ImplicitNotes

LS-DYNA IMPLICIT NOTES
Livermore Software Technology Corporation

www.lstc.com
January, 2000
For Version 950
Copyright  2000
All rights Reserved
Notes Developed by
Ala (Al) Tabiei, Ph.D.

The Center of Excellence in DYNA3D Analysis 
www.ase.uc.edu/~atabiei
&
Brad Maker, Ph.D. (LSTC’s Principle Code Developer)
Mailing Address
7374 Las Positas Road

Livermore, CA 94550
USA
Tel: (925) 449-2500

Fax: (925) 449-2507
EMAIL: sales@lstc.com
ACKNOWELEDGMENTS
The author wishes to thank several people for valuable comments and discussions on the notes.
Firstly, Dr. J. Hallquist advance course notes were of great help in constructing the lecture notes.
Many thanks goes to Dr. B. Maker for valuable comments, suggestions, and discussions. Thanks
also go to the LSTC staff for their support.
TABLE OF CONTENTS
Introduction 1-5
Finite Element Equation & Time Integration 2-13
Equilibrium and Nonlinearity 3-21
Incremental Equations and Linearization 4-25

Automatic Time Step Control 4-37
Nonlinear Solution Methods for Implicit Analysis 4-39
Finite Element Modeling Techniques 5-49

1. Engineering a FEA Model: 5-49
2. Element Selection 5-50
3. Mesh Density 5-60
4. Symmetry 5-62
5. Modeling for Physical Phenomenon 5-67
6. Ad – Hoc Guidelines 5-72
7. How to Tell If Your Results Are Correct 5-73
Contact and Friction 6-77

Contact Algorithm 6-77
Friction 6-86
Material Nonlinearity 7-88
Damping 8-94
Reduce Integration & Hourglass Phenomenon 9-96

Reduced and Selective Reduce Integration 9-96
Hourglassing 9-97
Quasi-Static Initialization and Quasi-Static Simulations 10-100

Quasi-Static Initialization 10-100
Quasi-Static Simulations With Explicit FE 10-101
Adaptive Meshing 11-104
Springback 12-108
Trouble Shooting 13-112

Workshop 14-118
INTRODUCTION
_______________________________________________________________________________________________
INTRODUCTION
LS-DYNA OFFERS NEW IMPLICIT ANALYSIS CAPABILITY
Version 950 includes the first release of implicit analysis capability using LS-DYNA. Initially
targeted at metal forming springback simulation, this new capability allows static stress analysis
without interfacing to LSTC's implicit software package LS-NIKE3D. Over time, increased
implicit functionality will be added to LS-DYNA, and the need for LS-NIKE3D will be eliminated.
LSTC anticipates that including both explicit and implicit capability in a single software package
will simplify both code development and analysis.
IMPLEMENTATION STRATEGY
The new implicit features in LS-DYNA were not simply created using subroutines from LS-
NIKE3D. Instead, the existing element and material routines in LS-DYNA were augmented to add
stiffness matrix calculations when the implicit mode is active. This approach has several
advantages:
keyword input and arbitrary ID numbering are available in implicit mode

existing optimized subroutines and data structures are used
parallel processing capability is immediately available
post-processing interfaces remain unchanged
This means that all of LS-DYNA's optimized programming and data structures are used in the
implicit mode, so high speed is preserved. Furthermore, pre- and post-processing interfaces are
virtually unchanged, so that existing support software can be used with LS-DYNA's new implicit
features. This also provides the first opportunity for LSTC customers to use an implicit code with
keyword or structured format input, and fully arbitrary ID numbering.
FEATURES AVAILABLE FOR IMPLICIT MODE
One disadvantage to this low-level, closely coupled implementation is that all of the functionality
of LS-NIKE3D is not immediately available in LS-DYNA. Instead, implicit stiffness matrix
calculations must be added individually to activate each LS-DYNA feature for implicit mode. The
summary below shows features currently available for implicit analysis. Contact LSTC or your LS-
DYNA distributor if you need a feature which is not shown, since this list is expanding rapidly.
_______________________________________________________________________________________________
LS-DYNA IMPLICIT NOTES 1-5
INTRODUCTION
_______________________________________________________________________________________________
LS-DYNA FEATURES CURRENTLY AVAILABLE FOR IMPLICIT ANALYSIS

(as of end of 1999)
solid element formulations 1, 2

solid material models 1,2,3,5,6,7,12,13,18,20,24
beam element formulation 1

beam material models 1, 20
shell element formulations 2,6,10,16

shell element material models 1,2,3,18,20,24,36,37
2-D solid element formulations 12,13,15

2-D solid material models 1,2,3,5,6,7,12,13,18,20,24
discrete springs
contact interface types 3, m3, 5, m5, 10, m10, 13:

SURFACE_TO_SURFACE
NODES_TO_SURFACE
ONE_WAY_SURFACE_TO_SURFACE
FORMING_SURFACE_TO_SURFACE
FORMING_NODES_TO_SURFACE
FORMING_ONE_WAY_SURFACE_TO_SURFACE
AUTOMATIC_SINGLE_SURFACE
2D_AUTOMATIC_SURFACE_TO_SURFACE
Implicit Analysis Features nonlinear equation solvers: Linear, Full Newton, Quasi-Newton, Arc
Length, sparse linear equations solvers: direct incore/out-of-core, CG and Lanczos iterative
automatic time step control, Newmark method for implicit dynamics, artificial stabilization for
multi-step springback analysis, eigenvalue analysis, seamless switching for springback analysis.
ACTIVATING IMPLICIT MODE
LS-DYNA's implicit mode may be activated in two ways. Using one of the new
*CONTROL_IMPLICIT keywords, a simulation may be flagged to run completely in implicit
mode. Alternatively, an explicit simulation may be seamlessly switched into implicit mode at a
specific time using the *INTERFACE_SPRINGBACK_SEAMLESS feature. (All of the implicit
controls and options are also available using structured format input). The seamless switching
feature is intended to simplify metal forming springback calculations, where the forming phase can
be run in explicit mode, followed immediately by an implicit static springback simulation. In case
of difficulty, full restart capability is supported.
Seven new keywords are available to support implicit analysis, along with corresponding new
control cards in the structured format input deck. Default values are carefully selected to minimize
input necessary for most simulations. These are summarized below:
_______________________________________________________________________________________________
INTRODUCTION
_______________________________________________________________________________________________
Activating Implicit Analysis
Three types of analyses can be performed

• fully explicit (default)
• fully implicit
• explicit followed by implicit ("seamless" springback)
New keywords have been added
*CONTROL_IMPLICIT_GENERAL
*CONTROL_IMPLICIT_LINEAR
*CONTROL_IMPLICIT_NONLINEAR
*CONTROL_IMPLICIT_AUTO
*CONTROL_IMPLICIT_STABILIZATION
*CONTROL_IMPLICIT_DYNAMICS
*CONTROL_IMPLICIT_EIGENVALUE
● not all features are available in implicit mode
Implicit Keywords
*CONTROL_IMPLICIT_GENERAL (required for implicit)

• activates implicit mode
• defines implicit time step size (standard LS-DYNA termination time is used too)
• optional if *INTERFACE_SPRINGBACK_SEAMLESS is used
*CONTROL_IMPLICIT_LINEAR (optional)
• parameters for linear equation solver, which inverts stiffness matrix: [K]{x}={f}
• does NOT invoke a "linear" analysis
*CONTROL_IMPLICIT_NONLINEAR (optional)
• parameters for nonlinear equation solver (Newton-based methods)
• controls iterative equilibrium search, convergence
• "linear" analysis selected here (a special case where no iterations are performed)
*CONTROL_IMPLICIT_AUTO (optional)
• activates automatic time step control
• back up, try again if equilibrium iterations fail
• default is fixed time step size, error termination if any steps fail to converge
_______________________________________________________________________________________________
INTRODUCTION
_______________________________________________________________________________________________
*CONTROL_IMPLICIT_DYNAMICS (optional)
• include inertia terms: M∆a n +1 + K∆u n +1 = f next+1 − f nint − Ma n
• problem "time" must now be real, physical time
• can improve convergence, especially when (even weak) rigid body modes are present
*CONTROL_IMPLICIT_EIGENVALUE (optional)
• signals LS-DYNA to perform eigenvalue analysis, then stop
• number of eigenvalues/vectors, optional frequency shift
*CONTROL_IMPLICIT_STABILIZATION (optional, metal forming only)

• allows multi-step springback by unloading internal stresses over several steps
• automatic springs attached to shell element nodes, slowly removed
• stabilization must be "completely removed" for accurate results
User's Manual contains helpful notes on each input parameter
A Note on Implicit Numerical Methods
Implicit governing equations contain two problems to solve

M∆a n +1 + K∆u n +1 = f next+1 − f nint − Ma n
Nonlinear Problem: *CONTROL_IMPLICIT_NONLINEAR

• find displacements u which satisfy equilibrium fext=fint
• both K, fext and fint can be nonlinear functions of u
• iterative search employed using Newton-based method
• interactive switch "<ctrl-c> nlprint" toggles diagnostic output
Linear Problem: *CONTROL_IMPLICIT_LINEAR

• solve system of linear algebraic equations
• must solve during every nonlinear iteration
• great CPU and memory cost makes this problem important
• interactive switch "<ctrl-c> lprint" toggles diagnostic output
Implicit Eigenvalue Analysis
Extract "n" eigenvalues by subspace iteration

• lowest "n" frequencies
• optional frequency shift: "n" eigenvalues nearest to shift frequency
• initial stress influences frequencies when S/R Hughes-Liu shell is used
Two additional databases output

• d3eigv: binary plot database with each mode shape
_______________________________________________________________________________________________
INTRODUCTION
_______________________________________________________________________________________________
• eigout: text file summarizing frequencies found
Simple input parameters
activates implicit analysis
*CONTROL_IMPLICIT_EIGENVALUE
activates eigenvalue solver
input number of eigenvalues, optional shift frequency
Implicit Boundary Conditions
Kinematic Boundary Conditions

• prescribed displacement, velocity (global directions only)
• constrained nodes (*NODE constraints, or SPCs)
• rigid bodies, extra nodes for rigid bodies, nodal rigid bodies
Applied Loads
• force, pressure, gravity, angular acceleration
• most loads which contribute only a RHS term, no stiffness contribution
Implicit Output Databases
Binary Output Databases

• plot, time history files in standard LS-DYNA format
• iteration plot database "d3iter" activated interactively "<ctrl-c> iter"
deformed geometry each nonlinear iteration
good for debugging convergence problems
ASCII Output Databases

