.3.5.3
Convergence Criteria
The CONVER commands specify the convergence criteria for the equilibrium iteration process. See 30.1.4 on page 488 for theoretical backgrounds. Multiple CONVER commands may be specied to use more than one criterion simultaneously. Diana terminates the iteration process if one of the specied criteria is satised. syntax BEGIN CONVER [ SIMULT ] } ... [ crit w ] [ OFF ] [ refer w ] [ noconv w ] { FORCE NEWREF TERMIN TOLCON=tc r DISPLA OLDREF CONTIN TOLABT=ta r ENERGY RESIDU END CONVER SIMULT will let Diana assume convergence when all specied criteria are satised simultaneously.
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
244
Nonlinear Analysis crit is the name of the criterion: FORCE based on the external loading and outof-balance forces (the default), DISPLA based on displacements, ENERGY based on internal energy, or RESIDU based on the out-of-balance force vector. OFF switches o the criterion crit . refer species whether or not Diana must determine a new reference norm: with option NEWREF a new norm is determined at the start of each step (the default), with option OLDREF no norm will be determined and the previously stored norm will be used. The reference norm must at least be determined once, before it is used. Only in cases where a new reference norm would result in an extremely high or low value, it can be advantageous to use the old reference norm. In other words: option OLDREF must be used with great care. noconv indicates what to do if no convergence occurs within the maximum number of iterations. With option TERMIN Diana stops the analysis run, no further steps will be executed (the default). However, further steps may be executed in subsequent runs. With option CONTIN the analysis run will be continued, you should check the relevance of the analysis results carefully in this case. Two parameters specify the tolerance values. The default values depend on the used criterion as indicated in the examples below. TOLCON=tc is the tolerance for convergence. When the norm has become less than tc the reference norm, Diana assumes sucient accuracy, stops the iteration process and continues with the next step. TOLABT=ta is the tolerance for divergence. When the norm exceeds a value of ta the reference norm, Diana assumes divergence and aborts the nonlinear analysis. Default criteria
[ commands ] BEGIN EXECUT [ commands ] BEGIN ITERAT CONVER END ITERAT END EXECUT *END file .dcf
[FORCE] [DISPLA]
[NEWREF]
[TERMIN]
245
If, like in the above commands, you only give the CONVER, or if you omit this command completely, then Diana will simultaneously apply the criteria based on external loading and on displacements. This is equivalent to the following commands:
file .dcf [ commands ] BEGIN EXECUT [ commands ] BEGIN ITERAT CONVER FORCE DISPLA END ITERAT END EXECUT *END
Default tolerances
*NONLIN [ commands ] BEGIN EXECUT [ commands ] BEGIN ITERAT CONVER FORCE DISPLA ENERGY END ITERAT END EXECUT *END
file .dcf
If, like in the above commands, you dont specify parameters for the tolerances, then the default values depend on the type of the criterion as shown below.
file .dcf *NONLIN [ commands ] BEGIN EXECUT [ commands ] BEGIN ITERAT BEGIN CONVER FORCE NEWREF TERMIN TOLCON=1.E-2 TOLABT=1.E+4 DISPLA NEWREF TERMIN TOLCON=1.E-2 TOLABT=1.E+4 ENERGY NEWREF TERMIN TOLCON=1.E-4 TOLABT=1.E+4 END CONVER END ITERAT END EXECUT *END
246
Nonlinear Analysis
12.3.6
If you specify a stop criterion via a STOP command, then Diana will stop the execution of the load or time steps in the current EXECUT command block. The analysis can be continued in a subsequent EXECUT command block. You may specify a stop criterion based on loading, or based on resulting strain or stress. By default, Diana will not apply a stop criterion. syntax BEGIN STOP [ OFF ] [ LOAD ] [ RESULT ] END STOP LOAD species stop criteria based on loading [ 12.3.6.1]. RESULT species stop criteria based on results [ 12.3.6.2]. Multiple criteria
*NONLIN [ commands ] BEGIN EXECUT [ commands ] BEGIN STOP BEGIN LOAD TOTAL 3.567 INCREM 0.27 END LOAD BEGIN RESULT STRAIN PRINCI INTPNT MINVAL=-3.5E-3 END RESULT END STOP END EXECUT *END file .dcf
If, like in the above commands, you specify multiple STOP commands then Diana will use the criteria simultaneously. 12.3.6.1 Loading Based Stop
Criteria based on loading are useful in combination with Arc-length methods [ 12.3.2.4 p. 230] or adaptive loading [ 12.3.3.2 p. 235]. If a loading based stop criterion is specied then Diana takes the number of steps from the SIZES command as a maximum number of steps. The stop criterion will stop the execution of the current EXECUTE command block. The analysis will continue with the next block (if any).
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IV)
12.3 Step Execution syntax BEGIN LOAD [ OFF ] { } TOTAL totlod r INCREM inclod r SIGN END LOAD TOTAL stops execution of steps when a total load totlod is reached. INCREM stops execution of steps when the incremental load is less than inclod . SIGN stops execution of steps when the sign of the load vector changes. This criterion is for use in combination with automatic loadingunloading or loadingunloading based on negative pivots [ 12.3.2.3 p. 229]. 12.3.6.2 Result Based Stop
247
With a stop criterion based on analysis results, Diana will stop the execution of the steps in the current EXECUT command block if the value of certain strains or stresses exceeds a specied extreme. syntax BEGIN RESULT [ OFF ] [ BEGIN SELECT END SELECT ] item w [oper w ] [comp w ] [loca w ] STRAIN MAXVAL=max r STRESS MINVAL=min r END RESULT
SELECT selects certain parts of the model for which the criterion for resulting strain or stress must be applied. For complete syntax description see 3.6.1 on page 71. By default Diana applies the specied results criterion on all elements of the model. STRAIN applies a result stop criterion based on the total GreenLagrange strain [Fig. 12.1a].2
2 The
248
Nonlinear Analysis
stop max min stop min (a) total strain stop max (b) total stress stop
Figure 12.1: Result based stop criteria STRESS applies a result stop criterion based on the total Cauchy stress [Fig. 12.1b].3 The speciers for operation (transformation), component selection and location, oper , comp and loca , are described in 12.4.2 on page 255 for strains and in 12.4.3 on page 260 for stresses. Be sure to select a result which is available in the model selection, or else the stop criterion will be void. MAXVAL max is the maximum value that the strain or stress may reach before Diana stops execution of steps. MINVAL min is the minimum value that the strain or stress may reach before Diana stops execution of steps.
file .dcf *NONLIN [ commands ] BEGIN EXECUTE [ commands ] BEGIN STOP BEGIN RESULT SELECT ELEMEN CONCRETE STRAIN PRINCI INTPNT MINVAL=-3.5E-3 END RESULT END STOP END EXECUTE *END
3 The
249
Due to the commands in the example above Diana will apply a stop criterion based on a maximum principal strain of 3.5103 in the integration points of elements in group CONCRETE. When the criterion is satised, Diana will terminate the analysis job and issue the following message on the standard output le:
file .out
EXECUTION STOPPED ON STRAIN TOTAL GREEN PRINCI INTPNT CRITERION IN ELEMENT 324 THE VALUE IS -0.3635E-02; CRITERION IS -0.3500E-02 CONTINUED ANALYSIS POSSIBLE
12.3.7
Logging
You may ask Diana to produce a certain amount of log information during step execution via the LOGGIN commands. syntax BEGIN LOGGIN [ OFF ] [ REPORT [ verbo w ] [ when w ] ] BRIEF EXECUT FULL STEP ITERAT [ PLASTI [ OFF ] ] [ CRACKI [ OFF ] ] END LOGGIN OFF turns logging o, i.e., Diana will not give any log information. When logging is on (the default), you may customize the logging information via the following commands: REPORT indicates what to be logged and when. verbo indicates the verbosity, i.e., the amount of log information to be produced: BRIEF gives a brief summary, FULL gives full log information including a summary. when indicates when log information will be produced: EXECUT at the termination of the last step, STEP at the termination of each of the executed steps, or ITERAT at the termination of each iteration. PLASTI turns the logging of the plastic points in the model on or o. CRACKI turns the logging of the cracks in the model on or o.
Diana-9.4 Users Manual Analysis Procedures (IV)
[ON]
[BRIEF]
[STEP]
[ON] [ON]
250 Default
*NONLIN [ commands ] BEGIN EXECUT [ commands ] END EXECUT *END
Nonlinear Analysis
file .dcf
If, like in the above example, you dont specify LOGGIN commands then Diana will briey log the occurrence of plasticity and cracking after each load step. This is also achieved via the following commands:
file .dcf *NONLIN [ commands ] BEGIN EXECUT [ commands ] BEGIN LOGGIN REPORT BRIEF STEP PLASTI CRACKI END LOGGIN END EXECUT *END
Comprehensive
*NONLIN [ commands ] BEGIN EXECUT [ commands ] BEGIN LOGGIN REPORT FULL ITERAT PLASTI CRACKI END LOGGIN END EXECUT *END
file .dcf
These commands ask for the most comprehensive logging: full information of plasticity and cracking, including a summary, at the end of each iteration.
12.3.8
In geotechnical analyses, the initial state may be set during initialization of the nonlinear analysis by means of the REFERE commands. Due to these commands, Diana will evaluate the initial state at the end of the current EXECUT block, after the postprocessing of the last step of this block.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IV)
12.3 Step Execution syntax BEGIN REFERE [ OFF ] } { WEIGHT PRESSU END REFERE WEIGHT causes initial state procedures, like application of the lateral pressure ratio K0 , to be applied using the results of the last step. See the section on input of the initial stress ratio in Volume Material Library, If the initial state is a result of some nonlinear steps, you may use the WEIGHT command to let Diana (re)apply initial state procedures. PRESSU applies the results of the last step to compute the initial values used for computation of the excess pore pressure ratio rp [ 12.4.3 p. 262] and the excess pore pressure. See also Part VII on owstress analysis. Default
*NONLIN [ commands ] BEGIN EXECUT [ commands ] REFERE END EXECUT *END file .dcf
[PRESSU]
251
These commands cause the last step of the current EXECUT block to be selected as the (new) reference for the determination of the initial geotechnical state. The same would occur if you had given the following commands:
file .dcf *NONLIN [ commands ] BEGIN EXECUT [ commands ] REFERE PRESSU END EXECUT *END
12.3.9
Save/Restore Steps
With the SAVE and RESTOR commands you may save or restore executed steps.
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
252
Nonlinear Analysis syntax [ RESTOR step n ] [ SAVE [ ]] steps n... ALL LAST CONVER RESTOR restarts the analysis starting from a previously saved step, where step is an absolute step number. This step must have been saved previously via the SAVE command.
[ALL]
SAVE indicates to save steps. steps are absolute step numbers of the steps to be saved. ALL will save all steps that are executed. LAST will save the last step executed. CONVER will save the last converged step.
12.4
You may indicate the analysis results to be output via the OUTPUT command block. Depending on the level where this block is specied, the selection is valid for all executed steps [Ch. 12 p. 209], or for the steps of the current EXECUT block [ 12.3 p. 219]. For the general, analysis type independent options for output of analysis results see 3.6 on page 69. syntax BEGIN OUTPUT [device w ] [outopt w... ] [params ] [ OFF ] [ BEGIN SELECT [ ] [ STEPS ] END SELECT ] [ LAYOUT ] item w . . . DISPLA VELOCI ACCELE STRAIN STRESS STATUS
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IV)
12.4 Output of Analysis Results PRESSU TEMPER CONCEN MATURI FORCE NODFOR ELMFOR PARAME FRACTU FSPRES END OUTPUT SELECT command block to customize the batch output. for model selection see 3.6.1 on page 71, for stress- and strain transformation 3.6.1.2 on page 73. STEPS selects load or time steps for output [ 12.4.1 p. 254]. LAYOUT optional commands to customize the layout and style of tabular output [ 3.6.3 p. 80]. item is the name of the analysis result to be output. See 3.6.2 on page 77 for complete syntax of this command. DISPLA for displacements, syntax and output is analogous to linear transient analysis [ 7.4.1 p. 134]. VELOCI for velocities, syntax and output is analogous to linear transient analysis [ 7.4.2 p. 135]. ACCELE for accelerations, syntax and output is analogous to linear transient analysis [ 7.4.3 p. 137]. STRAIN for strains [ 12.4.2 p. 255]. STRESS for stresses [ 12.4.3 p. 260]. STATUS for status [ 12.4.4 p. 265]. PRESSU for pressure [ 12.4.5 p. 269]. TEMPER for temperature [ 12.4.6 p. 270]. CONCEN for concentration [ 12.4.6 p. 270]. MATURI for maturity [ 12.4.6 p. 270]. FORCE for nodal forces and moments [ 12.4.7 p. 271]. NODFOR for element nodal forces [ 12.4.8 p. 272]. ELMFOR for internal element forces [ 12.4.9]. PARAME for model parameters [ 12.4.10 p. 274]. FRACTU for Linear Elastic Fracture Mechanics analysis parameters of crack tip elements [ 12.4.11]. FSPRES for dynamic pressures of uidstructure interface elements [ 12.4.12].
Diana-9.4 Users Manual Analysis Procedures (IV)
253
254
Nonlinear Analysis
file .dcf *NONLIN [ commands ] BEGIN EXECUTE [ commands ] BEGIN OUTPUT TABULA SELECT STEPS 2 4 6 MAX DISPLA TOTAL TRANSL GLOBAL END OUTPUT END EXECUTE *END
TRANSL Stepnr Stepnr Stepnr TDtZ 0.000E+00 ( 0.000E+00 ( 0.000E+00 ( 0.000E+00 ( 0.000E+00 ( 0.000E+00 ( 0.000E+00 ( 0.000E+00 ( 0.000E+00 ( 44 92 1 1) 1) 1) 1) 1) 1) 1) 1) 1) Nodnr Nodnr Nodnr 6 9 1
Nodnr 1 2 3 4 5 6 7 8 9
TDtX TDtY 1.349E-02 (100) 0.000E+00 ( 1) 1.349E-02 (100) 0.000E+00 ( 1) 1.349E-02 (100) 0.000E+00 ( 1) 1.288E-02 ( 99) -7.838E-05 ( 1) 8.943E-03 (100) -7.838E-05 ( 1) 2.932E-02 ( 44) 3.531E-03 ( 93) 2.108E-03 ( 20) -5.283E-05 ( 1) 6.586E-04 ( 39) 5.793E-03 ( 94) 1.422E-02 ( 45) 9.231E-03 ( 92)
This tabular output shows a maximum value for displacement component uX at node 6 to be 0.02932 for step number 44. The maximum value for uX at node 1 is 0.01349 for step 100. Notice that Diana writes the maximum value for all selected output points just below the header paragraph.
12.4.1
Step Selection
syntax
]{ MIN MAX
255
ALL will produce output for all steps that are executed (the default). LAST selects the nal step. The interpretation of the above options depends on the position of the current OUTPUT block. At the *NONLIN command level [Ch. 12 p. 209], steps are absolute step numbers, and ALL/LAST refer to all steps/the last step of the current analysis. At the EXECUT command level [ 12.3 p. 219], steps are relative step numbers, and ALL/LAST refer to all steps/the last step of the current EXECUT block. Diana will output the extreme values that occurred up to and including the selected steps if you specify one of the options: MIN for the minimum values, MAX for the maximum values.
12.4.2
Strains
[ form w ] GREEN FORCE MOMENT TRACTI DISSEI VOID [ oper w ] { comp w } { loca w } { opti w } LOCAL INTPNT AVERAG GLOBAL NODES PRINCI STRPNT VONMIS CROSSE REAXES CENTER VOLUME
syntax
STRAIN [ type w ] TOTAL ELASTI PLASTI CREEP CRACK CRKSUM TEMPER CONCEN SHRINK
STRAIN species strains as output item. Table 12.2 on the following page outlines the availability and applicability of the various strain output options for each of the element families. type species the strain type. TOTAL for total strains. ELASTI for elastic strains. PLASTI for plastic strains. CREEP for creep strains. CRACK for crack strains in each crack individually. CRKSUM for summed crack strains over all cracks. TEMPER for thermal strains. CONCEN for concentration strains.
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
[TOTAL]
256
Nonlinear Analysis
am be
item
type
form
oper
loca
opti
STRAIN TOTAL ELASTI PLASTI CREEP CRACK CRKSUM TEMPER CONCEN SHRINK GREEN FORCE MOMENT TRACTI DISSEI VOID LOCAL GLOBAL PRINCI VONMIS REAXES VOLUME INTPNT NODES STRPNT CROSSE CENTER AVERAG
us
a a a a ad a a a a a a a a a a a a a a
a a a a ad a a a a a c c a a a a a a a a a a a
pl
tr
a a a a ad a a a a a a a a a a a a a a a
.s
a a a a ad a a a a a a a a a a a a a a a
a a a a ad a a a a a a a a a a a a a a a
ae a a a a ae a a a a -
ae ae a a a a ae a a a a -
a a a a ad a a a a a a a a a a a a a a a
a a a a ad a a a a a a a a a a a a a a a a
a a f a a a -
a a a g g -
a j j j a h i h a h j h -
(a) All elements. (b) Not for class-I. (c) For all beam elements, classII and classIII only in combination with local coordinate system, results are derived from STRESS FORCE and STRESS MOMENT and translated assuming linear material behavior in the beam elements. (d) Only in integration points. (e) Only linear elastic. (f) Only for plane structural interface elements. (g) Only for base springs. (h) Only for bondslip reinforcements. (i) Only for bondslip reinforcements modeled by beam elements. (j) Not for bondslip reinforcements. (-) Not available or not suitable.
SHRINK for shrinkage strains. form species the strain formulation. GREEN for GreenLagrange strains [ 12.4.2.1]. FORCE for deformations due to normal and shear forces [ 12.4.2.2]. MOMENT for curvatures due to bending moments [ 12.4.2.2]. TRACTI for tractions in structural interface elements [ 12.4.2.3]. DISSEI for distributed seismic moments in plane structural interface elements [ 12.4.2.4].
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IV)
[GREEN]
12.4 Output of Analysis Results VOID for void-ratio [ 12.4.2.5]. oper species an operation to be performed on the primary strains [ 3.6.2 p. 79].
[GLOBAL]
257
comp selects strain components for output. Default is all available components. loca species the location for the strains to be output [ 3.6.2 p. 79]. opti are additional options. AVERAG is specic for nonlinear analysis with the plasticity fraction model [Vol. Material Library], it gives the average value of all the fractions of the fraction model. see 3.6.2 on page 79 for other options. 12.4.2.1 GreenLagrange Strains
comp XX YY ZZ
Exx Eyy
Total strains item type form oper STRAIN TOTAL GREEN LOCAL STRAIN TOTAL STRAIN TOTAL STRAIN TOTAL STRAIN TOTAL
XY
YZ
Gyz
ZX
Gzx
Ezz Gxy
xx yy zz xy yz zx GREEN GLOBAL EXX EYY EZZ GXY GYZ GZX XX Y Y ZZ XY Y Z ZX GREEN PRINCI E1 1 GREEN REAXES E1RA a 1 GREEN VOLUME Evol vol comp XX YY
E2
E3
2
E2RA
a 2
Primary elastic strains item type form oper STRAIN ELASTI GREEN LOCAL
ZZ
XY
YZ
ZX
e e e e e e xy yz zx xx yy zz STRAIN ELASTI GREEN GLOBAL EeXX EeYY EeZZ GeXY GeYZ GeZX e e e e e e XX Y Y ZZ XY Y Z ZX STRAIN ELASTI GREEN PRINCI Ee1 e 1 STRAIN ELASTI GREEN REAXES Ee1RA ea 1 STRAIN ELASTI GREEN VOLUME Eevol e vol
Ee2
Ee3
e 2
Ee2RA
e 3
ea 2
258
Primary plastic strains item type form oper STRAIN PLASTI GREEN LOCAL comp XX YY p xx p yy
Nonlinear Analysis
ZZ p zz
XY
p xy
YZ
p yz
ZX
p zx
STRAIN PLASTI GREEN GLOBAL EpXX EpYY EpZZ GpXY GpYZ GpZX p p p p p p XX Y Y ZZ XY Y Z ZX STRAIN PLASTI GREEN PRINCI Ep1 p 1 STRAIN PLASTI GREEN REAXES Ep1RA pa 1 STRAIN PLASTI GREEN VOLUME Epvol p vol Primary creep strains item type form oper STRAIN CREEP GREEN LOCAL STRAIN CREEP STRAIN CREEP STRAIN CREEP STRAIN CREEP comp XX YY c xx c yy
Ep2
Ep3
p 2
Ep2RA
p 3
pa 2
ZZ c zz
XY
c xy
YZ
c yz
ZX
c zx
GREEN GLOBAL EcXX EcYY EcZZ GcXY GcYZ GcZX c c c c c c XX Y Y ZZ XY Y Z ZX GREEN PRINCI Ec1 c 1 GREEN REAXES Ec1RA ca 1 GREEN VOLUME Ecvol c vol
Ec2
Ec3
c 2
Ec2RA
c 3
ca 2
Von Mises strains item type form oper STRAIN TOTAL GREEN VONMIS Eeq eq STRAIN ELASTI GREEN VONMIS Eeeq e eq STRAIN PLASTI GREEN VONMIS Epeq p eq STRAIN CREEP GREEN VONMIS Eceq c eq Crack strains comp NN NS NT item type form loca STRAIN CRACK GREEN INTPNT Eknn Gkns Gknt cr cr cr ns nt nn
259
ZZ k zz
XY
k xy
YZ
k yz
ZX
k zx
STRAIN CRKSUM GREEN GLOBAL EkXX EkYY EkZZ GkXY GkYZ GkZX k k k k k k XX Y Y ZZ XY Y Z ZX STRAIN CRKSUM GREEN PRINCI Ek1 Ek2 Ek3 k k k 1 2 3 STRAIN CRKSUM GREEN VONMIS Ekeq k eq Thermal strains item type form oper STRAIN TEMPER GREEN LOCAL comp XX YY T xx T yy
ZZ T zz
XY
T xy
YZ
T yz
ZX
T zx
STRAIN TEMPER GREEN GLOBAL EtmpXX EtmpYY EtmpZZ GtmpXY GtmpYZ GtmpZX T T T T T Y T XY Y Z ZX XX Y ZZ Concentration strains item type form oper STRAIN CONCEN GREEN LOCAL comp XX YY C xx C yy
ZZ C zz
XY
C xy
YZ
C yz
ZX
C zx
STRAIN CONCEN GREEN GLOBAL EconXX EconYY EconZZ GconXY GconYZ GconZX C C C C C Y C XY Y Z ZX XX Y ZZ Shrinkage strains item type form STRAIN SHRINK GREEN Eshr sh
12.4.2.2
Force deformations item type form oper STRAIN TOTAL FORCE LOCAL STRAIN TOTAL FORCE
Z
Pz
ux uy uz GLOBAL PX PY PZ uX uY uZ comp X Y Z
Kx Ky Kz
Nonlinear Analysis
Traction deformations item type form oper STRAIN TOTAL TRACTI LOCAL
Z
PTz
12.4.2.4
Distributed seismic moment [ 31.1.4] item type form STRAIN TOTAL DISSEI PSTOT PSNEG PSPOS PS PS PS+
12.4.2.5
Void Ratio
The void ratio can be output for elements with the Modied MohrCoulomb or Cam-clay material models [Vol. Material Library].
Void ratio item type form STRAIN TOTAL VOID VOID e
12.4.3
Stresses
[ oper w ] { comp w } { loca w } { opti w } LOCAL INTPNT NOBOND GLOBAL NODES PRINCI STRPNT VONMIS CROSSE INVARI CENTER REAXES ERATIO CRKIND BIAXFE CAPACI
syntax
STRESS [ type w ] [ form w ] TOTAL CAUCHY CRACK PIOLAK FORCE MOMENT DISFOR DISMOM TRACTI GRADIE
STRESS species stresses as output item. Table 12.3 on the facing page outlines the availability and applicability of the various strain output options for each of the element families.
[TOTAL]
type
261
us s be am
item type
form
oper
loca
opti
STRESS TOTAL CRACK CAUCHY PIOLAK FORCE MOMENT DISFOR DISMOM TRACTI GRADIE LOCAL GLOBAL PRINCI VONMIS INVARI REAXES ERATIO CRKIND BIAXFE CAPACI INTPNT NODES STRPNT CROSSE CENTER NOBOND
pl
pl
.s
.s
a a a a a a a a a al a a a a -
a a a a a a a a a a hl a a a a a a a
a a a a a a a a a a a al a a a a a -
a a a a a a a a a a al a a a a a -
a a a a a a a a a a al a a a a -
pl
tr
ym
tr
ae ae ae ae a a a a ae a a a -
ae ae ae ae a a a a ae a a a -
a a a a a a a a a a a al a a a a a
a a a a b b a a a a a a a al a a a a a -
a a a a a -
a a a f a f a -
a a a -
a a a j i c a i k i -
(a) All elements. (b) For composed solids only [Vol. Element Library]. (c) Only for bar reinforcements. (e) Only linear elastic. (f) Only for base springs. (h) Only for class II and class III beams. (i) Only for bondslip reinforcements. (j) Only for bondslip reinforcements modeled by beam elements. (k) Not for bond slip reinforcements. (l) Only linear elastic and viscoelastic. (-) Not available or not suitable.
TOTAL for total stresses. CRACK for crack stresses. form species the stress formulation. CAUCHY for Cauchy stresses [ 12.4.3.1]. PIOLAK for PiolaKirchho stresses [ 12.4.3.2]. FORCE for concentrated force or tractions [ 12.4.3.3]. MOMENT for concentrated moment [ 12.4.3.4]. DISFOR for distributed force [ 12.4.3.3]. DISMOM for distributed moment [ 12.4.3.4].
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
[CAUCHY]
262
Nonlinear Analysis TRACTI for tractions in structural interface and contact elements [ 12.4.3.5]. GRADIE for gradients of stresses in reinforcement bars [ 12.4.3.6].
[GLOBAL]
oper
species an operation to be performed on the primary stresses [ 3.6.2 p. 79]. ERATIO is specic for the excess pore water pressure ratio rp in geotechnical analysis which is a function of the eective isotropic stress invariant p and may be regarded as an utilization ratio for the excess pore pressure: p rp = 1 (12.6) pref Diana derives the value for pref from the initial stress eld as specied in a START command [ 12.3.1 p. 219]. CRKIND gives the crack indices Icr (the crack indicator) and Ftu (the tensile strength utilization) which correspond to indications whether a crack arises or not: ft (t) Icr (t) cr with Icr (t) = (12.7) I (t) I (t) (12.8) ft (t) where ft is the tensile strength of concrete, I is the maximal principal stress and cr and cr are safety factors. In order to avoid an innite value of the crack indicator, Diana will automatically set Icr to 100 if I 0.01ft . And in order to avoid negative values of the tensile strength utilization, Diana will automatically set Ftu to 0 if I 0. To determine the crack index Diana needs the values of the tensile strength in time ft (t). You must specify these as a material property [Vol. Material Library]. BIAXFE gives the following Safety Factors for concrete under static and dynamic loading conditions with reference to a biaxial failure envelope: F Sstatic usual = Rstatic usual /r F Sstatic unusual = Rstatic unusual /r F Sdynamic unusual = Rdynamic unusual /r F Sdynamic extreme = Rdynamic extreme /r Where r is the distance from the origin to the actual stress point (1 , 2 ): r = (1 2 + 2 2 ) (12.9) Ftu (t) cr with Ftu (t) = To determine the Safety Factors Diana needs additional material properties. For the required parameters and additional information on the Safety Factors, see Volume Material Library.
12.4 Output of Analysis Results Note that all Safety Factors are limited to 100. CAPACI gives the shear capacity and hydrostatic pressure capacity of stress against MohrCoulomb failure criterion [ 31.2.9 p. 519]. To determine the shear capacity and hydrostatic pressure capacity Diana needs additional material properties, see Volume Material Library. comp selects stress components for output. Default is all available components. loca species the location for the stresses to be output [ 3.6.2 p. 79]. opti are additional options NOBOND includes the contribution of prestress in posttensioned reinforcements to element forces and moments [Vol. Element Library]. see 3.6.2 on page 79 for other options. 12.4.3.1 Cauchy Stresses
comp XX YY
Sxx Syy
263
Primary stresses item type form oper STRESS TOTAL CAUCHY LOCAL
ZZ
Szz
XY
Sxy
YZ
Syz
ZX
Szx
xx yy zz xy yz zx STRESS TOTAL CAUCHY GLOBAL SXX SYY SZZ SXY SYZ SZX XX Y Y ZZ XY Y Z ZX STRESS TOTAL CAUCHY PRINCI S1 1 STRESS TOTAL CAUCHY REAXES S1RA a 1 Von Mises stress item type form oper STRESS TOTAL CAUCHY VONMIS Seq eq Stress invariants comp item type form oper P Q LODE STRESS TOTAL CAUCHY INVARI P Q Lode p q Excess pore pressure item type form oper STRESS TOTAL CAUCHY ERATIO Ser rp Crack stresses comp item type form loca NN NS NT STRESS CRACK CAUCHY INTPNT Sknn Skns Sknt cr cr cr nn ns nt
Diana-9.4 Users Manual Analysis Procedures (IV)
S2
S3
2
S2RA
a 2
264
Crack index comp item type form oper ICR FTU STRESS TOTAL CAUCHY CRKIND Icr Ftu Icr Ftu
Nonlinear Analysis
Biaxial failure envelope comp item type form oper STUS STUN DYUN DYEX STRESS TOTAL CAUCHY BIAXFE FSstus FSstun FSdyun FSdyex F Sstus F Sstun F Sdyun F Sdyex Shear capacity [ 31.2.9] comp item type form oper SHRCAP PPCAP STRESS TOTAL CAUCHY CAPACI SHRCAP PPCAP
12.4.3.2
PiolaKirchho Stresses
comp XX YY ZZ XY YZ ZX 1 2 3
Primary stresses item type form oper STRESS TOTAL PIOLAK LOCAL
Sxx Syy Szz Sxy Syz Szx STRESS TOTAL PIOLAK GLOBAL SPXX SPYY SPZZ SPXY SPYZ SPZX SXX SY Y SZZ SXY SY Z SZX STRESS TOTAL PIOLAK PRINCI SP1 S1 STRESS TOTAL PIOLAK REAXES SP1RA a S1
SP2
SP3
S2
SP2RA
a S2
S3
12.4.3.3
Forces
comp X Y Z
Nx Qy Qz
Concentrated forces type type form oper STRESS TOTAL FORCE LOCAL STRESS TOTAL FORCE
Qx Qy Nz
Nx Qy Qz Qx Qy Nz GLOBAL NX NY NZ NX NY NZ comp XX YY ZZ XY
Distributed forces type type form oper STRESS TOTAL DISFOR LOCAL STRESS TOTAL DISFOR REAXES
YZ ZX
na 1
na 2
265
Concentrated moments item type form oper STRESS TOTAL MOMENT LOCAL
Mx My Mz STRESS TOTAL MOMENT GLOBAL MX MY MZ MX MY MZ Distributed moments item type form oper STRESS TOTAL DISMOM LOCAL STRESS TOTAL DISMOM REAXES comp XX YY ZZ
Mxx Myy
XY
12.4.3.5
Tractions
form oper TRACTI LOCAL comp X Y Z
STx STy STz
12.4.3.6
Stress gradients [ 31.2.8] item type form oper STRESS TOTAL GRADIE LOCAL SG xx
12.4.4
Status
The postanalysis result STATUS is used in Diana to output the internal state parameters of the material model at the integration points of an element, for instance the cracks, the plastic points etc. For each internal state parameter, a syntax will be presented in the following. 12.4.4.1 Plasticity Status
The plasticity status indicates whether irreversible plastic strains have occurred at an integration point.
266
Nonlinear Analysis syntax STATUS [ type w ] { loca w } { opti w } PLASTI INTPNT STATUS PLASTI species the plasticity status as output item.
[INTPNT]
loca species the location for the plasticity status to be output [ 3.6.2 p. 79]. For plasticity status only output in the integration points is possible. opti are additional options [ 3.6.2 p. 79]. The following types of plasticity status can appear in the output: 1. The status is elastic, and has always been so. 2. The status is plastic. In this case the plastic strain is also available for output via the STRAIN PLASTI options [Table 12.2 p. 256]. 3. The status is elastic but was previously plastic. 4. The status of a Cam-clay model is critical [Vol. Material Library]. In the output, the plasticity status is indicated with a text string which depends on the device .
item type device STATUS PLASTI TABULA FEMVIE FXPLUS 1
ELASTIC
Elastic
2
PLASTIC
Plastic PL
3
PREVIOUSLY PLASTIC
Pr.plast E!
4
CRITICAL
Critical CR
12.4.4.2
Crack Status
The cracking status indicates whether a tension cut-o limit has been exceeded at an integration point. syntax STATUS [ type w ] { loca w } { opti w } CRACK INTPNT STATUS CRACK for crack status.
[INTPNT]
loca
species the location for the status to be output [ 3.6.2 p. 79]. For cracking status only output in the integration points is possible.
opti are additional options [ 3.6.2 p. 79]. The following types of crack status can appear in the output (indicating the branch on the tension softening diagram [Fig. 12.2]):
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IV)
12.4 Output of Analysis Results 1. The status is no crack if there is no crack yet. 2. The status is open for cracks on the fully open loading branch. 3. The status is closed for fully closed cracks, the material is elastic again in compression. 4. The status is active for cracks on the partially open loading branch. 5. The status is inactive for cracks on the fully open unloading branch. 6. The status is partial for cracks on the partially open (secant) unloading branch.
4
cr nn
267
4 6 cr u cr nn 3 2 5
Figure 12.2: Crack status linear tension softening In the output, the crack status is indicated with a text string which depends on the device .
item type device 1 2 STATUS CRACK TABULA NO CRACK FULLY
YET OPEN LOADING
Fully l O
CLOSED PARTIALLY FULLY PARTIALLY OPEN OPEN OPEN LOADING UNLOADING UNLOADING
Closed C Partia l P Fully u O! Partia u P!
FEMVIE FXPLUS
Crack display. In the iDiana Results environment you may use the PRESENT SYMBOL command to display the crack status [Vol. Pre- and Postprocessing]. To get the crack pattern you must display the crack strains via the PRESENT DISC command. 12.4.4.3 Liquefaction Status
Depending on the applied liquefaction model [Vol. Material Library], some status variables are optionally available for output at integration points. Because FX+ for Diana can not handle real value results in integration points, liquefaction result items can not be used in combination with this output device.
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
268
Nonlinear Analysis syntax STATUS [ type w ] { comp w } { loca w } { opti w } TOWHAT GAMMA INTPNT NISHI GSTAR BOWL USER TOWHAT gives output of the liquefaction status parameter S0 of the Towhata-Iai model. NISHI gives output of the internal status variable of the Nishi model. BOWL gives output of the status variables of the Bowl model: GAMMA gives the value of the Bowl internal variable . GSTAR gives the value of the Bowl internal variable G . USER gives output of the internal parameters of the user-supplied liquefaction model [ 12.4.4.5].
[INTPNT]
loca
species the location for the status to be output [ 3.6.2 p. 79]. For liquefaction status only output at the integration points is possible.
opti are additional options [ 3.6.2 p. 79]. 12.4.4.4 Contact Status syntax
STATUS [ type w ] { loca w } { opti w } CONTAC NODES STATUS CONTAC for contact status.
[NODES]
loca
species the location for the status to be output [ 3.6.2 p. 79]. For contact status only output in the nodes is possible.
opti are additional options [ 3.6.2 p. 79]. The following types of contact status can appear in the output: 1. The status is no contact when there is no contact. 2. The status is stick when there is friction between the contact elements. 3. The status is slip when the contact elements slip.
12.4 Output of Analysis Results In the output, the contact status is indicated with a text string which depends on the device .
item type device STATUS CONTAC TABULA FEMVIE FXPLUS 1
NO CONTACT
No conta
269
2
STICK
Stick ST
3
SLIP
Slip SL
12.4.4.5
For the general user-supplied material model, and for the user-supplied liquefaction model, [Vol. Material Library], the stress and strain vector can be processed in the usual way: componentwise or in the principal directions. The same holds for the traction vectors of the user-supplied interface model. The vector of internal state variables for user-supplied models USRSTA can only be processed componentwise via the keyword STATUS USER. Because FX+ for Diana can not handle real value results in integration points, usersupplied material result items can not be used in combination with this output device. syntax STATUS [ type w ] { comp w } { loca w } { opti w } USER ITEM INTPNT STATUS USER species the userdened internal parameters as output item. comp selects components for output. ITEM loca with is 01 to 99 for the components of the userdened status variables USRSTA.
[INTPNT]
species the location for the user status to be output [ 3.6.2 p. 79]. For user status only output in the integration points is possible.
12.4.5
Pressure
The undrained behavior of porous materials like soil can be simulated by switching o the drained behavior [ 12.3.4 p. 239]. See also the UNDRAI input item [Vol. Material Library]. You may select the excess uid pressure for output according to the following syntax:
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
270
Nonlinear Analysis syntax PRESSU [ type w ] { comp w } { loca w } { opti w } TOTAL EX INTPNT NODES CENTER PRESSU species the calculated pore pressure for undrained behavior as output item.
[TOTAL]
type species the type [ 3.6.2 p. 78]. comp selects components for output. EX is the only appropriate component for the excess pore pressure pe = ref .
[INTPNT]
loca
12.4.6
The temperature, concentration and maturity can be dened for each nodal point of an element as a function of time via input tables TEMPER, CONCEN, and MATURI respectively [Vol. Material Library]. These data items can be selected for output according to the following syntax: syntax item w [ type w ] { loca w } { opti w } TEMPER TOTAL NODES CONCEN INTPNT MATURI CENTER item species the data item to be output for the elements. TEMPER for the interpolated temperature at a certain time step. CONCEN for the interpolated concentration at a certain time step. MATURI for the interpolated maturity at a certain time step.
[TOTAL]
type
271
loca
species the location for the temperature, concentration or maturity to be output [ 3.6.2 p. 79].
12.4.7
The nodal forces are energetically conjugate to the nodal degrees of freedom of the nite element system.4 In a nonlinear analysis, Diana can output three types of nodal forces: the total external forces, the support reactions, and the residual forces. syntax FORCE [ type w ] [ form w ] [ oper w ] { comp w } { opti w } EXTERN TRANSL LOCAL REACTI ROTATI GLOBAL RESIDU FORCE species forces in the nodes as output item. type species the type of the nodal forces. EXTERN for the externally applied forces fext [Eq. (30.1) p. 479]. REACTI for the reaction forces in all supported nodes. These are dened as the out-of-balance forces at the boundary conditions of the structure, i.e., the supported degrees of freedom: the dierence between the externally applied forces fext and the internal forces fint at the supports only [Eq. (30.5) p. 480]. RESIDU for the residual forces g, also called the out-of-balance forces. These are dened as the dierence between the externally applied forces fext and the internal resistance forces fint [Eq. (30.5) p. 480]. form species the formulation [ 3.6.2 p. 78]
[TRANSL] [GLOBAL] [REACTI]
oper species an operation to be performed on the forces and moments [ 3.6.2 p. 79].
4 If the types of the nodal degrees of freedom are translations and rotations then the nodal forces are translational forces (concentrated loads) and rotational forces (concentrated moments) respectively.
272
Nonlinear Analysis comp selects force or moment components for output. Default is all available components. opti are additional options [ 3.6.2 p. 79].
Total forces item type form oper FORCE EXTERN TRANSL LOCAL comp X Y Z
FTx FTy FTz
Fx FORCE EXTERN TRANSL GLOBAL FTX FX FORCE EXTERN ROTATI LOCAL MTx Mx FORCE EXTERN ROTATI GLOBAL MTX MX Support reactions item type form oper FORCE REACTI TRANSL LOCAL
Fy FY
Fz FZ
My Mz
MTY MTZ
MY MZ
comp X Y Z
FBx FBy FBz
B B B Fx F y F z
FORCE REACTI TRANSL GLOBAL FBX B FX FORCE REACTI ROTATI LOCAL MBx B Mx FORCE REACTI ROTATI GLOBAL MBX B MX Residual forces item type form oper FORCE RESIDU TRANSL LOCAL
FBY FBZ
B B FY FZ
MBy MBz
B B My Mz
MBY MBZ
B B MY MZ
comp X Y Z
FRx FRy FRz
R R R Fx F y Fz
FORCE RESIDU TRANSL GLOBAL FRX R FX FORCE RESIDU ROTATI LOCAL MRx R Mx FORCE RESIDU ROTATI GLOBAL MRX R MX
FRY FRZ
R R F Y FZ
MRy MRz
R R My Mz
MRY MRZ
R R MY MZ
12.4.8
syntax
NODFOR [ type w ] [ form w ] [ oper w ] { comp w } { opti w } ELEMEN TRANSL LOCAL REINFO ROTATI GLOBAL TOTAL
12.4 Output of Analysis Results NODFOR species internal nodal element forces and moments in the nodes as output item. This command gives the contributions of the element or reinforcement internal nodal forces (or both), to a certain node. A selection of elements which form a section of the model gives the total internal forces that act on that section. type species the forces type. ELEMEN for contribution of elements only. REINFO for contribution of embedded reinforcements only. TOTAL for contribution of both elements and embedded reinforcements. form oper comp species the formulation [ 3.6.2 p. 78]. species an operation (transformation) to be performed on the forces and moments [ 3.6.2 p. 79]. selects force or moment components for output. Default is all available components.
[TRANSL] [GLOBAL] [TOTAL]
273
[all]
Z
FNz
n Fz
FNZ
n FZ
MNz
n Mz
MNZ
n MZ
12.4.9
syntax
ELMFOR [ type w ] [ form w ] [ oper w ] { comp w } { opti w } ELEMEN TRANSL GLOBAL REINFO ROTATI TOTAL ELMFOR species internal element forces and moments in the nodes of an element as output item. This command gives the contributions of the element or reinforcement internal forces (or both), to a certain node of an element. type species the forces type. ELEMEN for contribution of elements only.
Diana-9.4 Users Manual Analysis Procedures (IV) December 15, 2009 First ed.
[TOTAL]
274
Nonlinear Analysis REINFO for contribution of embedded reinforcements only. TOTAL for contribution of both elements and embedded reinforcements.
[TRANSL] [GLOBAL]
species the formulation [ 3.6.2 p. 78]. species an operation (transformation) to be performed on the forces and moments [ 3.6.2 p. 79]. selects force or moment components for output. Default is all available components.
[all]
Z
FEZ
e FZ
MEZ
e MZ
12.4.10
Model Parameters
For some material models Diana will calculate, or assume, default values for appropriate parameters. In most cases you can overrule these values by specifying the parameters explicitly in the input data. Typical examples are the crack bandwidth in models for cracking and the preconsolidation stress in models for clay [Vol. Material Library]. Sometimes, these parameters may be useful in postprocessing. Via the OUTPUT block you can ask Diana to output certain parameters, being the values that you have explicitly specied or else the calculated (default) values. Output of model parameters is not available for FX+ for Diana. syntax item w [ type w ] loca w { opti w } PARAME BANDWI INTPNT PRECON PARAME for output of a model parameter. Table 12.4 on the facing page outlines the availability and applicability of the various model parameters output for each of the element families. type species which parameter must be output. BANDWI gives the crack bandwidth hcr . This value is applied in various models for concrete cracking and can be input as data item CRACKB [Vol. Material Library].
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IV)
275
item type
tr
b -
a -
c c
a d
a -
a a
(a) All elements. (b) Only class III beams. (c) Only for regular plane strain. (d) Not for axisymmetric shell. (-) Not available or not suitable.
PRECON gives the preconsolidation stress pc . This value is applied in various models for clay and can be input as data item PRECON [Vol. Material Library]. loca species the location of the parameter to be output. For model parameters this is always, and by default, the element integration point.
Model parameters item type PARAME BANDWI Hcr hcr PARAME PRECON Pc pc
[INTPNT]
12.4.11
For crack tip elements [Vol. Element Library] Diana can calculate and output the parameters for Linear Elastic Fracture Mechanics analysis (LEFM), based on displacements, stresses and strains resulting from a nonlinear analysis. See Chapter 37 for background theory. syntax FRACTU FRACTU species the LEFM parameters as output item. No further options apply for this output. The LEFM parameters are always output as three components.
LEFM parameters item FRACTU K G KI G I
276
Nonlinear Analysis KI is the Mode-I stress intensity factor [Eq. (37.2) p. 564]. The stress intensity factor can only be calculated if the material parameters and the thickness (for plane stress elements) of all the elements in the crack front are identical. GI is the Mode-I energy release rate [Eq. (37.8) p. 565].
12.4.12
Dynamic Pressures
For uidstructure interface elements [Vol. Element Library] Diana can calculate and output the dynamic pressures. See [ 32.5 p. 532] for background theory. syntax FSPRES [ type w ] [ loca w ] { opti w } TOTAL NODES FSPRES species the dynamic pressures of the uidstructure interface elements as output item.
[TOTAL]
type
species the dynamic pressure type. TOTAL for the total dynamic pressures of the uidstructure interfaces.
[NODES]
loca
species the location for the dynamic pressures to be output [ 3.6.2 p. 79].
Part V
Stability Analysis
Chapter 13
13.1
Euler stability analysis gives information about linearized stability of a structure. It tells whether solutions from linear elastic analysis are stable or whether small disturbances to those solutions exist, requiring no extra external energy. This type of stability analysis does not allow for any physical nonlinearities, geometrical nonlinear (i.e., large deformation) eects are only partly taken into account. However, often it is a relatively simple and eective method to get a fair impression of a structures buckling modes. Background theory is given in 34.1 on page 542. In order to perform an Euler stability analysis with Diana you must take the following actions: 1. Prepare input data like for a linear analysis [Ch. 4 p. 89]. If appropriate, add initial imperfections [ 13.4]. 2. Perform a stability analysis. An Euler stability analysis is accomplished by Module euler as described in Chapter 14. 3. Output the buckling modes (optional) [ 14.2.3 p. 290].
Diana-9.4 Users Manual Analysis Procedures (V) December 15, 2009 First ed.
280
13.2
Perturbation Analysis
Perturbation analysis gives information about the initial postbuckling behavior of the structure. Postbuckling deformation elds can be calculated and eects resulting from interaction of selected buckling modes can be analyzed. This type of stability analysis does not allow for any physical nonlinearities, geometrical nonlinear eects (large deformation) are only taken into account partly. See 14.3.1 on page 292 for commands and 34.3 on page 546 for background theory. In order to perform a perturbation analysis with Diana you must take the following actions: 1. Perform a complete Euler stability analysis as described in the previous section. 2. Select the perturbation modes. 3. Solve for the second order elds. 4. Compute the potential coecients Aijk and Aijkl . 5. Plot the second order elds.
13.3
Continuation Analysis
Continuation analysis gives information about the equilibrium paths in the initial postbuckling region. Continuation analysis must always be preceded by a perturbation analysis. See 14.3.2 on page 295 for commands and 34.3 on page 546 for background theory. In order to perform a continuation analysis with Diana you must take the following actions: 1. Perform a complete perturbation analysis as described in the previous section. 2. Solve the reduced set of equilibrium equations. 3. Plot the postbuckling displacement elds.
13.4
Many of Dianas elements may be applied in Euler stability analysis with initial imperfections [Vol. Element Library]. Any imperfection on an element should be small in comparison with the size of the elements, for instance less than 1 % for isoparametric elements and less than 0.1 % for class-I and class-II beams . Thus imperfections on an isoparametric element of size 50 should be less than 0.5 . This is required to get an accurate estimation of the inuence of imperfections.
281
node n
1
/ nodes ng... / type w dirnr n / displs r... / INIVAR is the general table heading for input of initial values for variables. DISPLA is the subtable heading for initial (strain-free) displacements. Number field is the eld number for reference from the analysis commands for imperfections [ 14.2.2 p. 288]. node is a single node number. nodes is a series of nodes, it must be specied between slashes and may comprise numbers or groups or both. type is the type of variable: TR for translation or RO for rotation. dirnr species the direction, referring to table DIRECT [ 1.4 p. 14]. displa is the initial displacement. displs is a series of initial displacements, one for each node in nodes , it must be specied between slashes. Omission of displa or displs induces random displacements in combination with the random imperfections option [ 14.2.2 p. 288].
file .dat INIVAR DISPLA 3 1 TR 3 0.001 / 10-12 / TR 2 / 0.002(3) / 4 RO 5 0.00005
This example species initial displacement eld number 3. The initial translation of node 1 is 0.001 in direction 3. The initial translation of nodes 10 to 12 is 0.002 in direction 2. The initial rotation of node 4 is 0.00005 around direction 5. All other translations and rotations are initially equal to zero.
13.5
Applicability of Elements
This section outlines which of the stability analysis options may be applied for the various structural elements. Table 13.1 on the following page outlines the
Diana-9.4 Users Manual Analysis Procedures (V) December 15, 2009 First ed.
282
a b
c cd e
h hl hl -
h hi hi -
j j j j
h h h -
() All elements. (a) Only enhanced elements. (b) Not for cable elements. (c) Not for 2-D elements. (d) Not for class-II. (e) Not for class-I. (f) Not for orthotropic geometry. (g) Not for wrinkling. (h) Not for rubber. (i) Not for shell of revolution. (j) Only spline elements. (k) Not for layered elements. (l) Not for innite shell. (m) Not for elements with drilling rotation. (-) Not applicable.
appicability of the various options for stability analysis for structural elements: Euler stability analysis, inital displacements, initial imperfections and perturbation and continuation analysis. See Volume Element Library for description of basic variables and input of these elements. Truss elements. All truss elements support Euler stability analysis and can take nonlinear eects of initial displacements into account. Only the enhanced truss elements accept imperfections. The cable elements do not support perturbation and continuation analysis. Beam elements. All beam elements support Euler stability analysis. The class-I beam elements do not support perturbation and continuation analysis. The two-dimensional class-III beam elements do not account for nonlinear eects of initial displacements or imperfections. Plane stress elements. The regular plane stress elements support Euler stability analysis. These elements also account for nonlinear eects due to initial displacements and imperfections. None of the plane stress elements support perturbation and continuation analysis. Plane strain elements. The regular plane strain elements and the innite shell elements support Euler stability analysis. These elements also account for nonlinear eects due to initial displacements and imperfections. None of the plane strain elements support perturbation and continuation analysis. Axisymmetric elements. The regular solid ring elements and the shells of revolution support Euler stability analysis. These elements also account for
13.5 Applicability of Elements nonlinear eects due to intial displacements and imperfections. None of the axisymmetric elements support perturbation and continuation analysis. Plate bending elements. The plate bending elements do not support stability analysis. Flat shell elements. The spline elements are the only at shell elements that can be used in stability analysis. These elements accept initial displacements and imperfections and are also suitable for perturbation and continuation analysis. Curved shell elements. All curved shell elements support Euler stability analysis. These elements also account for nonlinear eects due to intial displacements and imperfections. Only the higher-order curved shell elements support perturbation and continuation analysis. The layered shell elements do not support perturbation analysis. Solid elements. The regular solid elements support Euler stability analysis. These elements also account for nonlinear eects due to intial displacements and imperfections. None of the solid elements support perturbation and continuation analysis. Structural interface elements. The structural interface elements do not support stability analysis. Spring elements. The spring elements do not apply for stability analysis but can be used in nite element models for such an analysis. Mass elements. The point mass elements do not apply for stability analysis but can be used in nite element models for such an analysis. Embedded reinforcements. In stability analysis, the inuence of embedded reinforcements is accounted for via the element stiness.
283
284
Chapter 14
286
Euler Stability Analysis Default. If you only give the *EULER command, then Diana will perform a normal Euler stability analysis, i.e., as if you specied the following.
file .dcf *EULER MODEL EIGEN *END
14.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual Euler stability analysis. syntax BEGIN MODEL [ OFF ] [ EVALUA ] [ ASSEMB ] END MODEL EVALUA to check and evaluate geometric and material properties for elements [ 3.4 p. 66]. ASSEMB to assemble the elements, i.e., creating appropriate system degrees of freedom [ 3.5 p. 68].
14.2
Buckling Modes
To solve the eigenvalue problem in an Euler stability analysis, i.e., to determine the buckling modes, you must give commands in the EIGEN block. syntax BEGIN EIGEN [ OFF ] [ STABIL ] [ IMPERF ] [ EXECUT ] [ OUTPUT ] END EIGEN
14.2 Buckling Modes OFF suppresses the execution of a buckling analysis. This may be useful to save computing time if you want to perform a postbuckling analysis [ 14.3], while the previously determined buckling modes still reside on the filos le. STABIL species the type of stability analysis [ 14.2.1]. IMPERF species the imperfections [ 14.2.2]. [EXECUT ] species how to solve the eigenvalue problem. In particular this involves the type of the solution procedure. Commands are identical to those for eigenvalue analysis with Module eigen [ 27.3 p. 445]. OUTPUT to specify the buckling modes to be output [ 14.2.3 p. 290].
file .dcf *EULER BEGIN EIGEN STABIL LOAD=1 EXECUT NMODES=2 OUTPUT DISPLA END EIGEN *END
287
The rst command starts Module euler. Then, the commands in the EIGEN block ask for buckling analysis where the STABIL command species the type of stability analysis. The LOAD=1 parameter asks for buckling eigenmodes to be calculated for linear elastic load set 1. The EXECUT command species how to execute the buckling analysis. Parameter NMODES=2 selects the rst two modes to be determined (if existing at all). Finally the OUTPUT command species the output to be produced: the buckling eigenmodes (as displacements).
14.2.1
Stability Analysis
The STABIL command block within the EIGEN command block of Module euler species how to setup the system of equations for stability analysis. syntax BEGIN STABIL [ OFF ] [ LOAD=loset n ] ] [ DISPLA ON OFF END STABIL
288
Euler Stability Analysis OFF suppresses the setup of the system of equations. This may save computing time when the system of equations is still available on the filos le from a previous stability analysis. LOAD=loset uses the linear elastic stresses of the specied load set for the geometric stressstiness matrix KG .Default is the lowest available load set number. The load set number corresponds to a load set in input table LOADS [ 2.3 p. 36].
[ON]
DISPLA species whether or not the initial displacement matrix KLL must be incorporated in the system of equations [Eq. (34.18) p. 543]. ON causes the initial displacement matrix to be used and solves the eigenvalue problem KL0 + crit KLL (ulin ) + KG (ulin ) U = 0
This is also the default if you omit the DISPLA command. OFF causes the initial displacement matrix to be neglected, i.e., only the linear stiness matrix KL0 and the geometrical stressstiness matrix KG are considered in solving the eigenvalue problem KL0 + crit KG (ulin ) U = 0
file .dcf BEGIN EIGEN STABIL DISPLA OFF END EIGEN
file .dcf BEGIN EIGEN BEGIN STABIL LOAD=2 DISPLA ON END STABIL END EIGEN
14.2.2
Imperfections
14.2 Buckling Modes Any imperfections setup in a buckling analysis, are automatically taken into account in subsequent nonlinear analyses. This allows for postbuckling analysis including critical imperfections. See 34.2 on page 545 for background theory. The commands for application of an imperfection pattern are as follows. syntax BEGIN IMPERF [ OFF ] BUCKLI [ MODE=buckl n ] [ MAX=msize r ] RANDOM [ FIELD=field n ] [ MAX=msize r ] USER [ FIELD=field n ] [ FACTOR=fac r ] END IMPERF OFF suppresses the application of an imperfection pattern. BUCKLI asks for a buckling mode as imperfection pattern. Generally, the most appropriate imperfection eld is the lowest buckling mode which Diana takes by default. With parameter MODE=buckl you may explicitly specify a mode. Diana will scale the imperfection eld so much that the largest translation or rotation equals the value that you specify with parameter MAX=msize . The other imperfections are scaled consistently with the critical buckling mode. RANDOM asks for a random imperfection pattern.1 Parameter FIELD=field refers to a eld number from table INIVAR [ 13.4 p. 280]. Default is the lowest available eld number. The node-type-direction combinations in this table specify the degrees of freedom that will have a random imperfection. The values displa or displs for the displacements must not be input in this case. The imperfections in the nite element model will not exceed the value that you specify with parameter MAX=msize . But generally, none of the imperfections will exactly equal this value. USER takes a user-specied imperfection pattern. Parameter FIELD=field refers to a eld number from table INIVAR [ 13.4 p. 280]. Default is the lowest available eld number. The node-type-direction combinations in this table specify the degrees of freedom that will have an imperfection. If you specify parameter FACTOR=fac then Diana will multiply the dis1 The
289
[msize =0.01]
[msize =0.01]
[fac = 1]
random pattern is reproducible, i.e., each analysis run will produce the same random
pattern.
Diana-9.4 Users Manual Analysis Procedures (V) December 15, 2009 First ed.
290
Euler Stability Analysis placement values from the table by the factor to get the actual imperfection.
14.2.3
Diana can output the buckling modes as displacements via the OUTPUT command block. For the general, analysis type independent options for output of analysis results see 3.6 on page 69. syntax BEGIN OUTPUT [device w ] [outopt w... ] [ params ] [ OFF ] [ BEGIN SELECT [ ] [ MODES ] modes n... ALL END SELECT ] [ LAYOUT ] item w [ type w ] [ form w ] [ oper w ] { comp w } { opti w } DISPLA BUCKLI TRANSL LOCAL ROTATI GLOBAL END OUTPUT SELECT customizes the batch output selection. for model selection see 3.6.1 on page 71.
[ALL]
MODES selects specic buckling modes for output: modes is a series of mode numbers, ALL selects all calculated modes. LAYOUT optional commands to customize the layout and style of tabular output [ 3.6.3.1 p. 82]. item is the name of the analysis result to be output. See 3.6 on page 69 for complete syntax of this command. DISPLA for displacements, which are the only available results of a stability analysis. type species the displacement type [ 3.6.2 p. 78]: BUCKLI] for buckling modes.
[BUCKLI]
291
If you only give a single OUTPUT command, like in the above example, or if you dont give the OUTPUT command at all, then Diana gives output of the translational displacements in global XY Z orientation, as if you had given the following commands.
file .dcf *EULER [ commands ] BEGIN EIGEN BEGIN OUTPUT DISPLA BUCKLI TRANSL GLOBAL END OUTPUT END EIGEN *END
Buckling modes. Diana normalizes the buckling mode i so much that the largest translation component u has the value of 1.
item type form oper DISPLA BUCKLI TRANSL LOCAL comp X Y Z
Dtx Dty Dtz
ux DISPLA BUCKLI TRANSL GLOBAL DtX uX DISPLA BUCKLI ROTATI LOCAL Drx x DISPLA BUCKLI ROTATI GLOBAL DrX X
uy uY y Y
uz uZ z Z
Buckling values. The appropriate buckling value is output together with the required buckling mode. The buckling values correspond with the parameter i . The lowest buckling value is the most critical buckling value. If a load f is applied to the structure and if the resulting stresses are used for the geometric stress stiness matrix, the lowest buckling value represents the ratio fcrit /f . Thus it indicates the factor of the load f that can be applied on the structure before buckling occurs. However, the validity of this statement is restricted by the assumptions made in Euler buckling theory [ 34.1 p. 542].
Diana-9.4 Users Manual Analysis Procedures (V) December 15, 2009 First ed.
292
Euler Stability Analysis Negative buckling values indicate that buckling occurs if the direction of the external load is switched. Hence, such modes do not really occur if the load as specied in the data le is applied on the structure.
14.3
Postbuckling Analysis
To perform a postbuckling analysis with Module euler you must give REDUCE and CONTIN commands. Note that a postbuckling analysis can only be performed if the results of a previous buckling analysis are available on the filos le. syntax BEGIN REDUCE [ OFF ] END REDUCE BEGIN CONTIN [ OFF ] END CONTIN REDUCE species how to perform the perturbation analysis [ 14.3.1]. The OFF option suppresses the execution of a perturbation analysis. This may save computing time when the reduced system of equations is still available from a previous perturbation analysis. CONTIN species how to perform the continuation analysis [ 14.3.2]. The OFF option suppresses the execution of a continuation analysis.
14.3.1
Perturbation Analysis
To specify the perturbation analysis, i.e., the reduction of the full set of equations, you must give commands in the REDUCE block. syntax BEGIN REDUCE [ BEGIN EXECUT [ OFF ] [ SELECT MODES modes n... ALL [ LOADFA=lambda r ] END EXECUT ]
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (V)
14.3 Postbuckling Analysis [ BEGIN OUTPUT [device w ] [outopt w... ] [ params ] [ OFF ] [ SELECT ] [ LAYOUT ] DISPLA [ type w ] [ form w ] [ oper w ] { comp w } { opti w } SECMOD TRANSL LOCAL ROTATI GLOBAL END OUTPUT ] END REDUCE EXECUT denes how to execute the perturbation analysis. OFF suppresses the execution of a perturbation analysis. This may save computing time when you only want to get output of a previous perturbation analysis. SELECT MODES selects buckling modes for the perturbation analysis. The modes must have been determined previously [ 14.2 p. 286]. You may explicitly specify is a set of modes , or you may require a perturbation analysis for ALL determined modes. All modes is also the default if you dont select modes. LOADFA=lambda species the relative load factor p of the perturbation point [Eq. (34.37) p. 547]. OUTPUT species the desired output of the perturbation analysis. For the general, analysis type independent options for output of analysis results see 3.6 on page 69. SELECT customizes the batch output selection. Appropriate for perturbation analysis is node selection [ 3.6.1 p. 71]. LAYOUT optional commands to customize the layout and style of tabular output [ 3.6.3.1 p. 82]. DISPLA SECMOD will output the second order displacements [ 14.3.1.1].
file .dcf BEGIN REDUCE BEGIN EXECUT SELECT MODES 1 2 LOADFAC=0.91 END EXECUT BEGIN OUTPUT DISPLA SECMOD END OUTPUT END REDUCE
[SECMOD] [ALL]
293
[p = 0.95]
294
14.3.1.1
ux DISPLA SECMOD TRANSL GLOBAL D2tX uX DISPLA SECMOD ROTATI LOCAL D2rx x DISPLA SECMOD ROTATI GLOBAL D2rX X
Default
*EULER [ commands ] BEGIN REDUCE [ commands ] OUTPUT END REDUCE *END
If you only give a single OUTPUT command, like in the above example, or if you dont give the OUTPUT command at all, then Diana gives output of the translational displacements in global XY Z orientation, as if you had given the following commands.
file .dcf *EULER [ commands ] BEGIN REDUCE [ commands ] BEGIN OUTPUT DISPLA SECMOD TRANSL GLOBAL END OUTPUT END REDUCE *END
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (V)
295
14.3.2
Continuation Analysis
To specify the continuation analysis you must give commands within the CONTIN block. syntax BEGIN CONTIN BEGIN EXECUT [ OFF ] [ BEGIN START [ LAMBDA=lambda r ] [ AMPLIT ampl r... ] END START ] }] [ STEPS { NSTEPS=nsteps n SIZE=size r [ NORM [ DISPLA ] CONVER=eps r ] END EXECUT [ BEGIN OUTPUT [device w ] [outopt w... ] [ params ] [ OFF ] [ BEGIN SELECT [ ] [ STEPS ] steps n... ALL LAST END SELECT ] [ LAYOUT ] DISPLA [ type w ] [ form w ] [ oper w ] { comp w } { opti w } POSTBU TRANSL LOCAL ROTATI GLOBAL END OUTPUT ] END CONTIN EXECUT species how to perform the continuation analysis. OFF suppresses the execution of the continuation analysis.