• NODOUT, ELOUT, GLSTAT, RCFORC, NODFOR, RBDOUT
• same format as explicit
FUTURE DEVELOPMENTS
The list of features available for implicit analysis within LS-DYNA is growing each week.
Customers are encouraged to notify either LSTC or LS-DYNA distributors about important
features needed for their applications. Development priority will be strongly customer driven.
LSTC anticipates that an important future application area will be implicit stress initialization,
followed seamlessly by an explicit simulation. This will likely become the most common
application of implicit LS-DYNA.
_______________________________________________________________________________________________
INTRODUCTION
_______________________________________________________________________________________________
PROGRESS OF ITERATIONS
Control parameters for the nonlinear equation solver are input using the
*control_implicit_nonlinear keyword. By defaults, the progress of the equilibrium search is not
shown to the screen. This output can be activated using the “NLPRINT” input parameter, or
interactively toggled on and off by entering “<ctrl-c>” nlprint. The box below shows a typical
iteration sequence, where the norms of displacement and energy are displayed. When these norms
are reduced below user prescribed tolerances, equilibrium is reached within sufficient accuracy, the
iteration process is said to have converged, and the solution proceeds to the next time step.
_______________________________________________________________________________________________
INTRODUCTION
_______________________________________________________________________________________________
The next box depicts a typical print-out of the defult BFGS nonlinear equation solver. Several
automatic stiffness reformation are performed, initialy due to divergence, and later when the defult
limit of 10 iterations is exceeded.
_______________________________________________________________________________________________
INTRODUCTION
_______________________________________________________________________________________________
IN DYNAMIC SIMULATION---TIME STEPS

IN STATIC SIMULATION---LOAD STEPS
(multiple steps may be used to divide the nonlinear behavior into manageable pices)
_______________________________________________________________________________________________
FE EQUATIONS AND TIME INTEGRATION
_______________________________________________________________________________________________
The equations of equilibrium governing the dynamic response of the system are:
[M ]{u}+ [C ]{u}+ [K ]{u}= {R ext (t )} (1)
where
[M] is the mass matrix,
[C] is the damping matrix,
[K] is the stiffness matrix,
{u}, {u}, and {u} are the nodal accelerations, velocities, and displacements vectors respectively,
{ }
R ext (t ) is the external forces vector.
Equation (1) can be rewritten in the form
{FI (t )}+ {FD (t )}+ {R int (t )}= {R ext (t )} (2)
where
{FI (t )}= [M ]{u}
{FD (t )} = [C ]{u}
{ }
and R int (t ) = [K ]{u}
are the inertia force, the damping force, and the internal forces vectors all of which are time
dependent. Equations (1), which represent a system of nonlinear second order differential
equations, are solved with the help of the LSDYNA Explicit or Implicit finite element codes using
direct time integration techniques.
Direct Time Integration

When solving dynamic problems with the finite element method, the solution is sought by dividing
the total response time of the system into much smaller time intervals called time steps or time
increments. The equilibrium equations are solved and the values of the unknowns are determined at
time t + ∆t based on knowledge of their values at time t (quasi-linearization).
Using these values we continue solving the linearized differential equations at t + 2∆t, and so on,
for the entire response time of the system.
When employing time integration, LSDYNA uses the implicit integration operator and the explicit
integration operator.
The implicit integration operator definition is completed by the Newmark formulae for
displacement and velocity integration
[
u t + ∆t = u t + ∆t u t + ∆t 2 (1 / 2 − β )u t + β u t + ∆t ] (3)
[ ]
u t + ∆t = u t + ∆t (1 − γ )u t + γ u t + ∆t ,
_______________________________________________________________________________________________
_______________________________________________________________________________________________
where β and γ are parameters of the system. Thus, expressing the velocities and accelerations at
t + ∆t in terms of the displacements at t + ∆t and substitution into equations (1) yields:
[K ]{u }= {F }
t + ∆t t + ∆t
(4)
where
[K ] = [K ([K ], [M ], [C ], ∆t )] is the effective stiffness matrix, and
{F }= {F ({F }, [M ], [C ], ∆t, {u }, {u }, {u })} is the effective load at time t + ∆t.
t + ∆t t + ∆t t t t
The explicit dynamic analysis is based on integrating the equations of motion for the system using
the explicit central difference formula
u t =
1
u
∆t 2 t −∆t
(
− 2u t + u t + ∆t )
(5)
u t =
1
u
2∆t t + ∆t
(
− u t −∆t )
Thus equations (1) take the form
[M ]{u t + ∆t / 2
}= {F } t
(6)
where
[M ] = [M ([M ], [C ], ∆t )] is the effective mass matrix, and

{F }= {F ({F }, [K ], [M ], [C ], ∆t, {u }, {u })} is the effective load at time t.
t t t t − ∆t / 2
Equations (4) are solved at time t + ∆t and the corresponding solution methods are called implicit,
while equations (6) are solved at time t, and the corresponding methods are called explicit.
The explicit central difference operator is very convenient when lumped mass matrix can be
assumed and velocity-dependent damping can be neglected. Then [M ] is a diagonal matrix and the
solution is achieved automatically without having to solve the system of equations.
On the other hand the explicit operator is only conditionally stable, which means that in order to
obtain accurate results, the time step must be smaller than a certain critical value which is defined
by the mass and stiffness properties of the complete element assemblage.
Depending on the particular problem, the stable time increment of the model may be very small
which would require too many steps to solve and therefore extremely large CPU time. In such
cases the implicit method, which is unconditionally stable, may provide better efficiency.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
EQUATION OF MOTION FOR TRANSLATIONAL VELOCITIES:
dv
M{ } = {R ext } − {R int } + {R bod } , however in here we have additional terms as follows:
dt
dv
M { } = {R ext } − {R int } + {R bod } + {R hgr } + {R cont }
dt
M = ∑ ∫ ρN I N J dV is the mass matrix
elements V
{R } =
ext
∑ ∫N
elements S
I pi d S is the externally applied loads vector
∂N I
{R int } = ∑ ∫σ
elements
ij
∂χ j
dV is the internal forces vector
V
{R bod } } = ∑ ∫N
elements V
I ρbi dV is the body forces vector
{R hgr } = ∑r
elements
hgr
is the hourglass resistant forces vector
{R cont } = ∑r
contact surfaces
cont
is the contact forces vector
V and S are the actual volume and external surface
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Flowchart for Explicit Time Integration
1
−
Initial conditions σ = σ (t = 0); u = u (t = 0), n = 0
0 2
1
u(t = 0) = [ Roext − Roint − Cu (t = 0)]
mii
1 1
n+ n−
Update: Velocities u 2
= u 2
+ ∆t un
1
n+
n +1
Displacements u = u + ∆t u
n 2
Compute internal nodal forces
Loop over elements: E = 1 to NELE

ε n+1 / 2 = B u E n+
1/ 2
Velocity-strains:
1 1
∇n + n+
Stress rates: σ = T (ε )
2 2
Update stress: σ n +1
= σ n + ∆t σ n +1 / 2
= ∫ B T σ n +1 d V
n +1
Internal forces: rint
V
n +1 n +1
Assemble: rint into Rint
n +1
Compute external nodal forces Rext
1 n +1 n +1
Accelerations: un+1 = [ Rext − Rint − Cu n +1/ 2 ]
mii
Output n Ån +1; go to 2
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Implicit Governing Incremental Equations
To obtain solution at load increment n+1 given solution at load increment n, the linearized
equations are:
M u n+1 + K t ( x )∆u 0 = R ( x ) n+1 − R ( x )

.. n ext n int n
M: lumped mass matrix

Kt : tangent stiffness matrix based on geometry at n
ext
R : external load based on applied load at step n+1, but geometry at n
int
R : internal load based on displacement state and stress at load step n
Using Newmark method lead to the following equation
∗ ∗
K ∆u 0 = R ( x ) n+1 − R ( x )
ext n n
where
∗ 1
K = Kt + M
β∆t 2
.
∗ un 1 1
R = R (x ) − M [ + ( − β )u n ]
int n ..
β∆t β 2
Solution of the above equation yields ∆u 0 . The coordinate vector is updated using
x n+1 = x n + ∆u 0
The iteration for equilibrium now begins using:
∗ n +1 ∗ n +1 n +1
K j ( x j )∆u i = R ( x i ) n+1 − R ( x i ) = Q i j≤i
n ext
Where i denote the iteration number.
Convergence is assumed if ε d and ε e is ≤ some defined values.
( )M
Advantages as DT becomes small
mass contribution to stiffness matrix grows rapidly: K ∗ = K + 1
β∆t 2
stabilizing effect to equilibrium iterations
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Implicit Flow Chart
1. Set n = 0 , u 0 = u (t = 0), u 0 = u (t = 0) select ∆t , β , γ

. .
int ∗ ∗
2. Form M , R 0 , K t 0 , K 0 , and R 0
−1
3. Calculate u 0 = M ( R 0 − R 0 )
.. ext int
4. Set i = 0
∗
5. Solve for ∆u i=0 using K ( x )∆u 0 = R ( x ) n +1 − R ( x )
n ext n int n
n +1
= x + ∆u 0
n
6. Update coordinate vector x
∗ n +1
7. Calculate σ n+1 , R ( x )
8. Set i = i + 1
9. Solve for ∆u i using
∗ n +1 ∗ n +1 n +1
10. K j ( x j )∆u i = R ( x i ) n+1 − R ( x i ) = Q i j≤i
n ext
11. If convergence is not attained and NO divergence
n +1 n +1
12. Set x i = x i + ∆u i and go to 8
13. Otherwise n ← n + 1 and
n +1 ∗ n +1 ∗
14. Set x i=0 = x i and R ( x i=0 ) = R ( x i )
n n
. ..
15. Calculate u n and u n
16. Go to 4 if t<T, otherwise terminate.
Divergence is defined as when the residual force increases during iterations
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Implicit vs. Explicit FEA
Explicit FEA codes have enjoyed much success

● ease of use
● robustness
● capacity
However, latest generation computers offer new opportunities for implicit FEA codes
● huge memory is now common (RAM > 1.0 Gbyte)
Several crucial enhancements have been added, allowing implicit codes to offer a credible
alternative to explicit
● material, contact algorithms (rigid tooling)
● fast linear equation solver
● robust nonlinear solver
● keyword input, unified I/O databases
Summery
Conservation of momentum: ρu, tt = ∇ •σ + f
• LS-DYNA - Explicit time integration Ma n = f next − f nint

➤ impact, penetration, high rate dynamics
➤ many small time steps
➤ Courant condition limits largest stable time step
● LS-DYNA - Implicit time integration M∆a n +1 + K∆u n +1 = f next+1 − f nint − Ma n

➤ static, eigenvalue, low rate dynamic analyses
➤ few large time steps
➤ Linear equation solver - stiffness matrix
➤ Nonlinear equation solver - user chooses step size
Quasi-Static analysis: "time" represents a monotonically increasing parameter which characterizes

the evolution of the loading.
LS-DYNA offers several implicit nonlinear solvers:

• Full Newton - reform K ∗ every iteration
• Modified Newton - reform K ∗ every j iterations
• Quasi-Newton - employ simple, approximate "update" to K ∗
In every case, accuracy is maintained by re-evaluating R ∗ each iteration
_______________________________________________________________________________________________
_______________________________________________________________________________________________
_______________________________________________________________________________________________
EQUILIBRIUM AND NONLINEARITIES
_______________________________________________________________________________________________
∫B

The concept of mechanical equilibrium is: σ dν = Rexternal
v
Geometric Nonlinearities
∫B σdV = Rexternal
T
v
↑↑
Geometric nonlinearities are caused by:

Nonlinear relationship between displacement increments and strain increments
Integration over current volume
Effects:
Stress- stiffening
Bifurcation, buckling, and collapse
Snap-through
Necking
Other large displacement effects
Material Nonlinearities
T
v
Material nonlinearities are caused by:

Dependence of stress on current strain
Effects:
Plasticity
Plastic hinge formation and plastic collapse
Rubber nonlinear elasticity
Other nonlinear material effects
Boundary Nonlinearities
∫B σ dV = Rexternal
T
v
Boundary nonlinearities are caused by:
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Dependence of Rexternal on current displacements
Effects:
Contact across body surfaces (by means of load transfer)
Nonlinear external loads
Pressure load nonlinearities
Follower forces
Linear Equilibrium
For each F , there will always be only one solution u (existence and uniqueness).
If F causes a displacement of u , then (aF ) causes a displacement (au ) (scaling).
If F causes a displacement u and P causes a displacement v then F + P causes a
displacement u + v (superposition).
In a linearized problem, the solution u is determined by the current value of the load
Rexternal .
Ku = Rexternal
Nonlinear Equilibrium
For a particular F , there may be none, one, many, or an infinite number of solutions u
(non-existence and non-uniqueness).
If F causes a displacement of u , then (aF ) probably doesn’t cause (au ) (no scaling)
If F causes a displacement u and P causes a displacement v, then (F + P ) probably
doesn’t cause (u + v ) (no superposition).
In a real (nonlinear) problem, the “unique” solution is determined by the entire load history
of Rexternal
∫ B T σdV = Rexternal
v
Examples of non-unique solutions
Shallow spherical caps

Plasticity
Linear vs. Nonlinear FEM
The FEM equation corresponding to linearized equilibrium is a matrix equation in N dof

unknowns: Ku = Rextermal
The FEM equation corresponding to true (nonlinear) equilibrium is a vector equation in
N dof unknowns: ∫ B T σdV = R
v
_______________________________________________________________________________________________
_______________________________________________________________________________________________
The solution of a linear FEM model involves the solution of the linear N dof × N dof matrix
equation: Ku = Rexternal
The solution of a nonlinear FEA model involves the simultaneous solution of N dof
T
nonlinear vector equations:
v
There are no good, general methods for solving systems of more than one nonlinear
equation. There never will be any good, general methods.
Æ Nonconvergence
What LS-DYNA Does:
Solves the equation iteratively by using some of the Newton methods (or variations of these
methods) to find an approximate solution that minimizes the residuals!
What are “ residuals” ?
Re-write the equilibrium equations as a function r of u :

r (u ) = ∫ B T σdV − R
v
r (u ) are the residuals (out-of-balance forces) at u . r (u ) is nonlinear!

In general, r (u ) ≠ 0 , but, in equilibrium, r (u ) = 0
Interpretation
The residuals are just the distribution and the magnitude of extra external force at each DOF
needed to bring the structure into equilibrium at u :
r (u ) = ∫ B T σdV − Rexternal ≠ 0 , u is not in equilibrium under Rexternal
v
0 = ∫ B σdV − {Rexternal + r (u )}, u is in equilibrium under Rexternal + r (u )

T
v
Newton Method
Assume you are not in equilibrium at displaced position u so that r (u ) ≠ 0

Find ∆u so that u + ∆u is in equilibrium ⇒ r (u + ∆u ) = 0
Expand r (u ) in a Taylor’s series about current displacement u :
∂r
r (u + ∆u ) = r (u ) + ∆u + ... = 0
∂u u
Throw away higher order terms and solve the resulting equations for ∆u :
∂r
∆u = − r (u )
∂u u
Substitute the equilibrium equation into Newton-Raphson scheme :
∂r
∆u = − r (u )
∂u u
_______________________________________________________________________________________________
_______________________________________________________________________________________________
K tan gent ∆u = − ∫ B T σdV + Rexternal this equation is linear!

v
The solution of a linear FEA model involves single solution of the linear N dof × N dof matrix
equation: Ku = Fexternal
The solution of a nonlinear FEA model involves many solutions of the linear N dof × N dof
matrix equation: K tan gent ∆u = − ∫ B σdV + Rexternal
T
v
_______________________________________________________________________________________________
INCREMENTAL EQUATIONS & LINEARIZATION
_______________________________________________________________________________________________
• MATRIX EQUATIONS
Mx(t ) + F (x, x ) − P(x, t ) = 0
x, x , x = n dimensional acceleration, velocity, and geometry vectors

M = n × n mass matrix
P = body force and external load vector
F = ∑ ∫ B tσdVe
e Ve
• NONLINEARITIES
Nonlinearities are due to geometric affects and inelastic material behavior
Fe (x, x ) = ∫ B tσ (ε , ε )dVe
Ve
Where ε and ε are the strains and the strain rates. If linear
Fe (x, x ) = Cx + K∆x
Where K, C and ∆ x are the stiffness matrix, damping and displacement vector. Nonlinearities also
arise in P due to the geometry dependant loadings.
• RESIDUAL VECTOR
Regardless of whether an implicit or explicit scheme is used, we require that
Q = Mx + F − P = 0
If linear
Q = Mx + Cx + K∆x − P = 0
• LINEARIZATION
The residual vector becomes an implicit function of x n+1 only. We seek he vector x n+1 such that
( )
Q x n+1 = 0
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Assume an approximation xkn+1 to x n+1 for k=1,2,3, …..

( )
In the neighborhood of xkn+1 we use the linear approximation to Q x n+1 given by:
( ) ( ) ( )
Q x n+1 = Q xkn−+11 ∆xk + J xkn−+11 ∆xk
n +1 n +1
xk = xk −1 + ∆xk
∂Q n+1
( )
J xkn−+11 = |x
∂x k −1
The Jacobian matrix is expressed as
∂x ∂F ∂x ∂F ∂P
J =M + + −
∂x ∂x ∂x ∂x ∂x
∂x ∂x ∂P
J =M + C + Kt −
∂x ∂x ∂x
where
∂F
C= is the tangent damping matrix
∂x
∂F
Kt = is the tangent stiffness matrix.
∂x
• STRAIN CALCULATION
To prevent locking during incompressible flow, we impose a constant volumetric strain over the
element. A consistent B matrix for the constant pressure assumption is employed in the tangent
stiffness matrix. B is defined by Hughes in: T. R. Hughes, “Generalization of selective Integration
Procedures to Anisotropic and Nonlinear Media”, Int. J. for Numerical Methods in Engineering,
Vol. 15 no. 9, pp. 1413-1418 (1980).
Based on midpoint geometry
1  ∂xi 2 ∂x j 2 
n+ 1 n+ 1
εij =  n+ 1 + n+ 1 
2  ∂x 2 ∂x 2 
 j i 
this strain measure is a valid large strain measure , i.e., rigid body rotations do not cause straining
• IMPLICIT ALGORITHM
( )
K t x n ∆uo = P x n( ) n +1
( )
− F xn
K t = Positive –definite tangent stiffness matrix.

∆uo = Desired increment in displacements.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
( ) n +1
P x n = External load vector at n+1 based on the geometry at time n.
( )
F x n = Stress divergence vector at time n.
Update displacement vector
n +1
x1 = x n + s0 ∆u0
and begin iterations for equilibrium
K t j ∆ui = P xi ( )
n +1 n +1
( )= Q
− F xi
n +1
i
n +1
where the subscript i denotes the iterate and j<i. s0 is a parameter between 0 and 1 found from a
line search.
After each iteration, convergence is checked, convergence is assumed if the conditions:
∆ui
< εd
u max
and
t
ui Q i
t
< εe
u0 Q i
are satisfied.
If convergence Is not attained , we update the displacement vector
xin++11 = xin+1 + si ∆ui
and perform anther iteration. Lack of convergence within allowable number of iteration
(default=10) or divergence
Q0n+1 < Qin++11
Causes K t to be reformed. Termination occurs if allowable number of reformations is reached

(default =15).
The foregoing iteration method with K t held constant is called the modified-Newton method.
When convergence problems arise the stiffness matrix is reformed.
Four methods for updating the stiffness matrix are available:
1. BFGS
2. Broyden’s first method
3. Davidon
4. DFP
_______________________________________________________________________________________________
_______________________________________________________________________________________________
These are called quasi-Newton methods.
Quasi-Newton methods involve
Line searches
Stiffness updates
Slightly more expensive than modified Newton but results in more stable program
• QUASI-NEWTON SCHEMES
The secant matrices K i are found via the Quasi-Newton equations
K i ∆ui−1 = ∆Qi
where
∆Qi = Qi−1 − Qi = ( Pi−1 − Fi−1 ) − ( Pi − Fi )
Two classes of matrix updates that satisfy the Quasi-Newton equations are of interest
Rank one updates K i = K i −1 + αzx t
Rank two updates K i = K i −1 + αzx t + βvy t
Recall the Quasi-Newton equation K i ∆ui−1 = ∆ui
Subsitituting
K i = K i −1 + αzx t
we find
K i−1∆ui −1 + αzx t ∆ui−1 = ∆Qi
By choosing
α = 1 / x t ∆ui−1
z = ∆Qi − K i−1∆ui−1
we can satisfy the equation.

Note that x is an arbitrary vector but is restricted such that
∆uit−1 x ≠ 0
Broyden set x = ∆ui−1 and obtained the update
∆Qi − K i−1∆ui −1 t
K i = K i−1 + ∆ui−1
∆uit−1∆ui−1
_______________________________________________________________________________________________
_______________________________________________________________________________________________
resulting in non symmetric secant matrices.

Davidon chose
x = ∆Qi − K i−1∆ui−1
and obtained the updated formula
∆Qi − K i−1∆ui −1
K i = K i−1 + (∆Qi − K i−1∆ui−1 )t
(∆Qi − K i −1∆ui−1 ) ∆ui −1
t
resulting in symmetric secant matrices.

The inverse forms are found by the Scherman-Morrison formula:
A−1ab t A−1
( A + ab t ) −1 = A−1 −
1 + b t A−1a
where A is a nonsingular matrix

We can find the inverse form for Broyden’s update by letting
∆Qi − K i −1 ∆u i −1
a=
∆u t ∆ui −1 i −1
b = ∆ui−1
in the Sherman-Morrison formula. Therefore,
(∆ui−1 − K i−−11∆Qi )∆uit−1ki−−11

K i−1 = K i−−11 +
∆uit−1K i−−11∆Qi
The inverse form for Davidon’s method can likewise be found:
(∆ui−1 − K i−−11∆Qi )(∆ui−1 − K i−−11∆Qi )t

K i−1 = K i−−11 +
(∆ui −1 − K i−−11∆Qi ) t ∆Qi
Again recall the Quasi-Newton equation K i ∆ui−1 = ∆Qi
Substituting K i = K i −1 + αzx t + βvy t ∆ui−1

We find K i−1∆ui −1 + αzx t ∆ui−1 + βvy t ∆ui−1 = ∆Qi
Let
_______________________________________________________________________________________________
_______________________________________________________________________________________________
α = 1 / x t ∆ui−1
β = 1 / y t ∆ui−1
z = − K i −1∆ui −1
v = ∆Qi
Here x and y are vectors that are non-orthogonal to ∆ui −1 i.e.,
∆uit−1 x ≠ 0 and
∆uit−1 y ≠ 0 and ∆uit−1 y ≠ 0
In the BFGS method
x = K i−1∆ui −1
y = ∆Qi
Leading to the following updates formula
∆Qi ∆Qit K i−1∆ui −1∆uit−1 K i−1

K i = K i−1 + −
∆uit−1∆Qi ∆uit−1 K i−1∆ui −1
That preserves the symmetry of the secant matrix. A double application of the Sherman-Morrison
formula leads to the inverse form.
Special product forms have been derived for the DFP and BFGS updates and exploited by Matthies
and Strang (1979).
K i−1 = ( I + wi vit ) K i−−11 ( I + vi wit )
The primary advantage of the product form is that the determinant of K i and therefore the change
in condition number can be easily computed to control updates.
Update vectors
Let,
δ i = ∆ui−1
γ i = Qi−1 − Qi
 ∆u γ 1 
vi = Qi − Qi −1 1 + ( i i ) 2 
 δ i Qi−1 
δ
wi = i
δ iγ i
Determinant
_______________________________________________________________________________________________
_______________________________________________________________________________________________
det( K i ) = det( K i −1 )(1 + wt v) 2

Change in condition number
{ }
4
(v t v) 12 ( wt w) 12 + v t vwt w + 4(1 + v t w) 1
2 
 
c= 
[
4(1 + v t w)
2
]
• LINE SEARCHES
• With Quasi-Newton updates line searches are necessary to achieve convergence.