296
Euler Stability Analysis START species the starting point in the load factormode amplitude space. Parameter LAMBDA=lambda is an estimation for the load factor . AMPLIT ampl are estimations for the respective mode amplitudes, one value for each selected mode. STEPS species the execution of steps in the continuation procedure. Parameter NSTEPS=nsteps indicates the number of steps, SIZE=size is the step size. NORM species the convergence criterion for the iteration process. Parameter CONVER=eps indicates the tolerance . In postbuckling analysis the only available criterion is on the displacement norm, therefore you may omit the DISPLA option. OUTPUT species the desired output of the perturbation analysis. For the general, analysis type independent options for output of analysis results see 3.6 on page 69. SELECT customizes the batch output selection. is a model selection [ 3.6.1 p. 71]. Appropriate for continuation analysis is node selection STEPS selects steps for output of analysis results. Where steps are numbers of selected steps, ALL will produce output for all steps that are executed (the default), LAST selects the nal step. LAYOUT optional commands to customize the layout and style of tabular output [ 3.6.3.1 p. 82].
[POSTBU]
[ = 1]
[ampl = 0]
[ = 105 ] [DISPLA]
[ALL]
In the above example a postbuckling continuation analysis of ten steps is executed, starting in the initial buckling point. A step size of 1.0 is applied while the convergence criterion is 0.001.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (V)
297
14.3.2.1
The POSTBU type specier gives output of displacements of a postbuckling continuation analysis. As outlined in the background theory for postbuckling analysis [ 34.3 p. 546], Diana calculates the total displacements eld upb for the actual postbuckling conguration according to upb = ulin + ai i + ai aj uij (14.1)
where uij is called the second order displacement vector and ai should be interpreted as amplitude of the respective mode.
comp X Y uy uY y Y
Z uz uZ z Z file .dcf
ux DISPLA POSTBU TRANSL GLOBAL DPtX uX DISPLA POSTBU ROTATI LOCAL DPrx x DISPLA POSTBU ROTATI GLOBAL DPrX X
Default
*EULER [ commands ] BEGIN CONTIN [ commands ] OUTPUT END CONTIN *END
If you only give a single OUTPUT command, like in the above example, or if you dont give the OUTPUT command at all, then Diana gives output of the translational displacements in global XY Z orientation, as if you had given the following commands.
298
Load factor. The postbuckling load factor is output together with the postbuckling displacement eld. This factor corresponds with the quotient /1 where is the actual postbuckling load factor and 1 is the lowest buckling value. The load corresponding to the actual postbuckling conguration is dened by fpb = crit = 1 f (14.2)
Part VI
Chapter 15
See Volume Analysis Examples for real-life examples of potential ow analysis and Chapter 35 for background theory. Convection-diusion problems. Using the options described for heat ow it is of course possible to solve analogous other convection-diusion problems. Staggered owstress analysis. The potential eld, resulting from a potential ow analysis may be input for a structural analysis. This implies a so-called staggered owstress analysis. Modules heatss, heattr, growss, and growtr can automatically convert a nite element model for structural analysis to a model for potential ow analysis. Modules linsta, nonlin, eigen, euler, modal, freque, spectr, and hftd convert the nite element model back to a structural model and use the ow analysis results as input, e.g. loading input [Ch. 22]. Fluidstructure interaction. By coupling of a potential ow analysis to a dynamical structural analysis, see Part III, Diana can model the added mass and for modules freque and hftd also the added damping by the uid or gas which surrounds the structure.
Diana-9.4 Users Manual Analysis Procedures (VI) December 15, 2009 First ed.
302
Chapter 16
16.1
16.1.1
Nodal Potentials
Fixed Nodal Potentials
Depending in the type of analysis, nodes with a xed nodal potential must be specied in table FIXTEM or FIXPOT. Non-zero values for xed potentials must be input via table BOUNDA [ 16.2 p. 306]. Any node specied in one of these tables but not in table BOUNDA, will have a xed potential of zero. A special case occurs in a beam cross-section analysis [Ch. 19]: you must specify one, and only one, xed node in table FIXPOT when doing such analysis. syntax tabnam w FIXTEM FIXPOT
1 80
node n
304
80
/ nodes ng... / tabnam is the table heading for input of nodes with a xed potential: FIXTEM for xed temperatures in heat ow analysis, or FIXPOT for xed potentials in potential ow analyses other than heat ow (for instance concentration ow). The nodes may be specied with a single number node (one per input line) or with a series of nodes between slashes which comprises numbers or groups or both. The single number format and the series format may also be combined like in the following example.
file .dat FIXTEM 10 / 15-19 /
In this example node 10 has a xed temperature and nodes 15 to 19 have xed temperatures as well.
16.1.2
EQUAL
1 80
/ nodes ng... / EQUAL is the table heading for input of nodes with equal potential. The nodes must be specied with a series nodes between slashes which comprises numbers or groups or both. It is not allowed to equalize nodes that belong to the same element.
file .dat EQUAL / 15-19 /
305
16.1.3
You may start a transient analysis with an initial potential eld. In a nonlinear steady-state state analysis you may specify an estimation for the rst solution of the potential eld, which also may be considered as an initial eld. There are two ways to specify such an initial potential eld: (1) directly via input table INIVAR or (2) as a potential eld found from a previous analysis. If you dont specify an initial eld in either way, then Diana assumes that all the nodal potentials are initially equal to zero. There are three forms of input syntax for table INIVAR: (1) one nodal potential per line, (2) nodes in a series of numbers or groups or both, with one potential valid for all the nodes in the series, (3) nodes in a series of numbers and/or groups, with a series of potentials: one value for each node. syntax INIVAR subtab w field n TEMPER POTENT
1 5 6 80
node n
1
phi r
5 6 80
/ nodes ng... / / phis r... / INIVAR is the table heading for initial nodal potentials. subtab is the subtable name, depending on the type of analysis: TEMPER for initial temperatures in heat ow analysis, or POTENT for initial potentials in potential ow analysis other than heat ow (for instance concentration ow). The field number is used for reference from the analysis commands. node is a single node number. nodes is a series of nodes, it must be specied between slashes and may comprise numbers or groups or both. phi is the value of the initial temperature T0 , or the initial potential 0 ; phis is a series of initial temperatures or potentials, one for each node in nodes , it must be specied between slashes.
file .dat
306
16.2
Boundary Conditions
There are two classes of boundary conditions, each of which is specied in a separate subtable of table BOUNDA. Nodal potentials: prescribed potential on nodes or boundary elements. Discharges: prescribed discharge Q (source) on nodes or ordinary elements. You may specify various cases. Each case may have nodal or element boundary conditions or both. In the analysis, each case represents a set of nodal potentials for the nodes where the potential is prescribed, and a set of total equivalent prescribed nodal uxes for the other nodes. If you dont specify a condition for a certain boundary, then Diana assumes that this boundary is completely insulated, i.e., the ux through it is equal to zero. syntax BOUNDA CASE case n NODAL
1 5 6 80
element boundary conditions BOUNDA is the table heading for boundary conditions input. CASE case is the number of a new boundary case. A case number specication line must follow immediately after the table heading. The case number may be respecied at the start of a new subtable. NODAL is the subtable for nodal boundary conditions [ 16.2.1]. ELEMEN is the subtable for element boundary conditions. See Volume Element Library for input syntax.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VI)
307
16.2.1
Nodal boundary conditions are input in subtable NODAL. There are three forms of input syntax: (1) one nodal condition per line, (2) nodes in a series of numbers or groups or both, with one condition value valid for all the nodes in the series, (3) nodes in a series of numbers and/or groups, with a series of condition values: one value for each node. syntax BOUNDA NODAL
1 5 6 80
node n
1
type c value r
5 6 80
/ nodes ng... / type c value r NODAL is the subtable heading for nodal boundary conditions. node is a single node number, nodes is a series of nodes, it must be specied between slashes and may comprise numbers or groups or both. type is the type of the boundary condition: T for prescribed temperature in heat ow analysis, Q for prescribed discharge (point source), or P for for prescribed potential in ow analysis other than heat ow. value is the value of the boundary condition; values is a series of condition values, one for each node in nodes .
file .dat BOUNDA CASE 1 NODAL 2 Q 0.01 4 T -0.12 / 20-35 / T
0.25
16.2.2
For transient analysis, the boundary conditions specied in table BOUNDA [ 16.2 p. 306] must be related to the time. For each case from table BOUNDA, linear diagrams can be specied for the multiplication factor of the boundary conditions of that case in a time interval: t1 t < t2 . The total boundary conditions on a certain time are a superposition of all the multiplications of the interpolated factors with the boundary conditions belonging to the specied
Diana-9.4 Users Manual Analysis Procedures (VI) December 15, 2009 First ed.
308
Input for Potential Flow Analysis cases. You may specify the diagram completely, i.e., times and multiplication factors, in table TIMEBO. Alternatively, you may specify only the times in table TIMEBO and import the multiplication factors from an external le. Complete table TIMEBO
1 80
syntax
BOUNDA case n TIMES times r... / FACTOR f r... / TIMEBO is the table heading for time dependent boundary conditions. BOUNDA case is the number of the boundary case, active in the interval. This number refers to a case number specied in table BOUNDA [ 16.2 p. 306].
(ti+1 ti )
TIMES times are the times t for the corresponding multiplication factors. Times must be specied in increasing order. FACTOR f are the multiplication factors for the corresponding times. You may specify multiple sets of times and factors, optionally with a dierent case number for each set. Within one set, the number of factors f must always match the number of times in times as shown in the following example.
file .dat TIMEBO BOUNDA 2 TIMES 0.00 0.10 / FACTOR 2.5 3.6 / BOUNDA 1 TIMES 0.13 0.16 0.16 0.45 / FACTOR 6.4 4.2 2.5 0.0 / BOUNDA 3 TIMES 0.45:0.501(0.01) / FACTOR 4.8 5.3 7.9 7.8 6.2 1.4 /
In this example no boundary case is active from t = 0.10 to t = 0.13. Factors imported TIMEBO
1 5 6 80
syntax
16.2 Boundary Conditions [ SKIP nlin n ] [ ] SCALE sfac r PEAK pval r TIMEBO is the table heading for the time dependent boundary conditions. BOUNDA case is the number of the boundary case, active during the following times. The case number case refers to a case number specied in table BOUNDA [ 16.2 p. 306]. TIMES times are the times t for the corresponding load factors. Times must be specied in increasing order. FACTOR IMPORT asks Diana to read the multiplication factors from an external le named file . Factors on this le may be separated by spaces, commas, tabs or newlines. There are two additional options. SKIP causes Diana to skip the rst nlin lines before starting to read factors. SCALE species a scale factor sfac . Diana will multiply each factor from the external le by this factor before applying the boundary condition. PEAK asks Diana to scale the values from the external le such that the greatest absolute vale of the load factors is equal to pval . You may specify multiple sets of times and le, optionally with a dierent load number for each set. Within one set, Diana reads as many factors from file as there are times in times .
file .dat TIMEBO BOUNDA 4 TIMES 0.00 0.10 0.13 0.16 0.16 0.45 / FACTOR IMPORT "bound.log" SKIP 2 SCALE 3.
[nlin = 0] (ti+1 ti )
309
[sfac = 1]
(pval > 0)
In this example Diana will read six factors from external le bound.log. This le could be like shown below.
bound.log Logging boundary Thu Jan 6 15:42:15 CET 2000 1 3 4 5.2 3.2 1.6
The rst two lines will be skipped. The actual factors for transient boundary case 4 will be: f1 = 3 1 = 3, f2 = 3 3 = 9, etc. Note that in this example the factor suddenly drops at time t = 0.16 from 3 5.2 = 15.6 to 3 3.2 = 9.6 .
Diana-9.4 Users Manual Analysis Procedures (VI) December 15, 2009 First ed.
310
Input for Potential Flow Analysis Value formats. All regular Fortran formats are allowed for the factors on the external le. Integers (without decimal point) are interpreted as reals. Exponents of ten may be specied via E or D format. For instance, 45, 45.0, 45.0E0, and 45.0D0 all represent the real value of 45.0 for the scale factor.
Chapter 17
1 The
modules described in this chapter may also be applied for concentration ow analysis.
312 Transient
*FILOS INITIA *INPUT *HEATTR *END
The next section gives some information on general aspects of modeling for a heat ow analysis.
17.1
This section describes the general aspects for heat ow analysis: shape and dimensions of the nite element mesh, conductivity, capacitance, sources, xed temperatures, boundary convection and radiation, initial temperature and ow elds, Finally we give an overview of special features: hydration heat, cooling pipes, solidication or evaporation, subsequent stress analysis, phased analysis, and a convective eld.
17.1.1
Meshing
The shape and dimensions of the analysis domain are modeled by a nite element mesh of general ow continuum elements: triangles, quadrilaterals, wedges, or bricks. Available are continuum elements for two-dimensional models, for axisymmetric models, and for three-dimensional models. See Volume Element Library for more information.
17.1.2
The conductivity and capacitance of continuum elements are specied in table MATERI. Specication is possible as a constant property, and besides also as a multilinear function of temperature, time, or both. The conductivity may also be anisotropic. See Volume Material Library for more information.
17.1.3
Sources and temperatures can be prescribed using table BOUNDA, and scaled as a multilinear function of time, using table TIMEBO [ 16.2.2 p. 307]. Point sources can be specied as boundary conditions via subtable NODAL of table BOUNDA [ 16.2.1 p. 307], while line, plane, and volume sources can be specied as boundary conditions for continuum elements via subtable ELEMEN [Vol. Element Library]. Temperature dependent sources (point, lines, planes) can be modeled by a nonlinear conduction coecient in boundary elements or interface
17.1 General Modeling Aspects elements [Vol. Material Library]. Prescribed temperatures at nodes can be specied via table FIXPOT [ 16.1.1 p. 303] and subtable NODAL of table BOUNDA [ 16.2.1 p. 307].
313
17.1.4
Convection or radiation at boundaries of the continuum model can be modeled by the convection coecient of boundary elements or the conduction coecient of interface elements [Vol. Material Library]. Line shaped elements are used at the faces of a two-dimensional continuum model and triangular or quadrilateral elements at the faces of a three-dimensional model. The convection, conduction and radiation coecient are specied as material properties [Vol. Material Library]. Radiation may also be simulated by dening an appropriate temperature dependent convection coecient. In Volume Analysis Examples you may nd two examples of NAFEMS [43] benchmarks run with Diana: nafet4 which shows a linear steadystate heat ow analysis with convection, and nafet2 which shows a nonlinear steadystate heat ow with radiation.
17.1.5
Interface elements. Input of the conduction coecient for interface elements is possible as a constant property or as a multilinear function of the temperature dierence, time, or both, where the dierence refers to the internal and the external temperature [Vol. Material Library]. The external temperature is dened by temperatures in the nodes of the external face. Prescribed external temperatures are input via subtable NODAL of table BOUNDA [ 16.2.1 p. 307]. Boundary elements. Input of the convection coecient for boundary elements is possible as a constant property and as a multilinear function of boundary temperature and/or time [Vol. Material Library]. The external temperature is now specied in subtable ELEMEN of table BOUNDA [Vol. Element Library]. Continuum elements. Input of the conduction coecient for continuum elements is possible as a constant property and as a multilinear function of boundary temperature and/or time [Vol. Material Library]. Example rodht in Volume Analysis Examples shows a nonlinear transient heat ow using temperature dependent conductivity and capacitance.
17.1.6
Table INIVAR is used for input of a temperature eld [ 16.1.3 p. 305]. In a nonlinear steadystate analysis, a specied temperature eld may be applied as a rst estimate of the solution. In a transient analysis, it may be applied as an initial temperature eld. For all heat ow analyses, Diana will evaluate
314
Heat Flow Analysis the model, conductivity and boundary conditions by default. In a transient analysis, capacitance is also evaluated by default. A steadystate analysis with Module heatss [ 17.2 p. 316], or execution of several time steps in a transient analysis with Module heattr [ 17.3 p. 322], results in computation of temperature and ow elds. Both these modules can be used for linear and iterative nonlinear analysis, i.e., for constant and for time/temperature dependent properties respectively. Transient analysis includes capacitance contributions and initial temperatures, if they are available. The output of temperature and uxes can be specied via the OUTPUT command block.
17.1.7
17.1.7.1
Special Features
Hydration Heat
Heat generation based on the degree of reaction is available for continuum elements, to model chemical reactions like cement hydration. Heat generation characteristics and dependency of conductivity and capacitance on this degree of reaction, are specied as material properties with multilinear diagrams [Vol. Material Library]. Intended usage and initialization of the degree of reaction are specied as subcommands of a nonlinear analysis with Module heattr [ 17.3.2.1 p. 326]. The output of resulting degrees of reaction during the transient nonlinear analysis can be specied via the OUTPUT command block. Example hydrat in Volume Analysis Examples shows a nonlinear transient analysis of heat generation caused by cement hydration. 17.1.7.2 Equivalent Age
As an alternative to the degree of reaction, Diana can calculate the equivalent age as a maturity variable to quantify the progress of the hydration reaction. For such calculation you must specify special material characteristics [Vol. Material Library]. With a special command for Module heattr you may invoke an equivalent age calculation [ 17.3.2.1 p. 326]. The output of the resulting new maturity variable during the nonlinear transient analysis can be obtained via the OUTPUT command block [ 17.4.6 p. 333]. 17.1.7.3 Cooling Pipes
Cooling pipes are used to reduce temperature gradients due to cement hydration. Input of uid properties and the conduction coecient of the pipe lining is specied as a material property [Vol. Material Library]. To model the heat convection implicitly or explicitly, Diana oers a set of cooling pipe elements [Vol. Element Library]. These element types are usually combined with threedimensional general ow continuum elements. The implicit element L4HT has separate nodes for the uid temperature. Usage results in nonsymmetric matrices which Diana will setup automatically.
17.1 General Modeling Aspects A transient analysis results in computation of temperatures of both concrete and uid [ 17.3 p. 322]. The explicit element L2HT uses internal uid temperatures, which are adapted by Module heattr after each time step. A pipe composed from explicit elements needs additional input from table COOLPI. Internal temperatures can be obtained via the OUTPUT command block. 17.1.7.4 Solidication or Evaporation
315
A material generally consumes heat at the temperatures of solidication or evaporation. This heat consumption can be modeled inside a transient nonlinear analysis [ 17.3.2.1 p. 326], by specication of a temperature dependent multilinear enthalpy function as a material property [Vol. Material Library]. Diana derives a nonlinear capacitance from this function. Example solidi in Volume Analysis Examples shows a nonlinear transient heat ow analysis of solidication. 17.1.7.5 Subsequent Stress Analysis
Dianas Module nonlin allows you to couple a thermal stress analysis to a previous heat ow analysis. This is a so-called staggered analysis [Ch. 22], where Module nonlin generates the input of load, temperature and maturity for a structural analysis from the results of the heat analysis. Part IV in this volume describes how to perform the subsequent nonlinear structural analysis. 17.1.7.6 Phased Analysis
Dianas Module phase enables the combination of changing geometry and boundary conditions with transient potential ow analysis [Ch. 25]. The module is used, for example, to analyze cement hydration during phased concrete casting. Table EQUAL is especially used in combination with phased analysis, to enable nodal splitting and coupling [ 16.1.2 p. 304]. 17.1.7.7 Convective Field
Temperature transport by for example groundwater ow can be analyzed by application of a convective eld, additional to the conduction. Diana can derive the convective eld from manual input, or by importing the ux eld resulting from a preceding groundwater ow analysis on the same mesh [Vol. Material Library]. The conduction of the material is input as constant diusivity. A ux multiplication factor, equal to the volumetric capacitance of the uid fraction is also material input. Example conow in Volume Geotechnical Analysis shows an analysis of temperature transport by forced convection.
316
17.2
Steady-state Analysis
Module heatss is used to perform a steadystate heat ow analysis. Basically there are two types of steadystate analysis: linear if the material properties are constant or nonlinear if they are variable. syntax *HEATSS [ MODEL [ OFF ] ] [ EXECUT [ ] ] LINEAR NONLIN OFF [ OUTPUT [ OFF ] ] MODEL to evaluate the nite element model [ 17.2.1].
[LINEAR]
EXECUT to execute the steadystate heat ow analysis. The LINEAR option asks for a linear steadystate analysis [ 17.2.2], the NONLIN option for a nonlinear steadystate analysis [ 17.2.3]. OUTPUT to specify analysis results for output [ 17.4]. Default
*HEATSS *END file .dcf
If, like in the above commands, you only give the *HEATSS command, then Diana will execute a complete linear steadystate analysis with the default output. This is equivalent to the following commands.
file .dcf *HEATSS BEGIN MODEL EVALUA ASSEMB MATRIX CONDUC BOUNDA END MODEL BEGIN EXECUT LINEAR SOLVE END EXECUT BEGIN OUTPUT TEMPER FLUX LOCAL NODES END OUTPUT *END
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VI)
317
17.2.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual analysis. syntax BEGIN MODEL [ EVALUA [ OFF ] ] [ ASSEMB [ OFF ] ] [ MATRIX [ CONDUC ] [ OFF ] ] [ BOUNDA [ OFF ] ] END MODEL EVALUA to check and evaluate geometric and material properties for elements [ 3.4 p. 66]. ASSEMB to assemble the elements, i.e., creating appropriate system degrees of freedom. MATRIX CONDUC to setup the element conductivity matrices. The OFF option switches o the creation of conductivity matrices. This may save computing time if the matrices are still available on the filos le and appropriate. BOUNDA to convert boundary conditions to right-hand-side nodal ux vectors. The OFF option switches o the creation of right-hand-side nodal ux vectors, which may be useful if these are still available on the filos le.
17.2.2
A linear steadystate analysis will do if temperature does not aect the material properties. Otherwise you must perform a nonlinear steadystate analysis [ 17.2.3]. syntax BEGIN EXECUT [ LINEAR ] [ SOLVE ] END EXECUT LINEAR will perform a linear steadystate heat ow analysis. SOLVE customizes the settings for the solution method [Ch. 26 p. 429].
Diana-9.4 Users Manual Analysis Procedures (VI) December 15, 2009 First ed.
318
17.2.3
A nonlinear heat ow analysis is required if the material properties vary with the temperature. You may perform such an analysis via the NONLIN commands. syntax BEGIN EXECUT BEGIN NONLIN [ BOUNDA { }] CASE=case n FACTOR=fac r [ ITERAT ] [ ESTIMA ] END NONLIN [ SOLVE ] END EXECUT NONLIN will perform a nonlinear steadystate heat ow analysis. BOUNDA species the boundary conditions. Parameter CASE=case refers to a boundary case number in table BOUNDA [ 16.2 p. 306]. Default is the lowest available case number. Parameter FACTOR=fac species a multiplication factor. ITERAT species the iteration method to be applied in the nonlinear solution procedure [ 17.2.3.1]. ESTIMA species a rst estimation for the temperature eld [ 17.2.3.2]. SOLVE customizes the settings for the solution method [Ch. 26 p. 429]. Default
*HEATSS [ commands ] BEGIN EXECUT NONLIN END EXECUT [ commands ] *END file .dcf
[fac =1.0]
If, like in the above commands, you give a single NONLIN command, then Diana will perform a nonlinear steadystate analysis without a rst estimation of the temperature eld, i.e., a zero temperature eld is the default, and with nonscaled boundary conditions from case 1. The default iteration procedure will be applied. This is equivalent to the following commands.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VI)
319
17.2.3.1
Iteration Method
With the ITERAT commands you may customize the iteration process that Diana will apply in the nonlinear solution procedure. See also 35.1.3.2 on page 552 for some background theory on these iteration methods. syntax BEGIN ITERAT [ MAXITE=mi n ] [ METHOD [ NEWTON ]
] REGULA MODIFI [ [ CONVER [ TEMPER ] ] TOLCON=eps r ] END ITERAT MAXITE=mi is the maximum number of iterations. METHOD species the iteration method: NEWTON indicates a NewtonRaphson iteration scheme (the only option), i.e., the conductivity matrix will be updated. MODIFI invokes the Modied NewtonRaphson method. REGULA invokes the Regular NewtonRaphson method (the default). CONVER species the convergence criterion for the iteration process: TEMPER indicates a convergence criterion on the norm of the temperature eld (the only option). Parameter TOLCON=eps is the tolerance on the reference norm. When the norm of the incremental temperature eld has become less than the reference norm Diana assumes sucient accuracy and stops the iteration process [Eq. (35.32) p. 552].
Diana-9.4 Users Manual Analysis Procedures (VI)
[mi =5]
[REGULA]
[ = 106 ]
320 Default
*HEATSS [ commands ] BEGIN EXECUT BEGIN NONLIN [ commands ] ITERAT [ commands ] END NONLIN END EXECUT [ commands ] *END
If, like in the above commands, you give a single ITERAT command, then Diana will apply a Regular NewtonRaphson iteration scheme with at most ve iterations. As convergence criterion, the norm of the temperatures will be applied with a tolerance of 106 . This is equivalent to the following commands.
file .dcf *HEATSS [ commands ] BEGIN EXECUT BEGIN NONLIN [ commands ] BEGIN ITERAT MAXITE=5 METHOD NEWTON REGULA CONVER TEMPER TOLCON=1.E-6 END ITERAT [ commands ] END NONLIN END EXECUT [ commands ] *END
17.2.3.2
With the ESTIMA commands you may specify a temperature eld which Diana will apply as a rst estimation of the solution. syntax BEGIN ESTIMA [ OFF ] [ TEMPER [ OFF ] [ INPUT
December 15, 2009 First ed.
] { }
321
CALCUL
END ESTIMA TEMPER indicates that the estimation is for a temperature eld (the only possibility). There are two methods to specify the temperature eld. For both options parameter FACTOR=fac is an optional multiplication factor. INPUT indicates a start with temperatures as specied in table INIVAR [ 16.1.3 p. 305] (the default). Parameter FIELD=field refers to a eld number in this table. Default is the lowest available eld number. CALCUL indicates a start with temperatures as calculated in a steadystate analysis. If temperatures are available from a previous steadystate analysis, these temperatures are used. Otherwise a linear steady state analysis is performed. Parameter CASE=case refers to a boundary case number in table BOUNDA [ 16.2 p. 306]. Default is the lowest available case number. Default
*HEATSS [ commands ] BEGIN EXECUT BEGIN NONLIN [ commands ] ESTIMA [ commands ] END NONLIN END EXECUT [ commands ] *END file .dcf
[fac =1]
[INPUT]
If, like in the above commands, you give a single ESTIMA command, then Diana will apply the potentials of the lowest eld number in input table INIVAR as a rst estimation of the temperature eld. This is equivalent to the following commands.
file .dcf *HEATSS [ commands ] BEGIN EXECUT BEGIN NONLIN
Diana-9.4 Users Manual Analysis Procedures (VI) December 15, 2009 First ed.
322
[ commands ] BEGIN ESTIMA TEMPER INPUT FIELD=lowest FACTOR=1.0 END ESTIMA [ commands ] END NONLIN END EXECUT [ commands ] *END
17.3
Transient Analysis
Module heattr is used to perform a transient heat ow analysis. In order to perform such analysis, you must additionally specify the time dependency of boundary conditions in table TIMEBO [ 16.2.2]. After the preliminary tasks like model evaluation and setup of the system of equations, a transient analysis basically involves two steps: (1) initiation of a linear or nonlinear transient analysis, (2) solution of the equations for the specied time steps. You may perform these steps in one Diana job or in various subsequent jobs. syntax *HEATTR [ MODEL [ OFF ] ] [ INITIA [ OFF ] ] [ EXECUT [ OFF ] ] [ OUTPUT [ OFF ] ] MODEL to evaluate the nite element model [ 17.3.1]. INITIA to initiate a transient heat ow analysis [ 17.3.2]. EXECUT to execute time steps in the transient heat ow analysis [ 17.3.3]. OUTPUT to specify analysis results for output [ 17.4].
17.3.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual analysis.
17.3 Transient Analysis syntax BEGIN MODEL [ EVALUA [ OFF ] [ ASSEMB [ OFF ] [ MATRIX [ OFF ] [ BOUNDA [ OFF ] END MODEL ] ] ] ]
323
EVALUA to check and evaluate geometric and material properties for elements [ 3.4 p. 66]. ASSEMB to assemble the elements, i.e., creating appropriate system degrees of freedom. MATRIX to setup the element conductivity and capacity matrices [ 17.3.1.1]. BOUNDA to convert boundary conditions to right-hand-side nodal ux vectors. The OFF option switches o the creation of right-hand-side nodal ux vectors, which may be useful if these are still available on the filos le. 17.3.1.1 Conductivity and Capacity Matrices
By default, Diana will set up the element conductivity and capacity matrices each time that you invoke Module heattr. You may explicitly ask Diana to setup the matrices via MATRIX commands. syntax BEGIN MATRIX [ OFF ] [ CONDUC [ OFF ] ] [ CAPACI [ OFF ] [ CONSIS LUMPED END MATRIX OFF switches o the creation of matrices. This may save computing time if these are are still available on the filos le and appropriate. CONDUC sets up the element conductivity matrices. CAPACI sets up the element capacity matrices. You may specify the type of matrices with CONSIS for consistent matrices (the default), or with LUMPED for lumped matrices.
[CONSIS]
]]
324 Default
*HEATTR BEGIN MODEL MATRIX END MODEL [ commands ] *END
If, like in the above commands, you give a single MATRIX command, then Diana will setup the conductivity matrices and the consistent capacity matrices, which would also occur if you didnt give the MATRIX command at all. This is equivalent to the following commands.
file .dcf *HEATTR BEGIN MODEL BEGIN MATRIX CONDUC CAPACI CONSIS END MATRIX END MODEL [ commands ] *END
17.3.2
The INITIA commands initiate the transient analysis. The main purpose of this task is to evaluate the initial state of the model, i.e., prior to the execution of the rst time step. syntax BEGIN INITIA [ TEMPER [ OFF ] [ INPUT ] { FIELD=field n FACTOR=fac r } { CASE=case n FACTOR=fac r } ]
CALCUL
17.3 Transient Analysis NONLIN END INITIA TEMPER indicates an initial temperature eld. There are two methods to specify the temperature eld. For both options parameter FACTOR=fac is an optional multiplication factor. INPUT indicates a start with temperatures as specied in table INIVAR [ 16.1.3 p. 305]. Parameter FIELD=field refers to a eld number in this table. Default is the lowest available eld number. CALCUL indicates a start with temperatures as calculated in a steadystate analysis. If temperatures are available from a previous steadystate analysis, these temperatures are used. Otherwise a linear steady state analysis is performed. Parameter CASE=case refers to a boundary case number in table BOUNDA [ 16.2 p. 306]. Default is the lowest available case number. SOLVE customizes the settings for the solution method to be applied for the linear steadystate analysis [Ch. 26 p. 429]. TIME sttime is a user specied starting time. LINEAR initiates a linear analysis, i.e., with constant material properties. NONLIN initiates a nonlinear analysis, i.e., with time or temperature dependent material properties [ 17.3.2.1]. Default
*HEATTR [ commands ] INITIA [ commands ] *END file .dcf
[sttime =0] [LINEAR] [INPUT] [fac =1]
325
If, like in the above commands, you give a single INITIA command, then Diana will apply the potentials of eld 1 in input table INIVAR as an initial temperature eld. This is equivalent to the following commands.
file .dcf *HEATTR [ commands ] BEGIN INITIA TEMPER INPUT FIELD=lowest FACTOR=1.0 LINEAR END INITIA [ commands ] *END
A nonlinear heat ow analysis is required if the material properties vary with the temperature or the time. You may initiate such an analysis via the NONLIN commands. syntax BEGIN NONLIN [ HYDRAT [ ]] DGRINI=r0 r OFF [ EQUAGE [ ]] EQAINI=te0 r OFF END NONLIN HYDRAT initiates a nonlinear transient analysis with heat generation due to cement hydration. Parameter DGRINI=r0 species the initial degree of reaction r0 . The OFF option suppresses the application of heat generation due to hydration, as if you omitted the HYDRAT command. EQUAGE invokes the calculation of the equivalent age teq . Parameter EQAINI=te0 species the initial equivalent age teq0 . The OFF option suppresses the calculation of the equivalent age, as if you omitted the EQUAGE command.
[r0 = 0.1] (0 r0 1)
[teq0 = 0]
17.3.3
With the EXECUT commands you ask Diana to execute time steps in a transient heat ow analysis. A command le may contain one or more EXECUT blocks. Details of various commands are given in the referred sections. syntax BEGIN EXECUT [ OFF ] [ ALPHA=alpha r ] [ SIZES sizes r... ] [ NONLIN ] [ SOLVE ] ]] [ SAVE [ STEPS steps n... ALL LAST NONE END EXECUT
17.3 Transient Analysis ALPHA alpha is the time integration parameter . Usual methods are: Euler forward ( = 0), CrankNicolson ( = 1 ), Galerkin ( = 2 ) and Euler 2 3 backward ( = 1) which is the default. See 35.1.3.1 on page 551 for some background theory on time integration. SIZES sizes are explicitly specied time step sizes, i.e., time increments. The number of values species the number of steps to be executed. NONLIN species options for nonlinear transient analysis [ 17.3.3.1]. SOLVE customizes the settings for the solution method [Ch. 26 p. 429]. SAVE species the steps for which analysis results must be saved for a subsequent structural analysis in a staggered owstress analysis [Ch. 22]. By default, in a model for staggered owstress analysis, Diana will save the results for all executed steps. STEPS steps indicates the steps for which the analysis results must be saved. Alternatively you may save ALL steps, only the LAST step, or NONE of the steps. The NONE option drops the possibility of a subsequent structural analysis! Default
*HEATTR [ commands ] EXECUT *END file .dcf
327
(0 1) [ = 1]
[SIZES 1.0]
[ALL]
Due to these commands Diana will execute a time step with a factor of 1, applying the default solution procedure, and giving the default output of analysis results. The same would occur if you had given the following commands.
file .dcf *HEATTR [ commands ] BEGIN EXECUT ALPHA=1.0 SIZES 1.0 SOLVE END EXECUT BEGIN OUTPUT TEMPER FLUX LOCAL NODES REACTI NODES INTTMP NODES END OUTPUT *END
328
Degrees of reaction (REACTI) will only be output in case of hydration heat analysis [ 17.4.5 p. 333]. Internal temperatures (INTTMP) will only be output in case of nonlinear transient analysis with cooling pipe elements [ 17.4.7 p. 334]. 17.3.3.1 Nonlinear Analysis Options
Via the NONLIN commands you may customize a nonlinear transient analysis, provided that such analysis has been initiated previously [ 17.3.2.1 p. 326]. syntax BEGIN NONLIN [ BEGIN ITERAT [ MAXITE=mi n ] [ METHOD [ NEWTON ]
] REGULA MODIFI [ CONVER [ TEMPER ] [ TOLCON=eps r ] ] END ITERAT ] [ HYDRAT [ ]] STEP ITERAT END NONLIN ITERAT species the iteration method to be applied in the nonlinear solution procedure. See also 35.1.3.2 on page 552 for some background theory on these iteration methods.
[mi =5]
MAXITE=mi is the maximum number of iterations. METHOD species the iteration method: NEWTON for a NewtonRaphson iteration scheme (the only option), i.e., the conductivity matrix will be updated.
[REGULA]
MODIFI invokes the Modied NewtonRaphson method. REGULA invokes the Regular NewtonRaphson method (the default). CONVER species the convergence criterion for the iteration process: TEMPER indicates a convergence criterion on the norm of the temperature eld (the only option). Parameter TOLCON=eps is the tolerance on the reference norm. When the norm of the incremental temperature eld has become less than the reference norm Diana assumes sucient accuracy and stops the iteration process [Eq. (35.32) p. 552].
[ = 106 ]
17.4 Output of Analysis Results HYDRAT indicates when the degree of reaction, and thus the element heat production, must be updated. This option only applies for hydration heat analysis. STEP for update after each time step (the default). ITERAT for update within the time step after each iteration. Default
*HEATTR [ commands ] BEGIN EXECUT [ commands ] NONLIN END EXECUT *END file .dcf
[STEP]
329
If, like in the above example, you only give a single NONLIN command then Diana will perform a nonlinear analysis applying a Regular NewtonRaphson iteration scheme with at most ve iterations. As convergence criterion, the norm of the temperatures will be applied with a tolerance of 106 . In case of an hydration heat analysis, the heat production will be updated after each time step. This is equivalent to the following commands.
file .dcf *HEATTR [ commands ] BEGIN EXECUT BEGIN NONLIN [ commands ] BEGIN ITERAT MAXITE=5 METHOD NEWTON REGULA CONVER TEMPER TOLCON=1.E-6 END ITERAT HYDRAT STEP END NONLIN END EXECUT *END
17.4
You can get output of analysis results from a heat ow analysis via the OUTPUT commands. For the general, analysis type independent options for output of analysis results see 3.6 on page 69.
Diana-9.4 Users Manual Analysis Procedures (VI) December 15, 2009 First ed.
330
Heat Flow Analysis syntax BEGIN OUTPUT [device w ] [outopt w... ] [params ] [ OFF ] [ BEGIN SELECT [ ] [ BOUNDA ] cases n... ALL [ STEPS ] END SELECT ] [ LAYOUT ] item w TEMPER FLUX FLOW REACTI EQUAGE INTTMP END OUTPUT SELECT command block to customize the batch output. for model selection see 3.6.1 on page 71. BOUNDA species a selection of boundary cases in linear steadystate analysis: cases is a series of case numbers referring to input table BOUNDA [ 16.2 p. 306], ALL selects all four boundary cases. STEPS selects time steps for output of transient heat ow analysis [ 17.4.1 p. 331]. LAYOUT optional commands to customize the layout and style of tabular output [ 3.6.3 p. 80]. item is the name of the analysis result to be output. TEMPER for nodal temperatures [ 17.4.2 p. 331]. FLUX for element uxes [ 17.4.3 p. 331]. FLOW for nodal ows [ 17.4.4 p. 332]. REACTI for degrees of reaction [ 17.4.5 p. 333], only available in hydration heat analysis. EQUAGE for equivalent age [ 17.4.6 p. 333], only available in hydration heat analysis. INTTMP for internal temperature of cooling pipe elements [ 17.4.7 p. 334], only available in transient nonlinear heat ow analysis with cooling pipe elements.
331
17.4.1
Step Selection
The STEPS command selects time steps for output of analysis results of a transient heat ow analysis. syntax STEPS [ steps n... ALL LAST steps are numbers of selected steps. These numbers are relative to the steps of the preceding EXECUT command block. ALL will produce output for all steps that are executed (the default). LAST selects the nal step.
[ALL]
17.4.2
Temperatures
The temperature eld is the solution vector of the system of equations for the heat ow analysis. Temperatures are scalar values calculated in the nodes, component selection is not applicable. syntax TEMPER { opti w } TEMPER gives the temperatures T . opti are additional options [ 3.6.2 p. 79].
item TEMPER PTE T
17.4.3
Fluxes
In a heat ow analysis, uxes are calculated for elements. They can be output in the integration points or in the element nodes. Diana also writes the integrated uxes over groups of boundary elements to the standard output le file .out. syntax FLUX [ oper w ] { comp w } [ loca w ] { opti w } LOCAL INTPNT GLOBAL NODES BOUNDA CENTER
oper species an operation to be performed on the uxes. LOCAL transforms uxes in continuum elements to local xyz directions. GLOBAL transforms uxes in continuum elements to global XY Z directions. BOUNDA gives the total discharge Q for boundary elements. This rate of ow is a scalar value, therefore component selection is not applicable.
[NODES]
loca species the location of the uxes to be output [ 3.6.2 p. 79]. opti are additional options [ 3.6.2 p. 79].
comp X Y Z
FLx FLy FLz
qx qX
qy qY
qz qZ
17.4.4
Flows
In a heat transfer analysis nodal ows are calculated. Nodal ows are scalar values, component selection is not applicable. syntax FLOW [ type w ] { opti w } REACTI RESIDU EXTERN FLOW gives the nodal ows.
[REACTI]
type
species the type of the nodal ow. REACTI for the reaction ow Q in all nodes with xed temperature. RESIDU for the residual ow Q, also called the outofbalance ow. This is dened as the dierence between the externally applied ow Q and the internal resistance ow. EXTERN for the externally applied ow Q.
333
17.4.5
Degrees of Reaction
In hydration analysis Diana calculates the degrees of reaction r [ 17.1.7.1 p. 314], which can be output in the element nodes and integration points. syntax REACTI [ loca w ] { opti w } INTPNT NODES CENTER REACTI gives the degrees of reaction r. loca species the location of the degrees of reaction to be output [ 3.6.2 p. 79]. opti are additional options [ 3.6.2 p. 79].
item REACTI DGR r
[NODES]
17.4.6
Equivalent Age
In hydration heat analysis Diana can calculate the equivalent age teq [ 17.3.2.1 p. 326]. This result can be output in the element integration points or in the element nodes. syntax EQUAGE [ loca w ] { opti w } INTPNT NODES CENTER EQUAGE gives the equivalent age teq . loca species the location of the equivalent age to be output [ 3.6.2 p. 79]. opti are additional options [ 3.6.2 p. 79].
Diana-9.4 Users Manual Analysis Procedures (VI) December 15, 2009 First ed.
[NODES]
334
item EQUAGE EqA teq
17.4.7
The internal uid temperatures of cooling pipe elements can be output in the element nodes. syntax INTTMP [ loca w ] { opti w } NODES INTTMP gives the internal temperature Tint .
[NODES]
loca species the location of the temperature to be output [ 3.6.2 p. 79]. opti are additional options [ 3.6.2 p. 79].
item INTTMP PTE Tint
Chapter 18
18.1
For a detailed analysis of horizontal and vertical components of groundwater ow with partly saturated areas Diana oers some special nonlinear groundwater ow options. First we describe the general topics: shape and dimensions, conductivity and storativity, sources and sinks, and seepage faces. Finally we give an overview of special features: turbulence, contamination or thermal transport, subsequent stress analysis, and phased analysis.
18.1.1
Meshing
The shape and dimensions of the analysis domain are modeled by a mesh of continuum groundwater ow elements: triangles, quadrilaterals, wedges, and bricks. Available are continuum elements for two-dimensional models, for axisymmetric models, and for three-dimensional models. See Volume Element Library for details. Diana assumes the direction of gravity in global Y direction for two-dimensional models, and in global Z direction for three-dimensional models.
18.1.2
The conductivity and storativity of continuum elements are derived from input of material properties. The elastic storativity and saturated conductivity are specied as constants. The conductivity can be isotropic or anisotropic. Nonlinear conductivity and storage are applied in partially saturated areas. Diana mostly derives the nonlinear phreatic storativity from the porosity and
Diana-9.4 Users Manual Analysis Procedures (VI) December 15, 2009 First ed.
336
Groundwater Flow Analysis a pressure dependent saturation, which is specied by the user as a multilinear function. It is also possible to specify directly a pressure dependent phreatic storativity. A relative conductivity can be specied as a multilinear function of the saturation or the pressure. See Volume Material Library for details. See also example sandcol in Volume Geotechnical Analysis, a benchmark of a transient nonlinear analysis of phreatic level rise in a sand column.
18.1.3
Sources and sinks are input via table BOUNDA. Their values can be scaled as a multilinear function of time, using table TIMEBO [ 16.2.2 p. 307]. Point sources can be specied as boundary conditions via subtable NODAL of table BOUNDA [ 16.2.1 p. 307], while line, plane, and volume sources can be specied as boundary conditions for continuum elements via subtable ELEMEN [Vol. Element Library]. Sources which depend on the hydraulic head (point, lines, planes) can be modeled by a nonlinear conduction coecient in boundary elements or interface elements [Vol. Material Library]. Prescribed hydraulic heads at nodes can be specied via table FIXPOT [ 16.1.1 p. 303] and subtable NODAL of table BOUNDA [ 16.2.1 p. 307].
18.1.4
Seepage Faces
At a seepage face the water ows out above the external phreatic level. Diana has special boundary elements available to model seepage faces [Vol. Element Library]. Line shaped elements are used at the faces of a two-dimensional continuum model and plane shaped elements at the faces of a three-dimensional model. The external phreatic level is specied in table BOUNDA [ 16.2 p. 306]. Diana imposes a zero pressure condition at the seepage face by use of a penalty conduction coecient. The default value may be adapted via material input [Vol. Material Library]. See also example seepfce in Volume Geotechnical Analysis, a benchmark of nonlinear steadystate analysis with partly unsaturated ow and a seepage face.
18.1.5
Table INIVAR is used for input of an initial hydraulic head [ 16.1.3 p. 305]. Diana will evaluate the model, conductivity and boundary conditions by default. A subsequent steadystate analysis with Module growss [p. 339], or execution of several time steps with module growtr [ 18.4 p. 346], results in computation of elds for hydraulic head and ux. Partly saturated parts always require an iterative nonlinear analysis [ 18.3.3 p. 341] [ 18.4.2 p. 348]. In transient analysis, storage contributions and initial elds are included. The output of hydraulic heads and uxes can be obtained via commands in the OUTPUT block. Output of the pressure component of the hydraulic head and of discharges is also possible.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VI)
337
18.1.6
18.1.6.1
Special Features
Turbulence
Turbulence can be included in a nonlinear analysis by input of the Forchheimer constants as a material property [Vol. Material Library]. 18.1.6.2 Contamination Transport
Contamination transport is analyzed similar to a temperature analysis with an imported convective eld from a groundwater ow analysis [ 17.1.7.7]. Example conow in Volume Geotechnical Analysis shows a steadystate analysis with convection of contaminants. 18.1.6.3 Stability and Consolidation
Dianas Module nonlin allows you to analyze for example the inuence of groundwater ow on the mechanical stability, by coupling a separate groundwater ow analysis to a subsequent eective stress analysia, a so-called staggered analysiss [Ch. 22]. The module generates structural loads in the input table LOADS from the computed hydraulic heads. Dianas Module nonlin also allows you to analyze for example consolidation of soil, by simultaneous solution of pore uid ow and stress, a so-called mixture analysis [Ch. 23]. 18.1.6.4 Phased Analysis
Dianas Module phase enables the combination of changing geometry and boundary conditions with transient groundwater ow analysis [Ch. 25]. The module is used for example to predict the transient changes in groundwater ow by introduction of sheet pilings, sinks, excavation etc. Table EQUAL is especially used in combination with phased analysis, to enable nodal splitting and coupling [ 16.1.2 p. 304].
18.2
A two-dimensional regional analysis of groundwater ow can predict the horizontal ow through aquifers, using the Dupuit assumption. These aquifers can be interacting by interjacent resistance layers (aquicludes). Nonlinear transmissivity and phreatic storage is introduced in unconned aquifers. The direction of gravity is always assumed in global Z direction, e.g. normal to the plane of the aquifer(s). In Volume Geotechnical Analysis you may nd two examples of aquifer analysis. Caquif shows a linear steadystate analysis in a conned aquifer. Aquifs shows a nonlinear steady-and transient analysis of a combination of a semiconned aquifer, an intermediate resistance layer and an unconned aquifer.
338
18.2.1
Meshing
Aquifers are modeled by two-dimensional quadrilateral and triangular aquifer elements. For these elements you must specify the top and bottom level. Interjacent resistance layers are modeled by ow interface elements. See Volume Element Library for details.
18.2.2
The saturated conductivity, the elastic storativity and the porosity (phreatic storativity) are input for aquifers. A conduction coecient is input for the resistance layers. Input is specied as material properties, see Volume Material Library for details.
18.2.3
Sources and sinks are input via table BOUNDA. Their values can be scaled as a multilinear function of time, via TIMEBO [ 16.2.2 p. 307]. Point sources can be specied as boundary conditions via subtable NODAL of table BOUNDA [ 16.2.1 p. 307], while line and plane sources can be specied on the aquifer elements via subtable ELEMEN. Prescribed hydraulic heads at nodes can be specied via table FIXPOT [ 16.1.1 p. 303] and subtable NODAL of table BOUNDA [ 16.2.1 p. 307].
18.2.4
Table INIVAR is used for input of an initial hydraulic head [ 16.1.3 p. 305]. Diana will evaluate the model, conductivity and boundary conditions by default. A subsequent steadystate analysis with Module growss [ 18.3 p. 339], or execution of several time steps with module growtr [ 18.4 p. 346], results in computation of elds for hydraulic head and ux. Unconned aquifers always require an iterative nonlinear analysis [ 18.3.3 p. 341] [ 18.4.2 p. 348]. In transient analysis are storage contributions and initial elds included. The output of hydraulic heads and uxes can be obtained via the OUTPUT command block.
18.2.5
18.2.5.1
Special Features
Contamination Transport
Contamination transport is analyzed similar to a temperature analysis with an imported convective eld from a groundwater ow analysis. 18.2.5.2 Phased Analysis
Dianas Module phase enables the combination of changing geometry and boundary conditions with transient potential ow analysis. The module is used,
18.3 Steadystate Analysis for example, to model the initial eects to groundwater ow due to introduction of sheet pilings, sinks, excavation etc. [Ch. 25].
339
18.3
Steadystate Analysis
Module growss is used to perform a steadystate groundwater ow analysis. Basically there are two types of steadystate analysis: linear if the material properties are constant or nonlinear if they are variable. syntax *GROWSS [ MODEL [ OFF ] ] [ EXECUTE [ ] ] LINEAR NONLIN OFF [ OUTPUT [ OFF ] ] MODEL to evaluate the nite element model [ 18.3.1]. EXECUT to execute the steadystate groundwater ow analysis. The LINEAR option asks for a linear steadystate analysis [ 18.3.2], the NONLIN option for a nonlinear steadystate analysis [ 18.3.3]. OUTPUT to specify analysis results for output [ 18.5]. Default
*GROWSS *END file .dcf
[LINEAR]
If, like in the above commands, you only give the *GROWSS command, then Diana will execute a complete linear steadystate analysis with the default output. This is equivalent to the following commands.
file .dcf *GROWSS BEGIN MODEL EVALUA ASSEMB MATRIX CONDUC BOUNDA END MODEL BEGIN EXECUT LINEAR SOLVE
Diana-9.4 Users Manual Analysis Procedures (VI) December 15, 2009 First ed.
340
END EXECUT BEGIN OUTPUT PRESSU POTENT FLUX LOCAL NODES END OUTPUT *END
18.3.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual analysis. syntax BEGIN MODEL [ EVALUA [ OFF ] [ ASSEMB [ OFF ] [ MATRIX [ OFF ] [ BOUNDA [ OFF ] END MODEL ] ] ] ]
EVALUA to check and evaluate geometric and material properties for elements [ 3.4 p. 66]. ASSEMB to assemble the elements, i.e., creating appropriate system degrees of freedom. MATRIX to setup the element conductivity matrices [ 18.3.1.1]. BOUNDA to convert boundary conditions to right-hand-side nodal ux vectors. The OFF option switches o the creation of right-hand-side nodal ux vectors, which may be useful if these are still available on the filos le. 18.3.1.1 Conductivity Matrices
By default, Diana will set up the element conductivity matrices each time that you invoke Module growss. You may customize the setup of conductivity matrices via the MATRIX commands. syntax BEGIN MATRIX BEGIN CONDUC [ OFF ] [ CONVEC [ OFF ] {
December 15, 2009 First ed.
}]
Diana-9.4 Users Manual Analysis Procedures (VI)
18.3 Steadystate Analysis CASE=case n FACTOR=fac r END CONDUC END MATRIX CONDUC will set up the conductivity matrices. The OFF option switches o the creation of conductivity matrices, this may save computing time if the matrices are still available on the filos le and appropriate. CONVEC will set up the full nonsymmetric conductivity matrices for convective elds as calculated, and stored on the filos le, during a previous potential ow analysis.1 The OFF option switches o the creation of these special conductivity matrices. CASE=case species the case number to be selected from the calculated element uxes if the previous calculation was linear steadystate. The case number refers to the case number in table BOUNDA [ 16.2 p. 306]. Default is the lowest available case number. FACTOR=fac takes into account the components of the rate of ow of the previous groundwater ow analysis. In this case fac is the multiplication factor for the calculated ow velocity.
341
[fac =1.0]
18.3.2
A linear steadystate analysis will do if potential does not aect the material properties. Otherwise you must perform a nonlinear steadystate analysis [ 18.3.3]. syntax BEGIN EXECUT LINEAR [ SOLVE ] END EXECUT LINEAR will perform a linear steadystate groundwater ow analysis. SOLVE customizes the settings for the solution method [Ch. 26 p. 429].
18.3.3
A nonlinear groundwater ow analysis is required if the material properties vary with the potential, for instance with unconned aquifers. You may perform such an analysis via the NONLIN commands.
1 Example
342
Groundwater Flow Analysis syntax BEGIN EXECUT BEGIN NONLIN }] [ BOUNDA { CASE=case n FACTOR=fac r [ ITERAT ] [ ESTIMA ] END NONLIN [ SOLVE ] END EXECUT NONLIN will perform a nonlinear steadystate groundwater ow analysis. BOUNDA species the boundary conditions. Parameter CASE=case refers to a boundary case number in table BOUNDA [ 16.2 p. 306]. Default is the lowest available case number. Parameter FACTOR=fac species a multiplication factor. ITERAT species the iteration method to be applied in the nonlinear solution procedure [ 18.3.3.1]. ESTIMA species a rst estimation for the potential eld [ 18.3.3.2]. SOLVE customizes the settings for the solution method [Ch. 26 p. 429]. Default
*GROWSS [ commands ] BEGIN EXECUT NONLIN END EXECUT [ commands ] *END file .dcf
[fac =1.0]
If, like in the above commands, you give a single NONLIN command, then Diana will perform a nonlinear steadystate analysis without a rst estimation of the potential eld and with non-scaled boundary conditions from the lowest case number. The default iteration procedure will be applied. This is equivalent to the following commands.
file .dcf *GROWSS [ commands ] BEGIN EXECUT
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VI)
343
18.3.3.1
Iteration Method
With the ITERAT commands you may customize the iteration process that Diana will apply in the nonlinear solution procedure. See also 35.1.3.2 on page 552 for some background theory on these iteration methods. syntax BEGIN ITERAT [ MAXITE=mi n ] [ METHOD [ NEWTON ]
] REGULA MODIFI [ [ CONVER [ POTENT ] ] TOLCON=tc n ] END ITERAT MAXITE=mi is the maximum number of iterations. METHOD species the iteration method: NEWTON indicates a NewtonRaphson iteration scheme (the only option), i.e., the conductivity matrix will be updated. MODIFI invokes the Modied NewtonRaphson method. REGULA invokes the Regular NewtonRaphson method (the default). CONVER species the convergence criterion for the iteration process: POTENT indicates a convergence criterion on the norm of the potential eld (the only option). Parameter TOLCON=tc is the tolerance on the reference norm. When the norm of the incremental potential eld has become less than tc the reference norm Diana assumes sucient accuracy and stops the iteration process.
[tc =103 ] [REGULA] [mi =5]
344 Default
*GROWSS [ commands ] BEGIN EXECUT BEGIN NONLIN [ commands ] ITERAT [ commands ] END NONLIN END EXECUT [ commands ] *END
If, like in the above commands, you give a single ITERAT command, then Diana will apply a Regular NewtonRaphson iteration scheme with at most ve iterations. As convergence criterion, the norm of the potentials will be applied with a tolerance of 103 . This is equivalent to the following commands.
file .dcf *GROWSS [ commands ] BEGIN EXECUT BEGIN NONLIN [ commands ] BEGIN ITERAT MAXITE=5 METHOD NEWTON REGULA CONVER POTENT TOLCON=1.E-3 END ITERAT [ commands ] END NONLIN END EXECUT [ commands ] *END
18.3.3.2
With the ESTIMA commands you may specify a potential eld which Diana will apply as a rst estimation of the solution. syntax BEGIN ESTIMA [ OFF ] [ POTENT [ OFF ] [ INPUT
December 15, 2009 First ed.
] { }
345
CALCUL
END ESTIMA POTENT indicates that the estimation is for a potential eld (the only possibility). There are two methods to specify the potential eld. For both options parameter FACTOR=fac is an optional multiplication factor. INPUT indicates a start with potential as specied in table INIVAR [ 16.1.3 p. 305]. Parameter FIELD=field refers to a eld number in this table. Default is the lowest available eld number. CALCUL indicates a start with potentials as calculated in a steadystate analysis. If potentials are available from a previous steadystate analysis, these potentials are used. Otherwise a linear steadystate analysis is performed. Parameter CASE=case refers to a boundary case number in table BOUNDA [ 16.2 p. 306]. Default is the lowest available case number. Default
*GROWSS [ commands ] BEGIN NONLIN [ commands ] ESTIMA [ commands ] END NONLIN [ commands ] *END file .dcf
[fac =1]
If, like in the above commands, you give a single ESTIMA command, then Diana will apply the potentials of the lowest eld number in input table INIVAR as a rst estimation of the potential eld. This is equivalent to the following commands.
file .dcf *GROWSS [ commands ] BEGIN NONLIN [ commands ] BEGIN ESTIMA POTENT INPUT FIELD=lowest FACTOR=1.0 END ESTIMA
Diana-9.4 Users Manual Analysis Procedures (VI) December 15, 2009 First ed.
346
[ commands ] END NONLIN [ commands ] *END
18.4
Transient Analysis
Module growtr is used to perform a transient groundwater ow analysis. In order to perform such analysis, you must additionally specify the time dependency of boundary conditions in table TIMEBO [ 16.2.2]. After the preliminary tasks like model evaluation and setup of the system of equations, a transient analysis basically involves two steps: (1) initialization of initial potentials and boundary conditions, (2) solution of the equations for the specied time steps. You may perform these steps in one Diana job or in various subsequent jobs. syntax *GROWTR [ MODEL [ OFF ] ] [ INITIA [ OFF ] ] [ EXECUTE [ OFF ] ] [ OUTPUT [ OFF ] ] MODEL to evaluate the nite element model [ 18.4.1]. INITIA to initiate a transient groundwater ow analysis [ 18.4.2]. EXECUT to execute time steps in the transient groundwater ow analysis [ 18.4.3]. OUTPUT to specify analysis results for output [ 18.5].
18.4.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual analysis. syntax BEGIN MODEL [ EVALUA [ OFF ] [ ASSEMB [ OFF ] [ MATRIX [ OFF ] [ BOUNDA [ OFF ] END MODEL ] ] ] ]
18.4 Transient Analysis EVALUA to check and evaluate geometric and material properties for elements [ 3.4 p. 66]. ASSEMB to assemble the elements, i.e., creating appropriate system degrees of freedom. MATRIX to setup the element conductivity and storativity matrices [ 18.4.1.1]. BOUNDA to convert boundary conditions to right-hand-side nodal ux vectors. The OFF option switches o the creation of right-hand-side nodal ux vectors, which may be useful if these are still available on the filos le. 18.4.1.1 Conductivity and Storativity Matrices
347
By default, Diana will set up the element conductivity and storativity matrices each time that you invoke Module growtr. You may explicitly ask Diana to setup the matrices, or customize their setting up, via the MATRIX commands. syntax BEGIN MATRIX [ OFF ] [ CONDUC [ OFF ] ] [ STORAT [ OFF ] [ CONSIS LUMPED END MATRIX OFF switches o the creation of conductivity and storativity matrices. This may save computing time if the matrices are still available on the filos le and appropriate. CONDUC sets up the element conductivity matrices. STORAT sets up the element storativity matrices. You may specify the type of matrices with CONSIS for consistent matrices (the default), or with LUMPED for lumped matrices. Default
*GROWTR BEGIN MODEL MATRIX END MODEL [ commands ] *END file .dcf
[CONSIS]
]]
If, like in the above commands, you give a single MATRIX command, then Diana will setup the conductivity matrices and the consistent storativity matrices, which would also occur if you didnt give the MATRIX command at all. This is equivalent to the following commands.
Diana-9.4 Users Manual Analysis Procedures (VI) December 15, 2009 First ed.
348
18.4.2
The INITIA commands initiate the transient analysis. The main purpose of this task is to evaluate the initial state of the model, i.e., prior to the execution of the rst time step. syntax BEGIN INITIA [ POTENT [ OFF ] [ INPUT ] { FIELD=field n FACTOR=fac r { } CASE=case n FACTOR=fac r } ]
CALCUL
[ SOLVE ] [ TIME=sttime r ] [ ] LINEAR NONLIN END INITIA POTENT indicates an initial potential eld. There are two methods to specify the potential eld. For both options parameter FACTOR=fac is an optional multiplication factor. INPUT indicates a start with potentials as specied in table INIVAR [ 16.1.3 p. 305]. Parameter FIELD=field refers to a eld number in this table. Default is the lowest available eld number. CALCUL indicates a start with potentials as calculated in a steadystate analysis. If potentials are available from a previous steadystate
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VI)
[fac =1]
18.4 Transient Analysis analysis, these potentials are used. Otherwise a linear steadystate analysis is performed. Parameter CASE=case refers to a boundary case number in table BOUNDA [ 16.2 p. 306]. Default is the lowest available case number. SOLVE customizes the settings for the solution method to be applied for the linear steadystate analysis [Ch. 26 p. 429]. TIME sttime is a user specied starting time. LINEAR initiates a linear analysis, i.e., with constant material properties. NONLIN initiates a nonlinear analysis. A nonlinear analysis is required if the material properties vary with the potentials, for instance with unconned aquifers. Default
*GROWTR [ commands ] INITIA [ commands ] *END file .dcf
[sttime =0] [LINEAR]
349
If, like in the above commands, you give a single INITIA command, then Diana will apply the potentials of the lowest eld number in input table INIVAR as an initial potential eld. This is equivalent to the following commands.
file .dcf *GROWTR [ commands ] BEGIN INITIA POTENT INPUT FIELD=lowest FACTOR=1.0 LINEAR END INITIA [ commands ] *END
18.4.3
With the EXECUT commands you ask Diana to execute time steps in a transient groundwater ow analysis. A command le may contain one or more EXECUT blocks. Details of various commands are given in the referred sections.
350
Groundwater Flow Analysis syntax BEGIN EXECUT [ OFF ] [ ALPHA=alpha r ] [ SIZES sizes r... ] [ NONLIN ] [ SOLVE ] [ SAVE [ STEPS ]] steps n... ALL LAST NONE END EXECUT ALPHA=alpha is the time integration parameter . is the time integration parameter . Usual methods are: Euler forward ( = 0), CrankNicolson 1 ( = 2 ), Galerkin ( = 2 ) and Euler backward ( = 1) which is the 3 default. See also 35.1.3.1 on page 551 for some background theory on time integration. SIZES sizes are explicitly specied time step sizes, i.e., time increments. The number of values species the number of steps to be executed. NONLIN species options for nonlinear transient analysis [ 18.4.3.1]. SOLVE customizes the settings for the solution method [Ch. 26 p. 429]. SAVE species the steps for which analysis results must be saved for a subsequent structural analysis in a staggered owstress analysis [Ch. 22]. By default, in a model for staggered owstress analysis, Diana will save the results for all executed steps. STEPS steps indicates the steps for which the analysis results must be saved. Alternatively you may save ALL steps, only the LAST step, or NONE of the steps. The NONE option drops the possibility of a subsequent structural analysis! Default
*GROWTR [ commands ] EXECUT *END file .dcf
(0 1) [ = 1]
[sizes =1.0]
[ALL]
Due to these commands Diana will execute a time step with a factor of 1, applying the default solution procedure, and giving the default output of analysis results. The same would occur if you had given the following commands.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VI)
351
18.4.3.1
Via the NONLIN commands you may customize a nonlinear transient analysis, provided that such analysis has been initiated previously [ 18.4.2 p. 348]. syntax BEGIN NONLIN [ BEGIN ITERAT [ MAXITE=mi n ] [ METHOD [ NEWTON ]
] REGULA MODIFI [ [ CONVER [ POTENT ] ] TOLCON=tc n ] END ITERAT ] END NONLIN ITERAT species the iteration method to be applied in the nonlinear solution procedure. See also 35.1.3.2 on page 552 for some background theory on these iteration methods. MAXITE=mi is the maximum number of iterations. METHOD species the iteration method: NEWTON for a NewtonRaphson iteration scheme (the only option), i.e., the conductivity matrix will be updated. MODIFI invokes the Modied NewtonRaphson method. REGULA invokes the Regular NewtonRaphson method (the default).