• After ∆u is determined, a line search may be performed
• search only if global response is hardening
• prevents divergence of Newton method
• find multiplier which gives best estimate of equilibrium
compute trial displacement u tr = ui + s∆u, 0 ≤ s ≤1
evaluate out-of-balance force R (u tr ) = f ext (u tr ) − f int (u tr )
search (iterate) to find s which minimizes out-of-balance force
if s falls below 0.001, discard ∆u and reform K
∆ui = sd i
where
d i = K i−1Qi
ui+1 = ui + si d i
Assume solution lies in direction d i . The line search then component si such that components of
(P-F) in the search direction is zero
d i .[P ( xi+1 ) − Fi ( xi+1 )] = 0
Broyden:
K i−1 = (1 + α i−1wi−1∆ui−1 )....(1 + α1w1∆u1t )(1 + α $ w$ ∆u$t )k$−1 where

wi = ∆ui − K i−1Qi + K i−1Qi+1 and α i = 1 / ∆uit (∆ui − wi ) are the update factors.
Broyden’s Algorithm:
_______________________________________________________________________________________________
_______________________________________________________________________________________________
1. Solve d = K $−1Qi
2. Do i = 1, k − 1
Recall β = ∆uit−1d
d = d + βα i −1wi−1
wk −1 = ∆ui−1 − ∆uk −1 / s + d
3.
α k −1 = 1 / ∆ukt −1 (∆uk −1 − wk −1 )
β = ∆ukt −1d
4. d k −1 = d + βα k −1wk −1
• ARC-LENGTH METHOD
It is well known that the foregoing approach fails in the neighborhood of limit points.
Geomtrically, the plane of λ =constant does not necessarily intersect the load deflection curve. In
the arc length method of Riks/Wempner a constraint equation is added to limit the load step to a
constant “arc length”, in load/displacement space.
Consider F − λp = 0
where λ is the proportional load factor. Newton’s method leads to the incremental equations
i
K∆x ( j ) = iλp − i F
in classical approach λ = constant and iterations proceed until equilibrium is reached.
• ARC-LENGTH METHOD OUTLINED
1. Nonlinear equilibrium equation

R − λp = 0
2. Linearized equation
K i ∆ui = λp − R(ui )
xi+1 = xi + ∆ui
equivalently,
ui+1 = ui + ∆ui
3. Constraint equation
ui ui + λ2 = s 2
t
4. Load level is split into 3 parts

λ = λm + λi + ∆λi
_______________________________________________________________________________________________
_______________________________________________________________________________________________
λm = load level representing an equilibrium state at the beginning of a load increment

λi = increment load level in iteration i
∆λi = change in load level in iteration i
5. Iterative displacements
∆ui = K −1 (λP − R ) = K −1 [∆λi P + (λi + λm )P − R ]
∆ui = ∆λi ∆ui + ∆ui
I II
∆ui = K −1 P
I
∆ui = K −1 [(λm + λi ) P − R ]
II
∆u I represents displacement due to constraint condition

∆u II represents usual displacement increment
6. Linearization of the constraint equation

f (u , λ ) = ui ui + λ2 − s 2 = 0
t
∂f i ∂f
∆ui + i ∆λi + f (ui , λi ) = 0
∂u ∂λ
or γ i ∆ui + α i λi + f i = 0
7. We can now solve for ∆λi

Substituting ∆ui = ∆λi ∆ui + ∆ui
I II
Into γ t ∆ui + α i ∆λi + f i = 0

f i + γ i ∆ui
t II
Yields ∆λi = −
α i + γ i ∆ui
t I
Input Parameters for Nonlinear Solver
*CONTROL_IMPLICIT_NONLINEAR:
NSOLVR: select Newton method

• .eq.1: linear aproximation (no equilibrium iterations)
• .eq.2: BFGS quasi-Newton method (DEFAULT)
•
ILIMIT: set limits for number of iterations before re-evaluating
• .eq.1: new each iteration = Full Newton Method
• .eq.10: use 10 inexpensive BFGS updates, reform if not yet converged
MAXREF: reformation limit before abandoning step

• if AUTO is active, dt will be reduced and step will be re-tried, so MAXREF can be smaller
(~5)
_______________________________________________________________________________________________
_______________________________________________________________________________________________
• if AUTO not active, error termination occurs when MAXREF is reached so MAXRED should
be larger (~15, default)
DCTOL, ECTOL: convergence tolerances

• use NLPRINT=1 or "<ctrl-c> nlprint" to monitor progress of iterations
Nonlinear Solver Screen Dump
BEGIN implicit time step 3

============================================================
time = 1.09990E+00 DCTOL=1.E-03 ECTOL=1.E-02
current step size = 3.67821E-01
Iteration: 1 *|du|/|u| = 1.0894498E-01 *Ei/E0 = 1.8731172E+00
DIVERGENCE (increasing residual norm) detected:

|{Fe}-{Fi}| ( 1.0547507E+07) exceeds |{Fe}| ( 9.1389570E+06)
automatically REFORMING stiffness matrix...
Iteration: 2 *|du|/|u| = 3.8969724E-03 *Ei/E0 = 3.3420090E-02

Iteration: 7 *|du|/|u| = 6.4089308E-03 *Ei/E0 = 1.7184228E-02 ILIMIT=11
ITERATION LIMIT reached, automatically REFORMING stiffness matrix...
Equilibrium convergence summary for time step 3 at time = 1.0999005E+00

Number of iterations to converge = 13
Number of iterations to converge = 2
Example: 1-D Nonlinear Spring
To illustrate the Newton method, consider a simple nonlinear spring:
12
force, F(u)
F=10 8 k=1
k(u) u
k=2
4
k=4
1 2 3 4 5
displacement, u
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Iteration #1:
u0 = 0, k (0 ) = 4, u1 = 104 = 2.5
If the nonlinear solution method is linear, stop here and assume solution is correct. Do not
bother to check for equilibrium.
In a nonlinear problem, evaluate equilibrium by determining the internal force and the out-of
balance force R:
u1 = 2.5, F (u1 ) = 7 12
force, F(u)
R1 = F ext − F int = 3
8
Check convergence of equilibrium iterations

by evaluating displacement, energy norms 1 2 3 4 5
displacement, u
∆u 2.5
= = 1.000
u 2.5
e ∆u ∗ R 2.5 ∗ 3
= = = 0.300
eo u1 ∗ Ro 2.5 ∗10
Iteration #2: solve K∆u=R, update u=u+∆u

u2 = 2.5, k (2.5) = 2 12
R 3
∆u3 = = = 1.5, u2 = u1 + ∆u2 = 4.0
force, F(u)
k 2 8
F (u2 ) = 9, R2 = F ext − F int = 1
4
evaluate new R, convergence norms

1 2 3 4 5
∆u 1.5 displacement, u
= = 0.375
u 4
e ∆u ∗ R 1.5 ∗ 1
= = = 0.060
eo u1 ∗ Ro 2.5 ∗ 10
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Iteration #3: solve K∆u=R, update u=u+∆u

u2 = 4.0, k (4.0) = 2 12
R 1
∆u3 = = = 1.5, u3 = u2 + ∆u3 = 5.0
force, F(u)
k 1 8
F (u2 ) = 10, R2 = F ext − F int = 0.
4
evaluate new R, convergence norms
∆u 1.50
= = 0.200 1 2 3 4 5
u 5.0
displacement, u
e ∆u ∗ R 1.0 ∗ 0.
= = = 0.000
eo u1 ∗ Ro 2.5 * 10
Selecting Implicit Time Step Size
When should load be applied in more than one step?

• to reduce nonlinearity to manageable level
• to accurately integrate material behavior
• to resolve high frequency response in dynamic problems
• input parameters: stepsize DT and termination time TERM
How many steps should be used?

• estimate nonlinearity in problem
• tradeoff:
few steps with many equilibrium iterations (unreliable)
many steps with few iterations (too expensive)
• observe progress of equilibrium search
reduce stepsize until norms decrease monotonically
increase stepsize if very few iterations are needed (<10)
• activate automatic time step controller
_______________________________________________________________________________________________
_______________________________________________________________________________________________
AUTOMATIC TIME STEP CONTROL
If convergence fails, LSDYNA backs up to beginning of step with smaller DT. Solution is retried.
If convergence again fails, solution is retried with a smaller step and so on. Finally DT is set to
DTMIN, if fails terminate.
An optional number of iterations, IOPT, per step is chosen. Let ILS be the number of iterations
during the last increment. If
ILS<IOPT
1
DT= DT+(DTMAX –DT) MIN {0.5, [( IOPT / ILS ) − 1]} 2
ILS>IOPT
1
DT=DT-(DT-DTMIN) [1 − ( IOPT / ILS ) 2 ]
Where DT, DTMAX, and DTMIN are the current step size, the maximum step permitted, and the
minimum step size permitted, respectively.
• Automatic time step control allows specification of the Optimum Number of Equilibrium
Iterations per step.
• This indicated how hard LSDYNA should work in each time step.
• If equilibrium is reached in fewer than optimum iterations, the size of the next step is increased.
• If equilibrium search requires more than the optimum number of iterations, then the next step
size is decreased.
CONTROL_IMPLICIT_AUTO:
Automatic Time Step Control
Automatic time step control adjusts stepsize during the simulation

• very persistent, reliable
After successful steps

• compare iteration count to target value ITEOPT
• increase/decrease size of next step if difference exceeds window ITEWIN
After failed steps

• decrease step size
• back up, repeat failed step with new DT
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Exponential algorithm for adjusting step size

• increase stepsize by 1/5 decade until DTMAX is reached
• decrease stepsize by 1/3 decade until DTMIN is reached
• error termination if convergence fails when DT=DTMIN
Default time step control strategy

• ITEOPT = 11, ITEWIN=5, DTMIN=0.001*DT0, DTMAX=10*DT0
• decrease step if more than 16 iterations were required
• increase step if less than 6 iterations were required
• problem: stepsize continues to decrease even though every step converges successfully
• some models simply require more than 16 iterations per step
Aggressive step control strategy

• ITEOPT=200, conservative value for DTMAX
• stepsize always increased if convergence is successful
• agressively pushes stepsize toward DTMAX
• stepsize still decreases if convergence fails, but increases rapidly thereafter
• recommended for many problems
_______________________________________________________________________________________________
_______________________________________________________________________________________________
NONLINEAR SOLUTION METHODS FOR

IMPLICIT ANALYSIS
Define control parameters for the implicit nonlinear equation solver: There are two groups
for nonlinear solution. One group with four solution methods. The second group is the same as the
first group, however, with the arclength method.
If NSOLVR is equal to:
EQ 2: BFGS updates
EQ 3: Broyden updates
EQ 4: DFP updates
EQ 5: Davidon updates
EQ 6: BFGS updates + arclength
EQ 7: Broyden updates + arclength
EQ 8: DFP updates + arclength
EQ 9: Davidon updates + arclength
• Standard Newton Method:
The nth load increment is applied-then using the tangential stiffness matrix the iterative displacent
are found and hence the residual forces.The new tangential stifness is then evaluated and the new
iterative displacement and redidual forces are found. The process is repated until convergence is
obtained. The figure below depicts the algorithm.
ˆ (u(i ) )∆u = R
K ˆ (i ) = f ext − f int (u(i ) )
ˆ (u1 )
K ˆ2
R
in
f n+1
ˆ1
R
Force Norm
ˆ (un )
K
f ex
n
un u1(n+1) u(2n+1) u(3n+1) u(n+1)

Displacement
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Equilibrium is reached when iterations converge: ∆ u → 0, R

ˆ (i ) → 0
• Modified Newton-Raphson Method
The difference here between this method and the one before is that there is no update of the
tangential stiffness matrix after the initial iteration. The figure below depints the algorithm.
• BFGS Method
Also called the quasi-newton method (QN). In here an approximation to the local tangent matrices
are found. No factorization is required. The approximation to the local tangent matrix was defined
in the previous section. The figure below depicts the algorethim.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
• Snap-through and Collapse
In many collapse problems the equilibrium path could look like the figure below. In this type of
behavior the BFGS method or in general the Newton method has deficulties with the limit points.
For the ascending branch of the equilibrium curve, load control is greatly the most efficient
method. As the plateau in the response is reached (point A), then difficulties with convergence are
experienced (with the BFGS method the solutiion will jump to point A’). If displacement control is
adopted then when point ‘B’ is reached again there will be convergence difficulties (solution could
jump to point B’). To overcome this problem an algorithen is added to LS-DYNA that provide the
correct structural response. This algorithm is called the Arc-Length Method.
• Arc-Length Method
The arc-length method basic idea is that a constaint equation is added so that load level is modified
at each iteration rather thann holding it constant. The method is graphicaly depicted in the figure
below. Two arc-length methods are available in LS-DYNA. These methods are called the Ramm
method (updated normal plane method) and the Crisfield method (spherical path method)
_______________________________________________________________________________________________
_______________________________________________________________________________________________
• The Ramm Method
In here as oppose to forcing the iteration process to follow a plane perpendicular to the initial
tangent for each load step, the iteration process is constrained to be normal to the current arc. The
method is graphicly depicted in the figure below.
• The Crisfield Method
In here the iteration process is constrained to follow a spherical path centred at point ‘o’ with
radius equal to the arc length. The method is graphicly depicted in the figure below.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Linear Equation Solver
ˆ ∆u = R
During each nonlinear iteration, the linear system K ˆ is solved.
Iterative Methods
• (preconditioned) conjugate gradient
• potentially low operation count
• convergence difficult for some problems
• promising future developments
x x • • • • • • • • • • • • • • • • • •
x • • • • • • • • • • • • • • • • • •
x x x • • • x x x • • • • • • • • •
x x • • • x x x • • • • • • • • •
x • • • x x x • • • • • • • • •
x x x • • • • • • x x x • • •
x x • • • • • • x x x • • •
x • • • • • • x x x • • •
x x x • • • • • • • • •
x x • • • • • • • • •
x • • • • • • • • •
x x x • • • x x x
x x • • • x x x
x • • • x x x
x x x • • •
x x • • •
x • • •
x x x
x x
x
Direct Methods
• gaussian elimination
• inexpensive backsolve (quasi-Newton)
• The sparse direct solver looks promising for large linear systems
• substantial storage savings
• huge operation count reduction
Example: Beam Tower Model (Reverse Engineering, Ltd.)
• Model Size
• node points
• beam elements
_______________________________________________________________________________________________
_______________________________________________________________________________________________
• 43,590 degrees of freedom
Linear Equation Solver Data

skyline sparse
memory (Mw) 18 <6
max column height 720 114
avg column height 418 20
HP-735 CPU (sec) 1902 25
Example: Trimmed Inner Panel
Model Size
• 32,540 node points
• 31,640 shell elements
• 195,212 degrees of freedom
Linear Equation Solver Data

skyline sparse
memory (Mw) 176 <95
max column height 1320 54
avg column height 899 52
HP-735 CPU >5 day 88 min
Iterative Linear Equation Solver
LS-DYNA offers six iterative linear equation solvers

*CONTROL_IMPLICIT_LINEAR keyword [K]{x}={f}
• LSOLVR = 10: “best” iterative solver (currently activates #16)
• LSOLVR = 11: Conjugate Gradient method
• LSOLVR = 12: CG with Jacobi preconditioning
• LSOLVR = 13: CG with Incomplete Choleski preconditioning
• LSOLVR = 14: Lanczos method
• LSOLVR = 15: Lanczos with Jacobi preconditioning
• LSOLVR = 16: Lanczos with Incomplete Choleski preconditioning
_______________________________________________________________________________________________
_______________________________________________________________________________________________
All iterative solvers use the sparse matrix storage scheme

• eliminates all zero entries inside bandwidth
• minimizes total storage requirement
• Boeing Harwell format for portability
Iterative solvers outperform direct solvers in some problems

• solid, massive structures
• narrow range of natural frequencies (good numerical condition number)
Example:
20x20x20 brick elements, 26,460 equations
Solver type DIRECT ITERATIVE
LSOLVR 0 (default) 16
Memory 120 Mb 9 Mb
CPU 400 sec 21 sec
F
Linear Equation Solver Memory and CPU
An optional print-out is available to show memory and CPU requirements of the linear equation
solver: [K]{x}={f}
• activated by keyword input *CONTROL_IMPLICIT_LINEAR: lprint
• activated interactively <ctrl-c> lprint
• memory estimate printed before attempting to allocate
• change memory using command line option
• ls-dyna3d i=inputfile.k memory=8000000
units are “words”

1 word stores 1 floating point number
single precision = 4 bytes per word
double precision = 8 bytes per word
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Linear Equation Solver Memory and CPU

============================================================
time = 1.55684E+02
current step size = 7.78421E+01
SPARSE LINEAR EQUATION SOLVER STORAGE data (Mwords)
( 225972 degrees of freedom)
pointer arrays: initial = 11.523
actual = 6.413
stiffness coefficients = 6.187
Factorization Workspace (estimated)
symbolic = 14.015
numeric = 18.335
Final Storage Requirements (10% for pivoting)
incore out-of-core
symbolic factorization = 5.276 5.276
numeric factorization = 69.772 5.292
numeric solution = 65.561 3.145
TOTAL = 87.648 23.168
TOTAL available = 98.196 98.196
an INCORE solution will be performed
Initialization CPU = 7.220E+00 seconds
Symbolic Factorization CPU = 1.065E+01 seconds
Numeric Factorization CPU = 8.539E+02 seconds
Forward/Backward CPU = 5.060E+00 seconds
1 Mword = 4 Mbytes (single)

1 Mword = 8 Mbytes (double)
============================================================
time = 1.55684E+02
current step size = 7.78421E+01
SPARSE LINEAR EQUATION SOLVER STORAGE data (Mwords)

( 225972 degrees of freedom) number of equations
pointer arrays: initial = 11.523
actual = 6.413
stiffness coefficients = 6.187
Factorization Workspace (estimated)
symbolic = 14.015
begin factorization here
numeric = 18.335
Final Storage Requirements (10% for pivoting)
incore out-of-core
symbolic factorization = 5.276 5.276
numeric factorization = 69.772 5.292
numeric solution = 65.561 3.145
TOTAL = 87.648 23.168 memory required
TOTAL available = 98.196 98.196
memory available
an INCORE solution will be performed
solution method
Initialization CPU = 7.220E+00 seconds
Symbolic Factorization CPU = 1.065E+01 seconds
Numeric Factorization CPU = 8.539E+02 seconds
Forward/Backward CPU = 5.060E+00 seconds CPU timings
_______________________________________________________________________________________________
_______________________________________________________________________________________________
REMARKS
• In nonlinear problems equilibrium iterations must always be permitted. Reformation of the

stiffness matrix at the beginning of every time step often speeds convergence of the equilibrium
search. Stiffness matrix reformation normally occurs once at the beginning of each step, and
within a time step if either the equilibrium iteration limit is reached or the solution diverges.
• To suppress the stiffness reformation at the beginning of each step, set the parameter “ISTIF”,
in *CONTROL_IMPLICIT_NONLINEAR, to a value, which exceeds the total number of steps
in the problem. This is often economical for contact dominated problems, where the solution
diverges immediately after the first equilibrium iteration, causing a (second) stiffness
reformation.
• To suppress equilibrium iterations during each step, set the parameter “NSOLVR” in
*CONTROL_IMPLICIT_NONLINEAR to a one. A single linear solution will be performed
each step. This solution may not represent an equilibrium state, hence results obtained may be
inaccurate.
• In problems dominated by contact it may be cost effective to limit the BFGS iterations to five
or less between stiffness reformations. This allows LSDYNA/IMPLICIT to better account for
rapidly changing contact areas. Enter a value of one for “ILIMIT” in
*CONTROL_IMPLICIT_NONLINEAR to obtain Full Newton Method with Line Search.
• The convergence strategy for “DNORM=2”, which is the defult, in
*CONTROL_IMPLICIT_NONLINEAR compares the current displacement increment to the
total displacement over the analysis. This criterion becomes less strict as the analysis proceeds.
By setting “DNORM=1” the strategy considers the total displacement over the step, hence is
more consistent as the deformation evolves.
• In problems where there is much rigid body motion the displacement tolerance may be
insufficient, and it may be advisable, in some problems, to tighten the energy tolerance to 10-3.
• The arc length solver is intended for use with buckling problems where the load-displacement
curve is not monotonic, a problem class where traditional displacement or energy norm driven
method fail. In this method, the solution is advanced in increments of constant arc length on the
load-displacement curve, regardless of its path. This may require load reduction or unloading as
the problem passes a bifurcation point.
• The user may define a single node whose load-deflection history controls the solution. If no
node is defined, norms of the global force and displacement vectors are used in the generalized
arc length method.
IGS
Geometric nonlinearity is introduced when displacements become “large” (example: tip
deflection of a cantilever beam is greater than the beam thickness), or rotation become “large”
(example: beam rotation theta is large enough that the approximation sin(theta)=theta is not
reasonable).
A second geometric nonlinearity is due to finite rigid body rotation. Both of these nonlinearity
are always accounted for in LSDYNA-Implicit.
The flag IGS is an option to include an additional term in the stiffness matrix which reflects a
change in apparent stiffness due to rotation of stress. This flag does not change the degree of
_______________________________________________________________________________________________
_______________________________________________________________________________________________
nonlinearity included in calculation of internal or external forces. IGS does not alter the
equilibrium solution. However, the path taken during the iteration process to reach the final
solution depends on the stiffness matrix. In some cases, adding this term will accelerate
convergence and in other cases this term will create a singular stiffness matrix and halt the
iteration process. For instance including this term will speed convergence for transversely
loaded plates with clamped boundaries, where transverse deflection generates sharp increase in
membrane stress. It is required by hyperelastic materials.
LSOLVR
The LSOLVR "linear solver" provides the solution to the linear system of algebraic equations
Kx=f. There are several methods to compute this expensive solution. LSOLVR selects
between them.
NSOLVR
In contrast, the nonlinear solver controls the iteration process as we seek to find the
displacements which balance internal and external forces.
There are several nonlinear solvers to choose from, many of which are based on Newton
iteration. The parameter NSOLVR is used to make this choice.
The simplest nonlinear solver option is "NSOLVR=1", which is to skip the equilibrium
iterations and accept the linear approximation to the solutions.
This choice is commonly referred to as a "linear" solution (not to be confused with the linear
equation solver LSOLVR above).
Screen Messages and Interactive Commands