[REGULA] [mi =5]
352
Groundwater Flow Analysis CONVER species the convergence criterion for the iteration process: POTENT indicates a convergence criterion on the norm of the potential eld (the only option). Parameter TOLCON=tc is the tolerance on the reference norm. When the norm of the incremental potential eld has become less than tc the reference norm Diana assumes sucient accuracy and stops the iteration process. Default
*GROWTR [ commands ] BEGIN EXECUT [ commands ] NONLIN END EXECUT *END file .dcf
[tc =103 ]
If, like in the above example, you only give a single NONLIN command then Diana will perform a nonlinear analysis applying a Regular NewtonRaphson iteration scheme with at most ve iterations. As convergence criterion, the norm of the potentials will be applied with a tolerance of 103 . Moreover, a default set of analysis results will be output. This is equivalent to the following commands.
file .dcf *GROWTR [ commands ] BEGIN EXECUT BEGIN NONLIN [ commands ] BEGIN ITERAT MAXITE=5 METHOD NEWTON REGULA CONVER POTENT TOLCON=1.E-3 END ITERAT END NONLIN END EXECUT *END
18.5
You can get output of analysis results from a groundwater ow analysis via the OUTPUT commands. For the general, analysis type independent options for output of analysis results see 3.6 on page 69.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VI)
18.5 Output of Analysis Results syntax BEGIN OUTPUT [device w ] [outopt w... ] [params ] [ OFF ] [ BEGIN SELECT [ ] ] [ BOUNDA cases n... ALL [ STEPS ] END SELECT ] [ LAYOUT ] item w PRESSU FLUX FLOW END OUTPUT SELECT command block to customize the batch output. for model selection see 3.6.1 on page 71. BOUNDA species a selection of boundary cases in linear steadystate analysis: cases is a series of case numbers referring to input table BOUNDA [ 16.2 p. 306], ALL selects all four boundary cases. STEPS selects time steps for output of transient analysis [ 18.5.1 p. 353]. LAYOUT optional commands to customize the layout and style of tabular output [ 3.6.3 p. 80]. item is the name of the analysis result to be output. PRESSU for nodal pressures (potentials) [ 18.5.2 p. 354]. FLUX for element uxes [ 18.5.3 p. 355]. FLOW for nodal ows [ 18.5.4 p. 356].
353
18.5.1
Step Selection
The STEPS command selects time steps for output of analysis results of a transient groundwater ow analysis. syntax STEPS [ steps n... ALL LAST ]
ALL will produce output for all steps that are executed (the default). LAST selects the nal step. The interpretation of the above options depends on the position of the current OUTPUT block. At the module command level (*GROWTR [ 18.4 p. 346]), steps are absolute step numbers, and ALL/LAST refer to all steps/the last step of the current analysis. At the EXECUT command level [ 18.4.3 p. 349], steps are relative step numbers, and ALL/LAST refer to all steps/the last step of the current EXECUT block.
18.5.2
Pressures
The potential eld is the solution vector of the system of equations for the groundwater ow analysis. This represents the pressures in the nodes as scalar values, therefore component selection is not applicable. syntax PRESSU [ form w ] { opti w } POTENT HEAD PRESSU gives the pressures . form indicates the desired theoretical formulation of the pressures to be output.
[POTENT]
POTENT gives the potential = p + h (the default). HEAD gives the pressure head p in detailed groundwater ow analysis. opti are additional options [ 3.6.2 p. 79].
item form PRESSU POTENT PRP PRESSU HEAD PRH p
The gradient of the potential eld is a vector element result. syntax PRESSU [ form w ] [ oper w ] { comp w } { loca w } { opti w } GRADIE LOCAL INTPNT GLOBAL NODES CENTER
18.5 Output of Analysis Results PRESSU gives the pressures . form indicates the desired theoretical formulation of the pressures to be output. GRADIE gives the spatial gradient of the pressure potential. oper species an operation to be performed on the pressure gradient. LOCAL transforms the pressure gradient in continuum elements to local xyz directions. GLOBAL transforms the pressure gradient in continuum elements to global XY Z directions. loca species the location of the pressure gradient to be output [ 3.6.2 p. 79]. opti are additional options [ 3.6.2 p. 79].
Pressure gradients item form oper PRESSU GRADIE LOCAL comp X Y Z
PGx PGy PGz
x d dX y d dY z d dZ
355
[GLOBAL]
[NODES]
18.5.3
Fluxes
In a groundwater ow analysis, uxes are calculated for elements. They can be output in the integration points or in the element nodes. Diana also writes the integrated uxes over groups of boundary elements to the standard output le file .out. syntax FLUX [ oper w ] { comp w } [ loca w ] { opti w } LOCAL INTPNT GLOBAL NODES BOUNDA CENTER FLUX gives the uxes q. oper species an operation to be performed on the uxes. LOCAL transforms uxes in continuum elements to local xyz directions. GLOBAL transforms uxes in continuum elements to global XY Z directions. BOUNDA gives the total discharge Q for boundary elements. This rate of ow is a scalar value, therefore component selection is not applicable.
Diana-9.4 Users Manual Analysis Procedures (VI) December 15, 2009 First ed.
[GLOBAL]
356
Groundwater Flow Analysis loca species the location of the uxes to be output [ 3.6.2 p. 79]. opti are additional options [ 3.6.2 p. 79].
comp X Y Z
FLx FLy FLz
[NODES]
qx qX
qy qY
qz qZ
18.5.4
Flows
In a groundwater ow analysis nodal ows are calculated. Nodal ows are scalar values, component selection is not applicable. syntax FLOW [ type w ] { opti w } REACTI RESIDU EXTERN FLOW gives the nodal ows.
[REACTI]
type
species the type of the nodal ow. REACTI for the reaction ow Q in all nodes with xed pressure potentials. RESIDU for the residual ow Q, also called the outofbalance ow. This is dened as the dierence between the externally applied ow Q and the internal resistance ow. EXTERN for the externally applied ow Q.
Chapter 19
also the examples crosels and crosrec in Volume Analysis Examples, respectively for a homogeneous and an inhomogeneous cross-section.
Diana-9.4 Users Manual Analysis Procedures (VI) December 15, 2009 First ed.
1 See
358
19.1
This section describes the general aspects for beam cross-section analysis: shape and dimensions of the nite element mesh, and denition of elastic properties.
19.1.1
Meshing
The shape and dimensions of a beam cross-section are modeled by a mesh of twodimensional, quadrilateral or triangular cross-section elements [Vol. Element Library]. To avoid singularity of the system of equations you must specify an arbitrary node with a xed potential in table FIXPOT [ 16.1.1 p. 303].
19.1.2
Elastic Properties
The Youngs moduli and Poissons ratios for dierent material sections of the cross-section are specied as material properties [Vol. Material Library].
19.2
Analysis Commands
Module bcross determines the beam cross-section properties via a dedicated steady-state potential ow analysis. syntax *BCROSS [ MODEL [ OFF ] ] [ EXECUT [ OFF ] ] [ OUTPUT [ OFF ] ] MODEL to evaluate the nite element model [ 19.2.1]. EXECUT to execute the beam cross-section analysis [ 19.2.2]. OUTPUT for output of shear stress distributions due to shear forces and torsional moment [ 19.3].
19.2.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual analysis. syntax BEGIN MODEL [ EVALUA [ OFF ] ] [ ASSEMB [ OFF ] ]
19.2 Analysis Commands [ MATRIX [ CONDUC ] [ OFF ] ] [ BOUNDA [ OFF ] ] END MODEL EVALUA to check and evaluate geometric and material properties for elements [ 3.4 p. 66]. ASSEMB to assemble the elements, i.e., creating appropriate system degrees of freedom. MATRIX CONDUC to setup the element conductivity matrices. The OFF option switches o the creation of conductivity matrices. This may save computing time if the matrices are still available on the filos le and appropriate. BOUNDA to generate the required boundary cases and convert these to righthand-side nodal ux vectors [ 19.2.1.1]. 19.2.1.1 Right-hand-side Nodal Flux Vectors
359
The steady-state potential ow analysis, dedicated to a cross-section analysis with Module bcross requires the following boundary cases: case 1 to simulate a shear force Qy , case 2 to simulate a shear force Qz , case 3 to simulate a torsion moment Mx , case 4 the previous three cases combined. By default, Diana will assume unit values for the cross-sectional loading (Qy , Qz , Mx ), while setting up the four boundary cases. However, you may overrule the default unit values by specifying parameters. In either case, the boundary cases will be converted to right-hand-side nodal ux vectors. syntax BEGIN BOUNDA [ OFF ] [ QY=qy r ] [ QZ=qz r ] [ MX=mx r ] END BOUNDA OFF switches o the creation of the nodal ux vectors. This may save computing time if these vectors are still available on the filos le and appropriate. QY=qy species the value for Qy . QZ=qz species the value for Qz . MX=mx species the value for Mx .
Diana-9.4 Users Manual Analysis Procedures (VI)
360 Default
*BCROSS [ commands ] BEGIN MODEL BOUNDA END MODEL [ commands ] *END
If, like in the above commands, you give a single BOUNDA command, then Diana will setup the right-hand-side nodal ux vectors, with unit values for the crosssectional loading, which would also occur if you didnt give the BOUNDA command at all or if you would have given the following commands.
file .dcf *BCROSS [ commands ] BEGIN MODEL BEGIN BOUNDA QY=1.0 QZ=1.0 MX=1.0 END BOUNDA END MODEL [ commands ] *END
19.2.2
The EXECUT commands cause the actual beam cross-section analysis to be executed. syntax BEGIN EXECUT [ SOLVE ] END EXECUT SOLVE customizes the settings for the solution method [Ch. 26 p. 429].
19.3
Module bcross gives two types of analysis results: cross-sectional properties [ 19.3.1], and shear stresses [ 19.3.2].
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VI)
361
19.3.1
Cross-sectional Properties
Module bcross always prints the cross-sectional properties on the standard output le. Table 19.1 shows the available properties, with the label that preTable 19.1: Output of cross-section properties
Label Iy Iz It ymax ymin zmax zmin Wymax Wymin Wzmax Wzmin Xc Yc Alpha A ky kz ey ez E.Iy E.Iz G.It E.A Item Iy Iz It ymax ymin zmax zmin Wy.max Wy.min Wz.max Wz.min Xc Yc A ky kz ey ez Formula
R RA z2 dA
2 A y dA
Moments of inertia
Extreme bers
Sections moduli
Center, axes and area Reciprocal shear shape factors Shear center
Pn
Pn
Pn
Pn
cedes them on the output line.2 Figure 19.1 on the following page shows some properties of an arbitrary cross-section. Note that angle has a negative value in this example.
19.3.2
Shear Stresses
You can get output of shear stresses from a cross-section analysis via the OUTPUT commands. For the general, analysis type independent options for output of analysis results see 3.6 on page 69.
2 See
362
z Yc y
zmax
Figure 19.1: Cross-section properties syntax BEGIN OUTPUT [device w ] [outopt w... ] [params ] [ OFF ] [ BEGIN SELECT [ ] [ CASES ] cases n... ALL END SELECT ] [ LAYOUT ] item w { comp w } [ loca w ] { opti w } . . . SHEAR INTPNT NODES CENTER END OUTPUT SELECT command block to customize the batch output. for model selection see 3.6.1 on page 71. CASES species a selection of boundary cases [ 19.2.1.1]: cases is a series of case numbers between 1 and 4, ALL selects all four boundary cases. LAYOUT optional commands to customize the layout and style of tabular output [ 3.6.3 p. 80].
19.3 Output of Analysis Results item is the name of the analysis result to be output. SHEAR for shear stresses due to the standard boundary cases [ 19.2.1.1]. Via the loca specier you can get the shear stresses in the integration points or in the nodes, opti are additional options [ 3.6.2 p. 79].
comp X Y
Sx Sy
363
[NODES]
item SHEAR
zx
zy
The shear stresses can be used as input in a linear static analysis to dene the shear distribution for beam elements.
364
Chapter 20
1 See
366
20.1
20.1.1
The lm is modeled by quadrilateral or triangular lubrication elements [Vol. Element Library]. The varying thickness of the lm is input in table GEOMET.
20.1.2
Viscosity
20.1.3
Boundary Conditions
Prescribed pressures at the boundaries are input in table FIXPOT [ 16.1.1 p. 303]. Non-zero values for these prescribed pressures must be specied in subtable NODAL of table BOUNDA [ 16.2.1 p. 307].
20.1.4
Velocity
The velocity dierence between the two faces of the element is input in subtable ELEMEN of table BOUNDA [Vol. Element Library].
20.2
Analysis Commands
Module reynol determines the pressure eld via a dedicated linear steady-state potential ow analysis. syntax *REYNOL [ MODEL [ OFF ] ] [ EXECUT [ OFF ] ] [ OUTPUT [ OFF ] ] MODEL to evaluate the nite element model [ 20.2.1]. EXECUT to execute the lubrication analysis [ 20.2.2]. OUTPUT for output of the pressure eld [ 20.3].
20.2.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual analysis.
20.3 Output of Analysis Results syntax BEGIN MODEL [ EVALUA [ OFF ] ] [ ASSEMB [ OFF ] ] [ MATRIX [ CONDUC ] [ OFF ] ] [ BOUNDA [ OFF ] ] END MODEL EVALUA to check and evaluate geometric and material properties for elements [ 3.4 p. 66]. ASSEMB to assemble the elements, i.e., creating appropriate system degrees of freedom. MATRIX CONDUC to setup the element conductivity matrices. The OFF option switches o the creation of conductivity matrices. This may save computing time if the matrices are still available on the filos le and appropriate. BOUNDA to convert the boundary cases from input table BOUNDA to righthand-side nodal ux vectors. The OFF option switches o the conversion of boundary cases. This may save computing time if the these are still available on the filos le and appropriate.
367
20.2.2
BEGIN EXECUT [ SOLVE ] END EXECUT SOLVE customizes the settings for the solution method [Ch. 26 p. 429].
20.3
You can get output of analysis results from a lubrication analysis via the OUTPUT commands. For the general, analysis type independent options for output of analysis results see 3.6 on page 69.
368
Reynolds Flow Lubrication Analysis syntax BEGIN OUTPUT [device w ] [outopt w... ] [params ] [ OFF ] [ BEGIN SELECT [ ] [ CASES ] cases n... ALL END SELECT ] [ LAYOUT ] item w PRESSU FLUX END OUTPUT SELECT command block to customize the batch output. for model selection see 3.6.1 on page 71. CASES species a selection of boundary cases [ 16.2 p. 306]: cases is a series of case numbers referring to table BOUNDA. The ALL option selects all boundary cases. LAYOUT optional commands to customize the layout and style of tabular output [ 3.6.3 p. 80]. item is the name of the analysis result to be output. PRESSU for nodal pressures [ 20.3.1 p. 368]. FLUX for element uxes [ 20.3.2 p. 369].
20.3.1
Pressures
The pressure eld is the solution vector of the system of equations for the lubrication analysis. Pressures are scalar values calculated in the nodes, component selection is not applicable. syntax PRESSU { opti w } PRESSU the pressures p. opti are additional options [ 3.6.2 p. 79].
item PRESSU PTE p
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VI)
369
20.3.2
Fluxes
In a lubrication analysis, uxes are calculated for elements. They can be output in the integration points or in the element nodes. syntax FLUX [ oper w ] { comp w } [ loca w ] { opti w } LOCAL INTPNT GLOBAL NODES CENTER FLUX gives the uxes q. oper species an operation to be performed on the uxes. LOCAL transforms uxes in to local xy directions. GLOBAL transforms uxes to global XY Z directions. loca species the location of the uxes to be output [ 3.6.2 p. 79]. opti are additional options [ 3.6.2 p. 79].
comp X Y Z
FLx FLy FLz
[NODES] [GLOBAL]
oper LOCAL
370
Part VII
Chapter 21
21.1
The analysis is performed with the usual Diana modules for structural analysis. User-selected structural elements are extended to so-called mixture elements by
374
Introduction to FlowStress Analysis addition of a pore pressure potential to the displacements as nodal degree of freedom. Interaction. Mixture elements are applicable for the analysis of the interaction between deformation of porous media and pore uid, according to the full Biot theory: a mechanical load causes ow and an imposed ow causes deformation. Furthermore Biots theory includes undrained compressibility and dierentiation between eective stress and inter-granular stress. Biots theory does not account for inertial eects during rapid loading in dynamic analysis, e.g. the velocity of soil and uid are not considered separately. Biots theory assumes saturation and therefore is not directly suited for analysis with unsaturated areas. Material models. All eective stress based models for elasticity and plasticity are applicable in combination with mixture elements. The available models cover the elastoplastic behavior of the common soil types. Anisotropic properties for permeability and stiness can be modeled. Element model and analysis. The implementation allows plane strain, axisymmetric, and three-dimensional analysis. Combination of elements with porous material (soil) and non-porous material (structure) is feasible. The mixture continuum elements and mixture interface elements can be used with a wide range of structural elements and structural material models. It is possible to perform a static and transient analysis, linear as well as physical nonlinear. Special options are available to initialize in-situ (overconsolidated) stresses and ow. Initial stress and pore pressure can also be derived from a preceding nonlinear analysis phase [ 24.5 p. 418]. The elements have restricted applicability for dynamic analysis (no separate uid velocities). Furthermore the elements are not applicable for eigenvalue analysis and modal analysis, because these analysis types are currently restricted to symmetrical matrices. Application of the elements is meaningless for buckling stability analysis and pure potential ow analysis.
21.2
The analysis is performed with the usual Diana modules for potential ow and structural analysis. Modules linsta and nonlin will automatically transform the hydraulic heads from a potential ow analysis to input for the structural analysis by generation of input for table LOADS. Interaction. Staggered analysis is applicable to analyze the eect of pore uid on deformation and eective stress: an imposed ow causes deformation, but a mechanical load does not cause ow. The implementation of ow analysis and load transformation account for unsaturated areas.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VII)
21.3 Staggered Solution with Temperature Material models. All eective stress based models for elasticity and plasticity are applicable in combination with a pore uid load. The available models cover the elastoplastic behavior of the usual soil types. Darcys law is used to calculate hydraulic heads in the porous part of the mesh. Anisotropic properties for permeability and stiness can be modeled. Permeability and hydrostatic capacitance are nonlinear in unsaturated areas or areas with turbulent ow. Element model and analysis. The implementation allows plane strain, axisymmetric, and three-dimensional analysis. The same mesh must be used for potential ow and structural analysis. Quadratic structural elements are interpreted as linear potential ow elements. Combination of elements with porous material (soil) and non-porous materials (structure) is feasible. It is possible to perform a static analysis, linear as well as nonlinear.
375
21.3
The analysis is performed with the usual Diana modules for potential ow and structural analysis. Quadratic structural elements are interpreted as linear potential ow elements. Modules linsta and nonlin will automatically convert the temperature from a potential ow analysis to input for the structural analysis by generation of input tables LOADS or TEMPER. The temperature analysis of hydrating concrete also yields a degree of reaction, from which Diana generates a table MATURI. Applicability. Staggered analysis is applicable to analyze the eect of temperature on deformation and stress: an imposed temperature causes deformation, but a mechanical load does not cause thermal eects. Material models. All structural material models are applicable in combination with a temperature load. A nonlinear temperature dependency may be introduced for most of these models. The capacity and conductivity may be nonlinear. Element model and analysis. The implementation allows plane strain, axisymmetric, and three-dimensional analysis. A mesh with quadratic structural elements must be used for potential ow and structural analysis. Quadratic structural elements are interpreted as linear potential ow elements. Combination of elements with conducting material and insulating material is feasible. It is possible to perform a static or transient analysis, linear as well as nonlinear.
376
Chapter 22
Staggered Analysis
If no measured potential eld is available, the input for a structural analysis may be found from a preliminary potential ow analysis [Part VI]. The combination of a potential ow analysis and a subsequent structural analysis is also known as staggered owstress analysis. Diana will automatically convert the results of a potential ow analysis to input data for a structural analysis. A peculiarity in staggered owstress analysis is that the model for the ow analysis always must consist of linearly interpolated elements whereas for the structural analysis either linearly or quadratically interpolated elements must be applied. Diana will automatically convert the elements whenever necessary.
22.1
A staggered owstress analysis comprises sequential execution of the following steps: 1. Prepare a nite element model, appropriate for ow and structural analysis [ 22.2]. 2. Perform a potential ow analysis [ 22.3]. 3. Perform structural analysis [ 22.5]. We introduce these steps by showing two typical command sequences, respectively for a steady-state and for a transient heat ow analysis. Steady-state
*FILOS INITIA *INPUT *HEATSS file .dcf
378
*LINSTA OUTPUT ... *END
Staggered Analysis
First the complete model is read into a newly initialized filos le [Ch. 3]. Then a steady-state heat ow analysis is performed via Module heatss [ 17.2 p. 316]. Module linsta will convert the results of the heat ow analysis, in this case temperatures, to input data for a linear static structural analysis and then perform such analysis [Ch. 4]. Transient
*FILOS INITIA *INPUT *HEATTR BEGIN EXECUTE SAVE STEPS 1 4 5 END EXECUTE *NONLIN BEGIN EXECUTE BEGIN TIME EXPLIC SIZES ... END TIME END EXECUTE OUTPUT ... *END file .dcf
A transient heat ow analysis, here performed with Module heattr [ 17.3 p. 322], must be combined with a transient nonlinear structural analysis. In the heat ow analysis you must save the results of specied steps via the SAVE command. A transient nonlinear structural analysis is then performed with Module nonlin [Ch. 12], but rst nonlin will automatically convert the heat ow results to loads or temperatures and maturity for structural analysis.
22.2
Model Denition
Like for other analyses, the nite element model for a staggered analysis must be read from an input data le which you may create interactively in the iDiana Results environment [Vol. Pre- and Postprocessing]. A model for staggered owstress analysis basically comprises two domains: one for ow analysis and one for structural analysis. In practice these domains will overlap completely, or at least for a considerable part. We will call this overlap the owstress domain.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VII)
22.2 Model Denition A staggered owstress analysis with Diana requires that the complete model is dened prior to the potential ow analysis, i.e., the ow-only domain, the structural-only domain, and the owstress domain altogether on one input data le. The ow-only domain must be modeled with ow elements and the structural-only domain with structural elements. For both domains you must supply appropriate material properties. For the owstress domain please note the following:
The owstress domain, where the elements are active in both potential ow and structural analysis, must be modeled with structural elements. The choice of elements is restricted to those listed in column Flowstress domain of Table 22.1.
379
pl. strain
axisymm.
solid
interface
380
Staggered Analysis
The elements in the owstress domain that are linearly interpolated stay linearly interpolated also in the potential ow analysis. The elements in the owstress domain that are quadratically interpolated, however, are converted to linearly interpolated elements as indicated in column Conversion to ow analysis. Therefore in the potential ow analysis of a staggered analysis the mid-nodes of quadratic structural elements are meaningless. For the elements in the owstress domain you must specify appropriate material properties, not only for structural analysis but also for ow analysis [Vol. Material Library]. Diana will consider any structural element without properties for ow analysis to belong to the structural-only domain.
22.3
The rst analysis in a staggered owstress analysis is the potential ow analysis. This may be a heat ow or concentration ow analysis [Ch. 17], or a detailed groundwater ow analysis [ 18.1 p. 335]. Diana recognizes the type of analysis from the module that you invoke. Prior to the actual ow analysis, any structural element with material properties appropriate for the analysis type will be converted internally to a ow element as indicated in Table 22.1. Transient analysis. For transient analysis Diana can convert the time dependent ow results to input for structural analysis [ 17.3.3 p. 326] [ 18.4.3 p. 349]. By default conversion will be performed for all time steps. Alternatively you may select certain time steps for which conversion must be performed.
file .dcf *HEATTR commands BEGIN EXECUT commands SAVE STEPS 1 3 4 END EXECUT
22.4
In a staggered owstress analysis Diana will automatically convert results of a potential ow analysis to input for a structural analysis. Note that results of linearly interpolated potential ow elements will be converted to input either for linear structural elements or for quadratic structural elements. Results of each boundary case in a steady-state ow analysis will be transformed to a
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VII)
22.4 Conversion to Structural Analysis corresponding load case for linear structural analysis. Results of saved steps in a transient ow analysis will be converted to temperature and maturity input for a transient nonlinear structural analysis. The transformation of a groundwater ow model to a structural model also has some additional points of interest [ 22.4.2].
381
22.4.1
Temperature or concentration load cause initial strains following from thermal or chemical expansion or shrinkage of the elements [Vol. Element Library]. In a nonlinear analysis the eect of the temperature or concentration on several material properties may be analyzed in combination with other nonlinearities [Part IV]. Maturity. For a possible transient nonlinear structural analysis Diana will apply a maturity model to convert the transient heat ow analysis results. The type of the maturity model depends on what was determined during the heat ow analysis: the degree of reaction in hydration heat analysis, or the equivalent age [ 17.3.2.1 p. 326]. This feature enables the analysis of chemical reactions like cement hydration (young hardening concrete) and vulcanization of rubber. See Volume Material Library for more information on maturity.
22.4.2
After a detailed groundwater ow analysis [ 18.1], Diana will convert the resulting potential heights in the owstress domain to pressure input for structural analysis. This conversion requires the uid density f which you must have specied as a material property [Vol. Material Library]. The conversion also requires the acceleration of gravity g. For the conversion to structural, Diana will always assume g = 9.81m/s2 , in a two-dimensional model working in the negative Y direction and in a three-dimensional model in the negative Z direction. The gravity acceleration vector g is applied to calculate the dead weight load fw (22.1) and the buoyancy (22.2). If you specify the model in units other than SI, then you must specify the applied units [ 1.1 p. 3]. Otherwise, the assumed acceleration of gravity will be applied incorrectly. Pore uid load. The pore pressure height eld, resulting from a groundwater ow analysis, causes external loading for both linear and nonlinear analysis. The existence of pore uid causes two loading types on a porous medium like soil: dead weight and buoyancy. The dead weight load fw depends on the porosity n, the uid density f , the acceleration of gravity g and the degree of saturation S: fw = n f S g
Diana-9.4 Users Manual Analysis Procedures (VII)
(22.1)
December 15, 2009 First ed.
382
Staggered Analysis This dead weight load is applied as a special element load, see WEIGHT input in Volume Element Library. The buoyancy is caused by the isotropic uid stress ii which depends on the pressure head p and volumetric uid weight f : ii = p f = p f |g| (22.2)
Diana uses the pressuresaturation diagram and the porosity for both the potential ow analysis and the calculation of the structural pore uid load. This load is applied as a special element load, see PRESSU input in Volume Element Library. You must specify the pressuresaturation diagram and the porosity as material properties, see the SATURA and POROSI input in Volume Material Library.
22.4.3
In some cases it may be instructive to see how Diana has converted the ow stress domain to a model for structural analysis. Therefore you may invoke Module linsta or nonlin to set up the structural model and then input to remake appropriate input tables.
file .dcf *NONLIN MODEL TYPE OFF EXECUT OFF OUTPUT OFF *INPUT REMAKE TABLE LOADS *END
22.5
Structural Analysis
After the ow analysis you may either invoke one of the structural analysis modules linsta, nonlin, eigen, or balanc to perform a structural analysis. Each of these modules will convert a potential ow model and its analysis results to a model for structural analysis.
Chapter 23
Mixture Analysis
This chapter describes how to use Diana for mixture analysis. In this context, mixture analysis means the simultaneous solution of deformation and pore pressures for the mixture of a uid and a porous solid like soil. Typically, transient mixture analysis is used to analyze consolidation or swelling caused by mechanical loading or unloading. In transient mixture analysis, two-sided interaction is introduced if porous media are subjected to short duration load and posses a low permeability compared to the loading rate. Static mixture analysis can also be used as a substitute for staggered analysis [Ch. 22], to calculate the deformation caused by steady-state pore uid ow directly.
23.1
Model Input
The input of the element mesh, supports and mechanical load is as usual for structural analysis. You must extend structural elements to mixture elements by specifying additional pore pressure potentials. These mixture elements require additional material input. Mixture analysis requires input of pore pressure constraints. Special features are available to incorporate gravitational load.
23.1.1
Mixture Elements
Mixture elements are plane strain, axisymmetric, and solid continuum elements with corresponding structural interface elements as described in Volume Element Library, but with a special mixture option. For continuum elements, due to this option, Diana adds a scalar pore pressure potential to the set of element degrees of freedom in each element node. For interface elements this potential is added to the nodes at one or at both sides of the element.
Mixture Analysis
All options of structural elements in static and transient analysis, like material models, can also be applied in combination with the mixture option. There is an exception for combination with assumed strain options: the constant shear and constant dilatation options can be combined with the mixture option, however: Incompatible enhanced assumed strain modes (EAS) cannot be combined with the mixture option. See Volume Element Library for more information on assumed strain options. A combination of mixture elements with structural interface elements and non-mixture elements like shells or beams can be used to analyze for example soilstructure interaction. 23.1.1.2 Extension of Structural Elements
To extend a structural element to a mixture element you must specify the mixture option in input table DATA. There is a slight dierence between the input syntax for continuum elements and for interface elements. Continuum elements DATA
1 5 6 12 13 80
syntax
datnr n
MIXTUR
MIXTUR indicates that the mixture option applies for all continuum elements referring to data number datnr , i.e., Diana will extend these elements to mixture elements [Fig. 23.1].
structural degrees of freedom pore pressure potential both (mixture)
Figure 23.1: Addition of pore pressure potential to continuum elements The combination of mixture elements with non-mixture elements is straightforward by using separate data numbers for mixture elements like in the following example.
385
In this example input, elements 1, 3, and 5 will be extended to mixture elements. Elements 12, 17, and 18 will remain purely structural. Interface elements DATA
1 5 6 12 13 80
syntax
datnr n
MIXTUR
[ ONESID TWOSID
MIXTUR indicates that the mixture option applies for all interface elements referring to data number datnr , i.e., Diana will extend these elements to mixture elements. An input option denes which side(s) of the interface element must be extended with the potential degree of freedom. With option ONESID, Diana will only extend the rst side to mixture, i.e., the side with the rst node of the connectivity [Fig. 23.2a]. Option TWOSID causes both sides to be extended to mixture [Fig. 23.2b]. The ONESID opstructure porous medium structural degrees of freedom pore pressure potential both (mixture)
[ONESID]
1 porous medium
(a) ONESID
Figure 23.2: Addition of pore pressure potential to interface elements tion is typically applied for an interface between soil and structure. The TWOSID option is applied for an interface between soil and soil, to model hydraulic fracture in joints or connections in phased analysis [Ch. 25].
386
Mixture Analysis
file .dat ELEMEN 1 CL20I 61 62 63 64 65 71 72 73 74 75 DATA / 1 3 5 / 1 / 12 17 38 / 2 DATA 1 MIXTUR ONESID 2 MIXTUR TWOSID
In this example input, elements 1, 3, and 5 will be extended to mixture elements at side one only. Elements 12, 17, and 18 will be extended at both sides. The other elements will remain purely structural. Material properties. For mixture elements, you must supplement the material properties for structural analysis with special properties to describe the properties of the porous medium. See Volume Material Library for input description.
23.1.2
Diana uses the pore pressure potential as the basic degree of freedom in the nodes. You can only specify values for the pore pressure potential to input pore pressure constraints. During analysis, Diana composes the pore pressure potential in a uid pressure p and a gravitational part f g (xxref ) [ 36.1.6 p. 556]. Diana can supply output of the pore pressure potential , the pore pressure p, and the excess pore pressure pe . Input for the pore pressure potential is specied via the regular input tables for structural stress analysis [Ch. 2]. You must specify the nodes with an imposed pore pressure potential in table SUPPOR. Constraints for the nodal pore pressure potential are specied in table LOADS analogously to the constraints for a displacement: nonzero values in subtable DEFORM and nodal discharge in subtable NODAL.1 Moreover, linear constraints for pore pressure potentials may be specied in table TYINGS.
1 Diana-9.4
387 syntax
node n
1
PR
5 6 80
/ nodes ng... / PR
node is a single node number, nodes is a series of nodes which must be specied between slashes and may comprise numbers or groups or both. PR is the degree of freedom type that indicates an imposed pore pressure potential for the specied node (s ). Nonzero potential LOADS DEFORM
1 5 6 80
syntax
node n
1
PR p r
5 6 80
/ nodes ng... / PR p r
1 5 6 80
node is a single node number, nodes is a series of nodes which must be specied between slashes and may comprise numbers or groups or both. PR is the degree of freedom type which indicates a pore pressure potential. p is the prescribed value of the pore pressure potential , ps is a series of values, one for each node in nodes .
80
node n
1
PR q r
5 6 80
/ nodes ng... / PR q r
1 5 6 80
/ nodes ng... / PR / qs r... / node is a single node number, nodes is a series of nodes which must be specied between slashes and may comprise numbers or groups or both. PR is the degree of freedom type which indicates a pore pressure potential. q is the value Q of the nodal discharge, qs is a series of values, one for each node in nodes . Nonzero potential and nodal discharge are considered to be loadings and therefore must be related to a load case, like in the following example.
file .dat SUPPOR / 3-27 / PR LOADS CASE 1 DEFORM 5 PR 2.57 CASE 2 NODAL / 55-60 / PR -12.36
Linear constraints. Linear constraints (tyings) are mainly used to keep pore pressure potentials of dierent nodes mutually equal. See 2.2 on page 18 for general syntax description. For the pore pressure potential, the degree of freedom must be specied as type PR without a direction number, like in the following example.
file .dat TYINGS EQUAL PR 2 10
December 15, 2009 First ed.
12 / 3-6 /
Diana-9.4 Users Manual Analysis Procedures (VII)
389
In the EQUAL subtable you must specify sets of a master and slave nodes.
23.1.3
The gravity vector g will generally cause a distributed dead weight load g with being the total mass density. In mixture analysis, the gravitational part of the pore pressure potential f g (x xref ) induces an additional buoyancy force, which results in weight reduction for the porous medium and an hydrostatic normal force on the boundaries, with x xref the position vector. Diana includes the buoyancy force automatically, if you specify a gravity load and a uid density. A regular gravity load specication, as described in 2.3.2 on page 39, can be extended with the origin xref of the position vector. In mixture analysis you may specify a position vector for one load case only, however, Diana will apply it for all load cases with a weight load. syntax LOADS WEIGHT
1 80
dirnr n gracce r [ox r oy r oz r ] WEIGHT is the subtable heading for dead weight load. dirnr is the direction number referring to table DIRECT. gracce is the acceleration of gravity g. ox , oy , oz respectively are the XY Z coordinates of the origin xref of the position vector. If you dont specify the origin, then Diana assumes the origin of the XY Z coordinate system by default.
23.2
In static analysis, the time derivatives are zero. This means that the two-sided coupling reduces to single-sided coupling: ow inuences stress but stress does not inuence ow. An exclusive static linear elastic analysis must be performed, if one-directional interaction is analyzed and if the model behaves linearly. A preliminary static linear elastic analysis for a transient analysis may be performed to check the model and to calculate an initial stress and ow eld. The commands for linear static analysis are as described in Chapter 4, with an additional option to output pore pressure potentials: the PRESSU option in the DISPLA command as shown below.
Diana-9.4 Users Manual Analysis Procedures (VII) December 15, 2009 First ed.
390
Mixture Analysis
file .dcf *FILOS INITIA *INPUT *LINSTA BEGIN OUTPUT DISPLA PRESSU PH PO END OUTPUT *END
The PH and PO components of the PRESSU option give output of the pore pressure potential and the total pore pressure p [ 23.4.1.1 p. 395].
23.3
For the coupled analysis of stress and pore uid ow, a nonsymmetric system of nodal displacements u and pore pressure potentials is solved simultaneously. For direct time integration in transient analysis and/or analysis of nonlinearity, a stepwise method is available from Module nonlin, with iteration in case of nonlinearity. You may use a static nonlinear analysis for instance to analyze the stability of a model subjected to ground water ow. A transient analysis is suitable for analysis of consolidation, liquefaction, or swelling of soil.
23.3.1
Initialization
The stepwise analysis is initialized for Module nonlin with TYPE, and EXECUT START commands. The commands indicate whether there is nonlinearity in the model, whether a non-default time integration scheme must be used, and whether an initial stress eld must be used. The following is an example of such commands.
file .dcf *NONLIN BEGIN TYPE PHYSIC METHOD HHT END TYPE BEGIN EXECUT BEGIN START BEGIN INITIA STRESS CALCUL LOAD=1 FACTOR=1.0 EQUILI END INITIA END START END EXECUT
The PHYSIC command in the TYPE block indicates physical nonlinear analysis.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VII)
23.3 Transient and Nonlinear Analysis Initial load. In geotechnical analysis it is common practice to initialize the drained stress and ow eld, using the previously calculated elastic results due to dead weight load. Additionally, Diana also allows you to derive the initial stress and (excess) pore pressure from a preceding nonlinear analysis phase [ 24.5 p. 418]. The EQUILI option in the START INITIA block not only initializes, in new elements, the stress and pore pressure potential eld calculated from the specied linear load set, but also directly applies the corresponding dead weight load. During stress initialization, Diana adapts the initial stress in new elements according to the lateral pressure ratio K0 . Time integration. In mixture analysis, the METHOD command is primarily used to specify the direct time integration scheme, like in the above example with HHT for the HilberHughesTaylor scheme [ 12.2.3 p. 216]. For theoretical backgrounds of time integration with respect to mixture analysis see 36.2.2 on page 559.
391
23.3.2
Initial Equilibrium
During initialization with results of a linear analysis, in most cases no deformation will occur due to appliance of dead weight. However, in case of nonlinearity (plasticity, partial saturation) or if the initial stress eld is no longer balanced by the load (phased analysis), then initial deformation will occur and the initial stress and (excess) pore pressure eld will alter to restore equilibrium during following static load steps. These load steps can be performed with the START STEPS option of the EXECUTE command block [ 12.3 p. 218]. Suppression of strain superposition
*NONLIN BEGIN EXECUT BEGIN PHYSIC SUPPRE STRAIN END PHYSIC END EXECUT file .dcf
Diana oers a special option to suppresses superposition of displacements and strains during the establishment of the initial equilibrium. The SUPPRE STRAIN command suppresses superposition of displacements and strains during the establishment of the initial equilibrium [ 12.3.4 p. 238]. Only the stress state will be aected in this case. Initial state determination
*NONLIN BEGIN EXECUT
Diana-9.4 Users Manual Analysis Procedures (VII) December 15, 2009 First ed.
file .dcf
392
BEGIN START BEGIN STEPS EXPLIC SIZES 1.0 END STEPS END START REFERE PRESSU WEIGHT BEGIN PHYSIC SUPPRE STRAIN DRAINE END PHYSIC END EXECUT
Mixture Analysis
With the EXECUT START STEPS command block you can apply stepwise a possible unbalanced part of the initial load [ 12.3.1.2 p. 224]. During the rst step, Diana will implicitly adapt the initial stress and pore pressure in case of initial nonlinearity. During all load steps, the behavior is default completely drained, except for layers with material property UNDRAI specied in the input data. You can enforce drained behavior in all layers by using the keyword DRAINE in the PHYSIC command block [ 12.3.4 p. 239]. Due to the REFERE PRESSU WEIGHT command Diana will re-apply the K0 procedure to new elements after step 1, and use the pore pressure potential after step 1 as the reference ref for the calculation of excess pore pressures pe = ref [ 12.3.8 p. 250].
23.3.3
By default, elements with the undrained material property will behave completely undrained during the execution of load and time steps. Elements without this property will behave drained during load steps and will show transient pore pressure dissipation during time steps. However, you may overrule the undrained material property by specifying the DRAINE command for execution of time steps [ 12.3.4 p. 239]. In this case, Diana will switch o the undrained behavior for all elements. even for elements with undrained material property. The time dependency of the load sets, using multilinear diagrams, must be specied in input Table TIMELO
file .dat TIMELO LOAD 1 TIMES 1.0 1.1 1000.0 / FACTOR 0.0 1.0 1.0 /
23.3.3.1
The LOAD command block within the EXECUT command block may be used to apply static load steps. During static load steps the behavior is completely drained or completely undrained. The following is an example of such commands.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VII)
393
The rst EXECUT LOAD block uses the default behavior (undrained in layers with the material property UNDRAI, drained in the other layers). The second EXECUT LOAD block uses drained behavior in all layers. 23.3.3.2 Time Steps
Time steps are executed via the EXECUT TIME commands [ 12.3.3 p. 234]. You must specify the PHYSIC DRAINE command to enforce pore pressure dissipation in mixture elements with the material property UNDRAI. The following is an example of such commands.
file .dcf *NONLIN BEGIN EXECUT BEGIN TIME BEGIN STEPS EXPLIC SIZES 0.004(10) 0.008(10) END STEPS END TIME BEGIN PHYSIC DRAINE END PHYSIC END EXECUT
394
Mixture Analysis
23.4
In a Diana Mixture analysis, the pore pressure potentials are the basic system degrees of freedom in the nodes, from which Diana derives the total pore pressure p and excess pore pressure pe . Diana can also output the corresponding discharges in the nodes. For mixture elements with undrained material property you can get pressures at element level. Also at element level, you can get the saturation and the Darcy ux. syntax [ *NONLIN ] BEGIN OUTPUT [ device w ] [outopt w... ] [ params ] [ ] item w DISPLA VELOCI ACCELE PRESSU FORCE SATURA DARCYF END OUTPUT The following describes only the special commands for Module nonlin to get output of mixture analysis results. See 12.4 on page 252 for general output commands for Module nonlin. item is the name of the result item to be output. See 3.6.2 on page 77 for complete syntax of this command. DISPLA gives the pore pressure potentials which are the basic system degrees of freedom in the nodes. From these potentials, Diana can derive and output the total pore pressure p and excess pore pressure pe [ 23.4.1.1]. VELOCI species the rst time derivative of the pore pressure potential [ 23.4.1.1]. ACCELE species the second time derivative of the pore pressure potential [ 23.4.1.1]. FORCE gives the corresponding discharges in the nodes [ 23.4.2]. PRESSU gives the pressures at element level for mixture elements with undrained material property [ 23.4.1.2]. SATURA gives the saturation at element level [ 23.4.3]. DARCYF gives the Darcy ux at element level [ 23.4.4].
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VII)
395
23.4.1
Pressures
There are two types of pressures available for output: the basic degrees of freedom in the nodes and the pressures in elements with undrained material property. 23.4.1.1 Nodal Pressures { comp w } { opti w } PH PO EX
syntax
item w [ type w ] [ form w ] DISPLA TOTAL PRESSU VELOCI INCREM ACCELE PHASE
DISPLA species the system degrees of freedom at the nodes as output item. VELOCI species the rst time derivative of the pore pressure potential at the nodes as output item. ACCELE species the second time derivative of the pore pressure potential at the nodes as output item. type species the type: TOTAL for the total value of the system degrees of freedom, INCREM for the incremental values, PHASE for the phased values, i.e. the change between two consecutive phases in a phased mixture analysis. form species the formulation of the system degrees of freedom: PRESSU for the pore pressure potential components. Without this formulation Diana gives the translational displacements. comp selects components for output. PH for the pore pressure potential . PO for the total pore pressure: p = + f g (x xref ). EX for the excess pore pressure: pe = ref .
[TOTAL]
396
comp PH PO
TDprph TDphpo
Mixture Analysis
item DISPLA DISPLA DISPLA VELOCI VELOCI VELOCI ACCELE ACCELE ACCELE
type TOTAL INCREM PHASE TOTAL INCREM PHASE TOTAL INCREM PHASE
form PRESSU PRESSU PRESSU PRESSU PRESSU PRESSU PRESSU PRESSU PRESSU
EX
TDphex
IDprph
pe
PDprph
TVprph
IVprph
PVprph
TAprph
IAprph
PAprph
23.4.1.2
Element Pressures
For mixture elements with an active undrained material property, Diana can only give meaningful output of pore pressure components at the element level. Therefore, if you switched o the undrained behavior during step execution, you must ask for pore pressure potential components at the element level. syntax PRESSU [ type w ] { comp w } { loca w } { opti w } TOTAL PH INTPNT PO NODES EX CENTER PRESSU species the calculated pressure for undrained behavior as output item.
[TOTAL]
type species the type: TOTAL applies for pressure due to undrained behavior, the only possibility. comp selects components for output. PH for the pore pressure potential . PO for the total pore pressure: p = + f g (x xref ). EX for the excess pore pressure: pe = ref .
397
item PRESSU
type TOTAL
EX
PRex
pe
23.4.2
Nodal Discharges
syntax
FORCE [ type w ] [ form w ] { opti w } EXTERN PRESSU REACTI RESIDU FORCE species the nodal discharge as output item. These nodal discharges correspond to the basic degrees of freedom. type species the type: EXTERN for the externally applied discharges, REACTI for the reaction values (the constraints), RESIDU for the residuals (in each node). form species the formulation of the discharges: PRESSU for the pressures. Without this formulation Diana gives the translational forces.
item FORCE FORCE FORCE type EXTERN REACTI RESIDU form PRESSU PRESSU PRESSU
[REACTI]
QTpr
Q
QBpr
QB
QRpr
QR
23.4.3
Saturation
syntax
SATURA [ loca w ] { opti w } INTPNT NODES CENTER SATURA species the element saturation as output item.
item SATURA
398
Mixture Analysis
23.4.4
Darcy Flux
syntax
DARCYF [ oper w ] { comp w } { loca w } { opti w } LOCAL INTPNT GLOBAL NODES CENTER DARCYF species the Darcy ux in the element as output item.
comp X Y
Qx Qy
Z
Qz
qx
QX
qy
QY
qz
QZ
qX
qY
qZ
Part VIII
Phased Analysis
401 Introduction to phased analysis. Phased analysis enables modeling of phased construction. It determines the eects of construction history and shows the critical construction stages. A phased analysis comprises several calculation phases. Between each phase the nite element model changes by addition or removal of elements and constraints. In each phase a separate analysis is performed, in which the results from previous phases are automatically used as initial values. These results are typically stresses, deformations, potentials, velocities etc. The start of each phase can include input of the model part which is changed compared to previous input or added. At the start of each phase, you must select the active part of the model and specify the superposition of the nodal results (displacements or potentials) from previous phases. After the start you may perform a common analysis using regular Diana analysis modules. Diana can perform phased analysis for linear, nonlinear and dynamic structural analysis [Ch. 24], and for transient potential ow analysis [Ch. 25]. Also a phased staggered owstress analysis is possible [Ch. 22].
402
Chapter 24
optional!
Nonlinear analysis
*FILOS
Diana-9.4 Users Manual Analysis Procedures (VIII)
file .dcf
404
INITIA *INPUT commands *PHASE commands *NONLIN commands *INPUT commands *PHASE commands *NONLIN commands ...
optional!
24.1
In phased analysis you must start with reading the model data into the filos le with Module input. After input of the model, but prior to the actual analysis of a phase, you must perform the initialization of the phase with Module phase as described in 24.2 on page 408.
24.1.1
Via Module input you must read the model for the rst phase. The minimal input applies to the active part of the mesh and the actual boundary conditions. Input may additionally contain elements and nodes connected to inactive parts, which may be activated in subsequent phases.
file .dcf *INPUT
24.1.2
Adapt Model
Each following phase it may be necessary to adapt the model on the filos le. For instance, changes in boundary conditions often require adaptation of the model data and changes in the active part of the mesh may require such adaptation if the new elements and nodes were not supplied before. However, you should note the following. During a phased analysis the integration points of elements must keep their location.
24.1 Input of the Finite Element Model Consequently, you may not alter element properties which aect the location of the integration points. For instance, the node coordinates, the cross-section of beam elements,1 and the thickness of curved shell elements must all remain unaltered during a phased analysis. Module input oers three possibilities to adapt the model data on the filos le:
The single READ command replaces an existing table by a new one. The APPEND option of the READ command appends additional data to an existing table. The DELETE command deletes a table from the filos le.
405
See 3.3 on page 64 for formal syntax description of these commands. Phased analysis does not aect the already read tables, which are not appended or read again.
file .dcf *INPUT READ TABLE SUPPOR LOADS READ APPEND TABLE COORDI ELEMEN DELETE TABLE TYINGS
24.1.3
Import Deformation
Via the SUPERP IMPORT command [ 24.2], you may import the displacements of nodes resulting from the previous phase as constraints (xed displacements) in a subsequent phase. If you invoke this option, then you must also supply a special form of subtable DEFORM of table LOADS which species how to import displacements. syntax LOADS CASE case n DEFORM IMPORT LODSET loset n NONLIN
1 5 6 80
snode n
/ snodes ng... /
1 Except
DEFORM is the heading for the subtable with xed displacements. Keyword IMPORT indicates that the deformations come from the previous phase. LODSET species that deformations come from the linear static analysis in the previous phase, where loset is the load set number. NONLIN species that deformations come from the stepwise analysis inside the previous phase. snode is a supported slave node for which the values of xed displacement must be imported, snodes are a series of supported slave nodes specied by numbers and/or group names. type is the selected type of the xed displacements, TR for translations only, RO for rotations only. Default: all supported displacement types in the slave nodes. DI=dirnr is the selected direction of the xed displacements, dirnr refers to a direction in table DIRECT. Default: all supported directions of the displacement type in the slave nodes. mnode is the master node, mnodes are a series of master nodes. Imported displacements of the master nodes are transferred to xed displacements of the slave nodes. Default: if no master nodes are specied, Diana assumes that the master nodes and slave nodes are equal. Example
DIRECT 1 1. 0. 0. 2 0. 1. 0. 3 0. 0. 1. SUPPOR 982 TR 2 RO 3 983 TR 1 TR 2 984 TR 1 TR 2 RO 3 986 TR 1 TR 2 RO 3 LOADS CASE 1 DEFORM IMPORT LODSET 1 982 TR DI=2 981 982 RO DI=3 981 983 984 985
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VIII)
file .dat
407
Table DIRECT species the Model XY Z-directions. Load case 1 imports the displacements from load set 1 of the linear static analysis in the previous phase: u982 = u981 , 982 = 981 , all xed translations of node 983 from node 983, Y Y Z Z all xed translations and rotations of node 984 from node 985 and all xed translations of node 986 from node 985. Load case 2 imports the displacements from the stepwise analysis in the previous phase.
24.1.4
Initial Velocity
Stepwise dynamic analysis is an extension of stepwise transient static analysis with momentum contributions using masses and nodal velocities. The initial nodal velocities of an element determine the initial momentum of the connected nodes. By default, Diana uses the element velocities of the last phase in which the element was active as initial velocities of the active phase. Diana applies the following decision scheme to determine the initial velocities in phased analysis, if non-default initial values are required.
If specied, initial velocities of the element are taken from the table DATA.
syntax DATA
1 5 6 12 13 80
[ VELMOD
v1 r v2 r . . . vn r ]
VELMOD species initial element velocities: v1 for the rst degree of freedom in model XY Z-orientation, v2 for the second, etc. up to vn for the last. The sequence of the degrees of freedom is as follows (with n is the number of element nodes): u1 u1 u1 u2 u2 u2 . . . un un un . . . 1 1 1 2 2 2 . . . n n n X Y Z X Y Z X Y Z X Y Z X Y Z X Y Z In words: rst XY Z translations for element nodes, then XY Z rotations for element nodes. Translations or rotations according to the basic displacements for the element. For instance only translations u for a Q8MEM element but translations u and rotations for a CQ40S element [Vol. Element Library]. Condition: one velocity for each degree of freedom. This input is typically used whenever active elements have some well known initial velocities like a zero velocity.
Diana-9.4 Users Manual Analysis Procedures (VIII) December 15, 2009 First ed.
408
24.2
Phase Initialization
syntax
*PHASE [ BEGIN ACTIVE [ ELEMEN elems ng... ] [ REINFO reins ng... ] END ACTIVE ] [ BEGIN RESFOR [ ELEMEN elems ng... ] FACTOR factor r END RESFOR ] SUPERP ALL NONE IMPORT [ RENEW ] ACTIVE selects a certain part of the model to be active in the new phase. By default, Diana activates the complete model, as read previously [ 24.1]. With commands in an ACTIVE block you may select only a part of the model to be active. ELEMEN species active elements where elems is a series of element numbers and/or group names. If you dont specify active elements, then Diana will assume all elements available from input to be active. REINFO species active reinforcements where reins is a series of reinforcement numbers and/or group names. If you dont specify active reinforcements, then Diana will assume all the reinforcement parts in active elements of all reinforcements available from input to be active. RESFOR selects a certain part of the inactive model. For this part the residual forces of the inactivated elements will act as an external load on the
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VIII)
24.2 Phase Initialization elements that are active in the new phase. By default no elements are selected. With commands in a RESFOR block you may select inactive parts of the model to contribute to the external forces on active elements. Note that the stiness of the deactivated elements is not contributing in the new phase. ELEMEN species the inactive elements for which the residual forces must act as an external load on the active elements. elems is a series of element numbers and/or group names of the inactive elements. FACTOR species the fraction of the residual element forces that is applied as external load on active elements. SUPERP species how nodal results from the previous phase shall be superposed to the phased results. This command only applies for linear static analysis. In nonlinear analysis a similar eect can be achieved via the SUPPRE command [ 12.3.4 p. 238]. ALL will add the incremental displacements (nodal analysis results) of a new phase to the total displacements of the previous phases. NONE suppresses the superposition of incremental displacements to the total displacements in structural analysis. This is particularly useful to determine initial stresses due to dead weight load. IMPORT suppresses the superposition of the incremental phased displacements as specied in table LOADS [ 24.1.3]. In this way, the correct total phased displacement is obtained while assembling parts of the model. RENEW enforces Diana to evaluate all active elements and to renew their stiness, mass and damping matrices. By default, Diana will not evaluate the changed elastic material properties for existing elements in a phased analysis. However, with the RENEW command you may enforce an evaluation of the elastic material properties for both new and existing active elements.
file .dcf *PHASE BEGIN ACTIVE ELEMEN 1-7 REINFO 3 END ACTIVE SUPERP ALL
[ALL]
409
410
24.2.1
Reinforcement
One reinforcement section generally consists of a number of particles, embedded in elements [Vol. Element Library]. Diana generates element-by-element locations from global location input for new reinforcements, using the locations of all available elements. Diana also performs a geometry evaluation for new particles of active reinforcements in active elements. You may add new reinforcement to existing elements by direct reinforcement input or by activation in Module phase. Diana will implicitly add particles of existing reinforcements in newly activated elements, provided that the previously generated element-by-element location already included these elements. Changes in geometrical properties (location, cross-section) of existing reinforcements are not allowed. If you adapt any material properties of existing reinforcements in the input data, then you must also give the RENEW command to let Diana set up the stiness contributions for all active reinforcement parts.
24.2.2
Displacement Constraints
When initializing for a new phase, Diana checks and processes all specied supports, tyings and prescribed displacements from the appropriate tables on the filos le for the active nodes. In other words: in a new phase Diana will create a completely new set of degrees of freedom. Input tables which aect the creation of degrees of freedom are SUPPOR, TYINGS and LOADS. For subtable DEFORM of table LOADS you must realize the following. In phased analysis, Diana interprets the specied prescribed displacements as total loads. You may add new constraints to the model by appending data to the appropriate tables on the filos le [ 24.1.2].
24.2.3
Loads
When initializing for a new phase, Diana checks and processes all specied external loads (concentrated nodal loads, element loads, weight) and internal loads (temperature and prestress) from the filos table LOADS for the active nodes, elements and reinforcements.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VIII)
24.3 Output of Displacements In phased analysis, Diana interprets both the specied external and internal loads as total loads. You may add new loads to existing loads by appending data to the table LOADS on the filos le. If you want to add, adapt or delete loads on currently active nodes, elements or reinforcements, you must reread or delete table LOADS.
411
24.3
Output of Displacements
In linear and nonlinear phased analysis, there is a special option for the DISPLA output item to get output of the incremental displacements of a phase. See 3.6.2 on page 77 for general syntax. syntax *module w LINSTA NONLIN BEGIN OUTPUT device w [ options ] DISPLA [ type w ] PHASE END OUTPUT
PHASE gives the displacement increment in the active phase only. If you dont specify this option then Diana will output the superposed total displacements after the active phase.
24.4
To explain the basics of phased analysis we will rst describe the procedure for an unchanged model, i.e., elements, reinforcements and supports active in the rst phase will remain active in all subsequent phases, neither elements, nor reinforcements, nor supports will ever be added to the model during the phased analysis [ 24.4.1]. Then we will explain the analysis procedure for a changing model [ 24.4.3].
24.4.1
Unchanged Model
To illustrate the way of thinking for incremental loading in phased analysis we will use a very simple model: a bar fully clamped at one end, and for a certain analysis phase i loaded with an axial force load F(i) at the other end [Fig. 24.1].
412
Phase i Input F(i) u(i) Input i+1 F(i+1) Displacement
Incr. displacement
direct
Figure 24.1: Phases with unchanged model Phase i. To achieve an equilibrium state, i.e., in the direct analysis of phase i, the force causes a displacement u(i) of the loaded end. From the displacement u(i) Diana derives the strain (i) and then the stress (i) . Formally we could denote the direct analysis process of phase i as u(i) F(i) (i) (24.1) (i) Phase i +1. Suppose that in the subsequent phase the load increases with F(i+1) to a total of F(i+1) . When performing a phased analysis with Diana it is important to know that In each analysis phase, you must apply the total load on the model. That is because during the analysis of the new phase, Diana applies not only the total load F(i+1) , but also the strain (i) and the stress (i) from the previous phase which could be considered as an initial tension stress, i.e., a prestress load. The force causes a displacement u(i+1) , and the stress a displacement u(i) which is a compression equal to the elongation in the previous phase. These two together give an incremental displacement u(i+1) = u(i+1) u(i) . In phased analysis, the primary analysis results of a new phase are the incremental displacements. From the incremental displacement u(i+1) Diana derives the incremental strain (i+1) and the incremental stress (i+1) . The primary analysis process
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VIII)
24.4 Incrementing Analysis Results of phase i + 1 thus yields the incremental results u(i+1) Increment: F(i+1) F(i) (i+1) (i+1)
413
(24.2)
To terminate a phase in the analysis, Diana adds the incremental results (24.2) to the total results of the previous phase (24.1), which yields the total results of the current phase: u(i+1) = u(i) + u(i+1) Total: F(i+1) (i+1) = (i) + (i+1) (24.3) (i+1) = (i) + (i+1) Direct analysis. Alternatively we could have obtained this total result directly in a regular analysis, without the phased analysis option, by applying the total load F(i+1) on the model u(i+1) (i+1) (i+1)
F(i+1)
(24.4)
This equation is very much like the one for the previous phase (24.1). It seems that the only advantage of phased analysis is that we can get output of intermediate and incremental results. Is that really the only advantage of phased analysis? No, there is more: in phased analysis with Diana you may change the model from phase to phase [ 24.4.3]. Example. We will demonstrate the principle of total load input for each phase with the example of the bar which is loaded by a force F = 2 in phase 1 [Fig. 24.2]. We assume that all relevant physical and material properties are equal to 1. The results of phase 1 are simply to calculate: the bar will be elongated by u = 2 and there is a uniform strain = 2 and stress = 2 along the bar axis. In phase 2 the total load increases to F = 3 which we must apply on the model. Diana will now analyze the incremental situation with the load F = 3 and an internal tension stress = 2 from the previous phase. Due to the load, the elongation will be 3 and due to the internal stress the shortening will be 2. In other words: the real load is the external load (F = 3) minus the internal load (F = 2) which is equal to 1. This results in an incremental elongation u = 1, which yields an incremental strain = 1 and an incremental stress = 1.
414
Input phase 1 F =2 l=1 A=1 E=1 Increment F =3
Input phase 2 F =3
=2
Figure 24.2: Increasing load F Diana now calculates the total results of phase 2 by adding the increments to the results of phase 1. This gives a total elongation u = 3, a total strain = 3 and a total stress = 3. Obviously, the same total results would have appeared in a regular, non-phased, analysis where the load F = 3 was applied directly on the model.
24.4.2
Initial Stress
We will now demonstrate that In phased analysis with Diana you must not only apply the total external load for each phase but also the total initial stress. In other words: Diana considers initial stresses as loading and, if appropriate, you must specify them again for each phase. To illustrate this we consider the same model as in the previous section, but with an additional initial stress 0 , i.e., a prestress load [Fig. 24.3]. For this example we assume specic properties of the bar: the length is l, the cross-section is A, and the Youngs modulus of elasticity is E. Phase 1. The initial displacements and strains for phase 1 are zero. Therefore the total displacements and strains are the same as the incremental ones. If the total stress for phase 1 is (1) then the incremental stress is
0 (1) = (1) (1)
(24.5)
415
Strain and stress (1) (1) u(1)
Displacement F(1) A E
l Input 2
0 (2)
Incr. displacement
direct
Figure 24.3: Phases with unchanged model and initial stress If we denote the results from the force load with a subscript F and from the prestress with a subscript 0 , like uF and u0 for the displacements, then the results for phase 1 are u(1) = u(1) = uF(1) + u0 (1) 0 0 F(1) + F(1) (1) = (1) = F(1) + (1) (24.6) = 0 + = 0 + E (1) (1) (1) (1) (1) If we consider the results due to the prestress as increments then this situation is comparable with that of phase i + 1 in the previous example (24.3). Substitution of the models properties in (24.6) gives 0 l u(1) = u(1) = F(1) l (1) EA E u(1) = (1) = (1) l = F(1) (1) A
(24.7)
Phase 2. We assume that in phase 2 the total external load is F(2) and the 0 total stress from the previous phase is (2) . The phase increment to be analyzed now is: an internal stress (1) from the previous phase, a total force F(2) , and an incremental stress 0 0 0 (2) = (2) (1) (24.8)
Diana-9.4 Users Manual Analysis Procedures (VIII) December 15, 2009 First ed.