• It helps to have a quick look at the progress of a run. An input parameters is added to activate
printing this information to the screen. The default values prevent printing. These two
parameters are:
• NLPRT on *CONTROL_IMPLICIT_NONLINEAR and
• LPRINT on *CONTROL_IMPLICIT_LINEAR.
• We recommend *****ALWAYS***** activating NLPRT, and also recommend activating
LPRINT the first time you run a large problem to get a sense of how much memory and cpu the
stiffness matrix factorization will require.
• Interactive switches are added to activate each type of printing. The switches have the same
name as the input parameters:
• Typing "<ctrl-c> nlprt" activates the nonlinear print flag.
• Typing "<ctrl-c> lprint" activates the linear print flag.
Typing the switch a second time deactivates the flag, so you can turn them on and off at will. One
caution is that you must enter these interactive commands during the nonlinear iteration phase of a
time step. If you happen to issue the command during another solution phase, say while the plot
file is being printed, then the command will be ignored.
_______________________________________________________________________________________________
FINITE ELEMENT MODELING TECHNOLOGY
_______________________________________________________________________________________________
FINITE ELEMENT MODELING TECHNIQUES
1. Engineering a FEA Model

2. Element Selection
3. Mesh Density
4. Symmetry
5. Modeling for Physical Phenomenon
6. Ad – Hoc Guidelines
7. How to Tell If Your Results Are Correct
1. ENGINEERING A FFA MODEL
PRIMARY CONSIDERATIONS
• Design Situation
• Physical Nature of Design Concerns
• Information Required
• Most Effective Level of Simulation Detail
DESIGN SITUATION
• New Design or Existing Problem
PHYSICAL NATURE OF DESIGN CONCERN (S)

• Does previous product knowledge exist that can be used to pinpoint high risk area
• What is the nature of the predominant physical phenomena that must be simulated
• What assumption can be made and what are their effects on simulation results
INFORMATION REQUIRED
• Full scale tests, material characterization, geometry, etc.
MOST EFFECTIVE LEVEL OF SIMULATION DETAIL

• Purpose of Analysis/Evaluation Focus
• Requirements for time Responsiveness
• Cost/Benefit Trade off
• Confidence Level of Model and Loading Assumptions
• Accuracy Level of Acceptability Criteria
_______________________________________________________________________________________________
_______________________________________________________________________________________________
2. ELEMENT SELECTION
• A General-purpose finite element code can be thought of as a ”toolkit” of structural entities

from which a physical and mathematical description of a structure can be built.
• In order to model a structure efficiently and accurately, it is important that the correct types of
element be used and that the mesh of elements constructed be dense enough to accurately
represent the strain distribution in the structure.
CHOOSING ELEMENT TYPES FOR MODELING A STRUCTURE
1-D rod like

structure
2-D structure
3-D shell
structure
3-D solid structure
STEPS IN DEFINING A MODEL:
1. Select the appropriate element type:

Most “economic” element, with least DOF per node which will give the desired results.
2. Determine the mesh density required based on:
Overall stress/ strain gradients expected (which depends on loading to be applied )
Location of areas of stress concentration (fillets, holes, etc.)
3. Tradeoffs:
“Economic” element and coarse meshes are easy to generate and run, however, they do not always
give the results desired.
4. Rule of thumb:
_______________________________________________________________________________________________
_______________________________________________________________________________________________
The time to perform a finite element analysis (preprocessing, solution, postprocessing) is roughly
proportional to the number of nodes in the model squared, Time ∝ N 2
*ELEMENT_DISCRETE
element ID, part ID

2 nodal point ID’s, orientation, scale factor on force
force vs. displacement is defined by *MAT_SPRING
force vs. velocity is defined by *MAT_DAMPER
*MAT_SPRING_option *MAT_DAMPER_option
elastic viscous
elastic-plastic (E, Et, SGMY) nonlinear viscous
nonlinear elastic
inelastic (allow tension or compression only)
general nonlinear (negative or zero slope allowed in the F vs. D curve)
Maxwell (exponential decay of stiffness)
*SECTION_DISCRETE
transnational or rotational
dynamic magnification factor
tension/compression limits
*ELEMENT_MASS
element ID, node ID, mass value
*ELEMENT_BEAM
element ID
part ID
_______________________________________________________________________________________________
_______________________________________________________________________________________________
3 nodal point ID’s
*SECTION_BEAM
Beam (axial, bending, torsional deformations)
Hughes-Liu (default)
CROSS SECTION
Rectangular, tubular, arbitrary
Area and inertia’s
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Hughes-Liu: type 1 (DEFAULT), recommended

● 6 DOF per node: (dx, dy, dz, rx, ry, rz)
● one integration point along length
● integrated cross section (not "resultant" beam)
● transverse shear correction factor
● moveable mid-plane option
● s, t directions defined by
orientation node n3
*ELEMENT_SOLID
element ID
part ID
4-noded tetrahedron
6-noded wedge
8-noded brick
*SECTION_SOLID
FORMULATION
8-noded reduce integration (default)
8-noded fully integrated (selective reduce integration)
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Constant Stress Solid: type 1 (DEFAULT), not recommended

● 3 DOF per node: (dx, dy, dz)
● single point integration with hourglass control
● requires hourglass control type 6
● least expensive
Fully Integrated S/R Solid: type 2, recommended

● 3 DOF per node: (dx, dy, dz)
● 2x2x2 selective/reduced integration (B-bar method)
● most expensive
*ELEMENT_SHELL
element ID
part ID
4 nodes for quads and 3 nodes for tris
thickness can be defined in element card
*SECTION_SHELL
FORMULATION
Hughes-Liu
Beltschko-Tsay (default)
BCIZ triangular
C-zero triangular
S/R Hughes-Liu
Fast S/R integration
You can define material orientation for orthotropic and composite materials.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Co-Rotational Coordinate
• Material rotation is assumed to coincide with rigid body rotation of the element local reference
coordinate.
• The mid surface of the quadrilateral element is defined by the location of the element’s four
corner nodes
• If the four nodes are co-planner then the unit vectors in the e1 and e2 directions are tangent to
mid surface and e3 is in the fiber direction “f”.
As the element deforms, an angle may develop between “f” and e3 with magnitude
e3 . f −1 < δ . Acceptable values of δ on the order 10^-2 and therefore the difference
between co-rotational coordinate and material rotation should be small which is the
consideration for finite strain.
Shell Element Technology
4-Noded 3-Noded
1. H.L. 1. BCIZ
2. B.T. 2. C-zero
3. B.T. Membrane
4. S/R H.L.
5. S/R H.L. (co-rotational)
6. B.L.
7. B.T. Membrane full integration
8. B. Wang Chiang
9. H.L. (co-rotational)
10. Fast fully integrated shell
_______________________________________________________________________________________________
_______________________________________________________________________________________________
3-Noded Shell Elements
1. BCIZ (due to Zienkiewicz and co-workers 1965)

• First triangular element implemented in DYNA3D
• Used for compatibility with earlier user models
• Co-rotational formulation
• Non-conforming shape functions
• Kirchhoff theory
• Local nodal rotations are interpolated by flexural shape functions similar to B. beam element
• Three point in-plane integration
• No hourglassing
• More expensive than C-zero
2. C-zero (Kennedy and B. 1986) (C-zero means constant strain element)

• Used for compatibility and mesh transition
• Co-rotational formulation
• Mindlin theory
• Interpolations functions are the area coordinate
• One point gauss
4-Noded Shell Elements
Belytschko-Tsay: type 2 (DEFAULT), not recommended

● requires stiffness form of hourglass control (type 4 or 5)
● available w/ or w/o warping stiffness (BWC flag on *CONTROL_SHELL)
● option to automatically switch formulation during seamless springback to fully integrated
element #6
● degenerates to triangle
● least expensive
_______________________________________________________________________________________________
_______________________________________________________________________________________________
S/R Hughes-Liu: type 6, recommended

• 2x2 selective-reduced integration
• 6 DOF per node: (dy, dz, rx, ry, rz)
• default element from LS-NIKE3D should be used in comparisons
• use "unique nodal normals" on *CONTROL_SHELL
• includes optional geometric stiffness terms
• can improve convergence if large membrane stresses present
• required for eigenvalue problems with initial stress
• degenerates to triangle
• most expensive
Belytschko-Wong-Chiang: type 10, not recommended

● requires stiffness form of hourglass control (type 4 or 5)
● improved stiffness compared to Belytschko-Tsay type 2
● degenerates to triangle
● inexperienced
● inexpensive
Fast Shell: type 16, recommended

● 2x2 integration with enhanced strain formulation
● Bathe-Dvorkin transverse shear correction eliminates W-mode hourglassing
● hourglass type 8 adds warping stiffness, may improve convergence
● least expensive of 2x2 elements: 2x-3x more than Belytschko-Tsay (not 4x)
_______________________________________________________________________________________________
_______________________________________________________________________________________________
● does not degenerate to triangle

● best for springback
*ELEMENT_TSHELL
element ID
part ID
8-noded
*SECTION_TSHELL
• one point integration or
• selective reduce integration
• Similar to solid element with enhancement based on H.L. and B.T. shell.
• To obtain shell like behavior it is necessary to have several integration point through the
thickness while employing plan stress constitutive equations.
• Uses reference coordinate system similar to B.T. shell (co-rotational).
• To ensure plan stress condition the normal stress is updated as a penalty condition to maintain
the shell thickness.
Implicit 2D Solid Elements
Plane Strain: type 13 , Axisymmetric: type 15

● special case of 4-node shell element
● construct in X-Y plane
● Y-axis is axis of rotation for axisymmetric type 15
● 2 DOF per node (dx,dy)
● MUST constrain remaining DOF (dz, rx, ry, rz)
● MUST select NIP=4
● very fast, good convergence behavior
X
_______________________________________________________________________________________________
_______________________________________________________________________________________________
GUIDELINES FOR MODELING WITH BEAMS
• Element is a slender rod – like structural component.

• Element geometry is described by its length and cross-sectional properties (area, moments of
inertia, etc.)
• Orientation of the cross section principal axes with respect to the global (X, Y, Z) coordinate
directions.
• Suspension system components in vehicle crashworthiness simulations.
• Explicit modeling of skeletal frames structures such as building steel frames in Civil
Engineering design.
• As connector element in complex mechanical structures to simulate
Bolts or Rivets
Eccentric Connections
Stiffening Ribs in Shell Structures
GUIDELINES FOR MODELING WITH 2-D PLANE ELEMENTS
In many situations, the analysis can be simplified to a 2D situation. The types of 2D continuum
elements, which are generally provided, are:
• Plane Strain/Stress Elements

• Axisymetric Elements
GUIDELINES FOR MODELING WITH SHELLS
Thickness to span ratio, flat shell:

h h
h<< a, h << b, generally and < 0.1
a b
A shell resists lateral and in-plane loads by bending, transverse shear stresses and in-plane action,
not through block-like compression or tension in thickness direction.
• Stress in the Z, or thickness, direction is zero.