(24.9)
This results in an incremental displacement u(2) from which Diana determines the incremental strains and stresses: u(2) 0 Increment: F(2) + F(2) + F(1) (2) (24.10) (2) And substitution of the models properties in (24.10) gives 0 l 0 l u = F(2) l F(1) l (2) + (1) (2) EA EA E E 0 0 (2) (1) F(2) F(1) Increment: (2) = EA EA E + E = F(2) F(1) 0 + 0 (2) (2) (1) A A
(24.11)
The total results for phase 2 are now equal to those of phase 1 (24.6), with the addition of the calculated increments (24.10): 0 u(2) = u(1) + u(2) = uF(2) + u(2) = (1) + (2) (2) 0 (2) = (2) + (2) = (1) + (2) + (2) 0
Total:
(24.12)
Substitution of (24.7) and (24.11) in (24.12) for the models properties gives 0 l u = F(2) l (2) (2) EA E 0 (2) F(2) (24.13) Total: (2) = EA E = F(2) (2) A Direct. In the equations (24.13) for the total results of phase 2, all incremental values have disappeared. We would have obtained the same results when applying the force and prestress load of phase 2 directly to the model. This proves that the input of the total prestress in phase 2 yields the correct results. Linear analysis. We will demonstrate the principle of total initial stress input for each phase with the example of the bar which is loaded by a force F = 2
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VIII)
417
Input phase 2 0 = 2 F =5
=21+2
Figure 24.4: Increasing load F and initial stress 0 and an initial stress 0 = 1 in phase 1 [Fig. 24.4]. Again we assume that all relevant physical and material properties are equal to 1. In phase 1, the force F will cause an elongation of 2 and the prestress will shorten the bar by 1. This yields a total elongation of u = uF u0 = 2 1 = 1 and a strain = 1. The total stress in phase 1 is = 2 and the incremental stress is = 1. In phase 2 the external load increases to F = 5 and the stress from the previous phase = 2 (assuming that the prestress is still 1), which we both must apply on the model. Diana will now analyze the incremental situation with the load F = 5 an internal tension stress = 2 from the previous phase and the incremental prestress 0 = 1, i.e., we increment the prestress from 1 to 2. Due to the load, the elongation will be 5 and due to the stress of the previous phase the shortening will be 3. This results in an incremental elongation u = 2, which yields an incremental strain = 2 and an incremental stress = 2. According to (24.12) the total elongation after phase 2 is u = 3, the total strain = 3, and the total stress is equal to the prestress of phase 1 plus the stress of phase 1 minus the prestress of phase 1 plus the stress increment for phase 2: = 2 + 2 1 + 2 = 5. These total results are equal to those that we would have obtained in a direct, non-phased, analysis of the situation after phase 2.
24.4.3
Model Changes
In phased analysis you may change the model from phase to phase. For instance, you may declare some elements and/or reinforcements to become inactive or
418
Phased Structural Analysis declare inactive elements to become active again, add or delete elements in a new phase,2 change material properties, or remove existing supports or apply new supports. When the model changes the basic principles for structural analysis are the same as described in the previous section. However, in this case the inactive elements do not contribute to the transit of force through the model, and strains and stresses in inactive elements do not change until these elements become active again. Example. The scheme in Figure 24.5 shows what happens in phased analysis when in phase 2 an element is added to the model. Note that the total stresses
Input phase 1 0 = 1 F =2 Results u=1 =2 =1
Input phase 2 0 = 1 0 = 1 F =2
Increment + F =2 = 2 0 = 1 +
=2 =2 =1 =1
Figure 24.5: Addition of an element and strains for phased and direct analysis are alike for both elements. However, the total horizontal displacement u of the bar tip in phased analysis is half the displacement in direct analysis. This illustrates that in phased analysis, Diana connects new elements to the deformed mesh of the previous phase.
24.5
You may perform a phased nonlinear analysis with load steps (static) or time steps (transient static, dynamic). The analysis may be linear transient [Ch. 7], or nonlinear [Ch. 12]. A phased physical nonlinear analysis can be performed without restrictions. Phased geometrical nonlinear analysis is possible, but
2 It
24.5 Fixed Displacements in Nonlinear Analysis nonconservative loading cannot be applied. For material models where the strain and stress history is of interest, Diana will take this in to account whenever appropriate. An important dierence with other initial loads in phased nonlinear analysis is that In phased nonlinear analysis initial xed displacements, once specied in a particular phase, remain valid for all subsequent phases. In other words: initial xed displacements must only be specied once [ 2.3.4 p. 41].
419
420
Chapter 25
optional!
file .dcf
422
*PHASE commands *GROWTR commands *INPUT commands *PHASE commands *GROWTR commands ...
optional!
25.1
Implementation Backgrounds
The steps described in the previous section can be involved in each phase of a potential ow analysis. Results from previous phases are stored at element level and used as initial values. In transient analysis with capacity contribution, initial values will inuence the time dependent solution.
25.1.1
Basic Equations
Constitutive law: Equilibrium law: q=k (25.1) (25.2)
q = c + f
With q the ux vector, k the conductivity tensor, the potential, c the capacity and f volume sources. The Finite Element equations derived from the basic equations are: Mu+Ku=F (25.3) With M the capacity matrix, K the conductivity matrix, the vector of nodal potentials and F the resulting vector of nodal uxes, derived from external (dis)charges and constraints.
25.1.2
Solution
The potential ow nite element equations are solved stepwise with direct time integration. This means that for each step t, the equation is solved at time t + t. Combination of the Finite Element equations (25.3) and direct time integration results in an equation of the following form. K u = F (25.4)
With K the eective matrix, u the vector of nodal potentials, and F the eective vector of nodal uxes.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VIII)
25.2 Input of the Finite Element Model(s) Special for phased analysis is that at the rst time step in a new phase, the previously calculated values, stored at element level, contribute to F . The following paragraphs for implicit and explicit time integration show how these initial values ut contribute. Implicit time integration. The implicit time integration is characterized by >0 With the time derivative: ut+t = The eective matrix: K = Mt+t + t Kt+t The eective nodal ux vector: F = t Ft+t + Mt+t ut In case of nonlinearity the iteration scheme is: K u = F i i+1 i The extrapolation for < 1 is: ut+t = ut + ut+t t (25.10) (25.9) (25.8) (25.7) ut+t ut t (25.6) and u = ut+t (25.5)
423
Explicit time Integration. The explicit time integration is chracterized by =0 With the time derivative: ut = The eective matrix: The eective nodal ux vector: F = t Ft t Kt ut + Mt ut (25.14) and u = ut+t (25.11)
ut+t ut t
(25.12) (25.13)
K = Mt
25.2
Input of the nite element model is analogous to phased structural analysis [ 24.1 p. 404].
Diana-9.4 Users Manual Analysis Procedures (VIII) December 15, 2009 First ed.
424
25.3
Phase Initialization
syntax
To initialize a new phase you must invoke Module phase. *PHASE [ ACTIVE ELEMEN elmnrs ng... ] ACTIVE selects a certain part of the model to be active in the new phase. By default, Diana activates the complete model, as read previously [ 24.1]. With the ACTIVE command you may select only a part of the model to be active [ 24.2]. ELEMEN for selection of active elements where elmnrs is a series of element numbers and/or or group names. Default: if no selection is specied, all elements available from input are assumed to be active.
file .dcf *PHASE ACTIVE ELEMEN 1-7
25.4
25.4.1
Transient Analysis
Initialization
Each transient analysis part of a phase starts with an initialization of new elements for transient analysis, using specied initial elds if available. During initialization the initial calculation time can be specied to achieve a consistent time oset compared with other phases.
25.4.2
Time Steps
The stored element results from previous phases are used as initial values for active existing elements during the rst time step. The calculation and output of time dependent results is performed for active elements.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (VIII)
425
25.4.3
Aging
In phased analysis there is one global calculation time, but the age of each element is computed separately. The element age is to be considered as the total period of activity. The initial element age always equals zero. Material properties can be time dependent or age dependent, depending on the intended use. Age dependency is specied in table MATERI by using the AGE keyword instead of the TIME keyword for the specication of time points in the multilinear diagrams for the time and/or potential dependent material properties. syntax MATERI
1 5 6 12 13 80
] timvls r... agevls r... ] matvls r... matvls r... matvls r... matvls r...
times or ages
TIME timvls are the times for which the material properties are specied. AGE agevls are the ages for which the material properties are specied. See Volume Material Library for a description of the actual material properties.
426
Part IX
Solution Procedures
Chapter 26
430
Solve System of Equations TOLERA=eps is the tolerance for the check on accuracy of the solution process. This parameter is not applicable with the GENEL method. For all methods Diana applies useful defaults. You may customize the methods by specifying options as described in the referred sections.
[ = 108 ]
26.1
By default, Diana will automatically choose an optimal solution procedure which is initially based on a Sparse Cholesky method [ 29.1 p. 466].1 If a factorization from a previous Sparse Cholesky solution exists, the automatic procedure uses this as a preconditioner in an iterative process [ 26.4]. This automatic procedure does not require any additional SOLVE commands. However, for special cases you may customize the solution procedure via AUTOMA commands. syntax BEGIN SOLVE BEGIN AUTOMA [ FACTOR [ OFF ] ] [ SUBSTR options ] END AUTOMA END SOLVE FACTOR OFF suppresses factorization which may be useful if only the right-handside (load) vectors changed. SUBSTR customizes the application of substructuring [ 26.5].
26.2
For Intel based Windows and Linux platforms Diana oers the Intel PARDISO, parallel direct sparse solver. See 29.1.3 on page 467 for background theory. To apply this method you must give PARDIS commands. syntax BEGIN SOLVE BEGIN PARDIS [ FACTOR [ OFF ] ] [ MAXTHR mthread ]
there is not enough memory to carry out the Sparse Cholesky factorization, then an out-of-core direct solution procedure is performed [ 26.3].
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IX)
1 If
26.3 Out-of-Core Direct Solution END PARDIS END SOLVE FACTOR OFF suppresses factorization which may be useful if only the right-handside (load) vectors changed. MAXTHR=mthread species the maximum number of threads to be used. If the user has not explicitly set the number of threads by MAXTHR then this value can be set by the operating system to the available numbers of processors on the system. It is therefore always recommended to control the parallel execution of the solver by explicitly setting MAXTHR. If less processors are available then specied, the execution may slow down instead of speeding up.
431
26.3
Diana oers an out-of-core direct solution method based on Gauss decomposition. To apply this method you must give GENEL commands. syntax BEGIN SOLVE BEGIN GENEL [ FACTOR [ OFF ] ] END GENEL END SOLVE FACTOR OFF suppresses factorization which may be useful if only the right-handside (load) vectors changed.
26.4
Iterative Solution
To customize the iterative solution procedure for the system of equations you may give ITERAT commands. See 29.3 on page 473 for background theory. syntax BEGIN SOLVE BEGIN ITERAT [ PRECON [ ] [ MAXITE=miter n ] ] ILU DIAGON [ OUTCOR ] [ MAXTHR mthread ]
Diana-9.4 Users Manual Analysis Procedures (IX) December 15, 2009 First ed.
[ILU]
PRECON species the preconditioning process. By default Diana applies ILU preconditioning. You may customize the preconditioning process via the following options. ILU for the Incomplete LU-decomposition preconditioning, generally known as ILU preconditioning. This method is particularly suited for wellconditioned solid models [ 29.3.3.1 p. 474]. DIAGON for diagonal preconditioning, also called Jacobi preconditioning [ 29.3.3.2 p. 474]. OUTCOR applies an out-of-core solution algorithm. The computing time for out-of-core solution is considerably more than for the default in-core solution. MAXITE=miter is the maximum number of iterations n. The default depends on the applied preconditioning. MAXTHR=mthread species the number of decomposition domains to be set up for the iterative solver. If the user has explicitly set the number of decomposition domains by MAXTHR larger than zero, the domain decomposition solver, a parallel iterative solver, will be used. The domain decomposition will be performed once. Therefore, multiple specications of MAXTHR will result in the number of domains that has been specied by the rst MAXTHR command. The parallel iterative solver will automatically use the number of decomposition domains that has been set up. SUBSTR customizes the application of substructuring [ 26.5]. Default
SOLVE ITERAT file .dcf
[n = 400]
[n = 2000]
If you only specify the ITERAT option then Diana applies the most appropriate method with ILU preconditioning, as if you had given the following commands.
file .dcf BEGIN SOLVE BEGIN ITERAT PRECON ILU MAXITE=2000 TOLERA=1.E-8 END ITERAT END SOLVE
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IX)
26.5 Substructuring
433
26.5
Substructuring
By default Diana investigates whether substructuring in a structural nonlinear analysis with Module nonlin could be protable. If so, then substructuring will be applied automatically. In other types of analysis, for instance linear structural or potential ow, no substructuring is applied by default. To customize the application of substructuring, you may give the SUBSTR command. See 29.4 on page 476 for background theory. syntax SUBSTR [ AUTO ON OFF AUTO automatically applies substructuring if this seems to be protable. This is the default option if you only give the SUBSTR command. ON explicitly applies substructuring. NUMSUB nsub is the number of substructures to be set up. The default for structural nonlinear analysis is one substructure. For all other applications the default is two substructures. For a structural nonlinear analysis the substructure(s) comprise the elements without nonlinear material properties. For all other applications the substructure(s) comprise all active elements. OFF explicitly turns o substructuring. ] [ NUMSUB=nsub n ]
[AUTO]
434
Chapter 27
Eigenvalue Analysis
In order to perform an eigenvalue analysis with Diana you must take the following actions: 1. Invoke Module filos to initialize an analysis database [ 3.2 p. 60]. 2. Invoke Module input to read the nite element model into the database [ 3.3 p. 64]. 3. Invoke Module eigen to perform an eigenvalue analysis. 4. You may get a plot of the vibration modes in the postprocessing working environment of iDiana [Vol. Pre- and Postprocessing]. The primary tasks for Module eigen are invoked via the following command sequence. syntax *EIGEN [ MODEL ] [ TYPE ] [ EXECUT ] [ OUTPUT ] *END MODEL evaluates and assembles the nite element model [ 27.1 p. 439]. TYPE to specify the type of the eigenvalue problem [ 27.2 p. 439]. EXECUT to execute the eigenvalue analysis [ 27.3 p. 445]. OUTPUT to specify the analysis results for output [ 27.4 p. 446].
436
Eigenvalue Analysis Default. If you only give the *EIGEN command, then Diana assumes the following default command sequence.
file .dcf *EIGEN MODEL TYPE FREEVI LINEAR MASS CONSIS ROTATI ON BEGIN EXECUT NMODES=1 NTRIAL=2 MAXITE=30 SHIFT=0. TOLERA=1.E-6 OUTCOR OFF END EXECUT *END
This yields the rst eigenpair (value and mode) of a free vibration eigenvalue problem. Fluidstructure interaction analysis. In case of a model containing uid, structural elements and uidstructure interface elements, automatically a uidstructure eigenvalue analysis will be performed. Diana recognizes this element combination and will perform the required preparation to perform a uid structure eigenvalue analysis. In a free vibration eigenvalue analysis the added mass eect of the uid [Eq. (32.92) p. 537] is taken into account. Results will be available for the structural part of the model. Lumped element matrices may not be used in a uidstructure response analysis. Free vibration problem. This example illustrates the commands for generalized eigenvalue analysis of a free vibration problem in a concise form. Natural frequencies (in hertz) and corresponding shape modes are obtained by means of Module eigen with the following commands.
file .dcf *EIGEN MODEL TYPE FREEVI BEGIN EXECUT NMODES=20 END EXECUT OUTPUT DISPLA *END
437
In the free vibration eigenproblem the system mass matrix has to be used, which is invoked by TYPE FREEVI, and by default assembles the consistent element mass matrices. The EXECUT command block starts the calculation of frequencies and corresponding mode shape vectors, where parameter NMODES=20 asks for twenty frequencies. The number of trial vectors is twice the required number of eigenvalues. Diana takes reasonable defaults for the maximum number of iterations and the convergence criterion. Output results like natural frequencies and mode shapes are obtained by the OUTPUT command. Standard eigenproblem. The following example shows how to setup a standard eigenvalue analysis [ 29.2.1 p. 468] in a more comprehensive form; now the mass matrix M must be replaced by the identity matrix I. In order to save processing time or improve the accuracy for the iterative solution procedure, a number of parameters may be overruled. Eigenvalues and corresponding eigenmodes are now obtained with the following set of commands.
file .dcf *EIGEN MODEL TYPE IDENTI BEGIN EXECUT NMODES=5 NTRIAL=10 MAXITE=50 TOLERA=1.E-9 END EXECUT OUTPUT DISPLA *END
The use of the identity matrix I is specied by TYPE IDENTI. In order to overrule the defaults in the solution procedure, optional parameters have been specied: NTRIAL=10 asks for ten trial vectors, MAXITE=50 for a maximum of fty iterations and TOLERA=1.E-9 species a convergence criterion on eigenvalues of 109 . Unsupported system with shift. Consider the following example of an unsupported system which possesses rigid body modes. The stiness matrix K is now positive semidenite and has zero eigenvalues as much as the number of rigid body modes. For a frequency analysis, the stiness matrix K must be shifted according to Equation (29.27) on page 472. Before being output, calculated frequencies are corrected according to the performed shift.
file .dcf *EIGEN MODEL TYPE FREEVI
Diana-9.4 Users Manual Analysis Procedures (IX) December 15, 2009 First ed.
438
BEGIN EXECUT NMODES=20 SHIFT=1. END EXECUT OUTPUT DISPLA *END
Eigenvalue Analysis
The SHIFT parameter performs a shift on the stiness matrix. For frequency analysis, shifting is done by adding = 1 the system mass matrix M: (K + M) = 2 M Now a set of twenty shifted frequencies and corresponding mode shape vectors are calculated. Output results like natural frequencies and mode shapes are obtained by the OUTPUT DISPLA command. The resulting frequencies have been corrected according to Equation (29.28) on page 472. Nonlinear system with shift. This example examines the properties of a tangential stiness matrix KT , resulting from a static nonlinear analysis. Due to a nonlinear phenomenon like softening behavior, the stiness matrix KT may be no longer positive denite and for the negative eigenvalues the corresponding eigenmodes must be determined. By means of a proper shift, suciently large to make the smallest eigenvalue positive, the eigenproblem can now be solved by the standard iteration method.
file .dcf *NONLIN commands *EIGEN TYPE IDENTI NONLIN BEGIN EXECUT NMODES=2 SHIFT=100. MAXITE=50 TOLERA=1.E-9 END EXECUT OUTPUT DISPLA *END
The SHIFT command performs a shift on the tangential stiness matrix KT , by adding = 100 to its diagonal terms. Provided that the shift is suciently large, the analysis procedure starts the calculation of two shifted eigenvalues and corresponding eigenmodes. Parameter MAXITE=50 species a maximum of fty iterations and parameter TOLERA=1.E-9 a tolerance of 109 as measure for the convergence condition. Output results are obtained by the OUTPUT DISPLA command, where the printed eigenvalues have been corrected according to Equation (29.28) on page 472.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IX)
439
27.1
Model Evaluation
The MODEL commands customize the evaluation of the nite element model prior to the actual eigenvalue analysis. syntax BEGIN MODEL [ OFF ] [ EVALUA ] [ ASSEMB ] END MODEL EVALUA to check and evaluate geometric and material properties for elements [ 3.4 p. 66]. ASSEMB to assemble the elements, i.e., creating appropriate system degrees of freedom [ 3.5 p. 68].
27.2
Via the TYPE commands you may specify the type of the eigenvalue problem, i.e., the type of matrices to be used in the generalized eigenproblem formulated in Equation (29.12) on page 468. syntax BEGIN TYPE [ OFF ] FREEVI IDENTI STABIL END TYPE OFF switches off the type specication which may save computing time if an eigenvalue analysis was performed previously. FREEVI use mass matrices for free vibration frequency analysis [ 27.2.1 p. 440]. If you omit the TYPE commands then Diana assumes this type of analysis by default. IDENTI use an identity matrix for the standard eigenproblem [ 27.2.2 p. 442]. STABIL use element geometric stress-stiness matrices for linearized buckling analysis [ 27.2.3 p. 444].
Diana-9.4 Users Manual Analysis Procedures (IX) December 15, 2009 First ed.
[FREEVI]
440
Eigenvalue Analysis
27.2.1
Free Vibration
K = 2 M
The mass matrix M may be consistent or lumped, with rotational terms included or excluded. For the stiness matrix K, Diana can apply the linear elastic stiness with a geometric-stress stiness matrix added optionally. Alternatively the tangential stiness from a previously executed nonlinear analysis can be applied for K. See 29.2.2 on page 468 for background theory. Linear elastic stiness BEGIN FREEVI [ LINEAR ] [ MASS [ ] [ ROTATI [ ]] CONSIS ON LUMPED OFF [ STRESS [ OFF ] [ ] ] INPUT { } LOAD=loset n FACTOR=fac r CALCUL { } LOAD=loset n FACTOR=fac r PHASE END FREEVI Tangential stiness BEGIN FREEVI NONLIN [ MASS [ ] [ ROTATI [ ]] CONSIS ON LUMPED OFF [ STRESS [ OFF ] [ PHASE ]] END FREEVI LINEAR applies the linear elastic stiness matrix KL0 . This solves the eigenproblem KL0 = 2 M NONLIN applies the tangential stiness matrix KT from a previously executed nonlinear analysis. This solves the eigenproblem KT = 2 M
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IX)
syntax
syntax
441
MASS species the kind of mass matrices to be applied for M: CONSIS for consistent matrices (the default). LUMPED for lumped matrices. ROTATI indicates whether rotational terms must be included in the mass matrices: ON for yes, OFF for no. If you dont specify the ROTATI option explicitly, then Diana will include rotational terms by default. STRESS adds the geometric stress-stiness matrix KG to the stiness matrix K. This solves the eigenproblem (K + KG ) = 2 M A typical example is the calculation of the natural frequencies of a prestressed structure, for instance a guitar string. INPUT indicates a geometric stress-stiness matrix, with stresses specied via a prestress load for load set loset , in subtable ELEMEN and/or REINFO of table LOADS [Vol. Element Library]. CALCUL indicates a geometric stress-stiness matrix, with stresses which Diana will calculate automatically in a linear static analysis for load set loset (the default). PHASE indicates a geometric stress-stiness matrix, with stresses present from a previous phase. To setup the geometric stress-stiness matrix KG from a stress eld you may specify a load set number loset via the LOAD parameter. This load set number corresponds to a load set in input table LOADS [ 2.3.7 p. 55]. Default is the lowest available load set number. The optional parameter FACTOR species a multiplication factor fac .
file .dcf *EIGEN TYPE FREEVI *END
[ON]
[CALCUL]
[fac =1.0]
file .dcf *EIGEN TYPE FREEVI LINEAR MASS CONSIS ROTATI ON *END
442
Eigenvalue Analysis
file .dcf *EIGEN BEGIN TYPE BEGIN FREEVI LINEAR MASS CONSIS ROTATI ON END FREEVI END TYPE *END
file .dcf *EIGEN BEGIN TYPE BEGIN FREEVI LINEAR MASS CONSIS ROTATI ON STRESS CALCUL LOAD=3 END FREEVI END TYPE *END
file .dcf *NONLIN commands *EIGEN MODEL OFF BEGIN TYPE BEGIN FREEVI NONLIN MASS CONSIS ROTATI ON END FREEVI END TYPE *END
27.2.2
Standard Eigenproblem
The IDENTI commands specify the standard eigenproblem with the identity matrix I K = I For the stiness matrix K, Diana can apply the linear elastic stiness with a geometric-stress stiness matrix added optionally. Alternatively the tangential stiness from a previously executed nonlinear analysis can be applied for K. See 29.2.1 on page 468 for background theory.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IX)
27.2 Type of Eigenvalue Problem Linear elastic stiness BEGIN IDENTI [ LINEAR ] [ STRESS [ OFF ] [ ] INPUT { LOAD=loset n FACTOR=fac r CALCUL { LOAD=loset n FACTOR=fac r PHASE END IDENTI Tangential stiness BEGIN IDENTI NONLIN [ STRESS [ OFF ] [ PHASE ]] END IDENTI LINEAR applies the linear elastic stiness matrix KL0 . This solves the eigenproblem KL0 = I NONLIN applies the tangential stiness matrix KT from a previously executed nonlinear analysis. This solves the eigenproblem KT = I STRESS adds the geometric stress-stiness matrix KG to the stiness matrix K. This solves the eigenproblem (K + KG ) = I INPUT indicates a geometric stress-stiness matrix, with stresses specied via a prestress load for load set loset , in subtable ELEMEN and/or REINFO of table LOADS [Vol. Element Library]. CALCUL indicates a geometric stress-stiness matrix, with stresses which Diana will calculate automatically in a linear static analysis for load set loset (the default). PHASE indicates a geometric stress-stiness matrix, with stresses present from a previous phase. To setup the geometric stress-stiness matrix KG from a stress eld you may specify a load set number loset via the LOAD parameter. This load set number corresponds to a load set in input table LOADS [ 2.3.7 p. 55]. Default is the lowest available load set number. The optional parameter FACTOR species a multiplication factor fac .
Diana-9.4 Users Manual Analysis Procedures (IX)
443 syntax ] }
syntax
[CALCUL]
[fac =1.0]
444
Eigenvalue Analysis
file .dcf *EIGEN TYPE IDENTI *END
file .dcf *EIGEN TYPE IDENTI LINEAR STRESS CALCUL LOAD=2 *END
file .dcf *EIGEN BEGIN TYPE BEGIN IDENTI LINEAR STRESS CALCUL LOAD=2 END IDENTI END TYPE *END
27.2.3
Linearized Buckling
Due to the STABIL commands Diana applies the geometric stress-stiness matrix KG to solve the linearized buckling eigenproblem K = KG See Chapter 14 for background theory. syntax BEGIN STABIL [ LOAD=loset n ] END STABIL LOAD=loset applies the stresses of the specied load set to setup KG . The load set number corresponds to a load set in input table LOADS [ 2.3.7 p. 55]. Default is the lowest available load set number.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IX)
445
file .dcf *EIGEN BEGIN TYPE BEGIN STABIL LOAD=1 END STABIL END TYPE *END
27.3
With EXECUT commands you may customize the actual execution of the eigenvalue analysis. syntax BEGIN EXECUT [ OFF ] ]] [ OUTCOR [ ON OFF [ NMODES=nmodes n ] [ SHIFT=mu r ] [ NTRIAL=ntrial n ] [ MAXITE=maxitr n ] [ TOLERA=epconv r ] END EXECUT OFF switches o execution of the eigenvalue analysis. This may save computing time if you require other or more output of a previously executed analysis.
Diana-9.4 Users Manual Analysis Procedures (IX) December 15, 2009 First ed.
446
Eigenvalue Analysis OUTCOR indicates application of an out-of-core solution algorithm. The ON option switches out-of-core solution on, this increases the computing time considerably. The OFF option applies in-core solution, which also is the default if you omit the OUTCOR command. The single OUTCOR command is equivalent to OUTCOR ON. NMODES=nmodes species the number of eigenpairs to be calculated. Diana calculates the eigenvalues in ascending order. At most n eigenpairs can be calculated, where n is the dimension of the system matrix, which is the number of equations in the problem. If you dont specify the number of eigenpairs to be calculated, then Diana calculates only one pair by default. SHIFT=mu performs a shift where mu is the shift factor on K [ 29.2.3 p. 471].
[ON]
(nmodes n)
[nmodes = 1]
NTRIAL=ntrial is the number of trial vectors which may not exceed the dimension n of the mass matrix M. If you dont specify the number of trial vectors then Diana takes ve more than the number of eigenpairs to be calculated. MAXITE=maxitr is the maximum number of iterations. TOLERA=epconv is a preassigned error tolerance on eigenvalues. for the convergence criterion
[ = 106 ]
27.4
Diana always writes the following general information resulting from an eigenvalue analysis to the standard output le file .out: eigenvalues [ 27.4.1] and relative errors [ 27.4.2]. For a free vibration eigenvalue analysis additional information may be written: generalized masses [ 27.4.3], participation factors [ 27.4.4], modal damping factors [ 27.4.6], and eective masses [ 27.4.5]. The participation vectors (only available for a free vibration analysis) and eigenmodes may also be output [ 27.4.7].
eigen.out
4 EIGENVALUES FOUND AFTER EIGEN-FREQUENCIES: 0.39927D-01( 1) 0.10175D+00( RELATIVE ERROR ||R|| / ||Kx||: 0.25672D-14( 1) 0.50024D-15( MODE 1 2 3 4 MODE 1 2 3 FREQUENCY 0.39927D-01 0.10175D+00 0.15139D+00 0.21294D+00 FREQUENCY 0.39927D-01 0.10175D+00 0.15139D+00 1 ITERATIONS 2) 2) 0.15139D+00( 0.12696D-14( 3) 3) 0.21294D+00( 0.21452D-14( MODAL DAMPING 0.34458D-01 0.27646D-01 0.25010D-01 0.12886D-01 PART.FAC.TZ 0.00000D+00 0.00000D+00 0.00000D+00 4) 4)
GEN. MASS 0.10000D+01 0.10000D+01 0.10000D+01 0.10000D+01 PART.FAC.TX 0.30082D+01 -0.93124D+00 0.28889D+00
447
27.4.1
Eigenvalues
The resulting eigenvalues are output in ascending order, their type depends on the type of eigenvalue analysis performed. See also 29.2.4 on page 472 for background theory.
The standard eigenproblem yields the real eigenvalues from Equation (29.9) on page 468. The free vibration eigenproblem yields the natural frequencies f = /2 from Equation (29.12) on page 468. These frequencies are output as far as they have been found. Note that the unit of the frequencies depends on the units of the input data [ 1.1 p. 3]. If you supply the input data in ISO units N, m, kg, then the frequencies are in Hz (hertz). However, if you use N, mm, kg, then the unit of the frequencies is Hz 1000. A linearized buckling analysis yields the buckling values from Equation (29.14) on page 469.
27.4.2
Relative Errors
The error measures i for each eigenvalue and eigenmodes approximation is output [Eq. (29.34) p. 473].
27.4.3
Generalized Mass
In a free vibration analysis Diana will output for each eigenfrequency fi the corresponding generalized mass mii [Eq. (29.17) p. 469]. In Diana all eigenvectors are normalized such that the corresponding generalized mass is equal to one.
Diana-9.4 Users Manual Analysis Procedures (IX) December 15, 2009 First ed.
448
Eigenvalue Analysis
27.4.4
Participation Factors
In a free vibration analysis Diana will output for each eigenfrequency fi the corresponding participation factor i [Eq. (29.19) p. 469] and the direction dependent participation factors i for translations and, if available, rotations in global X, Y , and Z direction [ 29.2.2.4 p. 470].
27.4.5
Eective Mass
In a free vibration analysis Diana will output for each eigenfrequency fi the corresponding eective masses mef f.i for translations in global X, Y , and Z direction [ 29.2.2.5 p. 470]. The eective mass is also printed as percentage of the total structural mass of the model for each eigenfrequency fi . Furthermore, the cumulative percentage of eective mass with respect to the total structural mass is printed. For uidstructure analysis the mass contribution of the uid may result in (cumulative) percentages larger than 100 %, because the percentage is given with respect to the total mass of only the structural part of the model.
27.4.6
In a free vibration analysis, when you have specied an element damping coefcient he based on strain energy, Diana will output for each eigenfrequency fi the corresponding modal damping factors hi [Eq. (29.25) p. 471].
27.4.7
Eigenmodes
For the eigenmodes and participation vectors you may specify an output selection via the OUTPUT command block. For the general, analysis type independent options for output of analysis results see 3.6 on page 69. syntax BEGIN OUTPUT [ device w ] [outopt w... ] [ params ] [ OFF ] [ BEGIN SELECT [ ] [ MODES ] modes n... ALL END SELECT ] [ LAYOUT ] item w DISPLA
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (IX)
27.4 Output of Analysis Results END OUTPUT SELECT customizes the batch output selection. for model selection see 3.6.1 on page 71. MODES selects specic eigenmodes for output: modes is a series of mode numbers, ALL selects all calculated modes. LAYOUT optional commands to customize the layout and style of tabular output [ 3.6.3.1 p. 82]. item is the name of the analysis result to be output. See 3.6 on page 69 for complete syntax of this command. DISPLA for for normalized eigenmodes and participation vectors. syntax DISPLA [ type w ] [ form w ] [ oper w ] { comp w } { opti w } TOTAL TRANSL LOCAL MATRIX ROTATI GLOBAL PARTIC DISPLA species eigenmodes (displacements) as output item after an eigenvalue analysis. type species the displacement type. TOTAL for the total displacements of a structure, i.e., the deformed geometry of the eigenmode. Prior to outputting them, Diana normalizes the eigenmodes in such a way that the largest translation displacement component has a value of 1. MATRIX for the total displacement of a structure, i.e., the deformed geometry of the eigenmode. Prior to outputting them, Diana normalizes the eigenmodes with respect to the matrix used in the eigenvalue analysis, i.e., the mass matrix for a free vibration eigenvalue analysis, the identity matrix for the standard eigenvalue analysis, or the geometric stressstiness matrix for a linearized buckling analysis. Diana calculates the normalized eigenvectors with respect to the mass matrix according to Equation (29.18) on page 469. For the other types of eigenvalue analysis the mass matrix will be replaced by the appropriate matrix. See also 29.2.2 on page 468 for the used matrices. PARTIC for the participation vectors, i.e., weighted eigenmodes, calculated according to Equation (29.20) on page 470. This output is only available fro a free vibration eigenvalue analysis.
Diana-9.4 Users Manual Analysis Procedures (IX)
449
[ALL]
[TOTAL]
450 form
[GLOBAL]
oper species an operation (transformation) to be performed on the displacements [ 3.6.2 p. 79]. comp selects displacement components for output. Default is all available components. opti are additional options [ 3.6.2 p. 79]. Default
*EIGEN [ commands ] OUTPUT *END file .dcf
[all]
If you only give a single OUTPUT command, like in the above example, or if you dont give the OUTPUT command at all, then Diana gives output of the translational displacements in global XY Z orientation, as if you had given the following commands.
file .dcf *EIGEN [ commands ] BEGIN OUTPUT DISPLA TOTAL TRANSL GLOBAL END OUTPUT *END
comp X Y Z
Dtx Dty Dtz
ux DISPLA TOTAL TRANSL GLOBAL DtX uX DISPLA TOTAL ROTATI LOCAL Drx x DISPLA TOTAL ROTATI GLOBAL DrX X
uy
DtY
uz
DtZ
uY
Dry
uZ
Drz
y
DrY
z
DrZ
451
Z
Dmtz
um.x DISPLA MATRIX TRANSL GLOBAL DmtX um.X DISPLA MATRIX ROTATI LOCAL Dmrx m.x DISPLA MATRIX ROTATI GLOBAL DmrX m.X
um.y
DmtY
um.z
DmtZ
um.Y
Dmry
um.Z
Dmrz
m.y
DmrY
m.z
DmrZ
m.Y
m.Z
comp X Y
Dptx Dpty
Z
Dptz
up.x DISPLA PARTIC TRANSL GLOBAL DptX up.X DISPLA PARTIC ROTATI LOCAL Dprx p.x DISPLA PARTIC ROTATI GLOBAL DprX p.X
up.y
DptY
up.z
DptZ
up.Y
Dpry
up.Z
Dprz
p.y
DprY
p.z
DprZ
p.Y
p.Z
452
Eigenvalue Analysis
Part X
Background Theory
455 The purpose of Part X is to provide the Diana user the theory on which the Diana code has been based. It is not intended to be a general reference for the Finite Element Method. For that, readers are referred to the well-known handbooks like for instance Bathe [5]1 , Zienkiewicz [62] or Hughes [31]. The theory description here is subdivided in two major parts: the general concept of the Finite Element Method and theoretical aspects of the derivation of the element stiness matrices. Chapter 28 presents a summary of the general concepts of the Finite Element Method. This chapter intends to give an understanding of the basic principles of nite element theory and how these have been implemented in Diana. Chapters 29 and 30 present the methods of solution of the system of equations. This should provide the user insight in the command structure of Diana. In short, these chapters include the composition of the element matrices, the composition of the load vector, the methods of processing boundary conditions and tyings, solving the displacement vectors, and iterative solution of nonlinear systems. The remaining chapters give some background theory of particular types of analysis: structural dynamics, stability, potential ow, and soilpore uid.
1 Notation
456
Chapter 28
where K is the system stiness matrix, u is a vector of the unknown nodal degrees of freedom such as displacements and rotations and f is the vector of the nodal forces corresponding with the degrees of freedom u. This chapter describes the composition of the system stiness matrix, the solution of the degrees of freedom and the calculation of the element results as strains and stresses.
28.1
Global Formulation
When considering a general three-dimensional body, denoted by V , the problem is identied by unknown displacements u and known body forces per unit volume g. External forces in the form of concentrated forces and known tractions t are applied to the part St of the boundary and are called the natural boundary conditions. The displacements u are specied as known values u on the part Su of the boundary and are called the essential boundary conditions. In the Finite Element Method the body V will be approximated as an assemblage of nite elements, which are connected by nodal points on the element boundaries.
28.1.1
Displacements
In order to solve the problem the displacements u has to satisfy a continuity and dierentiability to the necessary degree. On the boundary Su the displacements must satisfy the essential boundary condition u=u on Su (28.2)
The displacements of a particular point (x, y, z) are assumed to be continuous functions expressed in terms of discretized variables at the nodal points and are
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
458 approximated as
(28.3)
where N is the displacement interpolation matrix and u is a vector of nodal point variables such as components of displacements and rotations, and is denoted as the vector of degrees of freedom. The interpolation matrix N comprises interpolation or shape functions described in terms of independent variables, such as coordinates and are locally dened for the individual elements.
28.1.2
where L is a dierential operator dening a compatible strain eld. Now the strain eld can be written as the derivative of the vector u as = L = LNu = Bu u (28.5)
where the matrix B denes the straindisplacement relation for a particular point and is called the dierential matrix. Assuming linear elastic behavior, the relationship between stresses and strains in a particular point can be written in the form = D( 0 ) + 0 (28.6) where the matrix D is the stressstrain relation and is a function of material properties like Youngs modulus E and Poissons ratio . The vector 0 denotes the initial strains changes etc. and the vector 0 contains the initial residual stresses.
28.1.3
Equilibrium
LT + g = 0 LT = t n on V
on St
where g is the vector of the known body forces per unit volume, with V as the total volume or domain of the model. Vector t represents the known traction forces on the boundary St such as surface, edge and point loads. For the derivation of the equilibrium equations the stationarity condition of the total potential energy can be used.
459
28.1.4
A simpler way of introducing the equilibrium relationships of (28.7) can be done by invoking the principle of virtual displacements. This principle states that an elastic structure is in equilibrium under a given loading system if, for any virtual displacement from a compatible state of deformation, the virtual work is equal to the virtual strain energy. The virtual work equation can be written as T dV =
V V
uT g dV +
St
uT t dSt
(28.8)
where are the virtual strains which correspond to the virtual displacements u. Substituting (28.3) and (28.5) into (28.8) gives uT
V
BT dV = uT
V
NT g dV +
St
NT t dSt
= uT r
(28.9)
where r is the vector of the internal forces corresponding to the vector of the nodal degrees of freedom u. The principle of the virtual work states that (28.9) should be satised for any u so that BT dV = r
V
(28.10)
These equations do not ensure that the equilibrium is satised at any point, but only guarantee that the stresses satisfy equilibrium in a weighted average sense. Substituting (28.6) and (28.5), the left hand side of (28.10) can be written as BT dV =
V V
BT DB dV
u
V
BT D0 dV +
V
BT 0 dV = r (28.11)
(28.12)
BT DB dV
(28.13)
is the system stiness matrix, and f is the right hand side vector dened by f = fg + ft + f 0 f 0 + fc with fg = ft = f 0 =
V St V
(28.14)
NT g dV NT t dSt BT D0 dV
the contribution of the body forces. the contribution of the surface tractions. the eect of the initial strains.
December 15, 2009 First ed.
460 f 0 = fc BT 0 dV
General Concepts of FEM the eect of the initial stresses. the contribution of the concentrated nodal forces.
This provides a set of linear simultaneous equations which can be solved in a direct or indirect way: u = K1 f (28.15) The method of solution is outlined in Chapter 29.
28.2
Discretization to Elements
In the Finite Element Method the solution domain V is divided into a nite number of elements Ve , which are connected by nodal points at the inter-element boundaries. In this way the solution domain is discretized and represented as a patch of elements. The unknown displacements in each element are now approximated by continuous functions expressed in terms of nodal variables; the functions over each nite element are called interpolation or shape functions. See Volume Element Library for more background theory.
28.2.1
Displacements
In each element the displacements of an arbitrary point (x, y, z) can be measured in a convenient local Cartesian coordinate system and are approximated by shape functions and nodal variables uc (x, y, z) = Nue (28.16) where N is the interpolation matrix with shape functions N (x, y, z) and ue the element nodal displacement vector,1 expressed in local xyz axes. This element vector can be composed from the nodal variables of the system degrees of freedom vector u by ue = Te u (28.17) where Te is the element transformation matrix which transforms the corresponding system degrees of freedom to the local element degrees of freedom, oriented in the xyz coordinate system. The rest is identical to 28.1.1.
28.2.2
Using the straindisplacement law for compatibility and assuming that the shape functions N are known, the discrete form of the straindisplacement relation can be written as = Bue (28.18)
1 From now on, the distinction between element and structural matrices, vectors and scalars has been done by adding a subscript e to the element quantities whenever this distinction cannot be obviously recognized.
28.2 Discretization to Elements Likewise (28.6) for the entire domain, the relation between strains and stresses, including initial strains and initial stresses, can be written for an element as = D ( 0 ) + 0 (28.19)
461
where D is the rigidity matrix representing the stressstrain law, usually derived from Hookes law, varying from element to element. Often the matrix D is only dened in a local element Cartesian (xl , yl , zl ) coordinate system. In order to obtain the strains in this system, it is necessary to apply a strain transformation l = T (28.20)
where T is the strain transformation matrix. With (28.18) the local strain vector l can now be related directly to the local element degrees of freedom vector ue by l = T Bue = Bl ue (28.21)
28.2.3
Element Assembly
In Diana, the assembling of elements is performed due to the ASSEMB command [ 3.5 p. 68]. During this process, also the constraints and the linear constraints (tyings) are handled. 28.2.3.1 Linear Constraints
The general tying equation for eccentric connection of three translations and three rotations is 1 0 0 0 z y uxj uxi u 0 1 0 yi 0 x uyj z u 0 0 1 u 0 zi y x zj = (28.22) 1 0 0 xj xi 0 0 0 0 0 0 yi 0 1 0 yj zi 0 0 0 0 0 1 zj Where uxi denotes the x translation of the slave node, yj the y rotation of the master node, x the eccentricity in x direction, etc. etc.
28.2.4
Now the structure has been idealized as an assemblage of elements, the integral form of the virtual work (28.8) can be rewritten as a summation of the virtual work done by the individual elements having volumes Ve and boundary surfaces Se like
ne ne ne
T dV =
e=1 Ve e=1 Ve
uT ge dV +
e=1 Se
uT te dS
(28.23)
462
General Concepts of FEM where ne is the total number of elements, ge is the element body force per unit volume and te are the element tractions 2 per unit area acting along the element boundary Se . For each element, its surface boundary Se can be separated in an exterior part and an interior part having imaginary interfaces with adjacent elements. Equation (28.23) is of fundamental importance for the displacement based Finite Element Method and imposes some restrictions on the displacement functions. In the Principle of Virtual Displacements nite element approximation, we will attempt to ensure equilibrium, which for an element looks like uT (LT + ge ) dV
Ve Se
(LT te ) dS = 0 n
(28.24)
It can be proved that this theorem is only valid provided that all derivatives of u and are nite through V . In general the stresses do not achieve continuity across the element interfaces. However, if the shape functions are chosen such that the displacements match at the nodes and the adjacent elements (i and j) have identical displacements at their interface, then a continuity condition on the stresses in the mean is met in the form of
Sei+j
t uT (LT i LT j e ) dS = 0 n n
(28.25)
where e is the contribution of applied external loads. This expression is another t approximation of satisfying equilibrium and therefore the equilibrium equation (28.24) is true within a single element and up to its surface boundary Se . Assuming that the displacement functions satisfy the conditions of (28.24), the integrations may now be performed over the element volumes and surfaces. Substituting for the element displacements and strains respectively (28.16) and (28.18), the virtual work equation for an individual element can now be written as uT BT dV = uT NT ge dV + uT NT te dS (28.26) e e e
Ve Ve Se
The integral form of the element boundary tractions NT te dS can be replaced by a kinematically equivalent nodal force vector re corresponding with the element degrees of freedom vector ue . Reordering and substituting re for the boundary tractions, the virtual work equation can now be expressed in a form uT e BT dV
Ve Ve
NT ge dV
= uT re e
(28.27)
As this relation is valid for any virtual displacement ue , the equilibrium equation for an element can be written as BT dV
Ve
2 Here
NT ge dV = re
Ve
(28.28)
the term tractions is not limited to an actual boundary surface, but is also used for the interior surfaces.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
463
28.2.5
Equation (28.28) is valid for any stressstrain relation and in case of a linear elastic behavior, substituting (28.19) for the stresses yields Ke ue + fe = re where Ke =
Ve
(28.29)
BT DB dV
(28.30)
NT ge dV
Ve
BT D0 dV +
Ve
BT 0 dV
(28.31)
is the element contribution to the right hand side vector f . Going back to (28.23) and using the piecewise approximation for the displacements (28.16) and the discrete straindisplacement relation (28.18), the virtual work equation is now obtained by
ne ne ne
uT e
e=1
BT dV =
Ve e=1
uT e
NT ge dV +
Ve e=1
uT e
NT te dS
Se
(28.32)
Substitution of the stressstrain relation (28.19) in case of linear elastic behavior yields
ne
uT e
e=1 ne
BT DB dV
Ve
ue =
ne
uT e
e=1 ne
NT ge dV +
Ve e=1
uT e
NT te dS+
St e
(28.33)
uT e
e=1
BT D0 dV
Ve Ve
BT 0 dV
28.2.5.1
Transformation
BT DB dV
(28.34)
The pre- and post-multiplication with T transforms the element stiness from local to global coordinates. The creation of element stiness matrices Ke according to (28.34) is performed in Dianas Module linsta due to the MATRIX command [ 4.1 p. 91].
464
28.3
The load vector f is composed of the external nodal forces as specied in the input le and of the assembly of the element loads. These element loads can be subdivided into the following components 1. Equivalent nodal forces due to thermal eects, eects resulting from difference in concentration and initial strains. Summing these eects results in an equivalent initial strain, which can be transformed to nodal loads. 2. Equivalent nodal forces resulting from initial stresses. 3. Equivalent nodal forces resulting from loads on element boundaries. 4. Equivalent nodal forces resulting from acceleration eects (dead weight). Due to the LOADS command [Ch. 4.1 p. 91], Dianas Module linsta calculates the above contributions per element and after that, for each degree of freedom the contribution of the connected elements are superposed and added to the external nodal point loads thus forming the load vector f .
28.4
Equilibrium
Invoking the theorem of the virtual displacements, the equilibrium equations of the element assemblage are Ku = f (28.35) where the matrix K is the stiness matrix of the element assemblage
ne
K=
e=1
TT Ke Te e
(28.36)
(28.37)
Te T Te T Te T Te T
Ve Se
NT ge dV NT te dS BT D0 dV BT 0 dV
the contribution of the element body forces. the contribution of the element surface tractions. the eect of the element initial strains. the eect of the element initial stresses. the contribution of the concentrated nodal forces.
ne e=1 ne e=1
Ve Ve
Chapter 29
Here subscript u is associated with the unconstrained degrees of freedom and subscript c with the constrained degrees of freedom. Equation (29.1) can be reduced to u f Kuu uu = fu Kuc uc or K = (29.2) Now all boundary conditions, tyings and loadings have been treated, the remaining system of equations can be solved. For convenience we write this equation in a more simple form Ku = f (29.3) The solution of u from this system of equations is often the most computation intensive part of a large-scale Finite Element Analysis. Generally matrix K is a known sparse n n matrix1 with a symmetric structure, f is a known right-hand-side vector and u the unknown solution vector to be computed. The solution of the system of equations (29.3) is done via the SOLVE commands [Ch. 26]. Basically there are two methods: direct solution and indirect
1 sparse
466
Solution Procedures of Sparse Linear Systems or iterative solution. The sequel of this chapter rst gives an overview of sparse matrix terminology, followed by a description of the available solution methods.
29.1
A general approach to solve the linear system of Equation (29.3) is rst to factorize K: K LU K LDU (29.4) (29.5)
where L, U and D are lower, upper and diagonal matrices respectively. The process of Equation (29.5) is known as LDU decomposition. After the factorization, the solution vector u can be computed by successive forward and backward substitution to solve the triangular system of equations (29.3) Lw = f DUu = w (29.6)
Under the assumption that K is a positive denite symmetric matrix the decomposition becomes LLT or LDLT .
29.1.1
If K is a sparse matrix then generally ll-in occurs during the factorization and the factor L will be more dense than the original matrix K. Fill-in is caused in the Gaussian elimination if in the operation kij kij kik kkj kkk (29.7)
the entry in location (i, j) of the original matrix was zero. It has been proved, see George & Liu [19], that the number of operations to compute the triangular factor L of a symmetric matrix K is equal to 1 2
N 1
(29.8)
where L(, i) denotes the i-th column in the triangular factor L and (L(, i)) is the number of non-zero terms, including the ll-ins, in column i. However, to take advantage of the sparseness of the coecient matrix K, the equations must be arranged in a special order. To nd a good ordering for a sparse symmetric matrix K we have to determine a permutation P that minimizes the ll-ins in the factor L of PKPT . Diana uses the ordering Metis ordering algorithm proposed by Karypis & Kumar [32]. If K is a symmetric positive denite matrix, pivoting is not required during factorization to maintain stability. Therefore, once the order is known, the non-zero structure of the factor
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
29.1 Direct Solution Methods L can be predicted in advance. The solution of (29.3) involves the following steps. 1. Ordering. Find a suitable ordering P for K, i.e., determine a permutation matrix P such that the ll of the factor L of PKPT is minimized. 2. Symbolic factorization. Determine the non-zero terms in L of PKPT and setup the corresponding data structure to store the lower half of PKPT and the factor L. 3. Numerical factorization. Assemble the terms of K into the the data structure and compute the factor L of PKPT . 4. Triangular solution. Solve Ly = Pf and DLT x = y and then set u = PT x. The rst two steps are symbolic operations and have to be performed only once. NewtonRaphson solution methods for nonlinear systems of equations [ 30.1.1.1 p. 482] are usually related to the following iteration scheme:
iterate: solve: set: for k := 0, 1, 2, . . . do K(uk )uk+1 = fk uk+1 = uk + uk+1 end do
467
Only step three and four have to be performed for each iteration.
29.1.2
The Sparse Cholesky direct solver in Diana basically is an implementation of the inner product formulation of the Cholesky factorization, see George & Liu [19]. This algorithm fully uses the nonzero pattern of the system matrix at the expense of introducing indirections of the inner loops. Prior to the factorization, the Sparse Cholesky solver reorders the system matrix according to the Metis algorithm, see Karypis & Kumar [32]. This reordering minimizes the ll-in of the matrix during factorization. Furthermore, to minimize the indirect access to arrays, the implementation applies so-called i-nodes, i.e., zones in the matrix with identical sparsity patterns. The code of the Sparse Cholesky solver is optimized for cache based memory access and achieves a performance which in most cases is superior to the Generalized Element method solver. However, the Sparse Cholesky solver is an in-core implementation and generally needs more primary memory than the out-of-core Generalized Element solver.
29.1.3
Intel Math Kernel Library (Intel MKL) provides a direct sparse solver PARDISO which can be used for solving real symmetric and structurally symmetric sparse linear systems of equations.
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
468
Solution Procedures of Sparse Linear Systems The PARDISO solver shows both a high performance and memory ecient usage for solving large sparse symmetric and unsymmetric linear systems of equations by shared multiprocessors. The solver uses a combination of left and rightlooking supernode techniques, see [52] and [53]. For suciently large problems, the scalability of the parallel algorithm is nearly independent of the sharedmemory multiprocessing architecture and a speedup of up to ve using eight processors has been observed.
29.2
Eigenvalue Procedures
This section briey summarizes the eigenproblems that can be solved and does not describe the actual solution procedure. We will emphasize on the standard eigenproblem, the generalized eigenproblem and on shifting of eigenvalues. Some attention will be paid to the analysis results as eigenvalues and -vectors, their accuracy and the generalized mass, modal damping factors, participation factors, and participation vectors.
29.2.1
Standard Eigenproblem
where Knn is the symmetric stiness matrix of the nite element model.2 There are n eigenvalues and corresponding eigenvectors satisfying (29.9). The i-th eigenpair is marked as (i , i ) if the eigenvalues are ordered increasingly: 1 2 n1 n The solution for p eigenpairs can be written as K = (29.11) (29.10)
where np is the matrix with eigenvectors and pp is a diagonal matrix with the corresponding eigenvalues.
29.2.2
Generalized Eigenproblem
A very frequently occurring eigenproblem is the free vibration equation to be solved in the mode superposition method (see also Chapter 32) written as K = 2 M (29.12)
2 The dimension of the stiness matrix K is dened by the number of independent equations n and is denoted as order n.
29.2 Eigenvalue Procedures where K is the symmetric stiness matrix and M is the mass matrix of the nite element model, is the circular natural frequency in radians per second. Eigenvalue is equal to 2 . The eigenvector is the corresponding mode shape vector . The mass matrix M can be obtained in a consistent mass analysis or in a lumped mass analysis. Analogous to (29.11) the solution for p natural frequencies squared and the corresponding mode shape vectors of (29.12) can be written as K = M2 (29.13) Two other possible generalized eigenproblems can be encountered in stability analysis and heat ow analysis. In stability analysis the problem is solved by linearized buckling analysis with K = KG (29.14)
469
where K is the stiness matrix and KG the geometric stress-stiness matrix [Ch. 34]. In heat ow analysis the equation is K = C (29.15)
where K is the heat conductivity matrix and C the heat capacity matrix [Ch. 35]. From now on, the eigenproblem will be discussed as K = 2 M (29.16)
from (29.12). However, the analysis description is also applicable to the solution of the other problems. For example, to solve the standard eigenproblem in (29.9), the mass matrix must be replaced by an identity matrix I and in case of a linearized buckling analysis (29.14) the mass matrix M must be replaced by the geometric stress-stiness matrix KG . 29.2.2.1 Generalized Mass
For each calculated frequency fi , Diana determines the corresponding generalized mass mii by mii = T Mi (29.17) i With the eigenvectors i normalized such that mii = 1 29.2.2.2 Participation Factor (29.18)
For each calculated frequency fi , Diana determines the corresponding participation factor by T M i (29.19) i = i mii Where i is the unity vector, i.e., a vector with a unity displacement for each degree of freedom and the eigenvectors normalized according to (29.18).
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
470 29.2.2.3
For each calculated frequency fi , Diana determines the corresponding participation vector by p.i = i i (29.20) The sum of all possible participation vectors of a structure will give the unity vector, i.e., a unity displacement for each degree of freedom:
n
i i = i
i=1
(29.21)
29.2.2.4
For each calculated frequency fi Diana determines the corresponding direction dependent participation factors i for the translational and rotational degrees of freedom in global X, Y , and Z direction. tX i = tY i = tZ i = rX i ltX i mii ltY i mii ltZ i mii
(29.22)
rY i = rZ i =
Where lt are the coecient vectors for each translational degree of freedom, lr are the coecient vectors for each rotational degree of freedom according to li = T M r i (29.23)
Where r is the inuence vector which represents the displacements resulting from a static unit ground displacement in the direction of the corresponding translational or rotational degree of freedom. The eigenvectors are normalized according to (29.18). 29.2.2.5 Eective Mass
For each calculated frequency fi Diana determines the corresponding eective masses mef f.i for the translational degrees of freedom in global X, Y , and Z
29.2 Eigenvalue Procedures direction. mef f.tX .i = mef f.tY .i = mef f.tZ .i =
471
Where lt are the coecient vectors according to (29.23). 29.2.2.6 Modal Damping Factor
When you have specied an element damping factor he based on strain energy, then Diana will determine the modal damping factors hi for each calculated frequency fi according to
n
he.j T Kj i i hi =
j=1 n
(29.25) i Kj i
T
j=1
The element damping factor he indicates the proportion of the element strain energy that has to be used as damping. The modal damping factors hi are useful to determine the two modes that should be employed to calculate the Rayleigh damping coecients.
29.2.3
Shifting
In an eigenvalue problem, a user-specied shift factor [ 27.3 p. 445] on the stiness matrix K causes Module eigen to provide the eigenvalues and -modes close to as rst. There can be several reasons to apply a shift to an eigenvalue problem:
Zero/negative eigenvalues. Some solution methods are not designed explicitly to calculate zero or negative eigenvalues. Shifting the stiness matrix K may be the solution in these situations. Softening. In case of softening material behavior, only the rst negative eigenvalue is of interest [Ch. 27 p. 438]. Shifting the stiness matrix K may help you to nd this eigenvalue and -mode. Perturbation analysis. In case of perturbation analysis [ 34.3 p. 546], where the nonlinear interaction of eigenmodes is considered, the accuracy of the interacting eigenvalues and -modes can be improved by a shift of the stiness matrix K.
472
Solution Procedures of Sparse Linear Systems Application of a shift reformulates the eigenvalue problem. For positive the eigenproblem formulation becomes K = 2 M K = K + M 2 = 2 + (29.26) (29.27) (29.28)
in which K is the modied stiness matrix, the eigenvalue 2 is related by (29.28) and the eigenvector is unchanged.
29.2.4
The calculated natural circular frequencies of the generalized eigenproblem of (29.12) are expressed in radians per second. The corresponding period T may be computed with 2 T = 2 T = (29.29) The cyclic frequency f , which is usually referred to as the frequency of motion, is the reciprocal of the period T and is given by f= 1 = T 2 (29.30)
and expressed in hertz, whereby one hertz is one cycle per second (cps). For the standard eigenproblem as given in (29.9), the eigenvalue can be obtained by taking the frequency squared 2 = 4 2 f 2 (29.31)
The quantities and can be determined from relation (29.30) and (29.31) respectively. Shifted system. If the eigenproblem has been shifted by a factor according to (29.27), Diana corrects the actual frequencies according to (29.28) with = 2 and the eigenvalues with = 2 = 4 2 f 2 (29.33)
1 2
= 4 2 f 2
1 2
(29.32)
where and f are the shifted frequencies and and f are the output frequencies.
473
29.2.5
Accuracy
The error measure for the calculated eigenvalue and eigenvector approximation and is determined as = Ki i Mi K i
2 2
(29.34)
with i as relative error for the ith eigenpair. This quantity should be small if i and i are an accurate solution of an eigenpair.
29.2.6
Solution Techniques
Diana applies an implicitly restarted Arnoldi method to solve the eigenvalue problem, see Lehoucq et al. [37].
29.3
The direct solution method has some drawbacks. The most important one is that the background storage requirements can be extremely high for large threedimensional problems. Another disadvantage is that the Gauss decomposition without pivoting (i.e., interchanging rows and columns) is not numerically stable if the stiness matrix is not positive denite. For these reasons two iterative methods are available in Diana as alternatives for the direct solution method. The common idea of all iterative methods for solving the linear system of equations Ku = f is to generate a sequence of approximations ui to the solution vector u via the recursion ui+1 = ui + i Q(f Kui ) (29.35)
in which Q is the preconditioning matrix or the preconditioner. In some way, Q should resemble the inverse of the stiness matrix K. Two algorithms to compute the iteration parameters have been implemented in Diana: the Conjugate Gradient method (CG) and the Generalized Minimal Residual algorithm (GMRES). Diana uses CG for symmetric matrices and GMRES for nonsymmetric matrices.
29.3.1
Conjugate Gradient
The Conjugate Gradient method, see Hestenes & Stiefel [26], is currently the most popular and probably the best iterative method for systems with a symmetric positive denite stiness matrix, for example all linear elastic problems. The CG algorithm generates i such that all residuals ri are perpendicular. By making clever use of the symmetry of K it is possible to orthogonalize the residual ri against all previous residuals by making use of only the residuals of the two previous iterations.
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
474
Solution Procedures of Sparse Linear Systems If the stiness matrix is nonsymmetric or indenite, i.e., has negative eigenvalues, the Conjugate Gradient method need not converge.
29.3.2
The GMRES method, see Saad & Schultz [51], converges even if the stiness matrix is not positive denite, for instance if the stiness matrix is nonsymmetric. The iteration parameters are computed by orthogonalizing the residual explicitly against all previous residuals. To be able to do this, all residuals must be stored. Moreover, the number of computations per iteration increases since the orthogonalization process becomes more expensive every iteration. Therefore the iteration is restarted after a xed number of residuals has been added to the basis.
29.3.3
29.3.3.1
Preconditioning
Incomplete LU-decomposition
By default Diana applies Incomplete LU-decomposition preconditioning, generally known as ILU preconditioning, see Meijerink & Van Der Vorst [39]. The idea of ILU preconditioning is to approximate the system matrix K by the product of a lower diagonal matrix L and an upper matrix U Q1 = L U K (29.36)
If the factorization is carried out exactly, we get a direct solution method. The disadvantage of the exact factorization is that ll-in occurs: the matrices L and U contain far more non-zero entries than the original matrix K. In the ILU approximation we try to restrict the ll-in of L and U. First we limit the ll to the sparsity pattern of K, i.e., Lij = 0 and Uij = 0 only if Kij = 0. The ILU decomposition is uniquely dened by Q1 = ij
k
Lik Ukj
if Kij = 0
(29.37)
If the setup of Q fails, or if the subsequent iteration does not converge, we improve the preconditioner by allowing more ll-in. Therefore we use a droptolerance strategy: non-zero elements are only included in the incomplete factors if they are larger than a given threshold parameter (ILUT, see Saad [50]). This threshold parameter is determined adaptively: we decrease it until the iteration has converged. We notice that we obtain the exact factorization if the drop tolerance is small enough. 29.3.3.2 Jacobi Preconditioning
The most simple and probably the most widely used preconditioning technique is to scale the stiness matrix with a diagonal matrix D. For problems with a
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
29.3 Iterative Solution Methods diagonally dominant stiness matrix the choice D equal to diag(K), and hence Q equal to the inverse of diag(K), is both natural and good. This preconditioner is known as Jacobi preconditioning or diagonal scaling.
475
29.3.4
Termination Criterion
The natural choice for the termination criterion of the iterative process is based on the reduction of the residual ri , ri = f K u i (29.38)
with ui the approximation of the solution u after i iterations. The iteration is stopped if ri < f (29.39) with a user-dened tolerance.
29.3.5
The advantage of a direct solution method is that for a given system of equations it is possible to compute how many oating points operations are necessary to determine the solution vector. Unfortunately this is not true for iterative methods. The number of oating point operations needed depends upon the rate of convergence of the iteration process. 29.3.5.1 Reordering
For direct methods it is of utmost importance that the equations are reordered such that the bandwidth becomes minimal. The savings in both computer time and memory consumption can be huge, especially for three-dimensional problems. This is due to the fact that the stiness matrix is decomposed. The situation is dierent for iterative solvers. The diagonal preconditioning does not depend on the ordering of the equations. For the ILU preconditioning the eciency does depend on the ordering, but the improvements are less consistent than in the case of a direct method. Extensive experimentation has indicated that ILU performs best with Metis ordering. 29.3.5.2 Convergence
The rate of convergence of the iterative method is hard to predict for a particular problem. However, there are some rules of thumb to give. The iterative solution method is best suited for the solution of the large sized linear systems that arise in three-dimensional simulations. Especially large potential ow problems can often be solved very eciently. If the stiness matrix is ill-conditioned, the rate of convergence will be very low. This is often the case if structural elements like beams, plates or shells are used in the model.
476
Solution Procedures of Sparse Linear Systems Another notorious example is if the model involves elements based on a penalty formulation. Ill-conditioning of the stiness matrix may also be caused by improper or few boundary conditions. An ill-conditioned linear system is susceptible to rounding errors, these may even ruin the solution. This is also the case for the direct solution method, but more hidden and therefore even more dangerous.
29.4
Substructuring
Substructuring is a standard technique in Finite Element Analysis (see Bathe [4, 8.2.4]). The basic idea is to treat a group of elements as a single substructure (superelement). The use of substructures is attractive in various cases. For example, if many elements in a nonlinear model behave linearly, these elements can be put in a substructure. The internal degrees of freedom in the substructure are then removed by static condensation. Suppose that we have two substructures that only contain linear elements. The degrees of freedom in the substructures can then be divided in internal degrees of freedom and interface degrees of freedom. After a proper reordering, the stiness matrix of this system can be written as A1 0 B 1 (29.40) 0 A 2 B2 BT 1 BT 2 C
where A1 and A2 are the submatrices representing the connectivity of the internal degrees of freedom of the substructure. The rectangular matrices B1 and B2 are the connectivity of the internal degrees of freedom and the interface degrees of freedom of the substructures. We explicitly assume that the stiness matrices of the elements in a substructure are symmetric. Finally, C represents the connectivity of the interface degrees of freedom and the degrees of freedom of elements not in any substructure. By construction, there is no connection between the internal degrees of freedom of the two substructures. The matrix (29.40) can be factorized as follows 0 A1
T
0 A2 B2
T
0 C
0 0
0 A1 B1 1 I I
(29.41)
A1 B2 2
B1
0 0
where I is the identity matrix and C the so-called Schur complement C = C BT A1 B1 BT A1 B2 1 1 2 2 (29.42)
The terms BT A1 B are nothing but the substructures after static condensation of the internal degrees of freedom. Diana assumes that the matrix A is
29.4 Substructuring symmetric positive denite, and rst constructs its Cholesky factorization LLT . Then BT A1 B is eciently computed as follows: BT A1 B = BT (LLT )
1
477
B = L1 B
L1 B
(29.43)
After all substructures have been assembled, it remains to solve a system of the form C u = f (29.44) In Diana the solution of this system can be obtained by factorization of C , or by using an iterative solver. This approach to substructuring is closely related to domain decomposition (see Chan & Mathew [12]). The main dierence is that in substructuring the elements are partitioned, whereas in domain decomposition the degrees of freedom are partitioned. A nal word about the eectiveness of substructuring. Even if we assume that the cost of assembling the substructures is negligible (they can be reused as is), it is not always true that the factorization of C is cheaper than the factorization of the full matrix. The matrix C is not only smaller but also but also denser than the full matrix. Especially if the substructures are characterized by a high ratio of interface degrees of freedom to internal degrees of freedom, the use of substructures may be detrimental to the performance. In this case Diana will give a warning message and turn o substructuring.
478
Chapter 30
In nonlinear Finite Element Analysis the relation between a force vector and displacement vector is no longer linear. For several reasons, discussed in Volume Material Library and 30.2, the relation becomes nonlinear and the displacements often depend on the displacements at earlier stages, e.g. in case of plastic material behavior. Just as with a linear analysis, we want to calculate a displacement vector that equilibrates the internal and external forces. In the linear case, the solution vector could be calculated right away but in the nonlinear case it cannot. To determine the state of equilibrium we not only make the problems discrete in space (with nite elements) but also in time (with increments). To achieve equilibrium at the end of the increment, we can use an iterative solution algorithm. The combination of both is called an incremental-iterative solution procedure. In this chapter we will consider a vector of displacement increments that must yield an equilibrium between internal and external forces, and a stiness matrix relating internal forces to incremental displacements. In reality the physical meaning of items in the displacement vector can also be e.g. a velocity or a Lagrange multiplier. In this chapter the physical meaning of what we call the displacement and force vector and the stiness matrix is irrelevant. Most often it represents a continuous system that is approximated using the Principle of Virtual Work, Galerkin discretization or another method. A good starting point is to strive to an equilibrium state in which the internal force vector equals the external force vector, satisfying boundary conditions. fint = fext ui = u0 i
Diana-9.4 Users Manual Analysis Procedures (X)
(30.1) (30.2)
December 15, 2009 First ed.
(i prescribed)
480
Solution Procedures for Nonlinear Systems In nonlinear analysis the internal force vector usually depends nonlinearly on the displacements (e.g. nonlinear elasticity). It can also depend on the displacements in the history. This is e.g. the case if the material is path dependent such as in plasticity and if large displacements facilitate multiple equilibrium solutions. The external force vector can also be displacement dependent. This is the case in geometrically nonlinear analysis, if the magnitude or the direction of the loading depends on the displacements such as with pressure on a wall. We can now write fint (u, history) = fext (u) (30.3) The system described above is already discretized in space. To enable a numerical solution, a time discretization is performed as well. Here time can have a real physical meaning e.g. in a creep analysis or it can be a pseudo-time, only to describe a sequence of situations. Starting at time t with an approximated solution t u, a solution t+t u is searched for which (30.3) holds. Within the time-increment, only the displacements at start and end are known. The internal force vector, which may be path dependent, is calculated from the situation at time t, the time increment t and the displacement increment u. The external forces only depend on the current geometry. If we consider only one increment, the time increment and the situation at the start of the increment (history) are xed. The equilibrium equation within the increment then only depends on u. We can write the nonlinear problem as: nd u such that
t+t
u = t u + u
(30.4)
and, with g as the out-of-balance force vector (the residual forces). g(u) = fext (u) fint (u) = 0 (30.5)
Acceleration forces (dynamic analysis) are considered in Chapter 32 and are ignored in this chapter. Starting the analysis at time tbegin we can increment the time with a number of increments, until the desired end value tend is reached.
30.1.1
Iterative Procedures
A purely incremental method usually leads to inaccurate solutions in nonlinear analysis, unless very small step sizes are used. In an iterative process the errors that occur can be reduced successively. This in fact realizes an implicit procedure. The allowable step size is usually higher than in case of a process without iterations (e.g. an explicit process). The general procedure is the same for all iteration processes [Fig. 30.1]. In all procedures, the total displacement increment u is adapted iteratively by iterative increments u until equilibrium is reached, up to a prescribed tolerance. Indicating the iteration number with a right subscript, the incremental displacements at iteration i + 1 are calculated from ui+1 = ui + ui+1 (30.6)
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
481
Begin increment
Calculate out-of-balance force g = fext fint,1 Predict change in displacements u ui+1 = ui + ui+1
no
Figure 30.1: Iteration process The dierence between several procedures is the way in which u is determined. The iterative increments are calculated by use of a stiness matrix K that represents some kind of linearized form of the relation between the force vector and displacement vector. The used stiness matrix can change every iteration, the matrix that is used in iteration i is called Ki . A direct approach is to determine the iterative increments by ui = K1 gi i (30.7)
where gi is the out-of-balance force vector at the start of iteration i. In this case a linear set of equations is solved at every iteration. Next sections describe the methods that are available in Diana. First three pure
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
482
Solution Procedures for Nonlinear Systems iterative procedures are presented: the NewtonRaphson method, the QuasiNewton method and the Constant Stiness method. Next, two variations that can be used in combination with these procedures are considered: the Continuation method and the Line Search method. Finally, several criteria to stop the iteration loop will be discussed. Another variation of the iteration algorithm is the Arc-length method [ 30.1.5.2 p. 492]. This method adapts the increment size. 30.1.1.1 NewtonRaphson
Within the class of NewtonRaphson methods, generally two subclasses are distinguished: the Regular and the Modied NewtonRaphson method. Both methods use (30.7) to determine the iterative increment of the displacement vector. In a NewtonRaphson method, the stiness matrix Ki represents the tangential stiness of the structure: Ki = g u (30.8)
The dierence between the Regular and the Modied NewtonRaphson method is the point at which the stiness matrix is evaluated. Regular NewtonRaphson. In the Regular NewtonRaphson iteration the stiness relation (30.8) is evaluated every iteration [Fig. 30.2]. This means that the prediction of (30.7) is based on the last known or predicted situation, even if this is not an equilibrium state. The Regular NewtonRaphson method yields
f
t+t
fext
fint,1
g1
fext
u0 u1
u1
Figure 30.2: Regular NewtonRaphson iteration a quadratic convergence characteristic, which means that the method converges to the nal solution within only a few iterations. A disadvantage of the method is that the stiness matrix has to be set up at every iteration and, if a direct solver is used to solve the linear set of equations,
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
30.1 Incremental-Iterative Solution the time consuming decomposition of the matrix has to be performed every iteration as well. Moreover, the quadratic convergence is only guaranteed if a correct stiness matrix is used and if the prediction is already in the neighborhood of the nal solution. If the initial prediction is far from the nal solution, the method easily fails because of divergence. In short: The Regular NewtonRaphson method usually needs only a few iterations, but every iteration is relatively time consuming. Modied NewtonRaphson. The Modied NewtonRaphson method only evaluates the stiness relation (30.8) at the start of the increment [Fig. 30.3]. This means that the prediction is always based on a converged equilibrium state. Usually, Modied NewtonRaphson converges slower to equilibrium than
f
t+t
483
fext
fint,1
g1
fext
u0 u1
u1
Figure 30.3: Modied NewtonRaphson iteration Regular NewtonRaphson. However, for every iteration only the prediction of the iterative incremental displacements and the internal force vector has to be calculated, it is not necessary to set up a new stiness matrix. If a direct solver for the linear set of equations is used, it is not necessary to perform the decomposition again, only the relatively fast substitution part will do. In short: The Modied NewtonRaphson method usually needs more iterations, but every iteration is faster than in Regular NewtonRaphson. In situations where Regular NewtonRaphson does not converge anymore, the Modied NewtonRaphson process can sometimes still converge. Small variations of both processes are possible by using the linear or previous stiness for the rst prediction and by setting up the current stiness matrix after the rst prediction. If unloading occurs, it can be advantageous to return to the linear stiness, e.g. in a plasticity analysis.
The Quasi-Newton method (also called Secant method) essentially uses the information of previous solution vectors and out-of-balance force vectors during the increment to achieve a better approximation [Fig. 30.4]. Unlike Regular NewtonRaphson, the Quasi-Newton method does not set up a completely new stiness matrix every iteration. In this case the stiness of the structure is
f
t+t
fext
fint,1
g1
fext
u0 u1
u1
Figure 30.4: Quasi-Newton iteration determined from the known positions at the equilibrium path. If the iterative displacement increment is called ui and the change in out-of-balance force vector related to this increment gi = gi+1 gi , the Quasi-Newton relation is Ki+1 ui = gi (30.9)
With a matrix Ki that fullls (30.9), the next iterative increment is calculated from (30.7). For a system with more than one degree of freedom, the secant stiness matrix K is not unique. The methods implemented in Diana are known as the Broyden, the BroydenFletcherGoldfarbShanno (BFGS) and the Criseld methods. By substitution it can be seen that the following two matrices fulll the Quasi-Newton relation (30.9). Ki+1 = Ki + (gi Ki ui ) cT cT ui (30.10)
Ki+1 = Ki +
(30.11)
In (30.10) and (30.11) the vector c can be chosen freely. The Quasi-Newton methods can be used eciently because the inverse of the new stiness matrix
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
30.1 Incremental-Iterative Solution can be derived directly from the previous secant stiness and the update vectors by using the ShermanMorrison formula. Broyden. If in (30.10) c is substituted by u and Ki+1 is inverted, the Broyden method results: K1 = K1 + i+1 i (ui K1 gi ) uT K1 i i i uT K1 gi i i (30.12)
485
BFGS. More elaborative, (30.11) can yield the relation attributed to Broyden, Fletcher, Goldfarb and Shanno, and therefore known as the BFGS method: K1 = i+1 I+
T ui gi uT gi i
K1 I i
gi uT i uT gi i
ui uT i uT gi i
(30.13)
The inverse secant stiness matrices are not calculated explicitly, but the iterative displacements u are calculated directly by substitution of (30.12) or (30.13) in (30.7). By successive application of (30.12) or (30.13), the correct secant stiness can be calculated from the stiness K0 that was used at the start of the increment and an update vector for every iteration. For every intermediate iteration one additional update vector is to be stored with size number of degrees of freedom. The higher the iteration number, the more additional storage is needed and the more additional vector calculations are to be performed. Criseld. To avoid the increasing storage and computation time requirements for the Broyden and BFGS methods, Criseld [15, 9.8] suggested to use only the most recent correction vector. For a one-dimensional situation this method still behaves as in Figure 30.4, but the Quasi-Newton relation (30.9) is not matched. All three Quasi-Newton methods can be used irrespectively of the stiness matrix K0 used for the rst prediction. This could be a tangential stiness matrix, as used in Figure 30.4, as well as a linear elastic stiness matrix. These methods usually have a convergence rate between that of the Regular NewtonRaphson and the Modied NewtonRaphson schemes. This holds also for the time consumption. For large systems, especially when using a direct solver, the time used per iteration will be more closely to the Modied NewtonRaphson than to the Regular NewtonRaphson scheme. For the Broyden and the BFGS schemes the memory and time consumption will increase with the number of iterations. 30.1.1.3 Linear and Constant Stiness
The Linear and Constant Stiness iteration methods can be used if the other methods become unstable, or if it is desirable to keep certain characteristics.
486
f
t+t
fext g1
fint,1 t fext u0 u1
u1
Figure 30.5: Linear Stiness iteration Linear Stiness. The Linear Stiness iteration method uses the linear stiness matrix all the time [Fig. 30.5]. This method potentially has the slowest convergence, but it costs the least time per iteration since the stiness matrix needs to be set up only once. Moreover, in case of a direct linear solver, the costly decomposition has to be performed only once. The Linear Stiness method can also be advantageous if it is desirable that the stiness matrix remains symmetric where the tangential stiness matrix would become non-symmetric. The Linear Stiness method is usually very robust, but it is very well possible that it follows unstable equilibrium paths after bifurcations. Constant Stiness. The Constant Stiness method uses the stiness matrix left behind by the previous increment. This means that if NewtonRaphson iterations are used during the rst phase of an analysis and Constant Stiness iterations in a second phase, the stiness in the latter will be equal to the last calculated stiness in the rst. If the Constant Stiness iteration is used since the rst increment, this method equals the Linear Stiness method. The Constant Stiness method can be used if NewtonRaphson or Quasi-Newton methods fail after a number of successful increments.
30.1.2
Continuation
If a deformation process is relatively continuous, then the displacements of the previous increment can be used as a rst prediction for the current increment. The displacements are scaled by the ratio between the current and the previous loading factor or by the ratio of time increments in case of time stepping. Figure 30.6 shows the Continuation method combined with Linear Stiness iteration. Compared to Figure 30.5, this example needs fewer iterations. But be careful:
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
487
fext g1
fint,1
t
fext
u0 u1
u1
Figure 30.6: Continuation method (Linear Stiness) If the equilibrium path shows irregularities, it is not advisable to use the Continuation method. An irregularity occurs for instance at loadingunloading points. In that case the displacement increments of the previous step are not a good prediction for the increments in the current step. In the Continuation method, the rst prediction of u involves only a scalar multiplication on vector u. This hardly uses any computation time.
30.1.3
Line Search
All iteration methods described so far are based on a reasonable prediction, so that the iteration process converges to the exact numerical solution. If the prediction is too far from equilibrium the iteration process will not converge! This easily takes place in structures with strong nonlinearities, for instance cracking. Line Search algorithms can increase the convergence rate and are especially useful if the ordinary iteration process fails. The Line Search algorithm uses a prediction of the iterative displacement increment u as obtained by one of the ordinary iteration algorithms and scales this vector by a value to minimize the energy potential . Although energy potential may not be the correct terminology for the physical behavior, e.g. in case of plasticity, this poses no problem for the algorithmic implementation within an increment. While the local minimum of the energy potential represents the equilibrium, the minimum in the line search direction can be regarded as the best solution in the predicted direction. To determine an optimal magnication factor, the incremental displacement vector ui+1 is calculated from a scaled iterative increment ui+1 = ui + ui+1
Diana-9.4 Users Manual Analysis Procedures (X)
(30.14)
December 15, 2009 First ed.
488
Solution Procedures for Nonlinear Systems where ui+1 is derived from the selected iteration method, see Criseld [15, 9.2]. A minimum of in the line search direction requires that the derivative of to must be zero. s() = u = = g()T u = 0 u (30.15)
We can try to nd the solution to s() = 0 by calculating s at various values of . The rst two values are readily derived from the original iteration process. Once the search direction is calculated, the values s(0) and s(1) are calculated by the inner product of u with respectively the out-of-balance force at the start and at the end of the iteration. Usually the Line Search method in Diana is only used to help the ordinary iteration processes, therefore the line searches do not really continue until a value s = 0 is found, but the line search is terminated if the absolute value of s() has a value that is less than times the value s(0) [Fig. 30.7]. In most cases a value = 0.8 is sucient to stabilize the global s min max acceptable s region
s(0) s(0)
3 2 0 s(0)
1 1
Figure 30.7: Line Search iteration iteration process. If the values of s have an opposite sign, a Regula-Falsi method is used to predict the value where s() = 0. If both values have the same sign, extrapolation is used to determine a new value for . To avoid unrealistic values, Diana bounds between an upper and lower bound, min and max . Diana stops the Line Search process if the change in per iteration is less than the selected tolerance.
30.1.4
Convergence Criteria
The iteration process must be stopped if the results are satisfactory [Fig. 30.1 p. 481]. For this purpose, Diana oers several convergence norms. Besides
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
30.1 Incremental-Iterative Solution stopping the iteration in case of convergence, the iteration process is also stopped if a specied maximum number of iterations has been reached or if the iteration obviously leads to divergence. The detection of divergence is based on the same norms as the detection of convergence. Figure 30.8 species the items used to set up the various norms. f
t+t
489
fext g1
fext E0 E1 g0
u u0 u1
30.1.4.1
Force Norm
The force norm is the Euclidian norm of the out-of-balance force vector g. To check convergence, the force norm after the current iteration is checked against the norm of the initial unbalance g0 Force norm ratio =
T gi gi T g0 g0
(30.16)
Because the reference force norm is known before the rst prediction of displacements, the force norm ratio can be calculated directly after the rst prediction, i = 1 in (30.16). This means that if the rst prediction is correct (nearly linear behavior) the force norm can detect convergence right away and no unnecessary iterations have to be performed. 30.1.4.2 Displacement Norm
The displacement norm is the Euclidian norm of the iterative displacement increment. To check convergence, the displacement norm is checked against the norm of the displacement increments in the rst prediction of the increment. Displacement norm ratio = uT ui i uT u0 0 (30.17)
490
Solution Procedures for Nonlinear Systems From (30.17) it is clear that the ratio of the displacement norm after the rst prediction (iteration 0) equals 1 by denition. To check convergence, always one additional iteration is necessary. 30.1.4.3 Energy Norm
A third way to check convergence is the energy norm. This norm is composed of internal forces and relative displacements as indicated in Figure 30.8 with E0 and E1 . To determine convergence, the energy ratio is calculated as Energy norm ratio = uT (fint,i+1 + fint,i ) i uT (fint,1 + fint,0 ) 0 (30.18)
Note that here the internal force is used and not the out-of-balance force. Use of the out-of-balance force would be improper, for a Line Search procedure could then minimize the norm, see Equation (30.15), before the solution really converges to equilibrium. As with the displacement norm, the energy norm also requires an additional iteration to detect convergence. The choice of the proper norm and its convergence criterion depends on the type of analysis. Using a lot of prescribed displacements generally makes the displacement norm less useful. On the other hand, a structure that can expand freely will hardly build up any internal forces and the force norm may be less useful. Always be sure that the reference norm (the denominator in the ratios) has a reasonable value i.e., not close to zero. Experience shows that the convergence criterion for softening type behavior should be more strict than the criterion that can be used in a hardening type analysis. If there is any doubt about the criterion to be used, it is advisable to perform the analysis with two distinct criteria and check the dierences in results. If large dierences occur, at least the less strict norm was to large. 30.1.4.4 Residual Norm
The residual norm is also a Euclidian norm of the out-of-balance force vector g. Contraray to the force norm, the residual norm also takes the values in constrained degrees of freedom (supports and tyings) into account. To check convergence, Diana compares the change in the residual norm during the current iteration with the change in the residual norm during the rst prediction of displacements in the current step. Residual norm ratio =
T gi gi T gi1 gi1 T T g0 g0 gn gn
(30.19)
Where gn denotes the out-of-balance force vector in the last iteration of the previous step. In the rst step Diana takes its value as zero.
491
30.1.5
Incremental Procedures
The incremental-iterative solution procedure consist of two parts: the increment part and the iteration part. The iteration part was discussed in 30.1.1, in this section the increment part is treated. We rst describe the most simple types: load control and displacement control. Then the Arc-length method is discussed, a method that can adapt the step size depending on the results in the current step. The initial choice of the step size for every increment is an important factor in the incremental-iterative process. Therefore, two methods are presented to determine step sizes and two methods to choose between loading and unloading depending on the previous analysis results. Finally, we present the Automatic Incremental Loading method [ 30.1.5.4]. This method requires only the nal loading to be specied. From this, Diana automatically determines the intermediate step size. 30.1.5.1 Load and Displacement Control
In 30.1.1 we have presented iteration processes where the external load was increased at the start of the increment, by directly increasing the external force vector fext . This is usually called load control [Fig. 30.9a]. Another way to put an external load on a structure is to prescribe certain displacements uc . This is called displacement control [Fig. 30.9b]. In case of displacement control the f
t3 t2 t1
uc
t2
uc
t3
uc
t4
uc
Figure 30.9: Load and displacement control external force vector is not increased directly. To get a proper rst prediction of the displacements, the prescribed displacements must be incorporated in the external force vector. This eective force can be calculated by rewriting (30.7) and splitting the displacement increment vector in two parts: one referring to the unconstrained and an other referring to the constrained displacements, respectively uu and uc . The stiness matrix and force vector are split likewise: Kuu Kcu Kuc Kcc
0
uu uc
0
gu gc
0
(30.20)
492
Solution Procedures for Nonlinear Systems The unknown displacement increments uu can be calculated from the rst row in (30.20) as 1 u uu = (Kuu ) {Kuc uc + g0 } (30.21) 0 0 Comparing (30.7) and (30.21) indicates that Kuc uc can be regarded as the 0 eective force vector, equivalent with the prescribed displacements. In subsequent iterations, the iterative increments of the prescribed displacements are zero and hence the eective force vector vanishes. A similar eective force vector can be generated in case of inuence of time on the analysis e.g. prescribed temperature increments or viscoelastic material behavior. In this case, the eective force vector contains the eect on the internal force vector during the time increment if the displacements remain constant. The addition of this eective force vector in the rst prediction (zero iteration) will improve the convergence of the iteration process signicantly. In subsequent iterations, the time does not change anymore and also this eective load vector will vanish. In real-life analysis, the loading does not have to be restricted to load control, displacement control or time increments, but they can be combined in any way. In that case the real external load and the eective force vectors from prescribed displacement increments and time inuences must be used together. 30.1.5.2 Arc-length Control
In an ordinary iteration process the predictions for the displacement increments can become very large. This is the case especially if the loaddisplacement curve is almost horizontal. If a xed load increment is prescribed, this results in very large predictions for the displacements. The problem can be overcome with the use of an Arc-length method. Using the Arc-length method the snapthrough behavior of Figure 30.10a1 can be analyzed, just as displacement control f f
1 In this gure, the load factor is included in the denition of the arc length. This is not the case in Diana, but the results in case of more than one degree of freedom are similar as indicated in the gures for one degree of freedom.
30.1 Incremental-Iterative Solution could. Here, however, it is possible to dene a system of loads that could not be substituted by prescribed displacements. Moreover, the Arc-length method is also capable of passing snap-back behavior [Fig. 30.10b], where displacement control fails. The Arc-length method constrains the norm of the incremental displacements to a prescribed value. This is done by simultaneously adapting the size of the increment. Note that the size is adapted within the iteration process and is not xed at the moment the increment starts. For this purpose we dene the external force vector at the start of the increment as t fext and the increment of the external force vector as i . The load factor i multiplies a unit load f f and can change every iteration. Substitution in (30.7) results in f ui = K1 i + t fint fint,i i The solution ui is now split in two parts: uI = K1 i i
t
493
(30.22)
fint fint,i
and
uII = K1 i i f
(30.23)
The load factor i is still undened and can now be used to constrain the incremental displacement vector. Diana oers a quadratic and a linearized constraint, leading to the Spherical Path Arc-length method and the Updated Normal Plane method, see Criseld [15, 9.3]. Spherical Path. In the spherical constraint, the constraint equation is uT ui = l2 i (30.25)
where l is the required arc length. Substitution of (30.6) and (30.24) in (30.25) gives the value for i = with a1 = (uII )T uII i i a2 = 2(uI )T uII + 2(u)T uII i i i a3 = 2(u)T uI + (uI )T uI + (u)T u l2 i i i Normally, two solutions for fulll (30.25) but if the discriminant a2 4a1 a3 < 2 0 then Diana uses a linearized equivalent of the Spherical Path method as described by Forde and Stiemer [18]. To determine which of the two regular
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
a2
a2 4a1 a3 2 2a1
(30.26)
494
Solution Procedures for Nonlinear Systems solutions should be used, the angle between the displacement increment vector of the previous iteration and the current iteration is calculated for both solutions cos = (ui1 )T ui ui1 ui (30.28)
If one of the solutions yields a negative cosine and the other a positive, Diana chooses the solution with the positive cosine (acute angle). If both solutions yield acute angles, the solution closest to the linear solution = a3 /a2 is used. Updated Normal Plane. The second constraint is a linearized constraint. If (30.25) is matched for ui1 , then the constraint equation for ui = ui1 + ui can be written as (ui1 )T ui = 0 (30.29) where the quadratic term in ui is ignored. Substituting (30.24) into (30.29) leads to the expression for i i = (ui1 )T uI i (ui1 )T uII i (30.30)
Geometrically this constraint means that the iterative increment must be perpendicular to the total increment at the previous iteration. The solution is projected on the plane, normal to the previous solution, hence the method is referred to as the Updated Normal Plane method. Indirect Displacement Control. In the previous description of the constraint equations all displacements were gathered together. For global nonlinear behavior this is adequate, but for local collapse mechanisms the method can perform better if only a part of the displacements is considered. The constraint equations can remain the same as in (30.25) and (30.29), but instead of using the vectors u and u vectors v and v are considered, dened by 1 u1 2 u2 v = 3 u3 (30.31) ... n un In the extreme case that only one item in v is non-zero, the arc length is dened as the displacement of the corresponding degree of freedom. A constant arc length during the analysis will result in this case in equal displacement increments for this degree of freedom. Because the loading is dened as an external force, this type of control is called Indirect Displacement control. A variant of Indirect Displacement control is Crack Mouth Opening Displacement control, usually called CMOD. This can be used, just as in experiments, to
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
30.1 Incremental-Iterative Solution control the increase in crack width et cetera. In case of CMOD control, a vector is formed with new degrees of freedom that can e.g. represent the dierence in displacements on opposite nodes on a crack plane. ... ... v= (30.32) 1 up + 2 uq ... This vector is used in the constraint equations (30.25) or (30.29). As long as the displacement increments per step remain relatively small, the dierence between the Spherical Path method and the Updated Normal Plane method are small. More important than the choice between these two methods is the choice of the value for the arc length l. The available Arc-length methods are particularly useful if they are combined with adaptive load incrementation, as described in the next section. 30.1.5.3 Adaptive Loading and Time Increments
495
Up to here we have used the initial load, displacement or time increment as a xed value. In an analysis, we could e.g. reach a load level of 100 N, by dening ten increments of 10 N. In combination with Arc-length control, the size of the increment can change inside an increment, but the start value was still xed. The size of the increments is limited by the physical behavior in case of path dependency and by the convergence characteristics of the selected iteration process. Especially in the latter case, the allowable step size depends on the amount of nonlinearity in the increment. This is usually not known a priori, i.e., at the moment that the analysis starts, and therefore the optimum increment sizes cannot be xed beforehand. To allow for result dependent increment sizes adaptive loading can be used. An even more important question that can usually not be answered before the analysis is started is whether at a certain load level the load must increase further or must decrease, e.g. in case of a snap-through. Three adaptive loading methods are implemented in Diana, an iteration based method for all types of loading, an energy based method that can only be used in combination with Arc-length control [ 30.1.5.2 p. 492], and the Automatic Incremental Loading method [ 30.1.5.4]. For the rst two methods, two algorithms are available to decide whether the next step must be an increment or a decrement. The Automatic Incremental Loading method does not allow for loadingunloading switching but oers the possibility for error-controled time increments if the physical behavior explicitly depends on the rate-of-change of the solution. Iteration Based. In the iteration based method, the experience is used that for many analyses, the iteration process converges faster if the increment size
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
496
Solution Procedures for Nonlinear Systems is smaller. Based on a desired number of iterations the new increment size can be made larger than the previous if the actual number of iterations in the previous increment was smaller than the desired number and vice versa. The size of the new load increment t+t 0 is calculated by
t+t
0 =
l uT u0 0
Nd tN
(30.33)
Here t l is the length of the predictor displacements of the previous step. Further N d is the desired number of iterations and t N is the actual number of iterations in the previous increment. The power is usually set equal to 0.5 . The iteration based method is also very useful to pass sharp snap-throughs or softening behavior in crack propagation analyses. When parameter is set equal to zero, a constant arc length is applied throughout the whole analysis. Experience shows that this method is stable in case of softening behavior. Since l uT u0 0
t
t+t
0 =
and
t+t
l =
t+t
uT u0 0
(30.34)
it follows that the length of the incremental displacement vector remains constant. Energy Based. The energy based method can only be used in combination with the Arc-length method. This method calculates a load increment such that the vector product of the load increment and the displacement increment in the rst prediction equals the vector product of the nal load increment and displacement increment of the previous step [Fig. 30.11]. Here t W indicates a kind of nal energy increment of step 1 and t+t W a rst prediction thereof in step 2. The new loading factor is derived from
t n
t+t
0 =
(t un )T f uT 0 f
(30.35)
where index n indicates the last iteration (of the previous increment). In order to get a proper initial value for the vectors uI and uII must be calculated 0 0 with a tangential stiness matrix. LoadingUnloading. A simple way to choose between increments or decrements is based on the appearance of negative pivots in the global system of equations. Often a negative pivot indicates unstable structural behavior that is related with some type of snap-through. If this occurs the sign of the load increment must be changed from loading to unloading.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
497
t+t
Another method, proposed by Criseld, can only be used in combination with the Arc-length methods and is similar to the method used in the Spherical Path Arc-length method. The angle between the new prediction and the previous increment should be acute, so + t+t if t uT uII 0 n 0 t+t (30.36) = t+t if t uT uII < 0 n 0 In many occasions, both loading and unloading criteria will yield the same result. In case of multiple equilibrium paths (bifurcations) the methods may dier and one should be aware that an equilibrium path is followed and not necessarily a stable equilibrium path. 30.1.5.4 Automatic Incremental Loading
Automatic load stepping is a simple tool for adaptive load increments. Given a nal loading, the automatic load step controller tries to take as few load steps as possible and at the same time tries to limit the number of steps in the iterative procedure. The main advantage over the iterations based load controller is that the automatic load controller recovers from non-convergence in the iterative solver. An outline of the algorithm is as follows.
t+t
0 minsiz
t+t
t+t
0 maxsiz
(30.37)
First the full loading is applied in a single step. If the iterative procedure fails to converge, the load step is decreased by a factor cutb [ 12.3.2.5 p. 233] and the calculation is restarted. If, after successive failures, the load step becomes
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
498
Solution Procedures for Nonlinear Systems smaller than a user-specied part of the full loading, (the MINSIZ parameter [ 12.3.2.5 p. 233]) the automatic load controller gives up and issues an error message. On the other hand, if the iterative procedure converges fast with respect to the maximum number of iterations (user-specied parameter MAXITE [ 12.3.5 p. 240]), the step size is increased. Optionally, the maximum load step size can be limited to a user specied part of the full loading (the MAXSIZ parameter [ 12.3.2.5 p. 233]), for example if some intermediate results are needed for output purposes. Adaptive error controlled time increments. If the physical behavior of the model explicitly depends on the rate-of-change of the displacements or an internal variable (for instance dynamics, viscous behavior), then the computed solution at a certain time depends on the time step sizes used. If very small time increments are taken, this inuence is negligible. However, a priori it is not clear how small the step size should be. Moreover, it is often unattractive to use a xed time step size. Clearly, one would like to use small time steps if there are rapid changes in the system, and to increase the step size if the system slowly relaxes to an equilibrium state. To explain the adaptive time step strategy in Diana we consider the initial value problem du = f (u) (30.38) dt Suppose that we use a time stepping method to advance the solution from tn to tn+1 with a step size t = tn+1 tn . By doing so we introduce an error due to the nite time step size t. Let u(t) be the solution to (30.38) with initial condition u(tn ) = un . The time step error is then u(tn+1 ) un+1 = O(tm+1 ) (30.39)
with m the order of consistency of the method. Suppose that we also have a less accurate method, with order of consistency m (with m < m). In Diana a second order RungeKutta method is available, i.e., the SDIRK2 method [ 32.4.5 p. 531], that has an embedded rst order method. So we have a second order accurate solution un+1 , and a rst order solution un+1 . We want the error to be less than a prescribed error tolerance un+1 u(tn+1 ) un+1 un+1
rel
un+1 +
abs
(30.40)
This criterion is both used to reject time steps and estimate the next time step:2 un+1 un+1 2 (30.41) rel un + abs Some extra logic is introduced to avoid unnecessary step rejections, and to keep the step size constant to save LU-factorizations. Moreover, care is taken not to overstep changes in the loading. tn+1 = tn
2 See
1
30.2 Geometric Nonlinearity Automatic time stepping without the SDIRK2 time stepping method is very much like automatic load stepping [ 30.1.5.4 p. 497]. Conversely, if SDIRK2 is used while there is no explicit rate dependence in the model, then a time accurate solution is produced. However, if you are not interested in intermediate results it is better, for eciency reasons, not to use SDIRK2.
499
30.2
Geometric Nonlinearity
By default, Diana assumes that in a nonlinear analysis the model behaves geometrically linear. In this case, the equilibrium equations are based on the undeformed geometry and the strains are linear functions of the nodal displacements. This limits the applicability of the analysis to small displacements, small rotations and small strains. Figure 30.12 shows two examples of structures for which the small displacement and rotation assumptions do no longer hold. F t Fext Fext
(a) translations
(b) rotations
Figure 30.12: Large displacements Figure 30.12a shows a exible beam structure. The displacements will be considered large when the vertical displacement is slightly larger than the beam thickness t. In this particular case, the initial stiness of the structure is zero and the force F can be counter acted only by the development of tensile stresses in the beams that act under a small angle. This geometrically nonlinear phenomenon is known as stress stiening. The structure of Figure 30.12b shows large rotations. If the dimensions of the vertical bar are relatively short, large displacements are not necessarily encountered. Since the degrees of freedom are set up at the start of an analysis, the algorithm assumes for a geometrically linear case that the element contributions to the nodal forces are determined by the undeformed geometry. In this case the direction of the compressive force in the bar would be incorrect. In a geometrically nonlinear analysis, these large rotations are accounted for. It is clear that certain types of loading, e.g. pressures on a wall, are heavily inuenced by large displacements (change of area) as well as by large rotations (change of direction). In Diana these types of loading are known as nonconservative loading.
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
500
Solution Procedures for Nonlinear Systems Large strains should not be calculated as linear functions of the displacements, as this would lead to, for example, large calculated strains in the case of a rigid body rotation or non-unique strain measures. Even without large displacements large strains can show up, for instance in the vicinity of a crack tip, where the local strains can be very large. On the other hand, large displacements are possible without large strains e.g. in thin-walled exible structures. Diana oers two types of geometrically nonlinear analysis: a Total Lagrange and an Updated Lagrange description, see for instance Criseld [15, Ch. 5] or Bathe [5, 6.4]. A Total Lagrange description is useful if rotations and displacements are large and strains are small, and is even obligatory for large strain hyperelastic (rubber-like) material behavior. An Updated Lagrange description can be used advantageously in case of large plastic deformations. The choice of the geometrically nonlinear description determines the stress and strain measures that will be used. These can be other than force over area for stress or displacement over initial length for strain. In a combination of physically and geometrically nonlinear behavior, a stressstrain relation must be dened compatible with the stress and strain measures used in the geometrically nonlinear description.
30.2.1
Denitions
In the following sections, a leading superscript (t ) indicates the state of a quantity, and a leading subscript (t ) indicates the reference coordinate frame for derivatives. A coordinate frame is attached to the material. The material axes can both rotate and stretch during the transition from one coordinate frame to another [Fig. 30.13]. The material point position x and material axis dx are shown for x2 dt x
x
0
d0 x x x1
30.2 Geometric Nonlinearity the coordinate frame at state 0 and state t. The displacement u is denoted by
t
501
u = tx 0x
(30.42)
The deformation gradient F expresses the rotation and stretch of dx. The deformation gradient is tx t (30.43) F = t x0 = 0 0 x with d t x = t F d 0 x. In case of an arbitrary deformation, the rotation matrix R 0 is dened by the polar decomposition of the deformation gradient F=RU=VR (30.44)
Here U and V are symmetrical right and left stretch matrices and R is an orthogonal matrix. The determinant J of the deformation gradient gives a volume change t (30.45) V = J 0 V ; J = det t F 0 The velocity of a point is denoted as u = tx The velocity gradient is (30.46)
tu L=u = t (30.47) x The spin and rate of deformation D are dened in terms of the local velocity gradient 1 D = 2 (L + LT ) and = 1 (L LT ) (30.48) 2 The GreenLagrange strain is dened as
t 0E
1 2
t T t 0F 0F
(30.49)
In small deformation problems, the stress is usually dened as force over area. For geometrically nonlinear analyses, this denition is not unique because the area may change in magnitude and/or direction during deformation. An important stress measure is the Cauchy stress . This stress is dened as the force over area in the deformed conguration n dA = df (30.50)
where df is the force acting on an area dA with a unit normal vector n. Since the Finite Element Method is based on energy principles, the choice for a stress measure determines the strain measure and vice versa. The energy conjugate of the Cauchy stress is the linearized strain. The energy variation can be calculated from W = dV (30.51)
V
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
502
Solution Procedures for Nonlinear Systems In a Total Lagrange analysis, the stress must be related to the undeformed conguration and must be energy conjugated to the GreenLagrange strain. This stress measure is the 2nd PiolaKirchho stress S and is related to the Cauchy stress by t t 1 t t t FT (30.52) t 0 0 S = det 0 F 0 F Diana interpretes all input parameters which indicate a stress, such as a yield stress, as 2nd PiolaKirchho stresses. The Jaumann rate is an objective rate, which means that it transforms properly as a tensor under rigid body motions. The Jaumann derivative of the Cauchy stress is dened as = + (30.53)
Here is the time derivative of the Cauchy stress. The Jaumann stress rate can be considered as the stress rate in the coordinate system that rotates with the material. Stressstrain relations in an Updated Lagrange analysis are dened in this co-rotating coordinate system. The Jaumann stress rate is useful in the case of rate formulated material laws such as plasticity and viscoplasticity.
30.2.2
Total Lagrange
In a Total Lagrange description, strain and stress measures are dened with reference to the undeformed geometry.
30.2.3
Updated Lagrange
The Updated Lagrange analysis, as opposed to the Total Lagrange description, uses an updated reference geometry. Theoretically, many intermediate congurations could serve as a reference frame. However, in Finite Element Analysis it is usual to take the last known equilibrium state, i.e., the situation at the end of the previous step. The incremental equations of motion will be presented here in terms of the Updated Lagrangian formulation. It should be noted that although large displacements, rotations and strain are described correctly, still a constitutive relation appropriate for large strain behavior has to be used. 30.2.3.1 Weak Equation
Using the principles of virtual work, virtual power or using the Galerkin weighted residual method, the following weak form of stress equilibrium can be derived
503
v : dV v dV u
V
W2 = W3 =
S
t v dS + f v dV
with a test function v, Cauchy stress , acceleration u, body force f and surface traction t. If we consider the test function v to be a virtual displacement then W1 is the virtual internal force, W2 is the virtual inertial force, and W3 is the virtual external force. The Jaumann stress rate is related to the deformation rate as =+ = +L:D (30.58)
with fourth order constitutive tensor L and deformation rate D. Using this rate type constitutive model we can derive W1 from
t
W1 =
0
W1 dt
(30.59)
After taking the time derivative of (30.55) and using (30.58) the required ex pression for W1 reads [58] [30]: W1 =
V
D : (L I I) : D + (
v u ) :
dV
v (30.60)
with I the fourth order unit tensor with components Iijkl = ik jl . 30.2.3.2 Finite Element Equations
By using displacement and test functions that are interpolated from discrete values in the nodes of elements, in combination with numerical integration in space and time, the weak form can be transformed to a system of nite element equations that can be solved numerically. Interpolation. The displacement eld is discretized in space by means of interpolation functions n un (30.61) u=
n
504
Solution Procedures for Nonlinear Systems with un the displacement in a node. The velocity, acceleration and test function are interpolated analogously. The deformation rate D is interpolated with D= Bn un Bn = 1 (I + T) : ( n I) (30.62) 2
n
where T is the transpose operator with components Tijkl = jk il , and where v is the velocity. Incremental iterative solution. In the following, the subscript n in un is dropped, which means that u is redened to the vector containing the nodal displacements. The system of governing equations for a transient dynamic problem at time t after space discretization is written as M t u + C t u + t fint = t fext (30.63)
where M is the mass matrix, derived from (30.56); C is the damping matrix; fext is the external force vector, derived from (30.57). Vectors u, u, and u are the resulting acceleration, velocity, and displacement vectors. Vector fint is the internal set of forces opposing the displacements, derived from (30.55). During direct time integration, displacement increments u = t+t u t u are determined at discrete time points in an iterative fashion, with subscript (i) denoting the iteration number, starting with 1. aM + bC + t+t K u(i) u(i1) =
t+t
(30.64)
Internal force vector. The internal force vector t+t fint.(i1) is determined at state t + t, with derivatives to the reference coordinate frame at t + t.
t+t
fint.(i1) =
m
t+t t+t
B : dV m
(30.65)
This means that besides the stress also the strain interpolation B and volume m contributions dV must be updated in each iteration. B is the transposed m matrix of Bn . B = 1 (I m ) : (I + T) (30.66) m 2 The stress update is performed via integration of the Jaumann rate, assuming a constant velocity inside the step: u (30.67) t The stress at t based on the reference frame at t is rst rotated to the reference frame at t + t. Then the stress increment at t + t is determined and superposed. Finally the stress at state t + t in reference frame t + t is rotated to the end of the increment. This reads: u=
t+t t+t
= tR
tR
t t RT + t
t+t
dt
t
t RT
(30.68)
505
t+t
dt = L :
t
t+t D t
(30.69)
t+t
(F U1 )
(30.70)
represents the rotation of the reference coordinate frame at t to the reference coordinate frame at t + t. Stiness matrix. The stiness matrix consists of dierent contributions: K=
n m
(Kmn + KmnS )
(30.71)
B : (L I I) : Bn + I m
m n dV
(30.72)
where Bn comes from (30.62) and B from (30.66). The contribution from a m nonconservative traction load comes from (30.57): KmnS = m t
S
n dS
(30.73)
30.2.4
Nonconservative Loading
In Finite Element Analysis it usually is sucient to calculate the force vector at the start of the analysis, or at most to combine some basic force vectors at the start of an increment. This is an ecient way of analysis if the loading does not depend on the displacements. If the loading (magnitude or direction) does depend on the displacements of the model, the load vector must be set up based on the current displacements. An obvious example of this is water pressure on a containment wall in combination with large displacements. In Diana these types of loading are described as nonconservative. The part of the load that is dened as nonconservative is calculated in every iteration. Equilibrium is now based on the internal and external force vector which both are calculated after the prediction of the displacements. Since the forces now depend on the displacements, there formally is a contribution of the nonconservative loading to the tangential stiness matrix. However, this leads to an asymmetric matrix and Diana disregards it.
30.2.5
Contact Analysis
When performing a contact analysis, Diana automatically invokes the Constrained Minimization Solver, to solve the contact problem. This solver can not
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
506
Solution Procedures for Nonlinear Systems be used in combination with Line Search algorithms [ 30.1.3 p. 487]. Therefore, contact analysis and Line Search do not go together. Contact analysis is only possible with a symmetric tangent stiness matrix. Therefore, contact analysis cannot be performed when the tangent stiness matrix becomes nonsymmetric, for instance due to friction or dilatancy in plasticity analysis.
Chapter 31
31.1
Element Strains
To determine the strains of an element, the nodal results for this element are transformed to the local Cartesian coordinate system of the element and placed in the vector ue . As described in 28.1.2 on page 458, the straindisplacement relation is dened by the matrix B as follows = Bue (31.1)
This relation is valid at any point within the element. But as the Finite Element Method minimizes the error at the integration points, it is obvious that the strains will be determined at these points. So for each integration point the expression (, , ) = B(, , )ue (31.2) will be evaluated at the standard values for , and .
31.1.1
(31.3)
exx = + 2 xx 1 yy 1 zz 3 3 3 eyy = 1 xx + 2 yy 1 zz 3 3 3 ezz = 1 xx 1 yy + 2 zz 3 3 3 The engineering strains are dened as: ij = 2 ij For some calculations the strains are placed general three-dimensional strain situation is xx xy E = xy yy zx yz (31.5) (31.4)
31.1.2
Principal Strains
Diana calculates the principal strains 1,2,3 as the roots of (31.7) ordered such that 1 2 3 . xx xy xz yy yz = 0 (31.7) xy zx yz zz For plane strain and axisymmetric elements, however, the third principal strain 3 is always the out-of-plane strain, while the rst and second principal strains are the inplane principal strains ordered such that 1 2 .
31.1.3
Volumetric Strain
Diana calculates the volumetric strain vol by summation of the principal strains: evol = 1 + 2 + 3 (31.8)
31.1.4
Plastic fault slip in the subsurface can lead to seismic events. The seismic moment M0 is a quantitative measure of the seismic energy released during the rupture of a fault. Aki [1] denes M0 as the product of the amount of relative shear displacement r of the fault, the area of the fault rupture, and the shear modulus G of the adjacent rock formation through which the fault takes place. M0 can be calculated as integral over the slipping fault of the distributed seismic moment PS which Diana can calculate and output via the STRAIN DISSEI option. Such integration can be performed via the function
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
31.1 Element Strains in the iDiana Results environment [Vol. Preand Postprocessing]. The distributed seismic moment is dened as the product of shear stiness of the fault, i.e., the second stiness coecient D22 of the plane interface element, and the relative shear displacement r of the fault:
RESULT CALCULATE INTEGRATE
509
PS = D22 r
(31.9)
For this purpose it is strongly recommended to dene D22 as representative for the shear modulus of the adjacent rock formations according to D22 = E 2 (1 + ) h (31.10)
where E is the Youngs modulus of the adjacent rock, is the Poissons ratio of the adjacent rock, and h is the thickness of the fault for which a minimum value of 1 meter should be applied. See also the DSTIFF input item in Volume Material Library. It is the users responsibility to make sure that the input of the fault shear stiness in the Diana calculations is correct and based on Equation (31.10) for a proper determination of the seismic moment. Diana derives the relative shear displacement r from the traction vector u which, for plane structural interface elements, has three components ux (31.11) u = uy uz where x denotes the normal direction and y and z the rst and second tangential direction [Vol. Element Library]. You can specify the second tangential direction via the XAXIS input item in table GEOMET. The relative shear displacement r is derived as follows. uz u2 + u2 if uz = 0 z y uz (31.12) r= uy if uz = 0 Positive values of r represent normal fault slip, i.e., fault slip whereby the hanging wall moves downwards relative to the footwall. Negative values of r represent reverse fault slip, i.e., fault slip whereby the hanging wall moves upwards relative to the footwall. The sign convention for the distributed seismic moment is equal: positive values denote normal fault slip, whereas negative values denote reverse fault slip. When both types of fault slip occur simultaneously on the same fault plane, a distinction has to be made between negative and positive values during integration in order to avoid an underestimation of the seismic moment. Therefore, the result items PS and PS+ for respective negative and positive values of PS are output. The command RESULTS CALCULATE INTEGRATE SURFACE in the iDiana
Diana-9.4 Users Manual Analysis Procedures (X)
(h 1 m)
510
Element Results and Nodal Forces Results environment integrates then solely over positive | negative values, resulting in a positive | negative seismic moment representative for normal | reverse fault slip.
31.2
Element Stresses
= D ( 0 ) + 0 (31.13)
Here 0 and 0 are input entities, whereas is determined according to (31.1). Above expression is used to determine the stresses in the integration points only. Therefore the same accuracy argument holds as for the strains. For some calculations the stresses are placed in a stress matrix S which for the general three-dimensional stress situation is given by xx xy xz S = xy yy yz (31.14) zx yz zz
31.2.1
(31.16)
szz =
1 xx 3
1 3 yy
2 3 zz
31.2.2
Principal Stresses
Diana calculates the principal stresses 1,2,3 as the roots of (31.17) ordered such that 1 2 3 . xx xy xz yy yz = 0 (31.17) xy zx yz zz For plane strain and axisymmetric elements, however, the third principal stress 3 is always the out-of-plane stress, while the rst and second principal stresses are the inplane principal stresses ordered such that 1 2 .
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
511
31.2.3
Principal Shears
Diana calculates the principal shears q1,2,3 from the principal stresses 1,2,3 according to q1 = q2 = q3 =
1 2 1 2 1 2
(1 3 ) (1 2 ) (2 3 ) (31.18)
A maximum shear qmax is calculated from the principal shears according to q1 if 1 = 0 and 3 = 0 qmax = q3 if 1 = 0 (31.19) q2 if 3 = 0
31.2.4
Plastic Yield
Diana calculates the plastic yield Yp from the deviatoric and shear stresses according to Yp =
2 2 2 s2 + s2 + s2 + 2 xy + yz + zx 2y xx yy zz 3
(31.20)
31.2.5
Pressure
Diana can calculate and output a pressure p from three stress components according to 1 p = 3 (xx + yy + zz ) (31.21)
31.2.6
Stress Invariants
1 p = 3 (1 + 2 + 3 )
q= =
1 3
1 2
(1 2 )2 + (2 3 )2 + (3 1 )2 (1 + p )(2 + p )(3 + p ) q3
(31.22)
arcsin 27 2
In this form, p is the isotropic stress or mean stress, q is the equivalent shear stress, and is referred to as Lodes angle. The equivalent shear stress is a composition of deviatoric stress components. Under triaxial stress conditions where 2 = 3 , q is just the principal stress dierence: |1 3 | .
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
512
31.2.7
The civil engineer often is interested in the amount and direction of reinforcement needed in concrete structures (e.g. plates), to prevent failure of the structure. Diana allows for a relatively easy determination of the forces and moments which are to be supported by the reinforcement. The engineer may use this data in designing the reinforcement. The theoretical considerations are completely based on research by Merks [41]. Although the behavior of reinforced concrete is essentially nonlinear, a fair approximation is gained by calculating reinforcement moments and forces from a linear elastic analysis. Doing so, the following assumptions are used:
Only the reinforcement steel supports the tension forces. The concrete only supports compressive forces. Stresses from a linear elastic calculation can be used (thus, redistribution of stresses due to concrete damage, etc. is neglected). Reinforcement in two directions (not necessarily perpendicular). Straight reinforcement. As few reinforcement as possible.
By means of these assumptions, the moments and/or forces which need to be supported by the reinforcement can be determined. In this derivation of reinforcement moments/forces, several approximations are used. 31.2.7.1 Transformation Rules for Stresses, Forces and Moments
Stresses, etc. are rstly determined with respect to the element axes. For determination of stresses, forces and moments acting on the reinforcement, the component of these stress quantities are needed in the frame of the reinforcement however. For this purpose, transformation rules are needed. These rules are presented in this section. Consider the Cauchy stresses following from a linear elastic analysis: xx , yy , yx = yx (31.23)
In the following, we ever use the fact that stress components are symmetric: yx = xy . The stresses can be stored in a matrix: = xx xy xy yy (31.24)
This matrix contains the components of the Cauchy stress tensor with respect to the element x and y axes. Stress components in the frame dictated by the reinforcement, x and y , are required however [Fig. 31.1]. Note that is the angle between x and x and is the angle between x and y . The outer product
513
Figure 31.1: Reinforcement directions of x and y always coincides with the element z-axis. The stress components in the reinforcement frame can be shown to be: = 1 R R RT RT sin 2 (31.25)
where R and R are matrices dened by R = cos sin sin cos R = sin 0 cos 1 (31.26)
It can be shown that the same transformation rule holds for moments (mxx , myy , mxy ): 1 M = R R MRT RT (31.27) sin 2 where M is a matrix containing the moments mxx , myy and mxy = mxy , and that it also holds for forces (nxx , nyy , nxy ): N = 1 R R NRT RT sin 2 (31.28)
where N is a matrix containing the forces nxx , nyy and nxy = nyx . Denitions for primary forces and moments for elements are given in Volume Element Library. The quantities M and N represent moments and forces in the reinforcement frame per unit length perpendicular to the reinforcement axes, e.g. nxx is a force in reinforcement direction x per unit length perpendicular to x . 31.2.7.2 Reinforcement Moments Only
If membrane behavior can be neglected, the reinforcement only needs to support moments. These moments follow from the condition that as few reinforcement as possible is to be used. Following this condition, the moments with respect to the x y frame read m1 = mxx mxy m2 = myy mxy (31.29)
These formulas represent reinforcement moments per unit distance perpendicular to the reinforcement axes x and y [Fig. 31.2]. And thus, the following
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
514
m1 y m2
Figure 31.2: Reinforcement moments reinforcement moments need to be supported by the reinforcement in the upper and lower plane of the plate. m1,up = mxx + mxy m1,lo = mxx mxy m2,up = myy + mxy m2,lo = myy mxy (31.30) (31.31)
In m1,lo the character m stands for moment, the superscript indicates a reinforcement variable, subscript 1 states that the moment is in x direction (per unit length x ) and subscript lo states that it is the reinforcement moment for the lower plane. The upper and lower planes are dened in positive and negative element z-direction respectively. Additionally, the transverse shear forces qx and qy are used to dimension transverse shear reinforcement by means of the quantity q =
2 2 qx + qy
(31.32)
The latter ve stress quantities are used to dimension reinforcement in plates. Positive moments m1,up or m2,up in the upper plane require reinforcement respectively in the x and y direction. In the lower plane however, due to the denition of moments, negative moments m1,lo or m2,lo require reinforcement respectively in the x and y direction. 31.2.7.3 Reinforcement Forces Only
If membrane forces are present, it is no longer possible to express the reinforcement loading by moments. Reinforcement forces, expressing the forces which must be supported by the reinforcement, are calculated instead. These reinforcement forces per unit length perpendicular to the reinforcement directions x and y are given by n1,up = n1,lo = nxy nxx + 2 2 nxy nxx + 2 2 n2,up = n2,lo = nxy nyy + 2 2 nyy nxy + 2 2 (31.33) (31.34)
These formulas represent forces in reinforcement directions per unit distance perpendicular to the reinforcement directions [Fig. 31.3]. Note that the forces in
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
515
Figure 31.3: Forces in reinforcement directions upper and lower plane are the same, in the absence of bending moments. The latter four stress quantities are used to dimension reinforcement in membranes. Reinforcement is required for positive forces in the corresponding direction; e.g. a positive force n1,up denotes a tension force and hence reinforcement in x direction is necessary. 31.2.7.4 Combined Moments and Forces
If both membrane forces and bending moments are present, the reinforcement loading is expressed by forces equivalent with the combined results of these membrane forces and bending moments. These combined forces can also be determined if only bending moments are present (plate elements). This section outlines that the required reinforcement is not just the addition of reinforcement needed for separate membrane and bending behavior. As the membrane and torsional shear stresses may act in opposite direction, these stresses should be taken into account with care [Fig. 31.4].
z
upper
z
upper
y x xy xy mxy xy
lower
y x xy
lower
mxy
(a) membrane
(b) torsional
Figure 31.4: Shear stresses The combined reinforcement forces read, in x -direction
c nxx,up =
m nxx + xx 2 zd nxx m xx 2 zd
c nxy,up =
(31.35) (31.36)
c nxx,lo =
c nxy,lo =
c nxy,up =
(31.37) (31.38)
c nyy,lo =
c nxy,lo =
where zd denotes the absolute internal beam arm for the moment. This beam arm follows from zd = (ht co) zr (31.39) with ht the local thickness, co the coverage (distance from reinforcement center to outer plane) and zr the relative internal beam arm. The relative internal beam arm zr in the case of mainly bending is about 80 to 90 percent of the eective height (ht co). If normal forces are present then zr can be much less. In general, zt depends on the ratio of bending moments and membrane forces. Several examples for determination of zt are given further. Finally, the reinforcement forces per unit length perpendicular to the reinforcement directions x and y are given by
c c c n1,up = nxx,up + nxy,up c c c n1,lo = nxx,lo + nxy,lo c c c n2,up = nyy,up + nxy,up c c c n2,lo = nyy,lo + nxy,lo
(31.40) (31.41)
Additionally, the transverse shear forces qx and qy are used to dimension transverse shear reinforcement by means of the quantity: q =
2 2 qx + qy
(31.42)
The latter ve stress quantities are used to dimension reinforcement for combined bendingmembrane behavior. Reinforcement is required for positive forces c in the corresponding direction, e.g. a positive force n1,up denotes a tension force and hence upper plane reinforcement in x direction is necessary. 31.2.7.5 Examples for the Internal Beam Arm
The absolute internal beam arm zd follows from user-specied values for coverage co and relative internal beam arm zr . The value of zr to be specied strongly depends on the ratio of moment and normal force. This is illustrated in the following examples. Bending only na = nb and na zd = m [Fig. 31.5].
Bending plus compressive force In the case of a compressive normal force, an additional eccentricity should be taken into account1 [Fig. 31.6]: nb na = ne and na zd = m.
1
See for instance Dutch regulations VB74-84, art. E304.3.3, NEN 3880.
Diana-9.4 Users Manual Analysis Procedures (X)
517
m ht co zd
nb
na
Figure 31.6: Internal beam arm for bending plus compressive force Bending plus large tension force [Fig. 31.7].
m ne ht co zd na
1 nb + na = ne and (na 2 ne ) zd = m
nb
Figure 31.7: Internal beam arm for bending plus large tension force
31.2.7.6
Numerical Example
Consider the following stress situation with bending moments mxx = 500 kNm/m and with membrane forces nxx = 250 N/mm nyy = 50 N/mm nxy = 20 N/mm myy = 100 kNm/m mxy = 100 kNm/m
with plate thickness t = 550 mm, coverage co = 50 mm and relative beam arm zr = 0.8 . Suppose these stresses need to be supported by orthogonal reinforced concrete (with the rst reinforcement direction coinciding with x). If there are only bending moments, the required reinforcement should be determined from: m1,up = 500 + 100 = 600 kN m1,lo = 500 100 = 400 kN
Diana-9.4 Users Manual Analysis Procedures (X)
518
Element Results and Nodal Forces If there are only membrane forces, the required reinforcement should be determined from n1,up = n1,lo = 250 20 + = 115 N/mm 2 2 250 20 = 115 N/mm + 2 2 n2,up = n2,lo = 20 50 + = 15 N/mm 2 2 50 20 = 15 N/mm + 2 2
However, as both bending moments and membrane forces are present, the reinforcement must support the combined action. Thus, it must support the following combined forces (with relative distance zd = (550 50) 0.8 = 400 mm):
c nxx,up = c nxy,up = c nyy,up =
50 100000 = 275 N/mm 2 400 Hence, the following (extreme) values for the reinforcement forces are obtained, in the upper plane:
c n1,up = 1125 + 260 = 1385 N/mm c n2,up = 225 + 260 = 485 N/mm
Note that the combined action leads to dierent results than simply adding membrane and bending action. It can be seen that only the upper plane needs reinforcement.
31.2.8
For reinforcement bars Diana can calculate the gradient of the axial stress xx . This is the derivative of the stress with respect to the local x axis of the bar: xx =
December 15, 2009 First ed.
dxx dx
(31.43)
519
31.2.9
Diana calculates the shear-stress capacity and the hydrostatic pressure capacity of stress against MohrCoulomb failure criterion, which are dened as: = and = with q= and qmc = 1 ((1 2 )2 + (2 3 )2 + (3 1 )2 ), 2 6 sin 6c cos p+ . 3 sin 3 sin (31.46) q , qmc (31.44)
(31.45)
(31.47)
Where p is the hydrostatic pressure, and c and are the cohesion and the friction angle for the MohrCoulomb material, respectively.
31.3
31.3.1
Nodal Forces
Internal Nodal Forces
The internal nodal forces for an element can be dened as the forces (moments) the element exhibits at its environment through its nodes, which are caused by the internal element stresses only.2 As these forces actually are related to the elastic deformation we write fe,in = Ke uno e which can be rewritten as fe,in =
Ve
BT C0 dV +
Ve Ve
BT 0 dV
(31.48)
B dV
(31.49)
This integral can be solved numerically, as the terms in the integrand are all known at the integration points. Summing the internal nodal forces for all elements yields the vector of internal nodal forces fin . For nodes that are both unconstrained and non-loaded these are zero or very small. With internal nodal element forces Diana oers the possibility to perform the summation only for a selection of elements. With such a selection of elements you dene a section of the model as an assembly of elements. In that case the
2 This denition has been made for pragmatic reasons only and has not much meaning in linear elastic analysis.
520
Element Results and Nodal Forces internal nodal element forces on the edges of this section will in general be nonzero. These internal nodal forces might be interpreted as the internal forces that act on that section. For internal nodal element forces Diana also oers the possibility to perform the summation only for the embedded reinforcements or only for the mother elements containing the embedded reinforcements.
31.3.2
The external nodal loads are dened as the vector of applied loads, as the nodal point loads, pressure loads and gravity eects. For one element, the element loads are expressed by the vector fe,ex as
el fe,ex =
Ve
NpVe dV
Ae
NpAe dA
(31.50)
All terms of this equation can be determined by numerical integration per element. Summing the external loads over all elements gives the total external load el pt vector fex . The vector fex contains the nodal point loads, so the total external nodal loads can be dened as
el pt fex = fex + fex
(31.51)
31.3.3
Comparison with Equation (31.50) learns that fin fex equals the right hand part of this expression. Now a vector fr is dened as fr = fex fin (31.52)
where fr contains the so-called residual forces. It can be seen that, for unconstrained freedoms which have no external loads, the equilibrium condition for static linear elastic problems requires that the internal nodal loads summed over all connected elements should be zero, as has been argued in Chapter 28. However, where the constrained degrees of freedom are not included in Equation (31.50), they are in fr . For these constrained degrees of freedom, it can be seen that the residual load vector contains non-zero values. These are the reaction forces at these constraints. It can be concluded from the above description of fr that the residual forces at the unconstrained nodes should be zero or small compared to the residual forces or reaction forces on the constrained degrees of freedom. If not, something is wrong with the condition of the system of equations.