• Element thickness and loads can vary from node to node.
• Thickness to span rations must be considered at transition joints.
GUIDELINES FOR MODELING WITH THICK SHELLS
• Based on shell theory like a thin shell, but is like a 3D element and has 6 DOF per node (plane
stress constitutive equations).
• Provides a convenient way of transitioning from shell –like to solid-like structures.
• Not all general-purpose FE programs provide this type of element.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
• The element is in general too stiff and in many cases can be replaced by solid element.
• The types of elements available are topologically similar to solid elements.
GUIDELINES FOR MODELING WITH SOLID ELEMENT
• Solid elements are the most versatile type of element available in general finite element
program.
• Virtually any type of structure can be modeled with this type of element; however its use is
primarily for:
Fully 3 dimensional structures for which a simplification to one of the previously described
situations not judged to be valid due to:
• Complex 3D geometry.
• Stress fields, which are fully 3D.
• Loadings which are not symmetric.
Detailed analysis of stresses in critical areas such as:
• Faster locations
• Welds
• Fillets
3. MESH DENSITY
In this section we will discuss some of the factors which influence how the user of finite element
analysis can estimate how fine a mesh is required to yield accurate results.
Obviously, since the finite element technique involves discretization of some type of physical
continuum into a gridwork (i.e. ”mesh”), the way in which the users chooses to do this influences
the outcome of the analysis to some extent. Furthermore, there are basically an infinite variety of
valid ways of discretizing a given physical phenomenon and arriving at approximately the “right
answer”.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
GUIDELINES FOR MESHING
• Element size must be small enough that it adequately matches the shape of the structure
boundaries.
• Element internal strain distribution must approximate the actual strain field in the physical
component.
• The model must be constructed so that stresses are sampled at or near where expected
concentration of stress actually exists in the model.
• For example, if we construct a cantilever beam from 8-noded brick solid elements, and apply a
uniform tensile load to it, we obtain:
Fixed P
end
Cantilever Beam Model with a pure Tensile Load
• We would find that the predicted stress and strain are exactly the correct values.
• But if we apply a lateral shear loading to this model we would find that
Fixed
end
• We would find that the predicted MAXIMUM stress and strain values are not the maximum
values in this structure.
• Why?
• The answer is simply that the error is in the modeling of the structure.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
• A liner hexahedral solid can in fact predict the linear strain distribution. And as a result, this
model dose in fact predicts the CORRECT MAXIMUM DISPLACEMENTS for the tensile
load.
• As we know from simple beam theory the maximum bending stress is actually at c= b/2 and is
thus:
Mb
σ =
max 2 I
• But for stresses and strains:
The optimal stress recovery point for a linear solid element is the centroid of the element.
In this model, the stress recovery point lie a long the centroidal axis of the beam structure,
thus, the maximum stress at this point is:
MC
σ = And c = 0, thus, stress is zero
max I
GUIDELINES:
The following general guidelines can be stated relative to determining an appropriate mesh density:
• The mesh density should be based upon the degree of accuracy required from the analysis.
• If only displacements or mode shapes are required, it is not necessary to mesh finely around
areas of very localized stress concentration.
• When it is desired to model local features in detail, the size of the elements around that detail
should be an order of magnitude less than that of the detail itself. Must have smooth mesh size
transition.
• Spacing of nodes should be closer near the detail. Some studies based on analytical solutions of
stress concentrations around holes and fillets suggest that a logarithmic spacing is a good
guideline.
• If the structure is very complex, a “two step” procedure is often appropriate:
Make a reasonable overall model of the structure with a fairly coarse grid.
Then make a detail model of the region in question and apply displacements from the
overall model to the detail model as boundary conditions.
4. SYMMETRY
TYPES OF SYMMETRY
1. Symmetry in the Geometry
2. Symmetry in Loading
The types of symmetry, which can be (separately or in combination), are:
• Rotational Symmetry
_______________________________________________________________________________________________
_______________________________________________________________________________________________
The structure is symmetric about some axis of revolution (here we say the structure is
axisymmetric).
• Reflective Symmetry
The structure has one or more planes about which a mirror image exists.
• Antisymmetry
The structure itself is symmetric but the loading or restraints are reversed on opposite sides
of the plane (s) of symmetry.
• Cyclic Symmetry
• Simple cyclic symmetry, in which the smallest repetitive piece of the structure can be
modeled, loading is symmetric.
• Dihedral cyclic symmetry in which a repetitive segment exists which is in itself
reflectively symmetric about its own midsection, loading is not symmetric.
• Rotation cyclic symmetry, in which a repetitive section exists when rotated about an
axis of revolution, loading is not symmetric.
Example: A turbine disk
Boundary Conditions Which Must be Used With Reflective Symmetry:
When symmetry and antisymmetry are used, the proper boundary conditions (restrained degrees of
freedom) must be specified at nodes which lie in the plane of symmetry.
• Boundary conditions for symmetry are:

Translational displacements NORMAL to the plane of symmetry are zero. Rotation IN the
plane of symmetry are zero.
• Boundary conditions for antisymmetry are:
Translational displacement IN the plane of antisymmetry are zero.
Rotations NORMAL to the plane of symmetry is zero.
SIMPLE CYCLIC SYMMETRY

• The smallest repetitive piece of the structure can be modeled; loading is symmetric.
Example: A thin disk with holes, the section shown is modeled with symmetry boundary condition
applied in nodal local coordinate systems to the edges.
ROTATIONAL CYCLIC SYMMETRY

• This is the most general case. Cyclic symmetry in which a repetitive section exists when rotated
about an axis of revolution; loading is not symmetric but is periodic (sinusoidal) with K
periods.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
DIHEDRAL CYCLIC SYMMETRY

• Cyclic symmetry in which a repetitive segment exists which is in itself reflectively symmetric
about its own midsection; loading is symmetric.
• There are two mathematical models, called the ‘left model “ and “right model” although the
user may only need model one if the program has a built-in cyclic symmetry capability.
WARNING ABOUT SYMMETRY AND DYNAMICS PROBLEMS!
• If symmetry is used in dynamics, remember that due to the boundary conditions ONLY the
symmetric modes will be found if symmetry boundary conditions are imposed.
• If anti symmetry is used in dynamics, remember that due to the boundary conditions only the
antisymetric mode will be found if antisymetric boundary conditions are imposed.
• For rotational cyclic symmetry, only the modes involving the K th nodal diameter will be found
when the cyclic symmetry constraints are applied.
LS-DYNA INITIAL AND BOUNDARY CONDITIONS
*SET_NODES_option
• Define a set of nodes by a set identification number (SID)
• Define the nodes to be included (NID’S)
*SET_NODE_LIST
• Define the nodes, 8 per line
*SET_NODE_COLUMN
• Define the nodes, 1 per line
_______________________________________________________________________________________________
_______________________________________________________________________________________________
*DEFINE_BOX
• Define a box–shaped volume, everything inside the box can be used as input. Assign a box
identification number (BOXID)
• Define two extreme corners of the box
Xmin Xmax
Ymin Ymax
Zmin Zmax
*BOUNDARY_PRESCRIBED_MOTION
• Impose nodal motion on a node, set of node, or a rigid body
• Applicable to one degree of freedom
• Displacement, velocity, and acceleration (nodes only)
• Motion prescribed by a load curve
*DEFINE_CURVE
• Define a (load) curve to be used as input to other options
• Assign a load curve identification number (LCID)
• Define the point of the curve (Abscissa “x” and Ordinate “y”)
• Examples: force vs. time, velocity vs. time, force vs. displacement, etc.
*DEFINE_COORDINATE_option
• Used as input to other options
• Assign a coordinate system identification number (CID)
Define using:
NODES 3nodes: local origin, along local x-axis, in local x-y plane
SYSTEM x,y,z of three points (same as nodes)
VECTOR 2 vectors: local x-axis, local in –plane vector
*LOAD_option
• Define various loads (forces, moments, pressures, body, heat, etc.)
• Avoid single concentrated loads sine it is physically unrealistic and promote hourglass modes
• Requires a load curve
• Load can be scaled
*LOAD_BEAM
• Distributed traction load along any axis of a beam or set of beams
• Load is given in force per unit length
*LOAD_BODY
• Body force load due to a prescribed base acceleration, angular velocity
• Any degree of freedom: X,Y, Z, RX, RY, RZ
• All nodes or a subset of nodes
*LOAD_NODE and *LOAD_RIGID_BODY
_______________________________________________________________________________________________
_______________________________________________________________________________________________
• Apply a load to a node, or a rigid body

• X, y, or z force, x, y, or z moment
• Follower force
• Global or local coordinate system
DISTRINUTED PRESSURE
• Apply a distributed pressure to a:
• Segment (*LOAD_SEGMENT)
• Set of segments (*LOAD_SEGMENT_SET)
• Shell (*LOAD_SHELL_ELEMENT)
• Set of shells (*LOAD_SHELL_SET)
• Positive pressure acts in the negative normal direction of the shell/segment
_______________________________________________________________________________________________
_______________________________________________________________________________________________
5. MODELING FOR PHYISCAL PHENOMENON
(some examples)
INTRODUCTION
Some situations require special modeling techniques or considerations.
In this portion of the course, we will examine some of these and explain how to use proper
modeling techniques to insure correct results.
MIXING DIFFERENT TYPES OF ELEMENTS

• In some physical situations, it is desirable to be able to mix different types of elements in the
same model.
• When element types are mixed, it is important that degrees of freedom at the point that
dissimilar element types are joined be properly matched so that displacement compatibility and
proper load transfer paths are maintained.
• As an example, suppose we wish to model a situation where a support strut is welded to a thin
plate structure. In this case, the easiest way to do this would be to represent the strut by a beam
and the plate by shells:
3-D beam
shell
Physical modeling
• In this case both the space ( 3D ) beam, and thin shell elements have six degrees of freedom at
each node:
X, Y and Z Transnational Displacement DOF’s
X, Y and Z Rotational DOF’s.
• These two elements are thus entirely compatible
• Now suppose we have a similar situation, but this time support strut is to be welded to a
massive, thick plate structure:
_______________________________________________________________________________________________
_______________________________________________________________________________________________
3-D beam
shell
Physical modeling
• In this case 3D beam element has six degrees of freedom at each node:
X, Y and Z Transnational Displacement DOF’s
X, Y and Z Rotational DOF’s
But the solid elements have only the 3 translational DOF’s.
• These two elements are not entirely compatible, and in fact, modeling this way will cause a
”ball joint” to exist at the end of the beam attached to the solid since the solid will not transmit
moments to the end of the beam and vice versa.
To solve this problem, it is necessary to connect the beam end “J” to more than one point on the
solids structure to establish a ‘moment lever” by which the bending is transmitted to the solids.
This can be accomplished by either the use of rigid connectors.
Or by using a multipoint constraint to describe the relationship of the displacement, d, in X, Y and