Chapter 32
Dynamic Analysis
The governing equation of motion for a linear dynamic nite element system reads M + Cu + Ku = f (t) u (32.1) Where M is the mass matrix, C the damping matrix and K the stiness matrix of the nite element model, f (t) is the right-hand-side vector of forcing functions and u, u and u are the resulting acceleration, velocity and displacement vectors. The complete solution u consists of two contributions u = utr + ust (32.2)
The response utr depends on the conditions with which the response is initiated and is induced by the initial conditions
0
u = u(t0 )
and
u = u(t0 )
(32.3)
By contrast ust is the particular solution which satises the forcing function of f (t) of (32.1), but does not take account of the initial conditions. This particular solution ust is denoted as the steady-state response, whereas the additional solution utr is called the initial transient part of the response. In case of static loading, f (t) = 0, the transient response is completely determined by the initial conditions. Frequency response analysis only studies the steady-state response [ 32.2]. If the initial transient part is substantially important then you must analyze the complete transient response [ 32.4]. If nonlinear phenomena, like plasticity or cracking, are present then a direct time integration with Module nonlin must be used as described in Part IV.
32.1
General Phenomena
All types of structural dynamic analysis have to do with general phenomena like mass and damping, loading and base excitation as outlined in this section.
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
522
Dynamic Analysis
32.1.1
Mass
Either consistent or lumped mass matrices can be used in structural dynamic analysis. In practice, lumped or diagonal mass matrices are often employed due to their general economy and because they lead to some attractive time integration schemes like explicit methods. Diana uses an adaptive lumping technique to lump mass matrices. This technique was developed by Hinton et al. [28]. The (diagonal) terms of the lumped mass matrix are scaled proportionally to the diagonal entries of the consistent mass in such a way that the total element mass is conserved. Caution: Mass lumping may lead to inaccurate results in case of coarse meshes and/or irregular element shapes, for instance two-dimensional elements with very nonuniform thickness.
32.1.2
1
Damping
Either consistent or lumped damping matrices can be used in structural dynamic analysis. In practice the presence of damping reduces the steady-state response and damps out the transient response. A modal analysis [ 32.2.1] assumes the application of proportional viscous damping and that the damping matrix C satises the orthogonality condition. Modal damping can be employed for this and the magnitude of the damping has to be specied as a percentage of the critical damping factor T Cj = 2i i ij i (32.4)
where is the natural angular frequency and i is the damping ratio. The critical damping factor is ccrit = 2 km (32.5) Where k is the generalized stiness T Kj ij and m is the generalized mass i T Mj ij . i 32.1.2.1 Continuous Damping
When a form of continuous damping, like viscous dashpots, is included in the nite element model then the direct solution method [ 32.2.2 p. 525], or the direct time integration procedure [ 32.4 p. 528] must be used. Discrete places of damping are obtained by specifying properties for spring elements [ 6.2 p. 118]. 32.1.2.2 Rayleigh Damping
Applying a nonmodal solution technique, it is necessary to evaluate the damping matrix C explicitly and usually viscous damping eects can be included by
comprehensive discussion on damping in Finite Element Analysis is given by Spence & Kenchington [54].
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
1A
32.2 Frequency Response assumption of Rayleigh damping which is of the form C = aM + bK (32.6)
523
where a and b are constants to be determined from given damping ratios. 32.1.2.3 Structural Damping
Another type of damping, frequently employed in dynamic analysis, is structural damping also called hysteretic damping. This type of damping is proportional to displacement but in-phase to velocity of a harmonically oscillating system. In that way the equation of motion is expressed as M + K (1 + i) u = f (t) u (32.7)
where is the structural damping factor. Equations like (32.7) can only be solved in the frequency domain.
32.1.3
Loading
In a frequency response analysis with Diana the forcing function can be a (deterministic) harmonic one. That is, the forcing function is composed of sines and cosines with known amplitudes. Often, other types of loading, like short duration impulsive loadings or general forms of long term loads, may be present. For those other types a direct time integration, i.e., a transient response analysis, should be applied [ 32.4].
32.1.4
Base Excitation
The dynamic response like displacements, stresses of a nite element system may not only be induced by prescribed loadings, but also by motions of its supported points. By expressing the response relative to a xed ground point the equation of motion of (32.1) may be written in the form M + Cu + Ku = fef (t) = M su (t) u u (32.8)
where u is now the relative response and the external loading f (t) has been replaced by an eective loading due to the base excitation fef (t) in which usu represents the applied base acceleration vector.
32.2
2
Frequency Response
Frequency response analysis predicts the linear response of a system subjected to a continuous series of harmonic excitations. When the loading is a deterministic frequency content, a steady-state linear dynamic analysis may be performed.
more background theory on frequency response analysis see for instance the books by Craig [14] and by Clough & Penzien [13].
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
2 For
524
Dynamic Analysis Steady-state response is given as a frequency sweep through a specied range of frequencies. The analysis results are given terms of amplitudes and phase angles [ 32.2.3 p. 525]. If the forcing function of (32.1) is periodic, the system can be transformed in terms of frequency components f and solved in the frequency domain. In this case, the steady-state solution can be obtained in a direct way and the initial conditions are irrelevant. The steady-state response is determined for the same frequencies that dened the loading input. M (t) + Cu(t) + Ku(t) = f eit u u(t) = ue
it
(32.9) (32.10)
The response for each excitation frequency f can be determined from (32.9) where f is the complex load vector. The real part f is the in-phase component of the load and the imaginary part f is the out-of-phase component. The corresponding solution can now be written in the same form (32.10) with u as complex displacement vector with in- and out-of-phase components. Substitution of the solution (32.10) in the equation of motion (32.9) results in K 2 M + iC u = f u = K 2 M + iC
1
(32.11) (32.12)
f = H () f
This substitution may be repeated for each frequency component. The term H (), the response function, is the Fourier transformation of the response matrix. There are two methods to solve (32.11): an indirect method based on mode superposition [ 32.2.1], and a direct method [ 32.2.2].
32.2.1
Mode Superposition
The principle of the mode superposition technique may be employed providing the system of equations of (32.1) is linear, that is to say, M, C and K remain constant during the response. In a modal analysis, the response will be found by superposition of the response in each mode:
Np
u() =
i=1
i i ()
(32.13)
where i is the i-th eigenvector of the undamped eigenproblem [ 29.2 p. 468] and i () is the i-th generalized modal displacement. Substitution of this equation in (32.11) and pre-multiplication with the jth eigenvector yields an uncoupled system of equations, if the damping matrix C is orthogonal with respect to the eigenvectors. In this case the inversion of a diagonal matrix will do and the response may be determined from
Np
u = H()f =
i=1
December 15, 2009 First ed.
T f i
(32.14)
32.2 Frequency Response If the damping is not proportional but continuous, for instance when the model comprises discrete dampers like dashpots, the damping matrix C is not orthogonal and its diagonal structure is destroyed. In this case the damping is usually rather strong and, considering that the transformation of (32.14) is no longer correct, a direct solution method is more suitable.
525
32.2.2
Direct Solution
The direct solution method requires the solution of a complex system of equations. Complex arithmetic can be avoided by transformation of (32.11) to a system with real coecients: K 2 M C +C K 2 M u u = f f (32.15)
where is the excitation frequency, f and f are the imaginary and real parts of the complex load vector f , and u and u are the imaginary and real parts of the complex displacement vector u. The system (32.15) is regular but not always positive denite and can only be solved with an LDU-decomposition [ 29.1 p. 466]. This method has the advantage that it does not require the eigenfrequencies and eigenmodes to be determined. Unfortunately, there are also two disadvantages. In the rst place, a new system of equations must be solved for every excitation frequency which requires a lot of computing time, particularly if the number of excitations increases. Secondly, the system to be solved tends to a bad condition if an excitation exactly or nearly coincides with one of the (complex) eigenfrequencies. The disadvantages may be overcome by determination of the response function H() via the mode superposition technique.
32.2.3
Analysis Results
To obtain the response of a system subjected to a series of harmonic excitations only the steady-state response is of interest, it is calculated by Modules modal and freque. These modules present the resulting output in the frequency domain. Depending on the presence of damping and the type of the excitation function the response will not be in-phase or 180 out-of-phase with the excitation. The real part of a typical nodal degree of freedom direction i is given by ui (t) = ui cos (t i ) (32.16)
where ui is the amplitude and i is the phase angle of the response relative to the excitation. Diana represents all output results by their amplitude and corresponding phase angle with respect to the cosine function! In addition, dynamic stresses are derived directly from the calculated displacement response vector u. The steady-state solution of (32.1) may be assumed to have the form u(t) = ueit (32.17)
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
526
Dynamic Analysis where u = u + i u is the complex spatial displacement vector. The element stresses have the same form e (t) = e eit (32.18)
where = + i is the complex stress vector. This complex stress vector e is now determined via the usual relation e = DBue (32.19)
where D and B respectively represent the stressstrain and straindisplacement relation, and ue is the complex element displacement response vector. In the same way as the nodal response, only the real part of the element stress components in Diana are represented by their amplitude i and phase angle i with respect to the cosine function i (t) = i cos (t i ) (32.20)
32.3
Spectral Response
Spectral response analysis predicts the linear response of a system subjected to a base excitation spectrum. Dianas Module spectr may perform a spectral response analysis on a nite element model. This section summarizes the background theory of Module spectr.3 When designing a structure we are particularly interested in the maximum forces, which can be evaluated from the maximum relative displacements. The results of a spectral response analysis are given in terms of individual and combined modal forces. The implementation of spectr assumes that the contribution of damping is negligible and that the base acceleration spectrum has a working direction i, i.e., a base acceleration value usu . The system of equations for base excitation (32.8) may then be written as M (t) + Ku(t) = M i usu (t) u (32.21)
where i is the contribution of the degrees of freedom in the working direction of the excitation. The forces acting on the structure are the internal set of forces fk = Ku(t) and the pseudo-inertia forces fm = M u(t) + i usu (t) = Ma(t) Equilibrium of the forces gives fm = fk which results in a = 2 u
3 The
(32.22)
(32.23)
(32.24)
implementation of Module spectr is based on the Response Spectrum Method as described by Gupta [22].
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
32.3 Spectral Response The absolute maximum value of the pseudo-acceleration a is called the spectral acceleration SA SA () = max | a(t) | = 2 SD () (32.25) where the spectral displacement SD is SD () = max | u(t) | (32.26)
527
Spectral response analysis is based upon the mode superposition principle. Substitution of the mode superposition equation (32.13) in (32.21), premultiplication by T , and modal vectors scaled such that i T M i = 1 i gives and
2 T K i = i i
(32.27) (32.28)
2 i + i i = i usu
where is the generalized modal displacement and i is called the modal participation factor, which is equal to i = T M i i (32.29)
where i is the contribution of the degrees of freedom in the working direction of the excitation. Thus, the maximum displacement vector in the ith mode can be written as ui.max = i i SDi (32.30) Therefore, the maximum force vector for the ith mode can be calculated by fi.max = Kui.max = K i i SDi = M i i SAi (32.31)
It is obvious that the upper bound of the combined forces is given by the absolute sum of the modal forces
N
fabs =
i=1
fi.max
(32.32)
In general, it is unlikely that the maximum values of the modal forces would occur at the same time. Goodman, Rosenblueth & Newmark [21] showed that the probable maximum force or moment, in case the eigenfrequencies are not closely spaced, is approximately equal to the square root of the sum of the squares of modal values, the SRSS rule:
N
fs =
i=1
2 fi.max
(32.33)
A method to take the correlation among the modes into account is the Complete Quadratic Combination, the CQCrule:
N N
fs =
i=1 j=1
(32.34)
528
Dynamic Analysis In this approach the correlation among the modes is addressed explicitly by introducing correlation coecients ij , which vary between zero and unity. The equation for the correlation coecient due to Der Kiureghian is: ij = where rij = 8 i j (rij i + j )rij 2 2 )2 + 4 r (1 + r 2 ) + 4( 2 + 2 )r 2 (1 rij i j ij ij i j ij i j
3
(32.35)
(32.36)
and i , j are the damping ratios for mode i and modej, respectively. This equation implies that ij = ji and ij = 1 for i = j or for two modes with equal eigenfrequencies and equal damping ratios.
32.4
Transient Response
This section is an introduction to the background theory of dynamic analysis with time integration as performed by Module nonlin [Ch. 7 p. 127].4 Using Module nonlin, such analysis may be combined with physical or geometrical nonlinearities, see part IV. The analysis results are presented in the time domain. The system of governing equations for a transient dynamic problem at time t are generally written as M (t) + Cu(t) + fint (u, u, , , t, . . . ) = fext (t) u (32.37)
where M is the mass matrix, C the damping matrix and fext the external force vector or right-hand-side vector of forcing functions. Further, u, u and u are the resulting acceleration, velocity and displacement vectors, and are the strain and stress elds. Vector fint is the internal set of forces opposing the displacements. For linear situations fint = Ku(t) and (32.37) can be written as M (t) + Cu(t) + Ku(t) = fext (t) u (32.37a)
For geometrical or physical nonlinear analysis or both, fint must be calculated for the actual stress distribution satisfying all nonlinear conditions. fint = BT (32.38)
For the transient response of a nonlinear analysis, the solution of the second order dierential equation (32.37) is obtained by direct time integration techniques. The solution for the dynamics problem will be determined at a number
4 Elaborate
e 9], Zienkiewicz [63, 9], and Gradin & Rixen [20, Ch. 7].
discussions on dynamics can be found in Bathe [5, Ch. 9], Craig [14], Hughes [31,
32.4 Transient Response of discrete time points: t0 , t1 , t2 , . . . , t t, t, t + t, . . . , T . We assume that we have the solution at time t so that (32.37) holds M t u + C t u + t fint (u, u, , , t, . . . ) = t fext If the time integration procedure being used requires M t+t u + C t+t u +
t+t
529
(32.39)
fint (u, u, , , t, . . . ) =
t+t
fext
(32.40)
to obtain t+t u, then the time integration scheme will be called implicit. If, on the other hand, (32.40) is not required to obtain t+t u, the scheme will be called explicit. These methods use expressions which approximate the acceleration u and the velocity u in terms of incremental displacement components u = t+t u t u.
32.4.1
Newmark
A widely used integration scheme is the Newmark method, which consists of the following equations.
t+t t+t
u = t u + (1 ) t u + t+t u t u = t u + t u t +
t+t 1 2
(32.41) (32.42)
u + t+t u t2
To nd
t+t
u and
t+t
u=
(32.43)
u.
32.4.2
Euler Backward
u t
t+t
u= u=
(32.44) u tu t (32.45)
t+t
530
Dynamic Analysis
32.4.3
HilberHughesTaylor
The HilberHughesTaylor method [27] (also called -method) is an extension to the Newmark method. With the HilberHughesTaylor method it is possible to introduce numerical dissipation without degrading the order of accuracy. The HilberHughesTaylor method uses the same nite dierence formulas (32.41) and (32.42) as the Newmark method with xed and ( = 1 (1 2), = 2 1 2 4 (1 ) ). The time-discrete equation of motion is modied as follows: M t+t u + (1 + )C t+t u C t u+ (1 + ) t+t fint t fint =
t+(1+)t
fext
(32.46)
For = 0 the method reduces to the Newmark method. For 1 0, 3 = 1 (1 2), and = 1 (1 )2 the scheme is second order accurate and 2 4 unconditionally stable. Decreasing means increasing the numerical damping. This damping is low for low-frequency modes and high for the high-frequency modes.
32.4.4
Wilson-
The Wilson- method is basically an extension of the Newmark scheme with = 1 and = 1 (for which the Newmark method is conditionally stable) [5]. 2 6 In the Wilson- method the acceleration is assumed to vary linearly in time from t to t + t with 1 [Fig. 32.1].5 With this assumption we can derive u
t +t t +t t
u t + t
u t + t
u t
u and
t +t
u. 6 t
t
u= u=
6 2 t 2 3 t
t +t
u
t
u2t u
t
(32.47) (32.48)
t +t
t +t
u 2t u
t 2
5 Diana determines the state variables at the user-specied t + t which coincides with t + t for the purpose of postprocessing and for combination with other time integration methods.
531
M t +t u +t +t C t +t u +
t +t
fint =
t +t
fext
(32.49)
In this equation the external load vector on the right-hand side is just like the acceleration assumed to vary linearly in the time interval t t + t :
t +t
fext =
fext +
t +t
fext t fext
(32.50)
The Wilson- scheme is unconditional stable for 1.37. Like the Hilber HughesTaylor method, the Wilson- scheme has a low damping for low-frequency modes and a high damping for the high-frequency (spurious) modes. However, for all frequencies the damping is usually higher than for the HilberHughes Taylor method. Furthermore, the Wilson- scheme show a comparatively large initial overshoot as a response to a ramp load.
32.4.5
RungeKutta
RungeKutta methods are a large class of methods for solving initial value problems of the form dy = f (y, t) (32.51) dt By using the velocity v = u as an extra variable, the second order dierential equation (32.37) is split into a system of two rst order dierential equations u=v v = M1 fext (t) Cv fint (u, v, , , t, . . .) (32.52)
This system is of the same form as (32.51), so the RungeKutta method can be applied. A single step of an s-stage RungeKutta method is described by6
s ti
y = y + t
j=1 s
aij f
tj
y, tj (32.53)
t+t
y = y + t
j=1
bj f
tj
y, tj
with ti = t + ci t (32.54) The coecients aij , bj , and ci , that characterize the method, are conveniently represented in a Butcher tableau: ci aij bj
6 See
532
Dynamic Analysis If aij = 0 for j i the method is explicit, for example the classic 4th order method, which means that only function evaluations of f are needed, but that no systems of equations have to be solved. Unfortunately, explicit methods cannot be used for Equation (32.37) because of stability problems. In Diana a two-stage Single Diagonal Implicit RungeKutta method7 is implemented. This method is dened by the Butcher tableau 0 1 1 1 with = 1 1 . This method is unconditionally stable, second order accurate, 2 and it eciently damps high-frequency error modes. Moreover, the intermediate stag solution oers the possibility to estimate the time step error e :8 e=
t+t
y ty
t1
y ty
(32.55)
with = 1 5 2. The error estimate e can be used to control the time step 4 size [ 30.1.5.4 p. 497].
32.5
This section presents a brief overview of the background theory of the analysis of coupled uid and structural systems, the so-called uidstructure interaction analysis.9 Eects of large scale ow in the uid are excluded. Attention is paid to the discretization method, the numerical solution techniques and simplications. Figure 32.2 shows a general uidstructure interaction geometry. The solid extends throughout the region S , and F contains the uid. Surface I denes the uidstructure interface. The boundary of the uid B may be separated in a xed or prescribed part p , a part representing the bottom b , a part representing the free surface s and a part representing the innite extent e .
32.5.1
Solid
In the solid, the discretization in the familiar form is given below and will be assumed throughout:
ex MS u + CS u + KS u + fI = fS (t)
7 Also 8 See
(32.56)
known as SDIRK2. Hairer and Warner [23, IV.8], or Ellsiepen and Hartmann [16]. 9 For a more detailed description of the underlying theory see for instance Zienkiewicz & Bettes [61, 7] and Olson & Bathe [47].
533
p or b
Figure 32.2: Fluidstructure interaction where MS , CS and KS are mass, damping and stiness matrices respectively and u is a set of unknowns describing the displacements of the structure. The ex vector fI stands for forces due to the interface interaction with the uid and fS represents the external force contributions.
32.5.2
Fluid
The uid is characterized by a single pressure (or potential) variable p and the coupling with the structure is achieved by consideration of interface forces and a standard nite element idealization. Assuming the state of the uid is linear, the governing equation is the wave or acoustic equation
2
p=
1 p c2
(32.57)
where p is the pressure (compression positive) and c the wave speed, given by c2 = (32.58)
where is the bulk modulus and the density. Appropriate boundary conditions of the following form can be imposed. 32.5.2.1 Solid Boundary
The conditions applying to the surface I being the interface between the uid and structure, can be written as p = F nT uF F n and nS = pF nF on I (32.59)
where nF and nS are respectively the outward normal to the uid domain and the outward normal to the structural domain. The coupling between the uid domain and the structural domain is realized by continuity between the normal
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
534
Dynamic Analysis displacements with the condition uF = uS and is obtained by combining this condition with (32.59) p = F nT uS (32.60) F n 32.5.2.2 Prescribed Conditions p=p on p (32.61)
where p is a prescribed pressure often to be zero along part of the boundary p . 32.5.2.3 Free Surface p = F g u z on s (32.62)
where g is the gravity acceleration and z is directed normal to the free surface. Noting that 1 p uz = (32.63) F z Equation (32.62) can be written as p 1 = p z g on s (32.64)
which is the linearized free surface condition for rst order waves. 32.5.2.4 Radiation for Boundary of Innite Extent
If a boundary of innite extent has been placed suciently far away, it may be assumed that only plane waves exists. In the existence of only outgoing waves, incoming waves are supposed to be absent, giving a solution of the form p = f (x c t) (32.65)
where a positive x is the outward direction. The radiation boundary condition is now obtained by eliminating f and is given by p 1 = p x cs on e (32.66)
This condition is denoted as the Sommerfeld radiation condition and in general, will be applied in a plane normal to the direction of the wave speed. Due to the fact cs can correspond to several possible wave velocities, this relation of (32.66) at the radiation boundary should be applied then using a frequency () dependent velocity dened by cs = where h is the uid depth.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
g h tanh cs
(32.67)
535
The conditions applying to the surface b being the bottom of the uid reservoir, can be written as 1 B p = p on b (32.68) n c(1 + B ) where c is the wave speed given by (32.58) and B is the wave reection coecient of the bottom. The wave reection coecient B is the ratio of the amplitude of the reected hydrodynamic pressure wave to the amplitude of a propagating pressure wave incident on the reservoir bottom. The wave reection coecient B may range within the limiting values of 1 and 1. For rigid reservoir bottom materials B = 1 and for very soft reservoir bottom materials B = 1. 10
32.5.3
A standard nite element discretization has used approximating p in terms of nodal values p p NF pe (32.69) and the discretization gives a system of equations in a form MF p + CF p + KF p + rI = 0 (32.70)
where MF , CF , KF and rI are dened in terms of the following element matrices: [Mij ]F =
e
1 g
e s
Ni Nj d +
1 c2
Ni Nj d
F
(32.71)
1 cs
e e
Ni Nj d
(32.72)
e F
Ni Nj d
(32.73)
e I
Ni F nk uk d
e I
Niu nk p d
(32.75)
10 For a more detailed description see for instance Fenves & Chopra [17] and K ukarslan uc et al. [35].
Dynamic Analysis
(32.76)
e I
Nip F nk uk d
(32.77)
(32.78)
After assembling contributions from each type of element (i.e., solid, uid structure interface, uid, boundary uid elements), the following coupled system of equations is obtained MS OT F R M F u p + CS OT O CF u p + KS RT O KF u p = fS (t) 0 (32.79)
32.5.4
The technique to be used for solving the system of equations (32.79) strongly depends on the form of the forcing function fS (t). 32.5.4.1 Frequency Domain Analysis
If the forcing function of (32.79) has been expressed in, or can be transformed to a periodic form as fS (t) = S eit f (32.80) then for linear problems the steady-state solution will exist in the same form, thus u(t) = u eit and p(t) = p eit (32.81) Now a complex expression of the solution is obtained and can be written in the matrix form 2 MS + KS + iCS F R
2 2
RT MF + KF + iCF
u p
S f 0
(32.82)
from which the complex values of the amplitudes u and p can be found. A single set of complex equations for u is obtained by eliminating the pressure values p directly. The second subsystem of equations implied in (32.82) can be written 1 2 p = 2 MF + KF + iCF F R u (32.83)
537
p = 2 F HF () R u
On substitution for p the result from above into the rst subsystem of equations implied in (32.82) an additional uid matrix KI is obtained and is given as KI = 2 F RT HF () R (32.85)
In the above the matrix KI is a complex quantity and can be written in a form KI = 2 F or KI = 2 MF + i CF (32.87) (RT HF () R) + iF (RT HF () R) (32.86)
where MF and CF are denoted as the added mass matrix and the added damping matrix respectively. The structural matrix now takes the form 2 MS + MF + KS + i CS + CF f u = S (32.88)
32.5.5
If no surface waves are admitted and the eect of radiation waves at the innite boundary is ignored, i.e., p=0 respectively on s and e (32.89)
and compression eects are neglected, i.e., c = 0 on F , then the matrices MF and CF of (32.70) as well as the second of (32.79) become zero. The pressure vector p can now be obtained directly in terms of u as p = K1 F R u F (32.90)
On substitution into the rst of (32.79) the structural matrix now becomes of the general form MS + MF u + CS u + KS u = fS (t) where the added mass is simply given as MF = F RT K1 R F (32.92) (32.91)
Dynamic Analysis
If the structural damping is absent or not strongly, the solution in the frequency domain can be obtained by the mode superposition technique [ 32.2.1 p. 524]. Now the solution u will be obtained by superposition of the response in each mode: ... u(t) =
i=1
i i (t)
(32.93)
where i is the i-th eigenvector and i is the i-th generalized modal displacement. Therefore, it requires rst the solution of a sucient number of eigenvalues and corresponding eigenvectors of the problem in (32.91) with damping neglected (32.94) MS + MF u + KS u = 0 Next, the set of equations (32.91) are transformed to global coordinates and a decoupled set of equations is obtained provided that the damping matrix CS is proportional [ 32.1.2.3 p. 523]. 32.5.5.2 Time Domain Analysis
In case of an arbitrary transient loading, the response of the simplied problem in (32.91) can now be obtained by a direct time integration method [ 32.4 p. 528]. After the added mass matrix MF , given by (32.92), has been determined via the solution of (32.90), the transient analysis can be carried out in the usual way by Module nonlin [Ch. 12 p. 209].
Chapter 33
33.1
General Approach
The hybrid frequency time domain analysis comprises the following steps: 1. The pseudo-force at iteration i is computed in the time domain. The pseudo-force is dened as the dierence between the nonlinear and the linear interal force using the responses of iteration i 1. In this step the nonlinear behavior is taken into account. In the rst iteration the pseudo-force is equal to zero. 2. The pseudo-loads resulting from the pseudo-forces at iteration i are calculated and transformed to the frequency domain. 3. The system of linear equations is solved in the frequency domain. In this step the frequency dependent behavior is taken into account. 4. The solution in the frequency domain is transformed to the time domain.
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
540
Hybrid Frequency Time Domain Analysis 5. The convergence criteria are checked. If convergence is not achieved, return to the rst step. The new solution is used for the calculation of the pseudo-force. It should be noted that the total duration of the response is divided into segments of time and the solution is obtained for a segment at a time. A time segment consists of a number of time points representing a period of the signal. The number of time points within a time segment must be sucient to perform a Discrete Fourier Transformation (DFT). In general the number of time points is a power of two to perform an eecient DFT. After convergence has been achieved, the solution proceeds to the next time segment. The updating of the computed responses is done only in the present segment, not in the previous segments where convergence is reached already.
Chapter 34
Stability Analysis
The set of equations representing nodal equilibrium are written as r(u) = f (u) (34.1) where r represents the internal force vector, f the external force vector and u represents the vector of nodal degrees of freedom (displacements). Displacements ucrit are searched for such that r(ucrit ) = f (ucrit ) and r(ucrit + u) = f (ucrit + u) (34.3) i.e., incremental variations u to the solution ucrit exist such that the equations of equilibrium remain satised. The displacement vector ucrit is called a stability point in the space of possible displacement vectors. As ucrit + u is close to ucrit , (34.3) can be linearized with respect to the nodal degrees of freedom: r(ucrit ) + r u u
crit
(34.2)
f (ucrit ) +
f u
u
crit
(34.4)
Introducing the tangent stiness matrix K= leads to r(ucrit ) + Ku And subtracting (34.2) yields Ku f u u
crit
r u
(34.5)
crit
f (ucrit ) +
f u
u
crit
(34.6)
(34.7)
For conservative loading, the external force vector f does not depend on u and thus Ku 0 (34.8)
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
542
Stability Analysis
34.1
Only geometrical nonlinear eects are taken into account (any physical nonlinearities are neglected), and thus the tangent stiness matrix reads (see Bathe [5]) K = KL + KNL where KL =
V0
(34.9)
BT DBL dV0 L
with:
BL = BL0 + BL1
(34.10)
with BL1 due to an initial displacement eect. See Bathe [5] for a denition of the matrices BL0 and BL1 . Furthermore KNL =
V0
BT BNL dV0 NL
(34.11)
where BNL is dened in Bathe [5]. Both the matrices KL and KNL have up to second order displacement contributions. In linear buckling analysis only rst order displacement contributions are recollected however. This is described in the following. Suppose a solution ulin from (34.1) linearized is known, i.e., ulin results from KL0 ulin = f where KL0 =
V0
(34.12) (34.13)
BT DBL0 dV0 L0
with constant external loading f . It can be questioned whether solutions ucrit satisfying (34.2) and (34.8) exist such that ucrit = crit ulin (34.14)
Note that crit ulin would result from a loading fcrit = crit f on the linear system, as can be shown by multiplication of (34.12) by crit : crit KL0 ulin = crit f (34.15)
Parameter crit will be determined from the instability condition (34.8). Doing so, the following approximations are used:
Only rst order displacement contributions in matrix KL are taken into account. These are stored in the matrix KLL :
KLL (ulin ) =
V0
(34.16)
543
crit lin Application of these approximations leads to KL0 + crit KLL (ulin ) + KG (ulin ) where KG (ulin ) =
V0
(34.17)
u = 0
(34.18)
(34.19)
and KL0 =
V0
BT DBL0 dV0 L0
(34.20)
Equation (34.18) is satised by nontrivial solutions for u if det KL0 + crit KLL (ulin ) + KG (ulin ) (34.21)
And so, the general stability conditions (34.2) and (34.3) are replaced by the relatively simple conditions (34.15) and (34.21) in linear stability analysis. Equation (34.21) is solved as a generalized eigenproblem written as KL0 i = i crit KLL (ulin ) + KG (ulin ) i (34.22)
where i is the i-th buckling mode and i is the appropriate buckling value. The theory discussed in this section is referred to as linear buckling theory, or Euler buckling theory. Tutorial example. The theory of the linear Euler stability analysis will be illustrated with an example. Consider a simple arch structure made from two bars as shown in Figure 34.1 and loaded by a force F . Nodes 1 and 2 are
F 3 u d 1 c c 2
l0
Figure 34.1: Arch, loading and deformation xed in space. The deformed conguration is shown with a dashed line. The
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
544
Stability Analysis two displacements of the joint are the only degrees of freedom. For reasons of symmetry, it is assumed a priori that the joint does not displace laterally. A vertical displacement ucrit , corresponding with a force Fcrit , is searched for such that (34.2) and (34.8) are satised. The force Fcrit is called the critical load. Firstly, the equilibrium condition (34.1) will be examined in more detail. Load F is in equilibrium with the normal forces N in the bars as shown in Figure 34.2. Mathematically this reads:
F
N du
c c
2N sin = 2N
(34.23)
where l denotes the deformed length of the bars, in contrast with the initial length which is denoted by l0 . The forces N only depends on the shortening or elongation of the bars. If geometrical nonlinearities (large deformations) are taken into account, this relationship reads: N = A0 l l0 (34.24)
where A0 is the initial cross-section of the bars. The second PiolaKirchho stress is given by =E (34.25) where E denotes Youngs modulus (the bars are assumed to be isotropic) and represents the GreenLagrange strain given by =
1 2
c2 + (d u)2 1 2 l0
(34.26)
A0 l0
Ed2 2 l0
Edu Eu2 2 l2 l0 0
+ ( )
=0
(34.28)
34.2 Imperfections The rst term in the left-hand-side of (34.28) is the linear stiness; the second term is referred to as the stiness due to the initial displacements and the last term is the geometrical stressstiness of the structure. For Euler stability analysis the following approximations are made: 1. Small displacements, i.e., u2 is neglected. 2. where is the Cauchy stress following from a geometrically linear analysis: = =E (34.29) where is the geometrical linear (small) strain: = d 2u l0 (34.30)
545
These assumptions lead to a more simple expression for (34.28): 2 A0 Ed 3 (d 3u) = 0 l0 (34.31)
It can be seen that applying an external load Fcrit such that ucrit = 1 d leads to 3 instability because (34.31) becomes satised for such F . The required value for 1 Fcrit is determined by substituting u = 3 d into (34.27). It should be realized that due to the approximations the critical load Fcrit as obtained from an Euler stability analysis may be in error. In order to demonstrate this, the exact solution of the stability problem will now be derived. Elaborating (34.28) without applying any approximations leads to: EA0 2 2 c l0 + 3(d u)2 = 0 3 l0 Equation (34.32) is satised if: 1 ucrit = d 1 + 3 or: 1 ucrit = d 1 3 (34.33) (34.32)
with a load Fcrit following from (34.27). Thus, exactly solving the stability problem renders two solutions. The most critical of both is approximately 0.42 d whereas the linear buckling analysis rendered 0.33 d. Neglect of the initial displacement terms in the linear buckling analysis would have given d.
34.2
Imperfections
The constant part of the stiness matrix of the imperfect structure is denoted by KI . Once the imperfect geometry is known, this matrix KI can be deL0 L0 termined. The imperfect geometry is dened by a displacement eld uI from
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
546
Stability Analysis the perfect structure to the imperfect structure. Thus the matrix KI actually L0 depends on this displacement eld: KI = KI uI L0 L0 The eld uI can be chosen in dierent ways: Compatible with critical buckling mode. A buckling mode from an analysis without imperfections is used as imperfection pattern. Generally, the lowest buckling mode is most critical. For this pattern the critical buckling value may be decreased more substantially than for an arbitrary pattern. Random. Random imperfections may be used to simulate random imperfections of reality. User-specied. A specic eld, for instance taken from measurements, may be dened by the user. Summarizing, the buckling criterion for the imperfect structure reads: det KI (uI ) + KLL (ulin ) + KG (ulin ) L0 =0 (34.35) (34.34)
For a xed imperfection pattern (i.e., user-specied or random), an eigenvalue calculation on (34.35) is performed to calculate the critical buckling load. However, for the lowest buckling mode compatible imperfection pattern, a two-step strategy is necessary. First, an eigenvalue determination on (34.35) without imperfections ( uI = 0 ) is performed. The corresponding critical buckling mode is used to set up the imperfections uI . A second eigenvalue analysis of (34.35), now including uI , is performed to add the eect of the imperfections.
34.3
Postbuckling Analysis
The displacement vector ucrit is a stability point in the space of possible displacement vectors. The denition of ucrit is given by (34.3) which must be valid for arbitrary small u. The purpose of perturbation analysis is to calculate a postbuckling displacement eld upb satisfying (34.1) r(upb ) = f (upb ) but being dierent from the primary path u = ulin with being the load parameter. Assumption: there are M coinciding or nearly coinciding interacting buckling modes, denoted by k with k = 1, . . . , M . In the theory of mode interaction the initial postbuckling displacement eld upb is written as upb = ulin + ai i + ai aj uij (34.36)
where uij is called the second order displacement vector and ai should be interpreted as amplitude of the respective mode. In literature1 it is shown that uij
1 See
for instance Koiter [34], Van Erp [59], Byskov & Hutchinson [9].
Diana-9.4 Users Manual Analysis Procedures (X)
34.3 Postbuckling Analysis must be calculated by solving the system KL0 + p 1 KG (ulin ) uij = fij (34.37)
547
where p = 1 must be specied by the user. Applying the orthogonality conditions T KL0 uij = 0 with k = 1, . . . , M (34.38) k where fij is dened as the mode interaction load vector fij =
V0
BT ij + BT (i ) j + BT (j ) i dV0 L0 NL NL
(34.39)
where ij is the stress related to interaction of modes i and j and i is the stress related to mode i. Further, the potential can be written as a function of the load parameter and the mode amplitudes ai
M
P (ai , ) = where
1 2 I=1
aI aI + Aijk ai aj ak + Aijkl ai aj ak al
(34.40)
1 2
ij k dV0
V0
(34.41)
1 8
V0
(34.42)
are the third and fourth order potential terms, respectively. From the potential function the nonlinear equilibrium equations after buckling, the continuation analysis, can be calculated in a stepwise approach. Points of equilibrium are indicated by terms of ai and . From these data, the postbuckling displacement eld upb can be derived using (34.36).
548
Stability Analysis
Chapter 35
Basic Equations
ConvectionDiusion Equation
divq + + c = qV q = k (35.1) (35.2)
with the potential, q the specic ux vector, k the diusivity tensor, the eld convectivity vector, c the capacitance, and qV the external ux per volume.
35.1.2
Boundary Conditions
Three conditions can be applied to boundaries or parts thereof: essential, natural, or mixed. Essential (Dirichlet). The essential boundary condition is a prescribed boundary potential B : = B (35.3) Natural (Neumann). The natural boundary condition is a prescribed boundary ux qB : q n = qB (35.4) with n the vector pointing outwards normal to the boundary. Mixed (Neumann/Robin). The mixed boundary condition is a boundary ux which depends on a free boundary potential and a prescribed environment potential E : q n = K( E ) (35.5) with K the conduction coecient, which is typically used to model for example heat convection or radiation at boundaries.
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
550
Potential Flow Analysis If no boundary condition is specied then the natural condition of insulation is valid: qn = 0 (35.6)
35.1.3
The nite element formulation can be derived according to the Galerkin procedure. Equation (35.1) is multiplied by a test function and integrated over the considered volume V . divq + + c
V
dV =
V
qV dV
(35.7)
dV =
B
q n dB
(35.8)
dV +
V
c dV +
B
K dB (35.9) (K E + qB ) dB
B
=
V
qV dV +
We now assume the potential eld and the test function inside a nite element as a linear function of the values in the element nodes: (x) = N , (x) = N , (x) = B (x) = B (35.10) (35.11)
The vector contains the node potentials, the vector contains the node test function values, and N and B are interpolation matrices. Substitution of (35.10), (35.11), and (35.2) in (35.9) nally results in the following nite element equation: K + C = Q (35.12) with K=
V
BT k B dV +
V
NT B dV +
B
NT K N dB
(35.13) (35.14)
C=
V
NT c N dV NT qV dV +
V B
Q=
NT qB dB +
B
NT K E dB
(35.15)
where K is the conduction/convection matrix, C the capacity matrix, and Q the nodal discharge or external ux vector.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
551
Diana solves the potential ow nite element equations (35.12) stepwise with direct time integration, using a generalized trapezoidal rule. This means that for each step t, the equation is solved at time t + t. Various methods are available depending on the value of . = 0 for Euler Forward integration (explicit). This method has rst order accuracy and a limited step size for stable solution. With max for the largest eigenvalue, the condition is t < = =
1 2 2 3
2 (1 2)max
for
<
1 2
(35.16)
for CrankNicolson integration. This method has second order accuracy and no numerical damping. for Galerkin integration, a method with rst order accuracy.
= 1 for Euler Backward integration. This method has rst order accuracy without oscillations. It has numerical damping. Combination of the nite element equations and direct time integration results in an equation of the following form. K = Q
(35.17)
With K the eective matrix, the vector of nodal potentials, and Q the eective vector of nodal discharges. In Diana an implicit and an explicit integration scheme has been implemented. Implicit time integration. The implicit time integration is characterized by >0 With the time derivative:
t+ t t t+ t = t
and
= t+ t
(35.18)
(35.19)
The eective matrix: K = Ct+ t + t Kt+ t The eective nodal discharge vector: Q = t Qt+ t + Ct+ t t In case of nonlinearity the iteration scheme is:
K (i) (i+1) = Q(i)
(35.23)
December 15, 2009 First ed.
552
Potential Flow Analysis Explicit time integration. The explicit time integration is characterized by =0 With the time derivative: t+t t t = t The eective matrix is: K = Ct (35.25) and = t+t (35.24)
(35.26)
If the value of conductivity, capacity or boundary condition is a function of the potential, the set of equations must be solved iteratively. Iteration method. Diana uses an incremental iterative method to solve the set of nonlinear equations. This means that in iteration (i + 1) the incremental potential is calculated from K = Q K .(i) incr (i) (i) sec
(i+1) = (i) +
(35.28) (35.29)
The eective secant conduction matrix is updated in every iteration. The default Regular NewtonRaphson method updates the eective incremental conductivity in every iteration: Kincr = Ksec .(i) (35.30) The Modied NewtonRaphson method updates the eective incremental conductivity only at the start of a step: Kincr = Ksec .(1) (35.31)
Accuracy. During the iteration process, Diana uses the norm of the incremental vector of potentials to determine if accuracy is reached: || | | (1) (35.32)
Chapter 36
36.1
The equations and assumptions in this chapter relate the stresses and pore pressures to deformations and uid ow, using material parameters for elastic compressibility, drained elastoplasticity, density, porosity and permeability. Notations. For general notation conventions see the Glossary of Symbols on page xxiii. Following notations are specic for this chapter. The operator denotes the gradient of a function. For example p= p p p , , x y z
The gradient of a scalar is a vector, the gradient of a vector is a matrix. The divergence of a vector or matrix is denoted as div(u) = uy uz ux + + x y z
554
SoilPore Fluid Analysis Sign conventions. Stresses and strains are assumed to be positive for tension. Pressures are assumed to be positive for compression.
36.1.1
Assuming that all pores are lled with uid or air, the uid fraction depends on porosity and saturation. The apparent density can therefore be formulated as follows. = dry + S n f (36.1) where is the total density of the mixture, n the porosity (equal to the uid fraction), S the pore pressure dependent degree of saturation (0 S 1), f the density of the uid, and dry the density of the dry porous solid. Diana uses this equation to determine the deformation and saturation dependent density during the analysis, assuming a deformation dependent porosity. n=1+ ninit 1 J (36.2)
36.1.2
Undrained Compressibility
A compression modulus relates hydrostatic stress to elastic volumetric strain. The elastic compressible behavior of the undrained mixture is determined by the following compression moduli. KD Compression modulus of the (drained) porous soil skeleton, related to the eective stress. This compression modulus is accompanied by the shear modulus G for determination of drained elastic deformation. Kf Intrinsic compression modulus of the uid. Ks Intrinsic compression modulus of the non-porous solid. Kp Compression modulus of the solid fraction, related to the inter-granular stress. Assuming average stress distribution in the solid particles, this modulus is directly related to Ks : Kp = (1 n)Ks (36.4)
555
36.1.3
The total stress is separated into inter-granular stress and pore pressure. The inter-granular stress is closely related to Terzaghis eective stress. The eective stress governs the strength and deformation of the soil skeleton. = + S Ip = S Ip (36.5)
where is the total stress, the inter-granular stress, the eective stress, p the pore pressure, and a function of compressibility. Widely used is = KD /Ks , with for soils 0. Furthermore I is a unity matrix. Diana determines the pore pressure p from the pore pressure potential , using the initial position vector xinit xref and the displacement u: p = + f g (x xref ) with x = xinit + u (36.7) (36.6)
36.1.4
Momentum Conservation
The basic equilibrium law in stress analysis is momentum conservation. If a gravity eld is present, Diana uses the following momentum balance: div() + g + = 0 u (36.8)
where g is the gravity acceleration vector and u the acceleration of the solid. Convective terms and the relative accelerations between solid and uid are neglected. Using (36.5) this equation reads div( ) S (1 ) Ip + g + u = 0 (36.9)
36.1.5
Mass Conservation
The inuence of mass conservation depends on the degree of saturation of the porous medium: full or partial. Fully saturated. Mass conservation in a saturated context means that both uid ow and undrained compression or expansion will cause volumetric changes. The homogeneous form of the so-called storage equation is derived from the general mass conservation law div(u) + div(q) + p =0 Q (36.10)
556
SoilPore Fluid Analysis where u is the velocity vector and q the uid ux vector. Variables and Q are the Biot material parameters: n 1 = + (1 n) Q Kf 1 Ks Kp ; =1 (36.11)
For soils 1 and Q Kf /n. The value of 1/Q is often called the hydraulic capacitance. Partially saturated. For partially saturated areas, Diana uses the following extension to the storage equation, assuming that air pressures are zero. S div(u) + div(q) + p =0 Q (36.13)
Diana uses the pore pressure dependent degree of saturation S to determine the modied Biot parameter Q, with limS1 Q = Q nS 1 = Cs + + Kf Q ( n) S S+ Ks Cs p n
(36.14)
where the so-called moisture capacitance Cs is derived from the relation between degree of saturation and pore pressure. Cs = n S p (36.15)
36.1.6
Darcy Flow
Darcys law relates the uid ow in porous media to gradients in pore pressure potentials by permeability. q = k with k = k and = p f g (x xref ) (36.17) (36.16)
where q is the specic discharge, k the permeability, k the modied permeability, = the dynamic viscosity, the pore pressure potential, g the gravity acceleration vector, and x xref the current position vector.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
36.2 Finite Element Equations Note that Diana uses the permeability k , which relates the specic discharge to the pore pressure potential. However, in geotechnical practice also the permeability khead is in use, which relates the specic discharge to gradients in the pore pressure potential head. k = with q = khead head and = f |g| head (36.19) khead f |g| (36.18)
557
Dependency of the permeability. In general, the modied permeability depends on porosity and saturation. Diana supports a porosity dependent relative permeability kreln and a saturation dependent relative permeability krelS using k = kreln krelS kinit (36.20) Diana can derive the relative permeabilities from user-specied diagrams. In case of kreln Diana also supports the relation given by Kozeny & Karman in [36]: n kreln = ninit n 1n
2
ninit 1 ninit
(36.21)
Dynamic behavior. In case of dynamic behavior, Diana uses the following extended equation for the Darcy ow, assuming isothermal behavior, and omitting convective terms and relative accelerations between uid and solid. q = k ( f u) (36.22)
36.1.7
Elastoplastic Stiness
Elastoplasticity relates stress to strain , using parameters which are typically shear modulus G, compression modulus KD , cohesion c, friction angle , and dilatancy angle . = D( ) ( u) (36.23) where D is the drained elastoplastic stiness of the soil skeleton. See Volume Material Library for the eective stress models that are available in Diana.
36.2
The equations derived in this section, relate nodal displacements, pore pressure potentials and their velocities to nodal forces and uid discharges.
558
36.2.1
Space Discretization
The basic nite element assumption is the interpolation of coordinates, displacements and pore pressure potentials for each element from the values in nodes ui = Nu ui ; = N (36.24)
where is the vector of nodal values for the pore pressure potential , ui the vector of the nodal values for the displacement component ui , N the pore pressure potential interpolation vector, and Nu the displacement interpolation vector. Based on Equations (36.4) to (36.23) a nite element space discretization is derived via the standard Galerkin procedure. Mu u + Cu u + C + K = fb. Muu u + Kuu u Ku = fb.uu + fg.uu fu (36.25) (36.26)
where u is the nodal displacement vector, the nodal pore pressure potential vector, Muu the mass matrix, Mu the coupling mass matrix for ow by acceleration of displacement, Kuu the stiness matrix, K the permeability matrix, C the hydraulic capacity matrix, fb. the boundary ux vector, fg.uu the dead weight force vector, fu the buoyancy force vector, fb.uu the boundary force vector, Ku the coupling matrix for stress by pore pressure potential, and Cu the coupling matrix for ow by displacements. All these matrices and vectors are determined as follows: Kuu =
V
BT D Bu dV u 1 T N N dV Q BT k B dV BT m N dV u ; mT {1 1 1 0 0 0}
C =
V
K =
V
Ku =
V
Cu = KT u Muu =
V
NT Nu dV u BT f k Nu dV g Nu dV
Mu =
V
fg.uu =
V
fu =
V
f g x BT m dV u
36.2 Finite Element Equations where Bu is the strain vector interpolation matrix, D the material stiness matrix, N the pore pressure potential interpolation vector, Nu the displacement interpolation matrix, and B the pore pressure potential gradient interpolation matrix N .
559
36.2.2
Time Discretization
The space discretization is evaluated at or between several time steps, mostly using an Euler Backward time integration for consolidation analysis, and a nite dierence approximation for the time derivatives for dynamic analysis: Newmark , HilberHughesTaylor or Wilson- [ 32.4 p. 528]. This section illustrates how the Euler Backward approximation of velocities is incorporated in a nonlinear iteration scheme. Euler Backward assumes constant velocities within each time step u= u = t
t+t
u t u t t t
(36.36)
= = t
t+t
(36.37)
The preceding superscript t or t+t denotes the time of evaluation. Combination with (36.25), (36.26) and a nonlinear iteration scheme leads to the following incremental iterative formulation (subscript (i) denoting the iteration number, starting with 1).
t+t
1 C+K t fextern
(i1) t+t
(36.38)
t+t
fintern.(i1)
K=
(36.40)
fextern =
(36.41)
fintern
BT dV Ku u = u K + C + Cu t t
(36.42)
In general, accuracy is assured if the time step is small enough. However, if the time step is chosen too small, spatial instabilities will occur at the almost undrained and incompressible situation immediately after loading. Due to numerical diusion, these instabilities will disappear during the integration. The spatial instabilities are suppressed either by a larger time step or by a small element size. Vermeer & Verruyt [60] derived a criterion for one-dimensional consolidation with assumption of linearly interpolated elements and equally spaced time steps. For Euler Backward time integration this equation reads tmin l2 6 cc (36.43)
with cc the so-called coecient of consolidation, and l the element size. cc = k 1 1 + KD Q (36.44)
36.2.2.2
Dynamic Analysis
The derivation of the equations for dynamic behavior with dierent time integrations is basically analogous. Diana however does not include the Mu contribution in the total stiness matrix, but only in the internal force vector fintern . This means that always iteration is required in case of dynamic analysis with Mixture elements.
36.2.3
Undrained Behavior
Diana uses a penalty formulation to model completely undrained behavior [Vol. Material Library]. You can use the penalty method as a substitute for Mixture analysis to model the behavior of layers with practically no dissipation of excess pore pressure during the period of analysis. You can also use the penalty method in combination with a Mixture analysis, during the almost undrained and incompressible situation immediately after loading, to avoid spatial instabilities. The penalty method derives the excess pore pressure pe directly from the volumetric strain v . pe = Kf v (36.45) By default, Diana sets the value of the bulk modulus of the pore uid Kf equal to the drained bulk modulus KD times a large penalty factor: Kf KD . In practice, this implies that the uid fraction will prevent almost all volumetric deformations and that only the soil fraction will resist against deviatoric deformations.
561
36.2.4
As an alternative to coupled Mixture analysis, you can also apply a hydraulic pore pressure potential load h on the soil skeleton, which you may combine with undrained behavior via the penalty method. Diana uses: fh.u fu = Ku h f g x BT m dV u (36.46)
By applying reduced density red instead, you can simulate the buoyancy load: red = f
V
(36.47) (36.48)
f g Nu dV fh.u fu
36.2.4.1
Besides a Mixture analysis, Diana can also determine a nonuniform distribution of the hydraulic pore pressure potential via a separate potential ow analysis [Part VI]. The equations for determination of a hydraulic pore pressure load can simply be derived from the general equations by removing the deformation dependent part from the mass conservation law: div(q) + ph =0 Q (36.49) (36.50)
C h + K h = fb.
562
Chapter 37
37.1
In Linear Elastic Fracture Mechanics the problem of calculating the stresses around a crack in a structure is analyzed by using linear elastic constitutive relations. The solution to the linear elastic problem is singular at the crack tip. In this case, the stresses and strains around a crack tip tends to innity. The magnitude of this singularity can be described with the stress intensity factor. There are three basic stress intensity factors [Fig. 37.1]: KI for opening mode
t
564
Fracture Mechanics Analysis loading, KII for sliding mode loading and KIII for tearing mode loading. In case of an opening mode loading, the stresses just before the crack tip (in the crack plane) are: KI yy = (37.1) 2r where r is the distance from the crack tip. The stress intensity factor for an innite plate with a through the plate crack of size 2a and a far away stress eld yy = is: KI = a (37.2) In (37.2) we see that KI depends on the crack geometry and the loading. There are many textbooks about fracture mechanics and handbooks with stress intensity factors for specic structures and crack geometries.
37.1.1
Mesh Adaptation
As stated above the stresses and strains at the crack tip are singular. In the vicinity of the crack tip, this singularity can be described with: KI = f () 2r (37.3)
where KI is the stress intensity factor for Mode-I opening. Modeling this singularity in a nite element mesh with the usual elements, would require a considerably rened mesh near the crack tip. It can be shown however, for instance for two-dimensional quadratic elements, that when the midside nodes of the elements at the crack tip at a quarter distance from that crack tip, the eleare ment can describe this 1/ r singularity. The same holds for three-dimensional elements. For more information see for instance Zienkiewicz [62, pp. 189191], Barsoum [3], or Henshell & Shaw [25]. With midside nodes on quarterpoint position the singularity at the crack tip is best described.
37.2
The stress intensity factor can be calculated from the energy release rate. This energy release rate G, equals the release of the elastic energy Win minus the external potential Pex when the cracked area Acr increases with one unit area, with constant displacements at the outer boundary. G= Win Pex + Acr Acr (37.4)
See for instance Hellen [24]. For convenience, the contribution of the external potential is ignored in the sequel of this chapter. However, the contribution of thermal and chemical expansion, initial stresses and crack face loads have been included in Diana.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (X)
565
37.2.1
Two-dimensional Analysis
In case of a two-dimensional analysis, the increase in cracked area can be related to the increase in crack length a. If Win is the elastic energy and t is the thickness of a plate, then the energy release rate is dened as G= 1 Win t a (37.5)
at constant displacements. In nite element analysis the elastic energy W can be calculated from pre- and postmultiplication of the stiness matrix K by the displacement vector u. Win = 1 uT Ku (37.6) 2 Since the displacements are kept constant, the derivative in the right hand side of (37.5) can be calculated from: Win K = 1 uT u 2 a a (37.7)
The derivative K/a is determined analytically with reference to the known shape functions, see Van Den Boogaard [57]. Since the system stiness matrix is only aected by the elements that are at the crack tip, only these elements are to be re-analyzed. The relation between the stress intensity factor and the energy release rate is known for plane stress and plane strain conditions. In the case of Mode-I loading (opening) this relation is: GI = where for plane stress and for plane strain KI E (37.8)
E = E E =
(37.9)
E (37.10) 1 2 It is noted here that the calculation of KI from G only holds if we have a purely Mode-I loading. Therefore, Diana derives the stress intensity factor from the energy release rate as if the loading of the crack is purely Mode-I (opening).
37.2.2
Three-dimensional Analysis
The analysis of the energy release rate in two-dimensional structures is straightforward. For three-dimensional analysis the energy release rate can vary along the crack tip. Diana calculates the energy release rate in three-dimensional models (with solid elements) by independent virtual crack extension at dierent nodes at the crack tip. The virtual crack extension always takes place in the plane of the crack itself. It was shown that, for quadratic elements, the most accurate results are obtained if the corner nodes of every element are the
Diana-9.4 Users Manual Analysis Procedures (X) December 15, 2009 First ed.
566
Fracture Mechanics Analysis nodes that will have independent crack extensions and if the midpoint nodes of an element are given half the crack extension of a corner node. In this way, the crack grows around a chosen corner node in a shape that is triangular in the isoparametric coordinates [Fig. 37.2]. From the original structure and the
1 dl 2
dl
tip
Figure 37.2: Virtual crack extension in three-dimensional analysis perturbed structure, the decrease in elastic energy Win can be calculated in the same way as for two-dimensional structures. The integral of G a along the crack tip must be equal to this decrease in elastic energy. G a ds = Win
tip
(37.11)
In the left hand side of this relation, s is the length along the crack tip. The energy release rate can be calculated by assuming that G varies linearly between two corner nodes. This yields a set of equations in G, from which G can be calculated at the dierent corner points. For three-dimensional calculations Diana applies the plane strain equation (37.10). The implementation of the virtual crack extension method has been corrected for the thermal and concentration loads and initial stress at the crack tip elements, as well as for external loading on the crack faces like pressure loading.
Part XI
Appendix
Appendix A
570
A.1
Diana passes the source code of the user-supplied subroutines to the Fortran compiler of your computer system. When the compilation is terminated, the systems linker-loader is invoked to make the appropriate executables.1 These executables are then used instead of the standard versions. There are two ways to supply Diana with Fortran source code: directly included in the command le [ A.1.1], or via an external le [ A.1.2]. Validation. We advise you to ensure that your source code is syntactically correct conform the Fortran-77 standard [2], before supplying it to Diana. Moreover, you are encouraged to validate the source code by means of Dianas Fortran Validation Processor diva, which checks the argument lists, COMMON blocks, programming style and other. For use of diva see the Diana Programmers Manual.2
A.1.1
*FORTRAN *module w *END
Direct Inclusion
syntax
*FORTRAN invokes the Fortran-77 compiler to compile the source code. The source code must start with a Diana user-supplied subroutine. The usage of subroutines from Dianas service libraries is strongly encouraged [ A.4
1 Dianas user-supplied subroutine mechanism is based on the unix make utility. It uses source les, object les and executables as targets for make. 2 The Diana Programmers Manual is available interactively via the documentation processor ddoc.
A.1 Supplying the Source Code p. 576]. Other subroutines or functions which are not available in any of Dianas standard object service libraries, must follow the code of the user-supplied subroutine. If more than one user-subroutine is supplied, each of them must be preceded by a *FORTRAN command. Diana creates a source le for each set of source code. *module is the module command to start the analysis, with the analysis commands. The following example illustrates the direct inclusion of source code for usersupplied subroutines.
file .dcf *FORTRAN SUBROUTINE ELSEUS( EPS, NSTR, SE ) C C... USER SUPPLIED TANGENTIAL [SE] - MATRIX C INTEGER NSTR DOUBLE PRECISION EPS(NSTR), SE(NSTR,NSTR) C DOUBLE PRECISION YOUNG PARAMETER ( YOUNG=1.3D5 ) C IF ( NSTR .EQ. 4 ) THEN PRINT *, NUMBER OF STRESS COMPONENTS NOT CORRECT PRINT *, MODEL CANNOT BE USED WITH THIS ELEMENT CALL PRGERR( ELSEUS, 1 ) END IF C CALL RSET( 0.D0, SE, NSTR*NSTR ) CALL PRIVAL( YOUNG, YOUNG ) SE(1,1) = YOUNG + 2000.D0 * YOUNG * EPS(1) SE(2,2) = YOUNG + 2000.D0 * YOUNG * EPS(2) SE(3,3) = YOUNG + 2000.D0 * YOUNG * EPS(3) SE(4,4) = 0.5D0 * YOUNG C END *FORTRAN SUBROUTINE ELSGUS( EPS, NSTR, TSIG ) C C... TOTAL STRESSES C INTEGER NSTR DOUBLE PRECISION EPS(NSTR), TSIG(NSTR) C DOUBLE PRECISION YOUNG PARAMETER ( YOUNG=1.3D5 ) C
Diana-9.4 Users Manual Analysis Procedures (XI)
571
572
analysis commands
The rst *FORTRAN command invokes the Fortran-77 compiler for the rst usersupplied subroutine. The source code for this subroutine immediately follows the *FORTRAN command. It is subroutine ELSEUS for the user-supplied nonlinear elasticity material model [Vol. Material Library]. The second *FORTRAN command invokes the Fortran-77 compiler for a second user-supplied subroutine: ELSGUS. Commands for execution of a nonlinear analysis follow the *NONLIN command. This example creates the following les, used as targets for make: ELSEUS.f Fortran source code of subroutines ELSEUS. ELSEUS.o object code of subroutines ELSEUS. ELSGUS.f Fortran source code of subroutine ELSGUS. ELSGUS.o object code of subroutine ELSGUS. nl41 dummy executable of Module nonlin.
A.1.2
External File(s)
syntax
analysis commands
A.2 Setting the User Environment TAKE indicates to take code from le usrout . The le name must contain the extension .f for Fortran-77 source code, or .o for object code (result of previous compilation). The le usrout must start with a Diana user-supplied subroutine, its basename must be equal to the name of the user-supplied subroutine. You may choose upper or lower case for the basename, for instance ELSGUS.f or elsgus.f for the source code of subroutine ELSGUS. Externals which are not available in any of Dianas standard service libraries, must follow the code of the user-supplied routine. If more than one user-routine is supplied, each of them must be specied in a separate TAKE command. *module is the module command to start the analysis, with the analysis commands. The following example illustrates the use of externally supplied source and object code. We supply the two user-subroutines via external source les elseus.f and elsgus.f . Note that the two le names in lower case are specied in double quotes, else Diana would search for les ELSGUS.f and ELSEUS.f .
file .dcf *FORTRAN TAKE "elseus.f" TAKE "elsgus.f" *NONLIN *END
573
analysis commands
This example creates the following les, used as targets for make: elseus.o object code of subroutines ELSEUS. elsgus.o object code of subroutine ELSGUS. nl41 dummy executable of Module nonlin. Now if you modify the le elseus.f and run the same job again, Diana sees that the le elseus.f is younger than elseus.o and recompiles it.
A.2
Dianas user-supplied subroutine option knows which executable(s) depend on a particular subroutine. It automatically creates an executable version of the appropriate segment(s). To work properly, special attention must be paid to the user environment. This involves the setting of environment symbols and the management of les.
574
A.2.1
Environment Symbols
To use the user-supplied subroutine option you must login as a Diana user and then set some environment symbols. UNIX. To use the user-supplied subroutine option on a unix system you must rst invoke Dianas Makemac utility to set general symbols like compiler names and ags. In the Bourne shell do: . $DIAPATH/Makemac and in the C shell: source $DIAPATH/Makemac.csh Next set the user symbols BINSEG and SEGPATH: BINSEG contains the Diana installation directory for user-segments. Usually this symbol is set on a subdirectory diana/binseg of the users home directory, for instance in the Bourne shell: BINSEG=$HOME/diana/binseg; export BINSEG or in the C shell: setenv BINSEG $HOME/diana/binseg SEGPATH the search path for Diana segments. The name of the installation directory for user-segments must be before Dianas standard installation directory, for instance in the Bourne shell: SEGPATH="$BINSEG;$SEGPATH"; export SEGPATH or in the C shell: setenv SEGPATH "$BINSEG;$SEGPATH" It may be convenient to set the necessary environment symbols during the login procedure, consult your systems manager or Diana support for this. Windows. To use the user-supplied subroutine option on a Windows system you must run Diana from a command box (DOS box). If you want to keep the executables that Diana creates during the execution, then you should specify a directory where they can be stored via the symbol BINSEG, for instance: SET BINSEG=C:\USER\YOU\DIANA\BINSEG To let Diana search in this directory you must extend the SEGPATH symbol such that the BINSEG directory comes before the search path: SET SEGPATH=%BINSEG%;%SEGPATH%
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (XI)
575
A.2.2
Files
Dianas user-supplied subroutine option creates the following les on the current directory: Fortran source les, only created if the source code was directly included in the command le. Object les, result from the compilation of the Fortran source les.
Dummy executable segments, created after the real executable was moved to the installation directory for user-segments. These dummy les have length zero and are only used as a time stamp for the make utility. Maintenance. As long as the BINSEG symbol is in your search path SEGPATH, Diana takes executables from the BINSEG directory and not from its standard installation directory. Therefore: It is very important that you remove executables from your BINSEG directory as soon as you have nished the analysis with a specic user-supplied subroutine. If you dont remove your executables, Diana will use them in your next analysis projects as well.
A.3
Available Routines
The available routines are listed in the le $DIASLIB/userinfo. For Diana-9.4 this le looks like: # user suplied routine information # layout: # <routine_name>: <segment_to_remake> ... uslafr: ap/la02 elsgus: ap/nl41 elseus: ap/nl41 dmsgus: ap/nl41 dmseus: ap/nl41 bowliq: ap/nl41 nisliq: ap/nl41 towliq: ap/nl41 umat: ap/nl41 usrliq: ap/nl41 usryld: ap/nl41 usryou: ap/nl41 usrpoi: ap/nl41
Diana-9.4 Users Manual Analysis Procedures (XI) December 15, 2009 First ed.
576 usrten: usrcrk: usrcrv: usrifc: usrmat: usrmnl: usrtau: usrcrp: usrrub: usrbou: usrpar: usrhtp: ap/nl41 ap/nl41 ap/nl41 ap/nl41 ap/nl41 ap/nl41 ap/nl41 ap/nl41 ap/ls41 ap/nl41 ap/pe41 ap/pe41 ap/ht40
Each line contains the name and location of a pre-dened user-supplied subroutine. For instance the line starting with elsgus indicates that subroutine ELSGUS is user-supplied and called in the executable ap/nl41. The directory name ap refers to the Diana application executables and the le name nl41 to version 41 of Module nonlin.
A.4
The Diana programming environment comes with a set of service libraries of utility routines. The libraries are especially helpful when coding user-supplied subroutines because the routines are extensively tested and often optimized. When applying user-supplied subroutines, you are encouraged to call the routines of the Diana service libraries whenever appropriate. Dianas programmers service libraries are subdivided for some basic tasks such as matrix and vector manipulations, I/O operations etc. This section presents the subroutines and functions which are particularly useful in combination with the user-supplied subroutine option. See the Diana Programmers Manual for full description of the service library subroutines.3
A.4.1
Matrix Manipulation
Dianas Matrix library oers a set of subroutines to perform matrix manipulations like multiplication, triple products, inversion, and initialization. In this library we use the following naming convention: R----S----A----B----Result Result Result Result is is is is stored in output matrix R. superposed to input matrix R. stored back in input matrix A. stored back in input matrix B.
3 The Diana Programmers Manual is available interactively via the documentation processor ddoc.
A.4 Programmers Service Libraries -A----B-----T--INV--D----DET-----#-Input matrix A. Input matrix B. Preceding input matrix is processed transposed. Input matrix is inverted. Decomposition of input matrix. Determinant of input matrix. Dimension of input matrix.
577
For example: subroutine RAB performs the matrix multiplication R = A B. A.4.1.1 Simple Products Fortran
Multiply R = A B CALL RAB( a, n, m, b, l, r ) in in out in dbl dbl dbl int a(n,m) b(m,l) r(n,l) n m l Anm . Bml . Result Rnl . Dimensions n, m and l.
file .f DOUBLE PRECISION AMAT(4,3), BMAT(3,5), RMAT(4,5) CALL RAB( AMAT, 4, 3, BMAT, 5, RMAT )
See also AAB, BAB and SAB. Multiply R = R + A B CALL SAB( a, n, m, b, l, r, k ) in in in out in in dbl dbl dbl dbl int int a(n,m) b(m,l) r(k,l) r(k,l) n m l k Anm . Bml . Input Rnl . Added result Rnl . Dimensions n, m and l. Column size k of array r. Fortran
578
(k n)
The result matrix Rnl may be stored in a part of array r. result is erroneous.
If k < n the
file .f
DOUBLE PRECISION AMAT(3,2), BMAT(2,4), RMAT(6,8) CALL RSET( 0.D0, RMAT, 6*8 ) CALL SAB( AMAT, 3, 2, BMAT, 4, RMAT(1,1), 6 ) CALL SAB( AMAT, 3, 2, BMAT, 4, RMAT(4,1), 6 ) CALL SAB( AMAT, 3, 2, BMAT, 4, RMAT(1,5), 6 ) CALL SAB( AMAT, 3, 2, BMAT, 4, RMAT(4,5), 6 )
See also RAB. Multiply A = A B CALL AAB( a, n, m, b, l, w ) in in out in spc dbl dbl dbl int dbl a(n,m) b(m,l) a(n,l) n m l w(l) Anm . Bml . Result Anl . Dimensions n, m and l. Workspace. Fortran
This subroutine may be used instead of RAB to save memory space. The result matrix Anl is returned in array a itself, so the input matrix Anm is destroyed. The rows of array a must be dimensioned to the maximum of l and m.
579
See also RAB and BAB. Multiply B = A B CALL BAB( a, n, m, b, l, w ) in in out in spc dbl dbl dbl int dbl a(n,m) b(m,l) b(n,l) n m l w(n) Anm . Bml . Result Bnl . Dimensions n, m and l. Workspace. Fortran
?
This subroutine may be used instead of RAB to save memory space. The result matrix Bnl is returned in array b itself, so the input matrix Bml is destroyed. The following example calls subroutine PRIMAT to show how the various matrixes have been lled.
(m n)
580
Note that array BMAT is declared one-dimensional because it is used with two dierent column lengths. The address space of the output array BMAT is lled without gaps as shown in the examples output:
file .out
AMAT : 1 2 BMAT1 : 1 2 3 BMAT2 : 1 2 1 1.000E+00 1.000E+00 1 1.000E+00 1.000E+00 1.000E+00 1 6.000E+00 6.000E+00 2 2.000E+00 2.000E+00 2 2.000E+00 2.000E+00 2.000E+00 2 1.200E+01 1.200E+01 3 3.000E+00 3.000E+00 3 3.000E+00 3.000E+00 3.000E+00 3 1.800E+01 1.800E+01
See also RAB and AAB. Multiply R = A BT CALL RABT( a, n, m, b, l, r ) in in out in dbl dbl dbl int a(n,m) b(l,m) r(n,l) n m l Anm . Blm . Result Rnl . Dimensions n, m and l. Fortran
Subroutine RABT performs a simple matrix multiplication with the transpose of the second input matrix: l n R = n m A m l BT
file .f DOUBLE PRECISION AMAT(4,3), BMAT(5,3), RMAT(4,5) CALL RABT( AMAT, 4, 3, BMAT, 5, RMAT )
A.4 Programmers Service Libraries Multiply R = AT B CALL RATB( a, n, m, b, l, r ) in in out in dbl dbl dbl int a(n,m) b(n,l) r(m,l) n m l Anm . Bnl . Result Rml . Dimensions n, m and l. Fortran
581
Subroutine RATB performs a simple matrix multiplication with the transpose of the rst input matrix: l m R n = m AT n l B
file .f DOUBLE PRECISION AMAT(3,4), BMAT(3,5), RMAT(4,5) CALL RATB( AMAT, 3, 4, BMAT, 5, RMAT )
See also RABT. Multiply R = R + AT B CALL SATB( a, n, m, b, l, r, k ) in in in out in in dbl dbl dbl dbl int int a(n,m) b(n,l) r(k,l) r(k,l) n m l k Anm . Bnl . Input Rml . Added result Rml . Dimensions n, m and l. Column size k of array r. Fortran
Subroutine SATB superposes a matrix product with the transpose of the rst input matrix to the result matrix: l m k R = m l R + m n AT n l B
The result matrix Rml may be stored in a part of array r. If k < m the result is erroneous. See also RABT and SAB.
(k m)
Triple product R = AT B A CALL RATBA( a, n, m, b, r, w, k ) in in out spc in in dbl dbl dbl dbl int int a(n,m) b(n,n) r(k,m) w(m) n m k Anm . Bnn . Result Rmm . Workspace. Dimensions n and m. Column size k of array r.
Fortran
Subroutine RATBA performs a triple matrix multiplication: m m k 0 The result of the triple matrix product is returned in the rst m rows of array r(k,m). If k > m then the remainder is lled with zeros. In practice k = m in many cases. If k < m the result is erroneous. See also SATBA. Triple product R = R + AT B A CALL SATBA( a, n, m, b, r, w, k ) in in in out spc in in dbl dbl dbl dbl dbl int int a(n,m) b(n,n) r(k,m) r(k,m) w(n) n m k Anm . Bnn . Input Rmm . Added result Rmm . Workspace. Dimensions n and m. Column size k of array r. Fortran R = m n AT n n B n m A
(k m)
(k m)
The input and output matrix Rmm may be stored in a part of array r. The result of the triple matrix product is superposed to the rst m rows of this array. If k < m the result is erroneous. See also RATBA.
Diana-9.4 Users Manual Analysis Procedures (XI)
583
Invert 2 2 matrix CALL INV2( a, r, det ) in dbl a(2,2) out dbl r(2,2) out dbl det A22 . Result R22 = A1 . 22 Determinant det A.
Subroutine INV2 returns the inverse and the determinant of a 2 2 matrix A. If A is not positive denite, the inverse cannot be calculated and a fatal error occurs. Positive deniteness of A may be checked beforehand by subroutine DET2. See also INV3 and INVMTX. Invert 3 3 matrix CALL INV3( a, r, det ) in dbl a(3,3) out dbl r(3,3) out dbl det A33 . Result R33 = A1 . 33 Determinant det A. Fortran
Subroutine INV3 returns the inverse and the determinant of a 3 3 matrix A. If A is not positive denite, the inverse cannot be calculated and a fatal error occurs. Positive deniteness of A may be checked beforehand by subroutine DET3. See also INV2 and INVMTX. Invert matrix by partial pivoting CALL INVMTX( n, a, ai, dodet, det, silent, errflg in in out in out in out int dbl dbl log dbl log log n a(n,n) ai(n,n) dodet det silent errflg Dimension n. Ann . A1 . nn Compute determinant? Determinant of A. Work silently? Error encountered? ) Fortran
Subroutine INVMTX calculates the inverse of a matrix A by a partial pivoting algorithm: n n A errflg = .FALSE. n n A1
584
Using User-supplied Subroutines If the input A is (nearly) singular, the inverse cannot be calculated. In this case errflag is returned as .TRUE. and A1 is undetermined. The value of the determinant is returned in det only if dodet is input as .TRUE. and errflag is returned as .FALSE.. If silent is input as .TRUE. then no error messages will be given. In this case only errflag will indicate whether or not the inversion has succeeded. See also INV2, INV3, and INVPGB. Invert matrix by GaussJordan CALL INVPGB( a, n, w, v, eps, err ) in out in in spc spc out dbl dbl dbl int int dbl int a(n,n) a(n,n) eps n w(n) v(n) err Ann . A1 . nn Tolerance for pivot. Dimension n. Integer workspace. Real workspace. Error code. Fortran
Subroutine INVPGB calculates the inverse of a matrix Ann by a GaussJordan algorithm with partial pivoting: n n A err =0 n n A1
The inverse A1 is returned in array a which means that the original matrix is lost. The error code returns the status of the inversion: err = 0 indicates successful inversion, err = 1 indicates singularity. See also INVMTX. A.4.1.4 Decomposition Fortran Ann . Dimension n. Stretch matrix Snn . Rotation matrix Qnn . Square root of S. Singular values of A. Job information. Work space.
Polar decomposition CALL DGEPD( n, a, s, q, r, d, w, info ) in in out out out out out spc dbl int dbl dbl dbl dbl int dbl a(n,n) n s(n,n) q(n,n) r(n,n) d(n) info w(n)
Subroutine DGEPD performs a polar decomposition of the matrix A according to Strang [55]: Ann = Qnn Snn If info = 0 then the polar decomposition has been successful.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (XI)
585
Determinant of 2 2 matrix det = DET2( a ) in dbl a(2,2) out dbl det A22 . Determinant det A.
Subroutine DET2 returns the determinant of a 2 2 matrix A. See also DET3 and INVMTX. Determinant of 3 3 matrix det = DET3( a ) in dbl a(3,3) out dbl det A33 . Determinant det A. Fortran
Subroutine DET3 returns the determinant of a 3 3 matrix A. See also DET2 and INVMTX. A.4.1.6 Miscellaneous Fortran
Clear matrix CALL CLEAR( e, a, n, m ) in in out in dbl dbl dbl int e a(n,m) a(n,m) n,m Round-o value . Anm . Cleared Anm . Dimensions n and m.
Subroutine CLEAR clears a matrix Anm . It sets the elements of A to zero if their absolute value is : if |A(i, j)| then A(i, j) = 0
Fill symmetric matrix CALL FILMA( kod, a, n ) in in out in int dbl dbl int kod a(n,n) a(n,n) n Fill code. One half of Ann . Filled Ann . Dimension n.
Fortran
586
Subroutine FILMA lls the empty half of a symmetric matrix Ann : n Au n kod n >0 A n n Al
n kod n <0 A
The value of kod determines which half is to be lled. If kod > 0 the lower half is made identical to the upper, if kod < 0 the upper identical to the lower. If kod = 0 nothing is done. Set unity matrix CALL UNITMX( r, n ) out dbl r(n,n) in int n Unity matrix Inn . Dimension n. Fortran
A.4.2
Vector Manipulation
Dianas Vector library oers a set of subroutines to perform vector manipulations such as superposition, scaling, inner- and outer-products. In this library we use the following naming convention: U-----V-----P----M-----W-----SA.4.2.1 Result is vector u. Input is vector u. Addition (plus). Subtraction (minus). Result is vector w. Input is scalar S. Superposition Fortran
Add vectors u = v + w CALL UVPW( v, w, n, u ) in in out in dbl dbl dbl int v(n) w(n) u(n) n vn . wn . Result un . Dimension n.
587
Subroutine UVPW returns the sum of two column vectors v and w in a third column vector u: n u = n v + n w
See also UVMW and UVPWS. Subtract vectors u = v w CALL UVMW( v, w, n, u ) in in out in dbl dbl dbl int v(n) w(n) u(n) n vn . wn . Result un . Dimension n. Fortran
Suboutine UVMW subtracts a column vector w from a column vector v and returns the result in a column vector u: n u = n v n w
See also UVPW. Add scaled vector u = v + S w CALL UVPWS( v, w, n, s, u ) in in in out in dbl dbl dbl dbl int v(n) w(n) s u(n) n vn . wn . Scale factor S. Result un . Dimension n. Fortran
Subroutine UVPWS adds a scaled vector w to a column vector v and returns the result in a column vector u: n u = n v + S n w
Scale vector u = S v CALL UVS( v, n, s, u ) in in out in dbl dbl dbl int v(n) s u(n) n vn . Scale factor S. Result un . Dimension n.
Fortran
Simple inner-product P = uT v p = UV( u, v, n ) in in in out dbl dbl int dbl u(n) v(n) n p un . vn . Dimension n. Inner-product P .
See also PRODIN and VINPRO. Inner-product of row vectors P = A(i,) BT (j,) p = VINPRO( a, m, b, k, n ) in in in out dbl dbl int dbl a(n) b(n) m n k p Row i of Amn . Row j of Bkn . Dimensions m, n and k. Inner-product P .
Diana-9.4 Users Manual Analysis Procedures (XI)
Fortran
589
n k B
= m
i
This example returns the inner-product of the third row of AMAT and the second row of BMAT. See also UV. A.4.2.4 Outer-products Fortran
Outer product in 3-D space u = v w CALL OUTER( v, w, u ) in dbl v(3) in dbl w(3) out dbl u(3) v3 . w3 . Result u3 .
Subroutine OUTER calculates the outer product of two vectors v and w in threedimensional space: u w A v The result vector u is the vector perpendicular to v and w. The length of u is equal to the area A of the parallelogram spanned by v and w. See also AREA.
Area spanned in 3-D space A = v3 w3 a = AREA( v, w ) in dbl v(3) in dbl w(3) out dbl a v3 . w3 . Area A.
L2
Fortran
590
Subroutine AREA calculates the area A spanned by two vectors v and w in three-dimensional space: w A v It is assumed that the two vectors v and w have the same origin. See also OUTER. Arrow between two points u = p2 p1 CALL ARROW( p1, p2, u ) in dbl p1(3) in dbl p2(3) out dbl u(3) Coordinates x, y, z of point P1 . Coordinates x, y, z of point P2 . Vector u3 from P1 to P2 . Fortran
Subroutine ARROW returns the vector u between two points P1 and P2 in threedimensional space: p1 p2 P1 u P2
The vector is determined by subtraction of two vectors p1 and p2 . See also UVMW. A.4.2.5 Norms Fortran
Normalize 3-D vector CALL NORM( v, len ) in dbl v(3) out dbl v(3) out dbl len v3 . Normalized v3 . L2 -norm of input v3 .
Subroutine NORM normalizes a vector v in three-dimensional space to a length equal to one. If v is a null-vector, it cannot be normalized and NORM gives a fatal error. To prevent this, NULLVC should be called prior to NORM whenever v may be a null-vector.
591
C...
C...
See also NULLVC and NORM2. Check for 3-D null-vector isnull = NULLVC( v ) in dbl v(3) out log isnull v3 . Indicates if v3 is a null-vector. Fortran
Subroutine NULLVC checks if a vector in three-dimensional space is a null-vector. See also NORM. Normalize 2-D vector CALL NORM2( v, len ) in dbl v(2) out dbl v(2) out dbl len v2 . Normalized v2 . L2 -norm of input v2 . Fortran
Subroutine NORM2 normalizes a vector v in two-dimensional space to a length equal to one. If v is a null-vector, it cannot be normalized and NORM2 gives a fatal error. See also NORM.
A.4.3
Miscellaneous
This library contains a set of subroutines to perform various tasks like printing of values, initialization of arrays, error handling etc. A.4.3.1 Printing
This section describes subroutines to print values on Dianas standard output le.
592 Print real value CALL PRIVAL( val, label ) in in dbl val cha label*6
Subroutine PRIVAL prints a real value v preceded by a label. The value is printed in E12.3 format. See also PRIIVL, PRILVL and PRIVEC.
file .f DOUBLE PRECISON TEMP CALL PRIVAL( TEMP, TEMPER )
file .out
TEMPER= -2.730E+02
Print real vector CALL PRIVEC( v, n, label ) in in in dbl v(n) int n cha label*6 Real vector vn to print. Dimension n. Label to print.
Fortran
Subroutine PRIVEC prints a real vector n preceded by a label. The values are printed in 10E12.3 format. If n > 10 the vector is split in lines of at most ten values. See also PRIVAL, PRIIVL and PRILVL.
file .f DOUBLE PRECISON TEMP(4) CALL PRIVEC( TEMP, 4, TEMPER )
file .out
TEMPER: -2.730E+02 -1.710E+02 0.000E+00 1.000E+02
A.4 Programmers Service Libraries Print real matrix CALL PRIMAT( a, n, m, label ) in in in dbl a(n,m) int n m cha label*6 Real matrix Anm to print. Dimensions n and m. Label to print. Fortran
593
Subroutine PRIMAT prints a real matrix A preceded by a label. The values are printed in 10E12.3 format with row and column indices aside and above. If m > 10 the rows are split in lines of at most ten values. See also PRIVEC and MATPRI.
file .f DOUBLE PRECISON SE(4,4) CALL PRIMAT( SE, 4, 4, SE-MAT )
file .out
SE-MAT : 1 2 3 4 1 3.333E+04 8.333E+03 8.333E+03 0.000E+00 2 8.333E+03 3.333E+04 8.333E+03 0.000E+00 3 8.333E+03 8.333E+03 3.333E+04 0.000E+00 4 0.000E+00 0.000E+00 0.000E+00 1.250E+04
Print integer value CALL PRIIVL( val, label ) in in int val cha label*6 Integer value v to print. Label to print.
Fortran
Routine PRIIVL prints an integer value v preceded by a label. The value is printed in I6 format. See also PRIVAL, PRILVL and PRIIVC.
file .f INTEGER KODE CALL PRIIVL( KODE, KODE )
file .out
KODE = -273
594 Print integer vector CALL PRIIVC( v, n, label ) in in in int v(n) int n cha label*6
Subroutine PRIIVC prints an integer vector n preceded by a label. The values are printed in 20I6 format. If n > 20 the vector is split in lines of at most twenty values. See also PRIVEC and PRIIVL.
file .f INTEGER V(6) CALL PRIIVC( V, 6, VALUES )
file .out
VALUES: 101 202 303 404 505 606
Print integer matrix CALL MATPRI( a, n, m, label ) in in in int a(n,m) int n m cha label*6 Integer matrix Anm to print. Dimensions n and m. Label to print.
Fortran
Subroutine MATPRI prints an integer matrix A preceded by a label. The values are printed in 20I5 format with row and column indices aside and above. If m > 20 the rows are split in lines of at most twenty values. See also PRIIVC and PRIMAT.
file .f INTEGER KP(3,4) CALL MATPRI( KP, 3, 4, KP-MAT )
file .out
KP-MAT 1 2 3 1 2 3 333 333 333 4 444 444 444
A.4 Programmers Service Libraries Print logical value CALL PRILVL( val, label ) in in log val cha label*6 Logical value v to print. Label to print. Fortran
595
Subroutine PRILVL prints a logical value v preceded by a label. The value is printed in L3 format. See also PRIVAL, PRIIVL and PRILVC.
file .f LOGICAL DONE CALL PRILVL( DONE, DONEIT )
file .out
DONEIT= T
Print logical vector CALL PRILVC( v, n, label ) in in in log v(n) int n cha label*6 Logical vector vn to print. Dimension n. Label to print.
Fortran
Subroutine PRILVC prints a logical vector v preceded by a label. The values are printed in 40L3 format. If n > 40 the vector is split in lines of at most forty values. See also PRIVEC and PRIIVL.
file .f LOGICAL FL(6) CALL PRILVC( FL, 6, FLAGS )
file .out
FLAGS : T F F T T F
A.4.3.2
Initializing
This section describes subroutines to initialize various types of arrays and values. Especially for large arrays, these subroutines use less processing time than ordinary Fortran DO-loops.
Diana-9.4 Users Manual Analysis Procedures (XI) December 15, 2009 First ed.
596 Initialize real array CALL RSET( s, u, n ) in dbl s out dbl u(n) in int n
Subroutine RSET initializes the rst n elements of real array u to the value s. If n 0 the call to RSET has no eect. See also ISET and LSET. Initialize integer array CALL ISET( s, u, n ) in int s out int u(n) in int n Value s. Initialized array u(n). Number of array elements n to be initialized. Fortran
Subroutine ISET initializes the rst n elements of integer array u to the value s. If n 0 the call to ISET has no eect. See also RSET and LSET. Initialize logical array CALL LSET( s, u, n ) in log s out log u(n) in int n Value s. Initialized array u(n). Number of array elements n to be initialized. Fortran
Subroutine LSET initializes the rst n elements of logical array u to the value s. If n 0 the call to LSET has no eect. See also RSET and ISET. Move (copy) real array CALL RMOVE( v, u, n ) in dbl v(n) out dbl u(n) in int n Array v(n) to be moved. Array u(n) to receive. Number of array elements n to be moved. Fortran
Subroutine RMOVE copies the rst n elements of a real array v to the rst n elements of a real array u:
1
v
1
The input array v is left unchanged. If the arrays v and u overlap, the result is undened. See also IMOVE and LMOVE.
December 15, 2009 First ed. Diana-9.4 Users Manual Analysis Procedures (XI)
A.4 Programmers Service Libraries Move (copy) integer array CALL IMOVE( v, u, n ) in int v(n) out int u(n) in int n Array v(n) to be moved. Array u(n) to receive. Number of array elements n to be moved. Fortran
597
Subroutine IMOVE copies the rst n elements of an integer array v to the rst n values of an integer array u:
1
v
1
Routine IMOVE copies the rst n elements of an integer array v to the rst n values of an integer array u: The input array v is left unchanged. If the arrays v and u overlap, the result is undened. See also RMOVE and LMOVE. Move (copy) logical array CALL LMOVE( v, u, n ) in log v(n) out log u(n) in int n Array v(n) to be moved. Array u(n) to receive. Number of array elements n to be moved. Fortran
Subroutine LMOVE copies the rst n elements of a logical array v to the rst n elements of a logical array u:
1
v
1
The input array v is left unchanged. If the arrays v and u overlap, the result is undened. See also RMOVE and IMOVE. Random real CALL DSRAND( iseed ) ranvl = DRRAND( ) in int iseed out dbl ranvl Initializer. Random value. Fortran
Subroutine DRRAND returns a random real value in the range 0 to 1. To get a reproducable series of random values, the function value must be initialized by a call to DSRAND.
Diana-9.4 Users Manual Analysis Procedures (XI) December 15, 2009 First ed.
598
file .out
RANVL : 1.333E-01 3.599E-01 8.986E-01 7.494E-01 9.123E-01
Get machine parameters par = DPMPAR( kod ) in dbl kod out dbl par Indicator. Machine parameter.
Fortran
Subroutine DPMPAR provides double precision machine parameters. It returns a machine parameter depending on the value of kod: the machine precision if kod = 1, the lowest double precision value if kod = 2 or the highest double precision value if kod = 3.
file .f DOUBLE PRECISION DPMPAR CALL PRIVAL( DPMPAR( 1 ), DBLEPS ) CALL PRIVAL( DPMPAR( 2 ), DBLMIN ) CALL PRIVAL( DPMPAR( 3 ), DBLMAX )
file .out
DBLEPS= DBLMIN= DBLMAX= 2.500E-16 5.000-308 1.500+308
599
One-dimensional interpolation CALL INTER1( x, vx, wz, l, z, dzdx ) in in in out out in dbl dbl dbl dbl dbl int x vx(l) wz(l) z dzdx l x-coordinate. Array v(l) containing the x-values. Array w(l) containing the z-values. Interpolated value z = f (x). Gradient df (x)/dx. Dimension l.
Subroutine INTER1 interpolates in a piecewise linear one-dimensional diagram. See also INTER2. Two-dimensional interpolation CALL INTER2( x, y, ux, vy, wz, l, m, z, grad ) in in in in in out out in dbl dbl dbl dbl dbl dbl dbl int x y ux(l) uy(m) wz(l,m) z grad(2) l m x-coordinate. y-coordinate. Array v(l) containing the x-values. Array v(m) containing the y-values. Array w(l m) containing the z-values. Interpolated value z = f (x, y). Gradient . Dimensions l and m. Fortran
Subroutine INTER2 interpolates in a piecewise linear two-dimensional diagram. See also INTER1. A.4.3.4 Sorting Fortran
Sort real array CALL SORTR( kod, u, n ) in in in out int dbl int dbl kod u(n) n u(n) Indicates sort direction. Array u(n) to sort. Dimension n. Sorted array.
Subroutine SORTR sorts the rst n elements of array u. Argument kod indicates the sort direction: descending (high to low) if kod = 1, ascending (low to high) if kod + 1. See also SORTRI and SORTI.
600
file .out
VL-ASC: VL-DES: -4.200E-01 1.240E+02 6.300E-01 1.000E+01 1.000E+01 6.300E-01 1.240E+02 -4.200E-01
Sort real array with indices CALL SORTRI( kod, u, n, ind ) in in in out out int dbl int dbl int kod u(n) n u(n) ind(n) Indicates sort direction. Array u(n) to sort. Dimension n. Sorted array. Original positions in u(n).
Fortran
Subroutine SORTRI sorts the rst n elements of array u. Argument kod indicates the sort direction: descending (high to low) if kod = 1, ascending (low to high) if kod + 1. The original index of each sorted array element is returned in ind such that ind(i) contains the original index in u of u(i). See also SORTR and SORTII.
file .f DOUBLE PRECISION VALUE(4) INTEGER IND(4) DATA VALUE / 0.6300D0, 0.1240D3, -0.4200D0, 0.1000D2 / CALL PRIVEC( VALUE, 4, VL-ORG ) CALL SORTRI( +1, VALUE, 4, IND ) CALL PRIVEC( VALUE, 4, VL-ASC ) CALL PRIIVC( IND, 4, INDEX )
file .out
VL-ORG: VL-ASC: INDEX : 6.300E-01 -4.200E-01 3 1 1.240E+02 6.300E-01 4 2 -4.200E-01 1.000E+01 1.000E+01 1.240E+02
A.4 Programmers Service Libraries Sort integer array CALL SORTI( kod, u, n ) in in in out int int int int kod u(n) n u(n) Indicates sort direction. Array u(n) to sort. Dimension n. Sorted array. Fortran
601
Subroutine SORTI sorts the rst n elements of array u. Argument kod indicates the sort direction: descending (high to low) if kod = 1, ascending (low to high) if kod + 1. See also SORTII and SORTR.
file .f INTEGER VALUE(4) DATA VALUE / 63, 124, -42, 10 / CALL SORTI( +1, VALUE, 4 ) CALL PRIIVC( VALUE, 4, VL-ASC ) CALL SORTI( -1, VALUE, 4 ) CALL PRIIVC( VALUE, 4, VL-DES )
file .out
VL-ASC: VL-DES: -42 124 10 63 63 10 124 -42
Sort integer array with indices CALL SORTII( kod, u, n, ind ) in in in out out int int int int int kod u(n) n u(n) ind(n) Indicates sort direction. Array u(n) to sort. Dimension n. Sorted array. Original positions in u(n).
Fortran
Subroutine SORTII sorts the rst n elements of array u. Argument kod indicates the sort direction: descending (high to low) if kod = 1, ascending (low to high) if kod + 1. The original index of each sorted array element is returned in ind such that ind(i) contains the original index in u of u(i). See also SORTI and SORTRI.
602
file .out
VL-ORG: VL-ASC: INDEX : 63 -42 3 124 10 4 -42 63 1 10 124 2
A.4.3.5
Error Handling
If the control of execution runs in an erroneous situation, for instance in the ELSE branch of an IF-THEN-ELSE structure, it is convenient and good programming practice to call Dianas error handling subroutine PRGERR instead of a Fortran STOP statement. Abort in error situation CALL PRGERR( usnam, errnr ) in in cha usnam int errnr Subroutine name. Error number. Fortran
Subroutine PRGERR writes the subroutine name and the error number on Dianas standard output le, closes the filos le and aborts the executing of the current job.
file .f SUBROUTINE USRMAT( ... ) INTEGER NS C... C... IF ( NS .EQ. 3 ) THEN do something ELSE PRINT *, NUMBER OF STRESS COMPONENTS NOT CORRECT CALL PRGERR( USRMAT, 27 ) END IF
603
A.4.4
Database I/O
Dianas Database library oers a set of subroutines to access the data in the filos le, for instance getting data records from the various Data Structures. A.4.4.1 Access Fortran
Get record, check length CALL GTC( path, gval, len ) in cha path*(*) out gen gval(len) in int len Pathname. Data values. Expected length of data item.
Subroutine GTC checks the length of data item path and reads the data if the expected length is equal to the actual length. If the expected length is not equal to the actual length on the database, a fatal error message occurs. The type of output array gval is generic: it depends on the type of the data.
file .f DOUBLE PRECISION YOUNG, B(3,8) INTEGER USRIND(2) CALL GTC( ../MATERI/YOUNG, YOUNG, 1 ) CALL GTC( B, B, 24 ) CALL GTC( USRIND, USRIND, 2 )
604
Appendix B
BC3AG Axisymmetric groundwater ow, boundary line, 3 nodes, quadratic. BC3AHT Axisymmetric potential ow, boundary line, 3 nodes, quadratic. BC3GW Groundwater ow, boundary line, 3 nodes, quadratic. BC3HT Potential ow, boundary line, 3 nodes, quadratic. BCL6S2 Fluidstructure line interface, 5 nodes, quadratic-linear. BCL6S3 Fluidstructure line interface, 6 nodes, quadratic. BCQ8GW Groundwater ow, boundary quadrilateral, 8 nodes, quadratic. BCQ8HT Potential ow, boundary quadrilateral, 8 nodes, quadratic. BCT6GW Groundwater ow, boundary triangle, 6 nodes, quadratic. BCT6HT Potential ow, boundary triangle, 6 nodes, quadratic. BQ24S4 Fluidstructure quadrilateral
CHX69I Solid-line interface, 3-D, 23 nodes, quadratic. CHX96 CL10T Solid brick, 32 nodes, cubic. Curved truss bar, 2-D, 5 nodes,
606
quartic. CL12B CL12I CL12T CL15B CL15T CL18B CL18I CL20I CL24B CL24I CL30B CL32I Curved beam, 2-D, 4 nodes, degenerated cubic. Line interface, 2-D, 6 nodes, quadratic. Curved truss bar, 3-D, 4 nodes, cubic. Curved beam, 2-D, 5 nodes, degenerated quartic. Curved truss bar, 3-D, 5 nodes, quartic. Curved beam, 3 nodes, 3-D, quadratic. Curved line interface, 6 nodes, quadratic, linesolid connection. Curved line interface, 10 nodes, quartic. Curved beam, 4 nodes, 3-D, cubic. Line interface, to shell, 6 nodes, quadratic. Curved beam, 5 nodes, 3-D, quartic. Line interface, to shell, 8 nodes, cubic. Line contact interface, 2-D, 3 nodes. Line bounding, 3 nodes, quadratic, 2-D.
CQ18M Quadrilateral plane stress, 9 nodes, quadratic, Lagrange. CQ20A CQ20E CQ22A CQ22E CQ24C Quadrilateral axisymmetric, 8 nodes, quadratic, hyperelastic. Quadrilateral plane strain, 8 nodes, quadratic, hyperelastic. Quadrilateral axisymmetric, 9 nodes, quadratic, hyperelastic. Quadrilateral plane strain, 9 nodes, quadratic, hyperelastic. Quadrilateral contact interface, 3-D, 8 nodes.
CQ24GE Quadrilateral complete plane strain, 8 nodes, quadratic. CQ24GM Quadrilateral plane stress, 3-D, 8 nodes, quadratic. CQ24P CQ24T Quadrilateral plate bending, 8 nodes, quadratic, Mindlin. Quadrilateral bounding, 8 nodes, quadratic, 3-D.
CQ24TM Quadrilateral curved plane mass, 3-D, 8 nodes, quadratic. CQ36GE Quadrilateral complete plane strain, 12 nodes, cubic. CQ36T CQ40F CQ40L CQ40S CQ48F CQ48I CQ60S Quadrilateral bounding, 12 nodes, cubic, 3-D. Quadrilateral at shell, 8 nodes, quadratic, Mindlin. Quadrilateral curved shell, 8 nodes, quadratic, layered. Quadrilateral curved shell, 8 nodes, quadratic. Quadrilateral at shell, 8 nodes, quadratic, Mindlin + z d.o.f. Quadrilateral interface, 3-D, 16 nodes, quadratic. Quadrilateral curved shell, 12 nodes, cubic.
CL6TM Curved line mass, 2-D, 3 nodes, quadratic. CL6TR Curved truss bar, 2-D, 3 nodes, quadratic. CL8TR Curved truss bar, 2-D, 4 nodes, cubic. CL9AX Axisymmetric shell, 3 nodes, quadratic. CL9BE CL9PE Curved beam, 3 nodes, 2-D, quadratic. Innite plane strain shell, 3 nodes, quadratic.
CL9TR Curved truss bar, 3-D, 3 nodes, quadratic. CQ12C CQ16A CQ16E Quadrilateral base for composed solid, 12 nodes. Quadrilateral axisymmetric, 8 nodes, quadratic. Quadrilateral plane strain, 8 nodes, quadratic.
CQ8AG Axisymmetric groundwater ow, quadrilateral, 8 nodes, quadratic. CQ8AHT Axisymmetric potential ow, quadrilateral, 8 nodes, quadratic. CQ8CM Quadrilateral base for composed solid, 8 nodes. CQ8GW Groundwater ow, quadrilateral, 8 nodes, quadratic.
607
CQ8HT Potential ow, quadrilateral, 8 nodes, quadratic. CQ8KD Layered groundwater ow, quadrilateral, 8 nodes, quadratic. CQ8RE Reynolds ow, quadrilateral, 8 nodes, quadratic. CQ8TO Cross-section torsion, quadrilateral, 8 nodes, quadratic. CT12A CT12E Triangular axisymmetric, 6 nodes, quadratic. Triangular plane strain, 6 nodes, quadratic. CT6AHT Axisymmetric potential ow, triangle, 6 nodes, quadratic. CT6CM Triangular base for composed solid, 6 nodes. CT6GW Groundwater ow, triangle, 6 nodes, quadratic. CT6HT Potential ow, triangle, 6 nodes, quadratic. CT6KD Layered groundwater ow, triangle, 6 nodes, quadratic. CT6RE Reynolds ow, triangle, 6 nodes, quadratic. CT6TO Cross-section torsion, triangle, 6 nodes, quadratic. CT9CM Triangular base for composed solid, 9 nodes. CTE10G Groundwater ow, 3-D, pyramid, 10 nodes, quadratic. CTE10H Potential ow, 3-D, pyramid, 10 nodes, quadratic. CTE30 Solid pyramid, 10 nodes, quadratic. CTE33I Solid-point interface, 3-D, 11 nodes, quadratic. CTE39I Solid-line interface, 3-D, 13 nodes, quadratic. CTE48 Solid pyramid, 16 nodes, cubic. CTP15G Groundwater ow, 3-D, wedge, 15 nodes, quadratic. CTP15H Potential ow, 3-D, triangular prism (wedge), 15 nodes, quadratic. CTP45 Solid wedge, 15 nodes, quadratic. CTP48I Solid-point interface, 3-D, 16 nodes, quadratic. CTP54I Solid-line interface, 3-D, 18 nodes, quadratic. CTP72 HX24L HX25L Solid wedge, 24 nodes, cubic. Solid brick, 8 nodes, linear. Solid brick, 8 nodes, linear, hyperelastic.
CT12M Triangular plane stress, 6 nodes, quadratic. CT12O CT18C Triangular plane stress, 6 nodes, quadratic, orthotropic. Triangular contact interface, 3-D, 6 nodes.
CT18GE Triangular complete plane strain, 6 nodes, quadratic. CT18GM Triangular plane stress, 3-D, 6 nodes, quadratic. CT18P CT18T Triangular plate bending, 6 nodes, quadratic, Mindlin. Triangular bounding, 6 nodes, quadratic, 3-D.
CT18TM Triangular curved plane mass, 3-D, 6 nodes, quadratic. CT27GE Triangular complete plane strain, 9 nodes, cubic. CT27T CT30A CT30E CT30F CT30L CT30S CT36F CT36I CT45S Triangular bounding, 9 nodes, cubic, 3-D. Triangular axisymmetric, 15 nodes, quartic, Lagrange. Triangular plane strain, 15 nodes, quartic, Lagrange, hyperelastic. Triangular at shell, 6 nodes, quadratic, Mindlin. Triangular curved shell, 6 nodes, quadratic, layered. Triangular curved shell, 6 nodes, quadratic. Triangular at shell, 6 nodes, quadratic, Mindlin + z d.o.f. Triangular interface, 3-D, 12 nodes, quadratic. Triangular curved shell, 9 nodes, cubic.
HX27IF Solid-point interface, 3-D, 9 nodes, linear. HX30IF Solid-line interface, 3-D, 10 nodes, linear. HX8GW Groundwater ow, 3-D, brick, 8 nodes, linear. HX8HT Potential ow, 3-D, brick, 8 nodes, linear. ICL6H Potential ow, line interface, 6 nodes, quadratic.
608
ICQ16H Potential ow, quadrilateral interface, 16 nodes, quadratic. ICT12H Potential ow, triangular interface, 12 nodes, quadratic. IL4HT IPT2H IQ8HT IT6HT L12BE L12IF L13BE L16IF L20IF L2HT L4CT L4HT L4TB L4TM Potential ow, line interface, 4 nodes, linear. Potential ow, point interface, 2 nodes. Potential ow, quadrilateral interface, 8 nodes, linear. Potential ow, triangular interface, 6 nodes, malinear. Bending beam, 2 nodes, 3-D, Timoshenko or Bernoulli. Line interface, 4 nodes, linear, linesolid connection. Bending beam, 2 nodes, 3-D, isoparametric. Line interface, to shell, 4 nodes, linear. Line interface, to shell, 3+2 nodes, quadratic/linear. Cooling pipe, 2 nodes, linear. Line contact interface, 2-D, 2 nodes. Cooling pipe, 4 nodes, linear, nonsymmetric. Line bounding, 2 nodes, linear, 2-D. Line mass, 2-D, 2 nodes, linear.
Q12GME Quadrilateral plane stress, 3-D, 4 nodes, linear. Q12ME Quadrilateral plane stress, 4 nodes, linear, drilling d.o.f. Q12PL Q12TB Quadrilateral plate bending, 4 nodes, linear, Mindlin. Quadrilateral bounding, 4 nodes, linear, 3-D.
Q12TM Quadrilateral plane mass, 3-D, 4 nodes, linear. Q20SF Q20SH Q24IF Q24SF Quadrilateral at shell, 4 nodes, linear, Mindlin. Quadrilateral curved shell, 4 nodes, linear. Quadrilateral interface, 3-D, 8 nodes, linear. Quadrilateral at shell, 4 nodes, linear, Mindlin + z d.o.f.
Q48SPL Rectangular spline (strip), 8 nodes, 3 sections. Q4AGW Axisymmetric groundwater ow, quadrilateral, 4 nodes, linear. Q4AHT Axisymmetric potential ow, quadrilateral, 4 nodes, linear. Q4CMP Quadrilateral base for composed solid, 4 nodes. Q4GW Q4HT Q4KD Q4RE Q4TO Groundwater ow, quadrilateral, 4 nodes, linear. Potential ow, quadrilateral, 4 nodes, linear. Layered groundwater ow, quadrilateral, 4 nodes, linear. Reynolds ow, quadrilateral, 4 nodes, linear. Cross-section torsion, quadrilateral, 4 nodes, linear.
L4TRU Truss bar, 2 nodes, 2-D geometrically nonlinear. L6AXI L6BEN L6PE Axisymmetric shell, 2 nodes, linear. Bending beam, 2 nodes, 2-D, Timoshenko or Bernoulli. Innite plane strain shell, 2 nodes, linear.
L6TRU Truss bar, 2 nodes, 3-D geometrically nonlinear. L7BEN L8IF N4IF N6IF N6SPR Bending beam, 2 nodes, 2-D, isoparametric. Line interface, 2-D, 4 nodes, linear. Node interface, 2-D, 2 nodes, linear. Node interface, 3-D, 2 nodes, linear. Matrix spring, 3-D, 1 node.
Q56SPL Rectangular spline (strip), 10 nodes, 4 sections. Q8AXI Q8EPS Quadrilateral axisymmetric, 4 nodes, linear. Quadrilateral plane strain, 4 nodes, linear.
Q8MEM Quadrilateral plane stress, 4 nodes, linear. Q8OME Quadrilateral plane stress, 4 nodes,
609
linear, orthotropic geometry. SP12BA Base spring, 2 nodes, 3-D. SP1RO SP1TR SP2RO SP2TR SP6BA T15SF T15SH T18IF T18SF Rotation spring/dashpot, 1 node. Translation spring/dashpot, 1 node. Rotation spring/dashpot, 2 nodes. Translation spring/dashpot, 2 nodes. Base spring, 2 nodes, 2-D. Triangular at shell, 3 nodes, linear, Mindlin. Triangular curved shell, 3 nodes, linear. Triangular interface, 3-D, 6 nodes, linear. Triangular at shell, 3 nodes, linear, Mindlin + z d.o.f. linear. T6MEM Triangular plane stress, 3 nodes, linear. T6OME Triangular plane stress, 3 nodes, linear, orthotropic geometry. T9CT Triangular contact interface, 3-D, 3 nodes.
T9GME Triangular plane stress, 3-D, 3 nodes, linear. T9MEM Triangular plane stress, 3 nodes, linear, drilling d.o.f. T9PLA T9TB T9TM Triangular plate bending, 3 nodes, linear, Kirchho. Triangular bounding, 3 nodes, linear, 3-D. Triangular plane mass, 3-D, 3 nodes, linear.
T3AGW Axisymmetric groundwater ow, triangle, 3 nodes, linear. T3AHT Axisymmetric potential ow, triangle, 3 nodes, linear. T3CMP Triangular base for composed solid, 3 nodes. T3GW T3HT T3KD T3RE T3TO T6AXI T6EPS Groundwater ow, triangle, 3 nodes, linear. Potential ow, triangle, 3 nodes, linear. Layered groundwater ow, triangle, 3 nodes, linear. Reynolds ow, triangle, 3 nodes, linear. Cross-section torsion, triangle, 3 nodes, linear. Triangular axisymmetric, 3 nodes, linear. Triangular plane strain, 3 nodes,
T9WME Triangular plane stress, 3 nodes, nonlinear wrinkling. TE12L Solid pyramid, 4 nodes, linear. TE15IF Solid-point interface, 3-D, 5 nodes, linear. TE18IF Solid-line interface, 3-D, 6 nodes, linear. TE4GW Groundwater ow, 3-D, pyramid, 4 nodes, linear. TE4HT Potential ow, 3-D, pyramid, 4 nodes, linear. TP18L Solid wedge, 6 nodes, linear. TP21IF Solid-point interface, 3-D, 7 nodes, linear. TP24IF Solid-line interface, 3-D, 8 nodes, linear. TP6GW Groundwater ow, 3-D, wedge, 6 nodes, linear. TP6HT Potential ow, 3-D, wedge, 6 nodes, linear.
610
Appendix C
Element Library].
Mobile to specify physical properties required for mobile loads [ C.1]. Wind and Water to specify physical properties required for wind and water loads
C.1
Mobile
The Mobile aspect enables the specication of physical properties required for mobile loads [ 2.3.5 p. 42].
C.1.1
The Dutch VBB code concept enables the specication of geometrical parameters for mobile load according to the Dutch VBB code [ 2.3.5.4 p. 47].
1 See
612
Default position Width of carriage way . . . . . . . . . . . Length for bump and reduction . . Load reduction factor . . . . . . . . . . . . Bump factor . . . . . . . . . . . . . . . . . . . . Transverse position Width of carriage way . . . . . . . . . . . Distance between beams . . . . . . . . . Point i lorry inuence line . . . . . . . Factor for lorry forces . . . . . . . . . . . . Factor for distributed load . . . . . . . Length for bump and reduction . . Load reduction factor . . . . . . . . . . . . Bump factor . . . . . . . . . . . . . . . . . . . .
w l B S
w l b s
w b vi fF fq l B S
w b vi ff fq l b s
C.1.2
The Dutch VOSB code concept enables the specication of geometrical parameters for mobile load according to the Dutch VOSB code [ 2.3.5.4 p. 47].
Default position Width of carriage way . . . . . . . . . . . w Width of the lanes . . . . . . . . . . . . . . l Length for bump and reduction . . l Transverse position Width of carriage way . . . . . . . . . . . Width of the lanes . . . . . . . . . . . . . . Distance between beams . . . . . . . . . Factor for lorry forces . . . . . . . . . . . . Factor for distributed load . . . . . . . Length for bump and reduction . . BWIDTH w WLANE l LOALEN l [ 2.3.5.4 [ 2.3.5.4 [ 2.3.5.4 [ 2.3.5.4 [ 2.3.5.4 [ 2.3.5.4 [ 2.3.5.4 [ 2.3.5.4 [ 2.3.5.4 [ 2.3.5.4 [ 2.3.5.4 p. 47] p. 48] p. 48] p. 50] p. 48] p. 48] p. 48] p. 48] p. 49] p. 49] p. 50]
w l b fF fq l
w l b ff fq l
C.1.3
The European ENV code concept enables the specication of geometrical parameters for mobile load according to the European ENV 1991-3 code [ 2.3.5.5 p. 51].
Width of carriage way . . . . . . . . . . . w Factor for lorry forces . . . . . . . . . . . . fF Factor for distributed load . . . . . . . fq
C.1 Mobile
613
C.1.4
General
The General concept enables the explicit specication of geometrical parameters for mobile load [ 2.3.5.3 p. 45].
Width of carriage way . . . . . . . . . . . w Factor for lorry forces . . . . . . . . . . . . fF Factor for distributed load . . . . . . . fq BWIDTH w LODFAC ff DISFAC fq [ 2.3.5.3 [ 2.3.5.3 [ 2.3.5.3 p. 45] p. 46] p. 46]
614
Bibliography
[1] Aki, K. Generation and propagation of G waves from the Niigata earthquake of June 16, 1964: Part 2. Estimation of earthquake moment, released energy and stress drop from the G wave spectra. Tech. Rep. 44 (73-88), Bull Earthq. Res. Inst., Univ. of Tokyo, 1966. [2] ANSI. American National Standard Programming Language FORTRAN. Tech. Rep. ANSI.X3.9-1978, American National Standards Institute, 1978. [3] Barsoum, R. S. On the use of isoparametric nite elements in Linear Fracture Mechanics. Int. J. Num. Meth. Eng. 10 (1976), 2537. [4] Bathe, K.-J. Finite Element Procedures in Engineering Analysis. Prentice-Hall, 1982. [5] Bathe, K.-J. Finite Element Procedures. Prentice-Hall, 1996. [6] Becker, A. A. Understanding Non-Linear Finite Element Analysis. National Agency for Finite Element Methods & Standards (NAFEMS), Glasgow, 2001. [7] Bettes, P. B., and Zienkiewicz, O. C. Diraction and refraction of surface waves using nite and innite elements. Int. J. Num. Meth. Eng. 11 (1977), 12711290. [8] Biot, M. A. General theory of three-dimensional consolidation. J. Appl. Phys. 12, 2 (1941), 155164. [9] Byskov, E., and Hutchinson, J. W. Mode interaction in axially stiened cylindrical shells. AIAA Journal 15, 7 (1977), 941948. [10] Cardarelli, F. Scientic Unit Conversion. No. ISBN 3-540-76022-9. Springer-Verlag, London, 1997. [11] CEN. Eurocode 1: Basis of design and actions on structures Part 3: Trac loads on bridges. Tech. Rep. ENV 1991-3, European Committee for Standardization (CEN), 1995.
616
BIBLIOGRAPHY [12] Chan, T. F., and Mathew, T. P. Domain decomposition algorithms. Acta Numerica (1994), 61143. [13] Clough, R. W., and Penzien, J. Dynamics of Structures. McGraw-Hill, 1982. [14] Craig, R. Structural Dynamics. John Wiley & Sons, 1981. [15] Crisfield, M. A. Non-linear Finite Element Analysis of Solids and Structures, vol. 1: Essentials. John Wiley & Sons, 1991. [16] Ellsiepen, P., and Hartmann, S. Remarks on the interpretation of current non-linear nite element analyses as dierentialalgebraic equations. Int. J. Num. Meth. Eng. 51 (2001), 679707. [17] Fenves, G., and Chopra, A. K. Earthquake Analysis and Response of Concrete Gravity Dams. Tech. Rep. UCBEERC-8410. [18] Forde, B. W. R., and Stiemer, S. F. Improved Arc-length Orthogonality methods for nonlinear Finite Element Analysis. Computers & Structures 27, 5 (1987), 625630. [19] George, A., and Liu, J. W. Computer Solution of Large Sparse Positive Denite Systems. Prentice-Hall, 1981. [20] Geradin, M., and Rixen, D. Mechanical Vibrations Theory and Application to Structural Dynamics, 2nd ed. John Wiley & Sons, 1997. [21] Goodman, L. E., Rosenblueth, E., and Newmark, N. M. Aseismic design of elastic structures founded on rm ground. Proc. ASCE (November 1953), 349/1349/27. [22] Gupta, A. K. Response Spectrum Method In Seismic Analysis and Design of Structures. CRC Press, Inc., 1992. [23] Hairer, E., and Wanner, G. Solving Ordinary Dierential Equations II, 2nd ed. Springer-Verlag, 1996. [24] Hellen, T. K. On the method of virtual crack extension. Int. J. Num. Meth. Eng. 9 (1975), 187207. [25] Henshell, R. D., and Shaw, K. G. Crack tip nite elements are unnecessary. Int. J. Num. Meth. Eng. 9 (1975), 495507. [26] Hestenes, M. R., and Stiefel, E. Methods of conjugate gradients for solving linear systems. J. Res. Nat. Bur. Standards 49, 6 (1952), 409436. [27] Hilber, H. M., Hughes, T. J. R., and Taylor, R. L. Improved numerical dissipation for time integration algorithms in structural dynamics. Earthquake Engineering and Structural Dynamics 5 (1977), 283292.
BIBLIOGRAPHY [28] Hinton, E., Rock, T., and Zienkiewicz, O. C. A note on mass lumping and related processes in the Finite Element Method. Earthquake Engineering and Structural Dynamics 4 (1976), 245249. [29] Horvath, A. L. Conversion Tables of Units for Science and Engineering. No. ISBN 0-333-40857-8. Macmillan Reference Books, London, 1987. [30] Huetink, J. On the Simulation of Thermo-mechanical Forming Processes. PhD thesis, University of Twente, 1986. [31] Hughes, T. J. R. The Finite Element Method Linear Static and Dynamic Finite Element Analysis. Prentice-Hall, 1987. [32] Karypis, G., and Kumar, V. A fast and highly quality multilevel scheme for partitioning irregular graphs. SIAM J. Sc. Comp. 20, 1 (1999). [33] Kernighan, B. W., and Plauger, P. J. The Elements of Programming Style, 2nd ed. McGraw-Hill, 1978. [34] Koiter, W. T. Over de Stabiliteit van Elastisch Evenwicht. PhD thesis, Delft University of Technology, 1945. [35] Kucukarslan, S., Coskun, S. B., and Taskin, B. Transient analysis of dam-reservoir interaction including the reservoir bottom eects. Journal of Fluids and Structures 20 (2005), 10731084. [36] Lambe, T. W., and Whitman, R. V. Soil Mechanics. John Wiley & Sons, 1969. [37] Lehoucq, R. B., Sorensen, D. C., and Yang, C. ARPACK Users Guide: Solution of Large-Scale Eigenvalue Problems with Implicitly Restarted Arnoldi Methods. SIAM Series: Software, Environments, and Tools. SIAM: Society for Industrial and Applied Mathemathics, 1998. http://www.caam.rice.edu/software/ARPACK/UG/ug.html. [38] Li, Xikui, and Zienkiewicz, O. C. Multiphase ow in deforming porous media and nite element solutions. Computers & Structures 45, 2 (1992), 211227. [39] Meijerink, J. A., and van der Vorst, H. A. An iterative solution method for linear systems of which the coecient matrix is a symmetric M-matrix. Math. Comp. 31 (1977), 148162. [40] Meissner, L. P., and Organick, E. I. Fortran 77: Featuring Structured Programming. Addisson-Wesley, 1982. [41] Merks, P. J. G. Wapeningsgrootheden bij Plaatconstructies. Tech. Rep. BSW 88-05, Dutch Public Works, Rijswijk, The Netherlands, 1988. [42] NAFEMS. Guidelines to Finite Element Practice. National Agency for Finite Element Methods & Standards (NAFEMS), Glasgow, 1984.
Diana-9.4 Users Manual Analysis Procedures
617
618
BIBLIOGRAPHY [43] NAFEMS. The Standard NAFEMS Benchmarks. Tech. Rep. TNSB Rev. 3, National Agency for Finite Element Methods & Standards (NAFEMS), 1990. [44] NAFEMS. A Finite Element Primer. National Agency for Finite Element Methods & Standards (NAFEMS), Glasgow, 1992. [45] NEN. Concrete Bridges (VBB 1995) (Structural requirements and calculation methods). Tech. Rep. NEN 6723, Nederlands Normalisatie-instituut, 1995. [46] NEN. The Design of Steel Bridges (VOSB 1995) (Basic requirements and simple rules). Tech. Rep. NEN 6788, Nederlands Normalisatie-instituut, 1995. [47] Olson, L. G., and Bathe, K.-J. Analysis of uidstructure interactions. A direct symmetric coupled formulation based on the uid velocity potential. Computers & Structures 21 (1985), 2132. [48] Parks, D. M. A stiness derivative nite element technique for determination of crack tip stress intensity factors. Int. J. of Fracture 10, 4 (1974), 487502. [49] Roddeman, D. G., and Jansen, L. F. An a priori geometry check for a single isoparametric nite element. Computers & Structures 47, 1 (1993), 6972. [50] Saad, Y. ILUT: a dual threshold incomplete ILU factorization. Numerical Linear Algebra with Applications 1 (1994), 387402. [51] Saad, Y., and Schultz, M. H. GMRES: a Generalized Minimal Residual algorithm for solving nonsymmetric linear systems. SIAM J. Sci. Stat. Comput. 7, 3 (1986), 856869. [52] Schenk, O. Scalable Parallel Sparse LU Factorization Methods on Shared Memory Multiprocessors. PhD thesis. [53] Schenk, O., Gartner, K., and Fichtner, W. Ecient Sparse LU Factorization with Leftright Looking Strategy on Shared Memory Multiprocessors. BIT 40, 1 (2000), 158176. [54] Spence, P. W., and Kenchington, C. J. The Role of Damping in Finite Element Analysis. Tech. Rep. R0021, National Agency for Finite Element Methods & Standards (NAFEMS), Glasgow, 1993. [55] Strang, G. Linear Algebra and Its Applications, 3rd ed. Harcourt Brace Jovanovich College Publishers, 1988. [56] Terzaghi, K. V. Theoretical Soil Mechanics. John Wiley & Sons, 1943.
BIBLIOGRAPHY [57] van den Boogaard, A. H. The Computation of J-integrals Using the Finite Element Method. Tech. Rep. BI-93-022, TNO Building and Construction Research, Rijswijk, The Netherlands, 1993. [58] van den Boogaard, A. H. Arbitrary Lagrangian Eulerian Analysis in DIANA. Tech. Rep. 95-NM-R1002, TNO Building and Construction Research, Rijswijk, The Netherlands, 1995. [59] van Erp, E. M. Advanced Buckling Analysis of Beams with Arbitrary Cross-sections. PhD thesis, Eindhoven University of Technology, 1989. [60] Vermeer, P. A., and Verruyt, A. An accuracy condition for consolidation by nite elements. Int. J. Num. and Analytical Methods in Geomechanics 5 (1981), 114. [61] Zienkiewicz, O. C., and Bettes, P. B. Fluidstructure dynamic interaction and wave forces. An introduction to numerical treatment. Int. J. Num. Meth. Eng. 13 (1978), 116. [62] Zienkiewicz, O. C., and Taylor, R. L. The Finite Element Method, 4th ed., vol. 1: Basic Formulations and Linear Problems. McGraw-Hill, 1989. [63] Zienkiewicz, O. C., and Taylor, R. L. The Finite Element Method, 4th ed., vol. 2: Solid and Fluid Mechanics, Dynamics and Non-linearity. McGraw-Hill, 1991.
619
620
BIBLIOGRAPHY
Index
Page numbers. Bold face numbers in- Acceleration of gravity, see Gravity acceldicate pages with formal information about eration the entry, e.g., a syntax description (36). Accuracy Italic numbers point to an instructive exconsolidation analysis, 560 ample of how the concept in question might eigenvalue analysis, 447, 473 be used (132 ). Underlined numbers refer nonlinear potential ow analysis, 552 to theoretical backgrounds on the subject ACTIVE command, 408, 424 (95). Adaptive loading, 495 automatic, 233 iteration based, 227 Keywords. Sans serif type style refers to the interactive interface (EYE). Type- Adaptive time increments, 235, 495, 498 writer style refers to the batch interface AGE input, 425 Aging (YOUNG). phased potential ow analysis, 425 ALL option groundwater ow, 354 heat ow, 331 hybrid frequency time domain analproperty, 612, 613 ysis, 203 model selection, 72 A nonlinear analysis, 252, 255 postbuckling analysis, 296 A output, 165 ALPHA command, 232 spectral response analysis, 182 ALPHA parameter AAB subroutine, 578 nonlinear analysis, 217 ABS option transient groundwater ow, 350 spectral response analysis, 179 transient heat ow, 327 ABS rule, 179 -method, see HilberHughesTaylor ACCELE command, 155 Ambient inuence, 123 frequency response analysis, 165 AMPLIT command, 296 hybrid frequency time domain outAMPLIT option, 162 put, 202 Analysis commands, see Commands mixture analysis, 395 Anisotropy, 374, 375 nonlinear analysis output, 253 Append input, 405 spectral response analysis, 181 APPEND option transient analysis output, 137 input reading, 64 Acceleration output selection, 78 frequency response analysis, 165 phased analysis, 405 spectral response analysis, 181 postprocessing output, 84 transient analysis, 137
Diana-9.4 Users Manual Analysis Procedures December 15, 2009 First ed.
622
tabular output, 81 Aquifer, 341, 349 Arc-length control, 213, 230, 492 ARCLEN command, 227229, 230 AREA subroutine, 590 ARRAY option, 64 ARROW subroutine, 590 ASCII option, 78, 84 Aspect Ratio test, 66 ASSEMB command, 68, 461 buckling analysis, 286 cross-section analysis, 359 direct response analysis, 159 eigenvalue analysis, 439 hybrid frequency time domain analysis, 194 linear static analysis, 91 lubriation analysis, 367 modal response analysis, 152 nonlinear analysis, 210 spectral response analysis, 177 steadystate groundwater ow, 340 steadystate heat ow, 317 transient analysis, 129 transient groundwater ow, 347 transient heat ow, 323 Assumed strain, 208, 384 AUTO option, 433 AUTOMA command nonlinear initial state, 225 nonlinear load step, 226, 233 nonlinear time step, 235 solution procedures, 429, 430 transient load step, 131, 132 Automatic load increments, 233 Automatic solution, 430 Automatic time increments, 237 Automatic tying, 33 AUTOTY option, 68 AVERAG input, 54, 55 AVERAG option nonlinear strain, 257 AXES input, 27, 28, 30 AXES option, 80, 83 Axisymmetric elements options, 282 B
INDEX
BAB subroutine, 579 Backup of filos le, 61 BACKWA command, 216 Backward substitution, 466 BANDWI option, 274 Base excitation, 523 dynamic analysis, 118 spectral response analysis, 526 transient analysis, 134, 136 BASE subtable of LOADS, 118 Batch output, 69 *BCROSS command, 358 Beam cross-section analysis, see Cross-section analysis Beam elements output, 73 stability analysis, 280, 282 tying to plane stress, 35, 36 tying to solid, 35, 36 Bending moments frequency response analysis, 171 linear static analysis, 104 nonlinear analysis, 265 spectral response analysis, 189 transient analysis, 144 BETA parameter, 217 BETWEE subtable of TYINGS, 24, 25 BF parameter, 62 BFGS iteration, 242, 485 BFGS option, 242 BIAXFE option, 102, 142, 186, 262 Bifurcation, 213, 497 BINARY option, 78, 84, 85 BINSEG symbol, 574 Biot material parameters, 556 Biot theory (full), 374 BOND command, 238 Bottom absorption, 535 BOUNDA command cross-section analysis, 359, 359 lubriation analysis, 367 steadystate groundwater ow, 340, 342, 353 steadystate heat ow, 317, 318, 330 transient groundwater ow, 347
INDEX
transient heat ow, 323 BOUNDA input, 308, 309 BOUNDA table, 306 Boundary case, 306 Boundary conditions, 549 essential, 457 natural, 457 potential ow, 306 transient potential ow, 307 Boundary elements groundwater ow, 355 heat ow, 332 Bowl liquefaction status output, 268 BOWL option, 268 BRIEF command, 249 BROYDE option, 242 Broyden iteration, 242, 485 BTWMPC subtable of TYINGS, 26 BUCKLI command, 289 BUCKLI option, 290 Buckling analysis, 286, 542 Buckling modes, 291 Buckling values, 291, 447 Bump factor, 50 Bump factor, 612 BUMP input, 50 Buoyancy, 561 mixture analysis, 389 staggered analysis, 382 BWIDTH input, 46 Dutch Code, 48 European Code, 53 C CALCUL option direct response analysis, 160 eigenvalue analysis, 441, 443 hybrid frequency time domain analysis, 195 initial displacements, 222 initial stresses, 222 steadystate groundwater ow, 345 steadystate heat ow, 321 transient groundwater ow, 348 transient heat ow, 325 CAPACI command, 323 CAPACI option, 102, 143, 187, 263 Capacitance heat ow, 312 hydrostatic, 375 Capacity, 550 Capacity matrices, 323 CAPATT input phased analysis, 425 CASE input mobile load, 44 potential ow, 306 structural analysis, 37 transient analysis, 118 CASE parameter steadystate groundwater ow, 341, 342, 345 steadystate heat ow, 318, 321 transient groundwater ow, 349 transient heat ow, 325 CASES command cross-section analysis, 362 lubrication analysis, 368 CAUCHY option, 79 Cauchy stress, 501 frequency response analysis, 170 linear static analysis, 103 nonlinear analysis, 263 spectral response analysis, 187 transient analysis, 143 CEB-FIP code, 8 CELSIU input, 7 Cement hydration, see Hydration CENTER option, 79 CENTRI subtable of LOADS, 40 Centrifugal load, 40 CGS-units, 3 CHECK command element evaluation, 66 filos le, 63 Cholesky solver, see Sparse Cholesky CLASS input, 48 Class-I beams stability analysis, 280, 282 Class-II beams stability analysis, 280 Class-III beams automatic tying, 33 stability analysis, 282 CLASSE input, 52 Clay, 239
623
624
Egg Cam-clay model, 266 CLEAR subroutine, 585 CM input, 6 CMOD command, 232 CMOD control, 232, 494 CMS, see Component Mode Synthesis CO output, 271 CO parameter, 74 CODE input, 45 Dutch Code, 48 European Code, 52 COLLIN command, 82 COMBIN command, 83 COMBIN subtable of LOADS, 55 Commands, 59 COMPLE option, 162 Complete Quadratic Combination, 527 COMPOS command element evaluation, 67 output selection, 71, 76, 77 COMPOS subtable of GROUPS, 11 Composed beams generation, 67 output selection, 76 Composed solids generation, 67 output selection, 77 COMPRE command, 64 Compressibility undrained, 554 Compression modulus, 554 Compression of filos le, 61, 64 CONCEN command, 270 hybrid frequency time domain output, 202 nonlinear analysis, 213 CONCEN option, 255 Concentration, 270 Concentration ow, 311 Concentration load staggered analysis, 381 Concentration strain, 259 Concrete young hardening, 381 Concrete biaxial failure envelope linear static analysis, 103 nonlinear analysis, 264 spectral response analysis, 188 transient analysis, 143
INDEX
Concrete hydration, 375 CONDTT input, 425 CONDUC command cross-section analysis, 359 lubriation analysis, 367 steadystate groundwater ow, 341 steadystate heat ow, 317 transient groundwater ow, 347 transient heat ow, 323 Conduction, 550 Conductivity groundwater ow, 335 heat ow, 312 Conductivity matrices steadystate groundwater ow, 340 steadystate heat ow, 317 transient groundwater ow, 347 transient heat ow, 323 Conjugate Gradient, 473 Conservative loading, 541 CONSIS option cracking, 213, 214 direct response analysis, 160 eigenvalue analysis, 441 hybrid frequency time domain analysis, 195 nonlinear analysis, 217 transient groundwater ow, 347 transient heat ow, 323 Consistent mass analysis, 469 Consolidation of soil, 337, 373 mixture analysis, 383, 390 CONSTA command, 241 Constant dilatation mixture analysis, 384 Constant shear mixture analysis, 384 Constant Stiness iteration, 241, 486 Constrained Minimization, 505 CONTAC option, 268 Contact analysis, 242, 505 Contact status, 268 Contamination transport, 337 CONTIN command nonlinear iteration, 240 postbuckling analysis, 295 CONTIN option hybrid frequency time domain iteration, 200
INDEX
nonlinear iteration, 244 nonlinear load step, 228, 229 nonlinear time step, 236 Continuation analysis, 280, 295, 547 Continuation iteration, 240, 486 Control commands, 59 CONVEC command, 341 Convection, 313 Convectiondiusion, 549 CONVER command hybrid frequency time domain iteration, 200, 200 nonlinear groundwater ow, 343, 352 nonlinear heat ow, 319, 328 nonlinear iteration, 240, 243 CONVER option, 252 CONVER parameter, 296 Convergence eigenvalue analysis, 446 iterative solution, 475 Convergence criteria, 200, 200, 240, 243, 488 Convergence rate, 482, 485, 487 CONVTT input phased analysis, 425 Cooling pipe elements, 314, 330 internal temperature, 328, 334 COOR option, 80 COORD option, 83 COORDI table, 13 COPY command, 63 Copy filos le, see filos le CORROS command, 213 Coulomb friction, see Friction Coupled owstress analysis, 373 CP parameter, 63 CQC option spectral response analysis, 179 CQC rule, 179, 527 Crack bandwidth output, 274 Crack index nonlinear analysis, 262, 264 CRACK option nonlinear status, 266 nonlinear stress, 261 strain, 255 Crack status, 266 Crack strain, 258 Crack stress, 263 Crack tip, 500, 564 Crack tip elements, 109, 275, 563 CRACKB input, 274 CRACKI command, 213 nonlinear analysis logging, 249 CrankNicolson integration potential ow analysis, 551 transient groundwater ow, 350 transient heat ow, 327 CREEP command, 213 CREEP option, 255 Creep strain, 258 CRISFI option, 242 Criseld iteration, 242, 485 Critical damping factor, 155, 178, 522 CRKIND option, 262 CRKSUM option strain, 255 Cross-section beam composition, 76 beam elements, 73 Cross-section analysis, 357 input, 303, 358 output, 360 CROSSE command, 73, 76 CROSSE option, 79 Curvature frequency response analysis, 168 linear static analysis, 100 nonlinear analysis, 259 spectral response analysis, 184 transient analysis, 140 Curved shell elements automatic tying, 33 stability analysis, 283 CUTBCK parameter nonlinear load step, 234 nonlinear time step, 238 D D output buckling analysis, 291 eigenvalue analysis, 450 frequency response analysis, 163 linear static analysis, 97 spectral response analysis, 180 D2 output, 294
625
626
DAMPIN command direct response analysis, 160 hybrid frequency time domain analysis, 195 modal response analysis, 155 nonlinear analysis, 217 spectral response analysis, 178 Damping, 117, 522 added, 537 continuous, 522, 525 discrete, 522 modal, 155, 178, 522 Rayleigh, 217, 471, 522 strain energy based, 448, 471 structural, 523 Darcy ow, 556, 557 Darcy ux, 398 Darcys law, 375 DARCYF command, 398 Dashpots, see Damping, continuous Data structures acces, 603 DATA table, 384 Database, see filos le DAY input, 6 ddoc utility program, 570, 576 Dead weight, see Weight load Decomposition, 466 subroutines, 584 Default position property, 611, 612 DEFORM subtable of LOADS, 41 mixture analysis, 387 phased analysis, 406, 410 Deformation frequency response analysis, 168 linear static analysis, 99 nonlinear analysis, 259 spectral response analysis, 184 transient analysis, 140 Deformation load, see Prescribed displacements Degree of reaction, 326, 375, 381 output, 333 Degrees of freedom, 458 DELETE command, 65 phased analysis, 405 DELTAT parameter, 237 DENSIT input, 54, 55 Density apparent, 554
INDEX
reduced, 561 DET2 subroutine, 585 DET3 subroutine, 585 DETA parameter, 242 Detailed groundwater ow, 335, 354 Determinants subroutines, 585 Deviatoric deformation, 560 Deviatoric strains, 508 Deviatoric stresses, 510 DGEPD subroutine, 584 DGR output, 333 DGRINI parameter, 326 DI parameter dimensionality, 62 directory size, 13 DIAGON command, 432 DIAPATH symbol, 574 DIASLIB symbol, 575 Dierential matrix, 458 DIGITS command, 82 DIRECT command CMOD Arc-length control, 232 DIRECT input mobile load, 45 structural analysis, 54, 55 DIRECT option filos le, 64 Direct response analysis, 157 Direct solution, 431, 466, 525 DIRECT table, 14 Direct time integration, 528 Direction dependent participation factor, 470 Directions, 14 Dirichlet, 549 DISC option, 267 Discharge, 557 groundwater ow, 355 heat ow, 332 input, 306, 388 output, 397 prescribed, 307 DISFAC input, 46 Dutch Code, 50 European Code, 53 DISFOR option, 79 Disk space, 61 DISMOM option, 79 DISPLA command
INDEX
buckling analysis, 288, 290 eigenvalue analysis, 449 frequency response analysis, 163 hybrid frequency time domain output, 202 linear static analysis, 96 mixture analysis, 389, 395 nonlinear analysis, 222 nonlinear analysis output, 253 nonlinear iteration, 244 phased analysis, 411 postbuckling analysis, 293, 296 spectral response analysis, 180 transient analysis output, 134 DISPLA option, 239 DISPLA subtable of INIVAR stability analysis, 281 transient analysis, 120 Displacement, 460 frequency response analysis, 163 linear static analysis, 96 phased analysis, 412 prescribed, 41, 410 spectral response analysis, 180 transient analysis, 134 Displacement control, 230, 491 Displacement norm, 244, 489 DISSEI option, 508 linear static analysis, 99, 100 nonlinear analysis, 256, 260 spectral response analysis, 183, 185 Distance between beams, 612 DISTRI input, 45 Distributed seismic moment, see Seismic moment diva utility program, 570 Divergence, 201, 244, 483, 489 soilpore uid, 553 DM input, 6 Dm output, 451 Domain decomposition, 432, 477 DP output, 297 Dp output, 451 DPMPAR subroutine, 598 DRAINE command, 239 mixture analysis, 392 DRRAND subroutine, 597 DSRAND subroutine, 597 Dupuit assumption, 337 Dutch codes, see VOSB, VBB Dutch VBB code property, 611 Dutch VOSB code property, 612 DX parameter, 27, 28, 30 DY parameter, 27, 28, 30 DYN input, 6 DYNAMI command, 217 Dynamic analysis, 117 owstress, 557, 559, 560 linear transient, 127 mixture, 374 transient nonlinear, 217 Dynamic pressures frequency response analysis, 173 nonlinear analysis, 276 transient analysis, 148 DZ parameter, 27, 28, 30 E E output frequency response analysis, 167 linear static analysis, 99 nonlinear analysis, 257 spectral response analysis, 184 transient analysis, 140 Earthquake, 118 Ec output, 258, 259 ECCENT subtable of TYINGS, 28, 28 Eccentricity, 27, 33, 461 ECCMPC subtable of TYINGS, 29 Eceq output, 258 ECHO option, 64 Econ output, 259 Editor, see Text editor Ee output, 257 Eeeq output, 258 Eeq output frequency response analysis, 167 linear static analysis, 99 nonlinear analysis, 258 spectral response analysis, 184 transient analysis, 140 Eective force vector, 492 Eective mass, 448, 470 *EIGEN command, 435, 436 EIGEN command buckling analysis, 286
627
628
hybrid frequency time domain analysis, 196 modal response analysis, 152 spectral response analysis, 177 Eigenfrequency, 447, 472 Eigenmode, 449 Eigenvalue, 447, 472 Eigenvalue analysis, 435, 468 Ek output, 258 Ekeq output, 259 ELASTI command, 214 ELASTI option, 255 Elastic strain, 257 Elastoplasticity, 557 staggered analysis, 375 ELEMEN command output selection, 71, 72 phased analysis, 408, 409 reinforcement selection, 75 ELEMEN input mobile load, 44 ELEMEN option automatic tying, 68 linear static analysis, 107, 108 nonlinear analysis, 273 phased analysis, 424 spectral response analysis, 191, 192 transient analysis, 146, 147 ELEMEN subtable of GROUPS, 11 ELEMEN subtable of TYINGS, 35 Element assembling, 461 Element internal forces, 519 Element interpolation matrix, 460 Element load, 464 Element mass matrix direct response analysis, 159 eigenvalue analysis, 440 hybrid frequency time domain analysis, 194 Element rigidity matrix, 461 Element stiness matrix, 463 Element tractions, 462 Element transformation matrix, 460 Element types, 605 ELMFOR command, 108 hybrid frequency time domain output, 202 nonlinear analysis, 273 spectral response analysis, 192
INDEX
transient analysis, 147 ELSEUS user-supplied subroutine, 572 ELSGUS user-supplied subroutine, 572 ENERGY command nonlinear initial state, 225 nonlinear iteration, 244 nonlinear load step, 226, 229 transient load step, 131 Energy load increments, 496 Energy norm, 490 Energy release rate, 564 linear static analysis, 109 nonlinear analysis, 276 Engineering strains, 508 ENTHAL input phased analysis, 425 Enthalpy, 315 ENV 1991-3 code, 51, 52 ENV input, 45, 52 Environment symbols, 574 Ep output, 258 Epeq output, 258 EqA output, 334 EQAINI parameter, 326 EQUAGE command, 326, 330, 333 EQUAL subtable of TYINGS, 21, 22 mixture analysis, 389 EQUAL table, 304, 315 Equality tying, 21 EQUILI option, 222 mixture analysis, 391 Equilibrium, 458, 464 Equivalent age, 314, 326, 381 output, 333 Equivalent nodal forces, 464 EQUMPC subtable of TYINGS, 23 ERATIO option, 262 Error handling subroutine, 602 Error messages, 59 ERROR option, 80 ERRORS command, 59 Eshr output, 259 Essential boundary condition, 549 ESTIMA command steadystate groundwater ow, 342, 344 steadystate heat ow, 318, 320 ETAMAX parameter, 242 ETAMIN parameter, 242
INDEX
Etmp output, 259 Euler Backward integration nonlinear analysis, 216 potential ow analysis, 551 soilpore uid analysis, 559 structural dynamics, 529 transient groundwater ow, 350 transient heat ow, 327 Euler buckling, 286, 542 *EULER command, 285 Euler Forward integration potential ow analysis, 551 transient groundwater ow, 350 transient heat ow, 327 European code, see ENV European ENV code property, 612 EVALUA command, 66 buckling analysis, 286 cross-section analysis, 359 direct response analysis, 159 eigenvalue analysis, 439 hybrid frequency time domain analysis, 194 linear static analysis, 91 lubriation analysis, 367 modal response analysis, 152 nonlinear analysis, 210 spectral response analysis, 177 steadystate groundwater ow, 340 steadystate heat ow, 317 transient analysis, 129 transient groundwater ow, 347 transient heat ow, 323 Evaporation, 315 Evol output linear static analysis, 99 nonlinear analysis, 257 spectral response analysis, 184 transient analysis, 140 EX option, 395 Excitation hybrid frequency time domain analysis, 193 spectral response analysis, 175 EXECUT command buckling analysis, 287, 287 cross-section analysis, 358, 360 direct response analysis, 160 eigenvalue analysis, 445 frequency response analysis, 153 hybrid frequency time domain analysis, 196, 197, 199 lubrication analysis, 366, 367 mixture analysis, 391, 392 modal response analysis, 152, 155 nonlinear analysis, 209, 218 postbuckling analysis, 295 spectral response analysis, 177, 178 steadystate groundwater ow, 339 steadystate heat ow, 316 transient analysis, 128, 130 transient groundwater ow, 346, 349 transient heat ow, 322, 326 EXECUT option nonlinear analysis logging, 249 EXPLIC command hybrid frequency time domain analysis, 200 modal response analysis, 155, 161 nonlinear initial state, 225 nonlinear load step, 226, 227 nonlinear time step, 234, 235 transient load step, 131 transient time step, 132 Explicit time integration phased analysis, 423 transient analysis, 529 transient ow analysis, 552 EXPONE command, 132, 235, 236 Exponential time increments, 236 EXTERN option groundwater ow, 356 heat transfer, 332 load vectors, 106 F FACTOR command, 430, 431 phased analysis, 409 Factor for distributed load, 612, 613 Factor for lorry forces, 612, 613 FACTOR input frequency domain analysis, 124, 125 transient analysis, 121, 122 transient potential ow, 308, 309 FACTOR parameter buckling analysis, 289 direct response analysis, 160
629
630
eigenvalue analysis, 441, 443 hybrid frequency time domain analysis, 196 nonlinear initial state, 221 nonlinear time step, 237 steadystate groundwater ow, 341, 342, 345 steadystate heat ow, 318, 321 transient groundwater ow, 348 transient heat ow, 325 Factorization, 430, 431, 466 FAHREN input, 7 Fatigue failure, 111, 113 FB output frequency response analysis, 173 linear static analysis, 106 nonlinear analysis, 272 spectral response analysis, 190 transient analysis, 146 FE output, 107 FEMVIE output device, 78, 84 FI output, 107 FIELD parameter buckling analysis, 289 nonlinear initial state, 222 steadystate groundwater ow, 345 steadystate heat ow, 321 transient groundwater ow, 348 transient heat ow, 325 FILE option, 64 FILE parameter filos le, 63 input read, 64 output, 65, 78 FILMA subroutine, 586 *FILOS command, 62 linear static analysis, 90 Filos le access subroutines, 603 copy, 63 maintenance, 60, 60 Finite element formulation potential ow analysis, 550 FIRST option creep, 213 plasticity, 212 First order displacement, 542 FIX subtable of TYINGS, 30, 31 Fixed displacements, see Deformation load
INDEX
FIXMPC subtable of TYINGS, 32 FIXPOT table, 304 FIXTEM table, 304 FL output groundwater ow, 356 heat ow, 332 lubrication analysis, 369 Flat shell elements stability analysis, 283 FLB output, 333, 356 FLE output, 333, 356 FLOW command groundwater ow, 353, 356 heat ow, 330 heat transfer, 332 Flowstress analysis, 373 Flows groundwater ow, 356 heat transfer, 332 FLR output, 333, 356 Fluid density, 554 staggered analysis, 381 Fluid ow, 337 Fluidstructure interaction, 532 Fluidstructure interaction analysis eigenvalue, 436 frequency response, 150 hybrid frequency time domain, 193 spectral response, 176 transient analysis, 129 Fluidstructure interface elements, 148, 173, 276 Flux external, 550 groundwater ow, 355 heat ow, 331 lubrication analysis, 369 mixture analysis, 398 FLUX command groundwater ow, 353, 355 heat ow, 330, 332 lubrication analysis, 368, 369 FN output, 108, 147, 192, 273 Force frequency response analysis, 171 internal vector, 459 linear static analysis, 104 nonlinear analysis, 271, 272 nonlinear analysis output, 264
INDEX
spectral response analysis, 188 transient analysis, 144, 145 FORCE command frequency response analysis, 172 hybrid frequency time domain iteration, 200 hybrid frequency time domain output, 202 linear static analysis, 106 mixture analysis, 397 nonlinear analysis, 271 nonlinear iteration, 244 spectral response analysis, 190 transient analysis, 145 FORCE input structural analysis, 39, 56 units, 6 Force norm, 489 FORCE option, 79 Forchheimers law, 337 *FORTRAN command, 570, 571, 572 Fortran compiler, 570 Fortran standard, 570 Fortran validation, 570 Forward substitution, 466 FR output linear static analysis, 107 nonlinear analysis, 272 spectral response analysis, 191 transient analysis, 145 Fraction model, 257 FRACTU command, 109, 275 hybrid frequency time domain output, 202 Fracture mechanics, 109, 563 nonlinear analysis, 275 Free vibration, 152, 177, 436, 440, 447, 468 FREEVI command, 437, 440 FREQLO table, 123 *FREQUE command, 157 FREQUE command, 162 FREQUE input frequency domain analysis, 124, 125 FREQUE parameter, 155, 161 Frequency domain analysis input, 123 Frequency motion, 472 Frequency response analysis, 149, 157, 523 Frequencyload diagram, 123 FSdyex output linear static analysis, 103 nonlinear analysis, 264 spectral response analysis, 188 transient analysis, 143 FSdyun output linear static analysis, 103 nonlinear analysis, 264 spectral response analysis, 188 transient analysis, 143 FSPRES command frequency response analysis, 173 nonlinear analysis, 148, 276 FSstun output linear static analysis, 103 nonlinear analysis, 264 spectral response analysis, 188 transient analysis, 143 FSstus output linear static analysis, 103 nonlinear analysis, 264 spectral response analysis, 188 transient analysis, 143 FT input, 6 FT output nonlinear analysis, 272 transient analysis, 145 Ftu output, 264 FULL command, 249 FX+ output for postprocessing, 84 FXPLUS output device, 78, 84 G G input, 6 G output fracture mechanics, 109, 275 frequency response analysis, 167 linear static analysis, 99 nonlinear analysis, 257 spectral response analysis, 184 transient analysis, 140 Galerkin integration potential ow analysis, 550, 551 soilpore uid analysis, 558 transient groundwater ow, 350 transient heat ow, 327
631
632
GAMMA option, 268 GAMMA parameter nonlinear analysis, 216 nonlinear load step, 228 nonlinear time step, 236 Gauss theorem, 550 GaussJordan inversion, 584 Gc output, 258, 259 Gcon output, 259 Ge output, 257 GENEL command, 429, 431 General connections, 30, 31 General property, 613 Generalized eigenproblem, 468 Generalized mass, 447, 469 Generalized Minimal Residual, 474 GEOMET command, 210, 214 Geometric nonlinearity, 214, 499 Geometric stress-stiness matrix, 469 Geotechnics, 223, 262, 557 initial state, 219, 250 Gk output, 258 GLOBAL option, 79 Gp output, 258 GRADIE option groundwater ow, 355 linear static analysis, 102, 105 nonlinear analysis, 262, 265 spectral response analysis, 186, 190 Gradient of reinforcement stress, 518 linear static analysis, 105 nonlinear analysis, 265 spectral response analysis, 190 Gravity mixture analysis, 389 soilpore uid analysis, 556 Gravity acceleration, 6, 40 staggered analysis, 381 GREEN option, 79 GreenLagrange strain frequency response analysis, 167 linear static analysis, 99 nonlinear analysis, 257 spectral response analysis, 184 transient analysis, 140 GROUND input, 54, 55 Groundwater ow staggered analysis, 381 Groundwater ow analysis, 335 Groups, 9 GROUPS table, 11, 12 *GROWSS command, 339 *GROWTR command, 346 GSTAR option, 268 GTC subroutine, 603 Gtmp output, 259 H
INDEX
Hcr output, 275 HEAD option, 354 Heat ow, 469 Heat ow analysis, 311 *HEATSS command, 316 *HEATTR command, 322 HHT command mixture analysis, 391 nonlinear analysis, 217 HilberHughesTaylor integration, 217, 530, 559 Hookes law, 461 HOUR input, 6 Hybrid frequency time domain analysis, 193, 539 HYDRAT command, 326, 329 Hydration staggered analysis, 375, 381 transient heat ow, 314, 326, 328, 333 Hydraulic capacitance, 556 Hydraulic fracture, 385 Hydraulic head, 374 Hydraulic pore pressure, 561 Hydrostatic capacitance, 375 Hydrostatic pressure capacity, 102, 143, 187, 263, 519 linear static analysis, 103 nonlinear analysis, 264 spectral analysis, 188 transient analysis, 143 HYPER command, 214 Hyperelasticity, 214, 500 Hysteresis, see Damping, structural I IApr output mixture analysis, 396
Diana-9.4 Users Manual Analysis Procedures
INDEX
Icr output, 264 nonlinear analysis, 221, 223 ID output, 135 Initial displacement IDENTI command, 437, 442 stability analysis, 281 Identity matrix, 442, 469 transient analysis, 120 IDpr output Initial equilibrium, 391 mixture analysis, 396 Initial forces, 104, 188 Ill-conditioned system, 476 Initial potential eld, 305 ILU option, 432 Initial state, 219 ILU preconditioning, 432, 474 Initial state reference, 219, 222, 250 IMOVE subroutine, 597 Initial stress IMPERF command, 287, 288 linear static analysis, 103 Imperfections, 280, 288, 545 phased analysis, 414 user-specied, 546 spectral response analysis, 187 Implicit time integration Initial tractions, 105, 189 phased analysis, 423 Initial velocity, 120 potential ow analysis, 551 Initialization of filos le, 60 structural dynamics, 529 Initialization subroutines, 595 IMPORT input INIVAR table frequency domain analysis, 125 buckling analysis, 289 phased analysis, 406 potential ow analysis, 305 transient analysis, 122 stability analysis, 281 transient potential ow, 309 transient analysis, 119 Import of deformations, 405 Inner-products subroutines, 588 IMPORT option, 409 *INPUT command, 64 Impulsive loading, 523 phased analysis, 404, 424 IN input, 6 Input commands, 64 Incompatible modes, 384 Input data Incomplete LU-decomposition preconditiondynamic analysis, 117 ing, see ILU preconditioning nite element model, 3 INCREM option nonlinear analysis, 207 displacements, 134 potential ow analysis, 303 INCREM option structural analysis, 17 nonlinear stop criterion, 247 INPUT option output selection, 78 direct response analysis, 160 Incremental procedures, 242, 491 eigenvalue analysis, 441, 443 Indirect Displacement control, 230, 494 hybrid frequency time domain analINFLUE subtable of LOADS, 56 ysis, 195 Inuence eld, 56 initial displacements, 222 INFORM command, 64 initial stresses, 222 INISIZ parameter initial velocities, 222 nonlinear load step, 228, 229 steadystate groundwater ow, 345 nonlinear time step, 236 steadystate heat ow, 321 INITIA command transient groundwater ow, 348 filos le, 62 transient heat ow, 325 nonlinear analysis, 220, 221 Input reading, 64 transient groundwater ow, 346, 348 Instability, spatial, 560 transient heat ow, 322, 324 Insulation, 550 INITIA option, 78 Integration point coordinates Initial conditions tabular output, 80
633
634
Integration points output selection, 73, 75, 79 INTER1 subroutine, 599 INTER2 subroutine, 599 Interaction analysis, 384, 532, 553 Interconnection tying, 24 INTERF command, 214 INTERF option, 67 Interface elements, 385 stability analysis, 283 INTERN option r.h.s. vectors, 106 Internal beam arm, 516, 516 Internal element forces, 108 nonlinear analysis, 273 spectral response analysis, 192 transient analysis, 147 Internal force vector, 106, 459 Interpolation functions, 458, 460 Interpolation matrix, 460 Interpolation subroutines, 599 INTPNT command, 73, 75 INTPNT option, 79 INTTMP command, 330, 334 INV2 subroutine, 583 INV3 subroutine, 583 INVARI option, 79 Inversion subroutines, 583 INVMTX subroutine, 583 INVPGB subroutine, 584 IrD output, 135 ISET subroutine, 596 Isoparametric elements nonlinear analysis, 208 stability analysis, 280 ITEMS option, 64 ITERAT command nonlinear analysis, 219, 240 nonlinear groundwater ow, 342, 343, 351 nonlinear heat ow, 318, 319, 328 nonlinear initial state, 225 nonlinear load step, 226 nonlinear time step, 234 solution procedures, 429, 431 transient analysis, 130, 132 ITERAT option nonlinear analysis logging, 249 nonlinear load step, 227
INDEX
nonlinear time step, 235 transient heat ow, 329 Iterative solution procedure linear analysis, 431, 473 nonlinear analysis, 240, 480 potential ow analysis, 552 IVpr output mixture analysis, 396 J Jacobi preconditioning, 432, 475 Jaumann rate, 502 K K output fracture mechanics, 109, 275 frequency response analysis, 168 linear static analysis, 100 nonlinear analysis, 259 spectral response analysis, 184, 185 transient analysis, 140, 141 KELVIN input units, 7 KG input, 6 KGF input, 6 KLBF input, 6 KM input, 6 KN input, 6 KozenyKarman modied permeability, 557 KP input, 6 L Lagrange, 208, 215, 500, 502 LAMBDA parameter, 296 Large deformations, 544 LAST option groundwater ow, 354 heat ow, 331 hybrid frequency time domain analysis, 203 nonlinear analysis, 252, 255 postbuckling analysis, 296 Lateral pressure ratio, 251 LAYOUT command, 69, 82 cross-section analysis, 362 groundwater ow, 353
Diana-9.4 Users Manual Analysis Procedures
INDEX
heat ow, 330 hybrid frequency time domain output, 202 lubrication analysis, 368 nonlinear analysis output, 253 Layout of tabular output, 82 LB input, 6 LBF input, 6 LDU decomposition, 466 LEFM, see Fracture mechanics Length for bump and reduction, 612 LENGTH input, 6 LEVEL input, 55 Libraries of subroutines, 576 Line Search, 240, 242, 487, 506 Linear analysis static, 89 transient, 127 Linear buckling, see Euler buckling LINEAR command nonlinear iteration, 241 steadystate groundwater ow, 341 steadystate heat ow, 317 transient groundwater ow, 349 transient heat ow, 325 Linear constraints, see Tyings LINEAR option eigenvalue analysis, 440, 443 nonlinear iteration, 241, 242 Linear Stiness iteration, 241, 486 LINESE command, 240, 242 LINPAG command, 82 *LINSTA command, 89 LINSTR command, 211 LIQUEF command, 239 Liquefaction of soil, 239, 390 Liquefaction status, 267 LMOVE subroutine, 597 Load, 36, 464 import of deformations, 405 Load case, 37 Load combinations, 55 LOAD command initial nonlinear conditions, 220 mixture analysis, 392 nonlinear load step, 218, 225 nonlinear stop criterion, 246 transient load step, 130, 131 Load control, 491 Load cycles to failure, 112 LOAD input, 121, 122, 124 LOAD parameter, 444 buckling analysis, 288 direct response analysis, 160 eigenvalue analysis, 441, 443 hybrid frequency time domain analysis, 196 nonlinear initial state, 222 Load reduction factor, 612 Load set, 55 selection, 94 Load steps mixture analysis, 392 nonlinear analysis, 218, 225 transient analysis, 130, 131 Load vector output, 107, 191 setup, 464 LOADFA parameter, 293 Loading nonlinear stop criterion, 246 LOADNR parameter initial nonlinear conditions, 220 nonlinear load step, 226 transient load step, 131 LOADS command fatigue failure, 113 hybrid frequency time domain analysis, 194 linear static analysis, 91, 95, 464 nonlinear analysis, 210, 215 transient analysis, 129 LOADS table, 36 mixture analysis, 387, 388 phased analysis, 410 staggered analysis, 374, 375 transient analysis, 118 LOALEN input, 51 LOARED input, 50 LOCAL option, 79 Lode output linear static analysis, 103 nonlinear analysis, 263 spectral response analysis, 188 transient analysis, 143 Lodes angle, 511 LODFAC input, 46 Dutch Code, 50
635
636
European Code, 53 LODSET input, 406 LOG command, 59 LOGGIN command, 219, 249 Logging a job, 59 nonlinear analysis, 219, 249 Long term loading, 523 LORRY input, 45 LSET subroutine, 596 LTN input, 6 Lubrication, 365 Lumped mass, 469 LUMPED option direct response analysis, 160 eigenvalue analysis, 441 hybrid frequency time domain analysis, 195 nonlinear analysis, 217 transient groundwater ow, 347 transient heat ow, 323 Lumping, 522
INDEX
MASS input, 6 Mass matrices, 217 Master node, 18 MATERI table phased analysis, 425 Material codes, 5 Material properties phased analysis, 409 MATPRI subroutine, 594 MATRIX command cross-section analysis, 359 direct response analysis, 159 hybrid frequency time domain analysis, 194 linear static analysis, 91, 463 lubriation analysis, 367 nonlinear analysis, 210 steadystate groundwater ow, 340, 340 steadystate heat ow, 317 transient analysis, 129 transient groundwater ow, 347, 347 transient heat ow, 323, 323 M MATRIX option, 449 Matrix subroutines, 576 M input, 6 MATURI command, 214, 270 M output hybrid frequency time domain outfrequency response analysis, 171, 172 put, 202 linear static analysis, 104, 105 MATURI table nonlinear analysis, 265 staggered analysis, 375 spectral response analysis, 189 Maturity, 214, 270, 381 transient analysis, 144 staggered analysis, 381 M0 output, 104, 189 MAX option MA output, 271 hybrid frequency time domain analMachine parameters, 598 ysis, 203 make (unix utility), 570 linear static analysis, 95 Makemac utility, 570 nonlinear analysis, 255 Mass, 117 MAX parameter, 289 added, 537 MAXFRE command MASS command hybrid frequency time domain analdirect response analysis, 160 ysis, 199 eigenvalue analysis, 441 hybrid frequency time domain anal- MAXITE parameter eigenvalue analysis, 438, 446 ysis, 195 hybrid frequency time domain iteramodal response analysis, 153 tion, 200 nonlinear analysis, 217 iterative solution, 432 Mass conservation, 555, 561 nonlinear groundwater ow, 343, 351 Mass density, 117 nonlinear heat ow, 319, 328 Mass elements nonlinear iteration, 240, 498 stability analysis, 283
INDEX
transient iteration, 132 MAXLS parameter, 242 MAXSIZ parameter nonlinear load step, 228, 229, 234, 498 nonlinear time step, 236, 238 MAXTHR command iterative solver, 432 parallel direct sparse solver, 431 MB output frequency response analysis, 173 linear static analysis, 106 nonlinear analysis, 272 spectral response analysis, 190 transient analysis, 146 ME output, 107 Mesh quality, 66 Mesh renement, 25 METHOD command mixture analysis, 391 nonlinear analysis, 216 nonlinear groundwater ow, 343, 351 nonlinear heat ow, 319, 328 nonlinear iteration, 240, 241 Metis reordering, 467 MF parameter, 59 MI input, 6 MI output, 107 MI parameter, 63 MIN input, 6 MIN option hybrid frequency time domain analysis, 203 linear static analysis, 95 nonlinear analysis, 255 MINFRE command hybrid frequency time domain analysis, 199 MINSIZ parameter nonlinear load step, 228, 229, 234, 498 nonlinear time step, 236, 238 MITERA parameter creep, 213 plasticity, 213 Mixed boundary condition, 549 MIXTUR input, 384, 385 Mixture analysis, 239, 337, 373, 383, 553 Mixture elements, 208, 373, 383, 560 MM input, 6 MN output, 108, 147, 192, 273 Mobile load, 42 output of extreme values, 95 Mobile property, 611 MOBILE subtable of LOADS, 44 *MODAL command, 151 Modal damping factor, 448, 471 Modal response analysis, 151 Mode acceleration, 155 MODE parameter, 289 Mode superposition frequency response analysis, 524 spectral response analysis, 527 MODEL command buckling analysis, 286 cross-section analysis, 358, 358 direct response analysis, 159 eigenvalue analysis, 439 hybrid frequency time domain analysis, 194 hybrid frequency time domain analysis, 194 linear static analysis, 89, 90 lubrication analysis, 366 modal response analysis, 151 nonlinear analysis, 209, 210 spectral response analysis, 176, 176 steadystate groundwater ow, 339, 340 steadystate heat ow, 316, 317 transient analysis, 127, 129 transient groundwater ow, 346, 346 transient heat ow, 322, 322 MODEL OFF command, 84 MODES command buckling analysis, 290 eigenvalue analysis, 449 hybrid frequency time domain analysis, 199 modal response analysis, 155 postbuckling analysis, 293 spectral response analysis, 179 MODIFI option nonlinear groundwater ow, 343, 351 nonlinear heat ow, 319, 328 nonlinear iteration, 241 Modied NewtonRaphson iteration ow analysis, 552
637
638
nonlinear analysis, 241, 483 nonlinear groundwater ow, 343, 351 nonlinear heat ow, 319, 328 Mohr-Coulomb failure criterion, 102, 143, 187, 263, 519 Moisture capacitance, 556 MOMENT input, 39, 56 MOMENT option, 79 Momentum conservation, 555 MR output linear static analysis, 107 nonlinear analysis, 272 spectral response analysis, 191 transient analysis, 145 MSC, see Mobilized Shear Capacity MT input, 6 MT output nonlinear analysis, 272 transient analysis, 145 Multi-point tyings, 19 Multiplication subroutines, 577 MW parameter, 60 MX parameter, 359
INDEX
NF output, 113 NFATIG command, 113 Nishi liquefaction status output, 268 NISHI option, 268 NITERA parameter nonlinear load step, 228 nonlinear time step, 236 NMODES parameter eigenvalue analysis, 446 NOAXES option, 80 NOBOND input, 238 NOBOND option, 80, 263 NOCOOR option, 80 Nodal element forces, 107, 191, 272 transient analysis, 146 Nodal load, 38, 520 NODAL subtable of BOUNDA, 307 NODAL subtable of LOADS, 38 mixture analysis, 388 Node coordinates, 13 Node numbers input, 13 Nodes boundary conditions, 307 N equal potential, 304 xed potential, 303 N input, 6 initial potential, 305 N output NODES command frequency response analysis, 171 CMOD Arc-length control, 232 linear static analysis, 104 output selection, 71, 73, 76 nonlinear analysis, 264 spectral response analysis, 179 spectral response analysis, 188 NODES option transient analysis, 144 automatic tying, 68 N0 output, 104, 188 batch output, 79 Natural boundary condition, 549 nonlinear output, 271 NCLOAD command, 215 Negative pivots for (un)loading, 228, 229, NODES subtable of GROUPS, 11 NODFOR command, 107, 191 496 hybrid frequency time domain outNeumann, 549 put, 202 NEWMAR command, 216 nonlinear analysis, 273 Newmark integration, 216, 529, 559 transient analysis, 146 NEWREF option, 244 NOLOG command, 59 NEWTON option, 241 Nonconservative loading, 215, 230, 499, NewtonRaphson iteration 505 ow analysis, 552 nonlinear analysis, 228, 229, 241, 467, NONE input, 45 NONE option, 73 482 nonlinear groundwater ow, 343, 351 *NONLIN command, 209 mixture analysis, 390, 391, 392 nonlinear heat ow, 319, 328
INDEX
phased analysis, 418 NONLIN command eigenvalue analysis, 440, 443 steadystate groundwater ow, 341, 342 steadystate heat ow, 318 transient groundwater ow, 349, 350, 351 transient heat ow, 325, 326, 327, 328 NONLIN input, 406 Nonlinear analysis, 207, 209, 479 eigenvalue, 438 iteration, 467, 480 mixture, 390 phased, 418 potential ow, 552 NORM command, 296 NORM subroutine, 590 NORM2 subroutine, 591 Norms for convergence, 200, 243, 488 Norms subroutines, 590 Notation convention, 553 NSTEPS parameter nonlinear load step, 228, 229 nonlinear time step, 236, 237 postbuckling analysis, 296 NTRIAL parameter, 446 NULLVC subroutine, 591 NUMBER command hybrid frequency time domain analysis, 200 NUMBER parameter, 155 NUMSUB parameter, 433 O OLDREF option, 244 ONESID input option, 385 Ordering, 475 Out-of-balance forces, see Residual forces Out-of-core solution, 431 OUTCOR command eigenvalue analysis, 446 iterative solution, 432 OUTER subroutine, 589 Outer-products subroutines, 589 OUTPUT command, 69, 78 buckling analysis, 287, 290 eigenvalue analysis, 448 fatigue failure, 113 frequency response analysis, 153, 161 groundwater ow, 352 heat ow, 329 hybrid frequency time domain analysis, 197 linear static analysis, 89, 92 nonlinear analysis, 209, 219 postbuckling analysis, 293, 296 spectral response analysis, 178 steadystate groundwater ow, 339 steadystate heat ow, 316 tabular output, 80 transient analysis, 128, 131, 132 transient groundwater ow, 346 transient heat ow, 322 Output components, 79 Output selection, 69 Overconsolidation, 223, 374 KLB input, 6 OZ input, 6 P P input prescribed potential, 307 units, 6 P output frequency response analysis, 168 linear static analysis, 99, 100, 103 nonlinear analysis, 259, 263 spectral response analysis, 184, 188 transient analysis, 140, 143 PA output, 138 PApr output mixture analysis, 396 Parallel direct sparse solver, 430, 467 Parallel iterative solver, 432 PARAME command nonlinear analysis, 274 PARDIS command, 430 solution procedures, 429 PARDISO, 430, 467 Partial pivoting, 584 PARTIC option, 449 Participation factor, 448, 469 Participation vector, 449, 470 Pc output, 275
639
640
PD output linear static analysis, 97 transient analysis, 135 pD output, 97 PDL input, 6 PDpr output mixture analysis, 396 PEAK input frequency domain analysis, 125 transient analysis, 122 transient potential ow, 309 Peat, 239 Penalty conduction coecient, 336 Penalty formulation, 560 Periodic response, see Steady-state Permeability, 375, 556 Perturbation analysis, 280, 292, 471, 546 displacements, 294 PG output, 355 PH option, 390, 395 *PHASE command potential ow analysis, 424 structural analysis, 408 PHASE option displacements, 134 velocities, 136 PHASE option accelerations, 137 direct response analysis, 160 eigenvalue analysis, 441, 443 hybrid frequency time domain analysis, 195 nonlinear analysis, 223 output selection, 78, 411 Phased analysis, 78, 401 acceleration, 138 displacement, 97, 135, 412 ow, 421 groundwater ow, 337, 338 heat ow, 315 linear, 97 mixture, 374, 385, 391 structural, 403, 404 velocity, 137 PHYSIC command mixture analysis, 390 nonlinear analysis, 210, 212 nonlinear analysis options, 218, 219, 238
INDEX
transient analysis, 130 Physical properties specication, 611 PiolaKirchho stress (2nd), 264, 502 PIOLAK option, 79 PIVOTS option, 228, 229 Plane strain elements fracture mechanics, 565 stability analysis, 282 Plane stress elements fracture mechanics, 565 stability analysis, 282 PLASTI command, 212 nonlinear analysis logging, 249 PLASTI option status, 266 strain, 255 Plastic strain, 258 Plastic yield, 511 Plasticity status, 265 Plate bending elements stability analysis, 283 PO option, 390, 395 Point i lorry inuence line, 612 Point source, 307 Polar decomposition, 584 Pore uid, 553 mixture analysis, 374, 383 staggered analysis, 381 Pore pressure, 555, 560 mixture analysis, 373, 386, 395 nonlinear analysis, 262, 263, 269 staggered analysis, 381 POROSI input staggered analysis, 382 Porosity, 374, 553, 554 mixture analysis, 383 staggered analysis, 382 Position vector, 389, 556 POSTBU option, 296 nonlinear analysis, 222 Postbuckling displacements, 297 load factor, 298 Postbuckling analysis, 280, 292, 546 POTENT command steadystate groundwater ow, 345 transient groundwater ow, 348 POTENT option, 354
INDEX
POTENT subtable of INIVAR, 305 Potential ow analysis phased, 422 staggered, 377 Potentials, 303 PPCAP output linear static analysis, 103 nonlinear analysis, 264 spectral response analysis, 188 transient analysis, 143 PR input, 387, 388 PR output mixture analysis, 397 nonlinear analysis, 270 PrD output, 135 PRECON command, 432 PRECON input, 275 PRECON option, 275 Preconditioning, 432, 474 Preconsolidation stress output, 275 PRESCR option, 96 output selection, 78 Prescribed displacements, 41 output, 97 phased analysis, 410 Prescribed ux, 307 Prescribed potential, 307 Prescribed temperature, 307 PRESENT command, 267 PRESSU command groundwater ow, 353, 354, 355 initial state reference, 251 lubrication analysis, 368, 368 mixture analysis, 392, 396 nonlinear analysis, 270 PRESSU input, 382 PRESSU option mixture analysis, 389, 390, 395, 397 output selection, 79 Pressure, 511 groundwater ow, 354 lubrication analysis, 368 mixture analysis, 395 Pressure gradient, 354 Pressure head, 354 Prestress, see Reinforcement PREVIO option, 241, 242 PRfs output frequency response analysis, 174 nonlinear analysis, 148, 276 PRGERR subroutine, 602 PRH output, 354 PRIIVC subroutine, 594 PRIIVL subroutine, 593 PRILVC subroutine, 595 PRILVL subroutine, 595 PRIMAT subroutine, 579, 593 PRINCI option, 79 Principal shear, 511 Principal strain, 79, 508 Principal stress, 79, 510 Printing subroutines, 591 PRIVAL subroutine, 571, 592 PRIVEC subroutine, 592 PRP output, 354 PrV output, 137 PS output linear static analysis, 100 nonlinear analysis, 260 spectral response analysis, 185 PSD, see Power Spectral Densities Pseudo-acceleration, 527 PSI parameter, 242 PT output frequency response analysis, 169 linear static analysis, 100 nonlinear analysis, 260 spectral response analysis, 185 transient analysis, 141 PTE output heat ow, 331, 334 lubrication analysis, 368 PV output, 137 PVpr output mixture analysis, 396 Q Q input, 307 Q output frequency response analysis, 171 groundwater ow, 356 heat ow, 332 linear static analysis, 103, 104 mixture analysis, 398 nonlinear analysis, 263, 264 spectral response analysis, 188
641
642
transient analysis, 143, 144 Q0 output, 104, 188 QBpr output mixture analysis, 397 QRpr output mixture analysis, 397 QT output, 104, 188 QTpr output mixture analysis, 397 Quarterpoint position, 564 Quasi-Newton iteration, 241, 484 QY parameter, 359 QZ parameter, 359 R R1 parameter, 74 R2 parameter, 74 RAB subroutine, 577 RABT subroutine, 580 Radiation, 313 RANDOM command, 289 Random generator, 597 Random imperfection, 281, 289, 546 RATB subroutine, 581 RATBA subroutine, 582 RATIO parameter, 67 RAYLEI option, 217 Rayleigh damping, see Damping REACTI command, 330, 333 REACTI option forces, 106, 190 groundwater ow, 356 heat transfer, 332 output selection, 78 Reaction forces, 520 linear static analysis, 106 nonlinear analysis, 271, 272 spectral response analysis, 190 transient analysis, 146 READ command, 64 phased analysis, 405 REAXES command, 73, 74 REAXES option, 79 REDUCE command hybrid frequency time domain analysis, 198 modal response analysis, 154 postbuckling analysis, 292
INDEX
REFERE command mixture analysis, 392 nonlinear analysis, 219, 250 Reference norm, 489 Renement of mesh, 25 Regional groundwater ow, 337 REGULA option nonlinear groundwater ow, 343, 351 nonlinear heat ow, 319, 328 nonlinear iteration, 241 Regular NewtonRaphson iteration, 482 ow analysis, 552 nonlinear analysis, 212, 213, 241 nonlinear groundwater ow, 343, 351 nonlinear heat ow, 319, 328 REINFO command interface elements, 67 output selection, 71, 75 phased analysis, 408 REINFO option bonding, 238 linear static analysis, 108 moments and forces, 79 nonlinear analysis, 273, 274 slip, 239 spectral response analysis, 191, 192 transient analysis, 146, 147 REINFO subtable of GROUPS, 11 REINFO subtable of LOADS, 38 Reinforcement output selection, 74 phased analysis, 410 stability analysis, 283 Reinforcement forces, 79, 512 linear static analysis, 104 spectral response analysis, 188 transient analysis, 144 Reinforcement moments, 79, 512 linear static analysis, 105 spectral response analysis, 189 transient analysis, 144 RELATI option, 134, 136 REMAKE command, 65 Remaking input, 64 RENEW command, 409 Reordering, see Ordering REPORT command, 249 RESFOR command, 408 RESIDU command, 244
INDEX
RESIDU option forces, 106, 190 groundwater ow, 356 heat transfer, 332 output selection, 78 Residual forces, 520 linear static analysis, 107 nonlinear analysis, 271, 272, 480 spectral response analysis, 191 transient analysis, 145 Residual norm, 244, 490 Resonance, 157 RESPON command hybrid frequency time domain analysis, 197 modal response analysis, 153 spectral response analysis, 177 Response analysis frequency, 149 hybrid frequency time domain, 193 spectral, 175 Response Spectrum Method, see Spectral response analysis RESTOR command filos le, 63 nonlinear analysis, 252 nonlinear initial state, 225 nonlinear load step, 226 nonlinear time step, 234 transient load step, 131 transient time step, 132 Restore filos le, 63 Restore step in nonlinear analysis, 252 RESULT command, 246, 247 RESULT option, 83 *REYNOL command, 366 Reynolds ow, see Lubrication Rigidity matrix, 461 RMOVE subroutine, 596 RO input stability analysis, 281 structural analysis, 41 Robin, 549 Rock, 553 ROTATI option Arc-length control, 231 direct response analysis, 160 eigenvalue analysis, 441 hybrid frequency time domain analysis, 195 nonlinear analysis, 217 output selection, 78 PRESSU command hybrid frequency time domain output, 202 RSET subroutine, 571, 596 RungeKutta method, 531 S S output cross-section analysis, 363 frequency response analysis, 170 linear static analysis, 103 mixture analysis, 397 nonlinear analysis, 263 spectral response analysis, 187 transient analysis, 143 S0 output, 103, 187 SAB subroutine, 578 SATB subroutine, 581 SATBA subroutine, 582 SATURA command, 397 SATURA input staggered analysis, 382 Saturation, 554 mixture analysis, 397 staggered analysis, 381 Save filos le, 63 nonlinear analysis steps, 251 SAVE command filos le, 63 nonlinear analysis, 252 nonlinear initial state, 225 nonlinear load step, 226 nonlinear time step, 235 transient groundwater ow, 350 transient heat ow, 327 transient load step, 131 transient time step, 132 SCALE input frequency domain analysis, 125 transient analysis, 122 transient potential ow, 309 Scaling subroutines, 588 SDIRK2 command, 217, 238
643
644
SDIRK2 method, 238, 498, 532 SEC input, 6 Secant method, see Quasi-Newton SECANT option cracking, 213, 214 nonlinear iteration, 241 SECMOD option, 293 SECOND option, 212 Second order displacements, 294, 542, 546 Seepage face, 336 SEGPATH symbol, 574 Seismic Moment, 100, 185, 260, 508 SELECT command automatic tying, 68 buckling analysis, 290 cross-section analysis, 362 eigenvalue analysis, 449 fatigue failure, 112 frequency response analysis, 162 groundwater ow, 353 heat ow, 330 hybrid frequency time domain analysis, 201 linear static analysis, 93 lubrication analysis, 368 nonlinear analysis, 253 nonlinear stop criterion, 247 output selection, 69, 71 postbuckling analysis, 293, 296 spectral response analysis, 179 transient analysis, 133 Seq output frequency response analysis, 171 linear static analysis, 103 nonlinear analysis, 263 spectral response analysis, 188 transient analysis, 143 Ser output, 263 Service libraries, 576 SG output linear static analysis, 105 nonlinear analysis, 265 spectral response analysis, 190 Shape functions, 458, 460 SHAPE parameter, 67 Shape test, 66 Shear capacity, 102, 143, 187, 263, 519 linear static analysis, 103 nonlinear analysis, 264
INDEX
spectral response analysis, 188 transient analysis, 143 SHEAR command, 363 Shear modulus, 554 Shell elements automatic tying, 34 tying to solid, 35, 36 Shift eigenvalues, 437, 446, 471 nonlinear system, 438 SHIFT parameter eigenvalue analysis, 438, 446 SHRCAP output linear static analysis, 103 nonlinear analysis, 264 spectral response analysis, 188 transient analysis, 143 SHRINK command, 214 SHRINK option, 256 Shrinkage, 214, 381 Shrinkage strain, 259 SI-units, 3, 5, 7 SIDE1 command, 232 SIDE2 command, 232 Sign convention, 554 SIGN option nonlinear step size, 228, 229 nonlinear stop criterion, 247 Simple soil, 214 SIMULT command, 243 Single-point tyings, 19 SIZE parameter nonlinear load step, 234 nonlinear time step, 238 postbuckling analysis, 296 SIZES command nonlinear load step, 227 nonlinear time step, 235 transient groundwater ow, 350 transient heat ow, 327 Sk output, 263 SKIP input frequency domain analysis, 125 transient analysis, 122 transient potential ow, 309 Slave node, 18 SLIP command, 239 SMOOTH option, 80 Smoothing
INDEX
strain output, 80 stress output, 80 Snap-back behavior, 213, 230, 493 Snap-through behavior, 230, 492, 495, 496 Softening, 471 Soil, 383, 553 drained/undrained, 239, 269, 560 pore uid, 381 SOIL command, 214 Soilstructure interaction, 384 Solid elements automatic tying, 34 fracture mechanics, 565 stability analysis, 283 Solidication, 315 Solution iterative procedure, 240, 479, 480, 552 phased transient analysis, 422 transient ow analysis, 551 Solution methods, 429, 465 SOLVE command, 429 cross-section analysis, 360 linear static analysis, 89 lubrication analysis, 367 nonlinear analysis, 219, 223 steadystate groundwater ow, 341, 342 steadystate heat ow, 317, 318 transient analysis, 131 transient groundwater ow, 349, 350 transient heat ow, 325, 327 Sommerfeld radiation, 534 SORTI subroutine, 601 SORTII subroutine, 601 Sorting subroutines, 599 SORTR subroutine, 599 SORTRI subroutine, 600 Source code, 570 SP output, 264 Space discretization, 558 SPACE option, 64 Sparse Cholesky solver, 430, 467 Sparse matrix, 465 *SPECTR command, 176, 193 Spectral acceleration, 527 Spectral displacement, 527 Spectral response analysis, 175, 526 SPHERI command, 230 Spherical Path, 228230, 493 Spline elements stability analysis, 283 Spring elements stability analysis, 283 Spurious modes, 208 SRSS option spectral response analysis, 179 SRSS rule, 179, 527 ST output frequency response analysis, 172 linear static analysis, 105 nonlinear analysis, 265 spectral response analysis, 189 transient analysis, 144 STABIL command, 287 Stability analysis, 279, 541 Euler, 285 linearized buckling, 444, 447, 469 Stability of soil, 373 Staggered analysis, 301, 315, 337, 373, 377 Standard eigenproblem, 442, 447, 468, 469 START command mixture analysis, 391, 392 nonlinear analysis, 130, 218, 219, 262 postbuckling analysis, 296 transient analysis, 119 Static analysis, 89 Static condensation, 476 Status, 265 STATUS command, 266, 268, 269 hybrid frequency time domain output, 202 Steadystate groundwater ow, 339 Steady-state heat ow, 316 Steady-state response, 521 STEP option nonlinear analysis logging, 249 transient heat ow, 329 STEPS command heat ow, 331 hybrid frequency time domain analysis, 200 hybrid frequency time domain output, 202, 203 initial nonlinear conditions, 220, 224 mixture analysis, 392 nonlinear analysis output, 253, 254 nonlinear load step, 226
645
646
nonlinear time step, 234 postbuckling analysis, 296 postbuckling analysis output, 296 transient analysis output, 133 transient groundwater ow, 353, 353 transient heat ow, 330 transient load step, 131 transient time step, 132 STEPS option mixture analysis, 391 transient groundwater ow, 350 transient heat ow, 327 Stiness matrix, 464 STOP command nonlinear analysis, 219, 246 transient analysis, 131 Stop criterion, 246 Line Search, 242 Storage equation, 555 STORAT command, 347 Storativity, 335 Storativity matrices, 347 Strain, 460, 507 frequency response analysis, 166 linear static analysis, 97 nonlinear analysis, 255 spectral response analysis, 182 transient analysis, 138 STRAIN command frequency response analysis, 166 hybrid frequency time domain output, 202 linear static analysis, 98 nonlinear analysis output, 253, 255 nonlinear stop criterion, 247 spectral response analysis, 182 transient analysis, 138 Strain energy damping, 448, 471 Strain matrix, 508 STRAIN option, 239 mixture analysis, 391 Straindisplacement relation, 458 Stress, 460, 510 frequency response analysis, 169 linear static analysis, 100 nonlinear analysis, 260 separation, 555 spectral response analysis, 185
INDEX
transient analysis, 141 Stress amplitude, 111 Stress analysis, coupled to groundwater ow, 337 heat ow, 315 STRESS command direct response analysis, 160 eigenvalue analysis, 441, 443 frequency response analysis, 169 hybrid frequency time domain analysis, 195 hybrid frequency time domain output, 202 initial stresses, 222 linear static analysis, 101 nonlinear analysis output, 260 nonlinear stop criterion, 248 spectral response analysis, 185 transient analysis, 141 Stress gradients linear static analysis, 105 nonlinear analysis, 265 spectral response analysis, 190 Stress intensity factor, 563 linear static analysis, 109 nonlinear analysis, 276 Stress invariants, 79, 511 linear static analysis, 103 nonlinear analysis, 263 spectral response analysis, 188 Stress matrix, 510 STRESS option, 239 Stress points beam elements, 73 output selection, 79 Stress stiening, 499 Stressstrain relation, 458 STRPNT command, 73 STRPNT option, 79 Structural analysis input, 17 phased, 403 Structural dynamics, 521 Structural elements staggered analysis, 379 Structural loads, 36 SUBFIL option, 64 SUBSTR command, 430, 432, 433 Substructuring, 433, 476
INDEX
Summed crack strains, 259 Superelement, 476 SUPERP command, 409 Superposition subroutines, 586 SUPPOR table, 17 mixture analysis, 387 phased analysis, 410 Supports, 17 phased analysis, 410 SUPPRE command, 239 mixture analysis, 391 Swelling of soil, 383, 390 SYMBOL option results presentation, 267 Symbols, glossary of, xxiii T T input prescribed temperature, 307 units, 6 TA output, 138 TABLE option, 65 TABULA output device, 78, 80 Tabular output, 80 TAKE command, 573 TANGEN option, 241, 242 Tangent stiness, 541, 542 freeze, 241 nonlinear iteration, 241, 242 TApr output mixture analysis, 396 TD output, 134 TDph output, 396 TDpr output mixture analysis, 396 TE output, 271 TEMPER command heat ow, 330, 331 hybrid frequency time domain output, 202 nonlinear analysis, 213, 270 steadystate heat ow, 321 transient heat ow, 325 TEMPER input, 7 TEMPER option output selection, 79 strain, 255 TEMPER subtable of INIVAR, 305 TEMPER table staggered analysis, 375 Temperature heat ow, 331 nonlinear analysis, 270 Temperature load staggered analysis, 381 Temperature stability, 373 TERMIN option, 200, 244 TEXT command, 81 Thermal strain, 259 THETA parameter, 217 Threshold angle nonlinear analysis, 213 TIME command mixture analysis, 393 nonlinear time step, 218, 234 transient time step, 130, 131 Time dependency phased analysis, 425 staggered analysis, 381 Time discretization, 559 Time domain, see Transient analysis TIME input, 6 phased analysis, 425 Time integration, 528 mixture analysis, 390, 391 nonlinear analysis, 216 phased analysis, 422 transient ow analysis, 551 TIME parameter nonlinear analysis, 220 transient groundwater ow, 349 transient heat ow, 325 Time segment, 199 Time steps linear transient analysis, 131 mixture analysis, 393 nonlinear analysis, 234 Timeload diagram, 121 TIMEBO table, 308 TIMEDE command, 217 TIMELO table, 121 mixture analysis, 392 TIMES input transient analysis, 121, 122 transient potential ow, 308, 309 TIMESE command
647
648
hybrid frequency time domain analysis, 199 TOLABT parameter, 201, 244 TOLCON parameter, 201, 244 nonlinear groundwater ow, 343, 352 nonlinear heat ow, 319, 328 TOLDIR parameter, 213 TOLERA parameter, 438, 446 creep, 213 d.o.f. coincidence, 68 plasticity, 213 solution procedures, 430 TOLTCO parameter, 213 TOTAL command, 215 Total Lagrange, 208, 215, 502 TOTAL option displacements, 134 velocities, 136 TOTAL option accelerations, 137 nonlinear stop criterion, 247 nonlinear stress, 261 output selection, 78 strain, 255 temperature, concentration or maturity, 270 TOTALT parameter, 237 TOTCRK command, 213 TOWHAT option, 268 Towhata-Iai liquefaction status output, 268 TR input stability analysis, 281 structural analysis, 41 TRACTI option, 79 Traction frequency response analysis, 169, 172 linear static analysis, 100, 105 nonlinear analysis, 260, 264, 265 spectral response analysis, 185, 189 transient analysis, 141, 144 Trac load ENV 1991-3 code, 51 Transformation element matrix, 460 stress and strain, 512 TRANSI command nonlinear analysis, 210 transient analysis, 130
INDEX
Transient analysis input, 119 linear, 127 mixture, 390 nonlinear, 207, 207, 216 Transient ow analysis phased, 422 staggered, 378 Transient groundwater ow, 346 Transient heat ow, 322 Transient response, 521, 528 TRANSL option Arc-length control, 231 output selection, 78 Transverse position property, 612 TrD output, 134 Truck load, 45, 47 ENV 1991-3 code, 52 Truss elements stability analysis, 282 TrV output, 136 TRVINF input, 49 Turbulence, 337, 375 TV output, 136 TVpr output mixture analysis, 396 TWOBEA input, 49 TWOSID input, 385 Tyings, 18, 20, 461 automatic, 33 mixture analysis, 388 phased analysis, 410 TYINGS table, 18, 20 mixture analysis, 388 phased analysis, 410 TYPE command CMOD Arc-length control, 232 eigenvalue analysis, 437, 439 mixture analysis, 390 nonlinear analysis, 209, 210 transient analysis, 128, 130 U Unconned aquifer, 341, 349 UNDRAI input, 239, 269 mixture analysis, 392 UNITMX subroutine, 586 Units, 3, 381
INDEX
eigenfrequencies, 447 UNITS table, 5 Unity matrix, 586 UNLOCK command, 63 UPDATE command Arc-length method, 230 Lagrange formulation, 215 Updated Lagrange, 208, 215, 502 Updated Normal Plane, 230, 494 USER command, 289 User environment, 573 USER option, 268, 269 Userdened derived results beam elements, 105 plate elements, 105 shell elements, 105 User-supplied material model, 269 User-supplied subroutines, 569, 576 USRRBE option linear static analysis, 102, 105 USRRSH option linear static analysis, 102, 105 USRSTA input interface elements, 269 user-supplied material, 269 UV subroutine, 588 UVMW subroutine, 587 UVPW subroutine, 587 UVPWS subroutine, 587 UVS subroutine, 588 V V output, 164 spectral response analysis, 181 VARIAB option, 64 VBB code, 47 VBB input, 45, 48 Vector subroutines, 586 VELMOD input, 407 VELOCI command frequency response analysis, 164 hybrid frequency time domain output, 202 mixture analysis, 395 nonlinear analysis, 222 nonlinear analysis output, 253 spectral response analysis, 181 transient analysis output, 135 VELOCI subtable of INIVAR, 120 Velocity frequency response analysis, 164 spectral response analysis, 181 transient analysis, 135 VINPRO subroutine, 589 Virtual crack extension, 565 Virtual displacements, 459 Virtual strain energy, 461 VISCOE command, 214 Viscoelasticity, 214 VISCOP command, 214 Viscoplasticity, 214 Viscosity dynamic, 366 Viscous damping, see Damping, Rayleigh VOID option, 257, 260 VOID output, 260 Void ratio, 260 VOLUME option, 79 Volumetric deformation, 560 Volumetric locking, 208 Volumetric strain, 79, 508 linear static analysis, 99 spectral response analysis, 184 transient analysis, 140 VONMIS option, 79 Von Mises strain, 507 frequency response analysis, 167 linear static analysis, 99 nonlinear analysis, 258 output selection, 79 spectral response analysis, 184 transient analysis, 140 Von Mises stress, 510 frequency response analysis, 171 linear static analysis, 103 nonlinear analysis, 263 output selection, 79 spectral response analysis, 188 transient analysis, 143 VOSB code, 47 VOSB input, 45, 48 Vulcanization, 381 W WATER subtable of LOADS, 54 Water waves, 54
649
650
WEIGHT command, 251 mixture analysis, 392 Weight load, 39, 464 mixture analysis, 389, 391 WEIGHT subtable of LOADS, 39 mixture analysis, 389 Width of carriage way, 612, 613 Width of the lanes, 612 WILSON command, 217 Wilson- integration, 217, 530, 559 Wind load, 54 WIND subtable of LOADS, 54 WLANE input, 48 Whler diagram, 111 o Y YD input, 6 YEAR input, 6 Z ZERO option, 213 ZR parameter, 74
INDEX