Z at the nodes J, K, L, M and N.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
GEOMETRIC DISCONTINUITES
• As another case where geometric discontinuity can cause problems, consider the case of a
branch in a thin shell structure.
• Here the problems that can arise are in the interpretation of stress result at the intersection.
Most finite element programs produce contour plots by extrapolation of the element interior
(i.e. integration point) data to the nodes and AVERAGING that data for all elements connected
to given node. It is the nodal data that is then contoured.
• This is acceptable for planar or 3D solid element types, but is unacceptable for shells, since:
Shell element generally has its own coordinate system.
• In general, and especially when the shells have an intersection angle greater than say 5 degrees,
the like components of stress are in significantly different directions.
Shell 3
Shell 1 σx3
σx1
Shell 2
Node 1
σx2
• Membrane stress components, local shell x direction σ x 3 ⊥ σ x1 and ⊥ σ x 2

• As can be seen from the above, averaging the X, Y and in-plane shear components without
special consideration would be in error; however, scalar quantities such as vonmises or
_______________________________________________________________________________________________
_______________________________________________________________________________________________
maximum principal stresses could be averaged if they are computed from components averaged
in a common coordinate system.
• It is possible to make the program average the stress properly by placing two sets of nodes at
the joining line and constraining the co-located joints to have the same displacements. This
method will give correct nodal averaged values since the elements on one branch will not have
results averaged with those of the other branch.
MATRIAL DISCONTINUITIES
• When two or more materials exist in a given model, the same problem can occur in stress
averaging at the nodes on the material interface.
steel
brass
Incorrect stress Actual stress

distribution using distribution
nodal averaging
• One solution is to place two sets of nodes at the same location at the interface and couple them
as was described for the shell problem previously discussed.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
ECCENTERIC SHELL STEFFENERS
Beam elements are frequently used to represent stiffeners for plate/shell structures. If beams are
directly connected between shell mid-surface node points then:
• Beam cetroidal axis will be locked at the mid-surface of the shell.

• This is an incorrect representation for eccentric stiffeners, as shown above.
• Structural behavior is significantly different when stiffener is eccentric with respect to the shell
mid-surface.
Eccentric shell stiffeners can be modeled by:
• Modeling shells with nodes at mid-surface of physical shell surface.

• Modeling stiffeners as beam elements with separate nodes located along centroidal axis of the
beam cross section.
• Connecting corresponding shell and beam node pairs with rigid link elements.
TIME VARYING LOADS
• Frequently, the operating environment of structural components involves its response to time
varying loads (dynamic and impulsive).
• The determination of acceptable displacement and stress response can often be obtained with
simple quasi-static analysis.
• The choice of performing static or transient dynamic analysis depends on the frequency
characteristic of the applied time varying loads and the responding structural components.
• Rule of thumb: when the frequency of excitation is less than one third of the lowest natural
frequency of the structural then static analysis can be performed.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
6. AD HOC GUIDELINES
• Finite Element mesh should be uniform where this is practical however, nonuinformity is often
necessary to achieve refinement of the gridwork in regions of rapid changes in loading and /or
behavior.
• Analyses to obtain displacement (or force) behavior do not require as refined a gridwork as is
required to obtain strain (or stress) behavior.
• Prediction of modes of vibration usually requires a more refined idealization than dose the
prediction of frequency of vibration.
• The use of constant strain triangle finite element should be avoided wherever possible.
• A mesh planned for static analyses will be satisfactory for dynamic unless high frequency
response content is significant.
• Determining that the solution is insensitive to spatial and temporal mesh size provides an
adequate check of modeling except in rare cases.
• Rapid changes in element size should be minimized.
• For complex structures consider a “stage” analysis. Analyze the entire structure with a coarse
mesh and then study the areas of interest with a fine mesh, using as boundary the results of the
coarse analysis.
• Three elements through the wall thickness of thick-walled structures will usually give
reasonable result.
• If the structure and loading posses symmetry, advantage should be taken into account of this
fact. However, one should be careful in buckling and frequency analyses where asymmetric
modes might be important.
• When modeling reduced portions of the structure for evaluation of local effects, make the
boundary of this local area far enough away from the area of interest to avoid boundary effects
on the information at the area of interest.
• Aspect ratio limits vary from element to element; however, a general guideline is to keep
element aspect ratios below 10 for deformation analysis and below 5 for stress analysis.
• Element of extreme shape (high aspect ratio rectangles, large angled triangles, etc.) are
unfavorable with respect to minimizing error and should be avoided.
• Solutions for structures with uniform stiffness distribution are more accurate than those for
structure with stiff and flexible regions (e.g., portal frame in which axial deformation of
horizontal member is considered). The same rule may be worded differently as: solutions for
structures with a large frequency range are less accurate than those for structures with a small
range.
• For convergence studies, one must refine (i.e., subdivide) the original mesh, otherwise a new
approximating sequence is begun.
• The recommended first checks on predicted behavior are equilibrium checks using loads,
reactions and forces for quasi-static and energy balance check for impact problems.
• Vibration analyses using lumped mass model usually yield slightly lower frequency values than
are obtained from consistent mass models. LSDYNA uses lumped mass matrix.
• “Consistent mass model give frequency values higher than exact solution.
• Analyses including nonlinearity in geometry or material usually requires a more refined
idealization than linear analyses.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
SOLUTION ACCURACY
There are basically 3 ways to determine if sufficient accuracy is being achieved:
• Compare the results with “textbook” analytical solutions for similar stress concentration, using
the far – field stress in the model and an appropriate stress concentration factor from an
analytical/empirical solution (normally not available).
• Compare the results with test data. This is the best way to gain confidence in the modeling
technique for a specific type of component. Using test correlation, the user will have a high
degree of confidence in future modeling of similar types of components using the specific
modeling technique.
• If neither of the above two methods is viable, a finer mesh of the structure can be constructed
and the results compared to the original mesh. This process of refining can be repeated until the
results converge to an acceptable tolerance. The drawback to this method is obviously the
amount of repetitive modeling and computer run time, but it can be worth while for complex
components where no test prototype is readily obtainable.
7. HOW TO TELL IF YOUR RESULTS ARE CORRECT
• One serious error that is commonly made in using finite element analysis is a failure to
adequately verify that the results from the analysis are reasonable.
• There are a number of things that users can do to verify that the model is yielding reasonable
results. Will discuss some of the things you can do to examine the accuracy of the results.
USER MUST UNDERSTAND ENGINEERING PRINCIPLES
• Interpretation of result requires understanding the basic fundamentals of engineering

disciplines.
• While color graphics are useful for presenting results, they should not be used as a substitute
for examination of the numbers, which result from the analysis.
• DO NOT USE COLOR FILL CONTOUR PLOTS as means of verifying correctness of results.
While they are very useful for looking at trends and for impressing people, this type of plot
tends to mask some errors, since the plot will not show localized trend. This is because
“smoothing” the result for display tends to hide localized anomalies.
WHAT TO LOOK FOR

• The first thing to examine would be the FUNDAMENTAL result of the analysis.
• After the fundamental result is found to be acceptable, there are three basic categories of data
which must be examined to assess solution validity:
Global Summations, Nodal Results, Elemental Results
_______________________________________________________________________________________________
_______________________________________________________________________________________________
GLOBAL SUMMATIONS
• These give the user a good idea of how well equilibrium requirements have been met, and are
indicator of round-off error in the overall solution.
• Types of global sums which can be used in structural analysis are:
GLOBAL ENERGY
GLOBAL FORCE SUM (Quasi-Static Structural Analysis):
This output provides an absolute check on whether or not the structure is in equilibrium.
• GLOBAL ENERGY SUM (Static or Dynamic Structural Analysis).
NODAL RESULTS
• Displacements
• Summation of Forces
Examine the sum of forces at free nodes to verify that they are zero and at restrained nodes
to verify that they are reasonable non-zero values.
ELEMENT RESULTS
• As a result, the elemental results are generally less accurate than the fundamental nodal result
because they are DERIVED results, not FUNDAMENAL results
• Assessing the accuracy of elemental results can be done by:
• Looking at the variation of values around critical locations (numerically, by X-Y graph plots or
iso-value line contour plots). High variation, shown by jagged graph or iso-value contour liner,
indicates too coarse a mesh.
• Qualitatively comparing the stress distribution with that predicted by previous test or known
analytical solutions for similar stress concentration features (holes, fillets, etc.).
_______________________________________________________________________________________________
_______________________________________________________________________________________________
_______________________________________________________________________________________________
_______________________________________________________________________________________________
_______________________________________________________________________________________________
CONTACT AND FRICTION
_______________________________________________________________________________________________
CONTACT IMPACT ALGORITHMS
Only one methods currently available for LS-DYNA/IMPLICIT
1. Penalty method (general contact with sliding, separation, and friction).
A contact search algorithm is either nodal based or segment based.

The choice of master and slave surfaces is arbitrary when the symmetric penalty treatment is
employed, otherwise;
More coarsely meshed surface should be master
If there is a large difference in mass density then the side with lightest material density should
be master.
Penalty Method:
Used in explicit and implicit codes

The method consists of placing normal interface springs between all penetrating nodes and
contact surfaces.
In implicit we add spring stiffness matrix to the global stiffness matrix.
In here a force modulus is computed for each slave and master segment based on the thickness
and bulk modulus of the element in which it resides.
Results in less mesh hourglassing.
∆t is unaffected by the existence of the interface since the interface stiffness is chosen to be
approximately the same order of magnitude as the stiffness of the interface elements.
SLIDING WITH SEPARATION AND FRICTION
Node-to-Surface
One part is referred as the master part and the other as the slave part
A node may belong to several interfaces
An interface node can be in rigid body
Breaks down with bad or disjointed mesh
Penetration is reduced by the penalty method
The algorithm have the following characteristics
Accuracy (very good)

Generality (not as good)
Efficiency (very good)
Compatibility (not as good)
_______________________________________________________________________________________________
_______________________________________________________________________________________________
The spatial distribution of forces is not smooth and may induce hourglass deformation. This
Interface simulates impacts between a master surface and a list of slave nodes.
Contact surface can be used for:

Contact simulation of a complex fine mesh on a simply convex surface.
ALGORITHM
1. For each slave node

2. Find the closest master node
3. Find the closest master segment
4. Check if slave node has penetrated the master segment
5. Find the contact point
6. Compute the penetration
7. Apply forces to reduce the penetration
Normally when the algorithm fails is at step 2.
REDUCTION OF PENETRATION
Let:
N Slave Node
C Contact point
Mt Element Corner Node
mi Masses of Element Corners
K Spring Stiffness
_______________________________________________________________________________________________
_______________________________________________________________________________________________
∆ Penetration
Penetration is reduced by the introduction of a massless spring of stiffness K between node N and
contact point C. A force F = K∆ is applied on node N in direction NC.
Apply F1 , F2 , F3 , F4 on each master node in opposite direction such that
F1 + F2 + F3 + F4 = − F
Forces Fi are functions of the position of contact point C and are evaluated by
Fi = − N i (ξ c , ηc ) F
Where
Ni are the standard shape functions
ξ c , ηc are the contact point natural coordinates
SPRING STIFFNESS K
Depends on master side only

K should be limited for stability consideration
K dose not depend on element size
_______________________________________________________________________________________________
_______________________________________________________________________________________________
If segment belongs to shell element:
1
K = αEt
2
Where E is young modulus, t is shell thickness, and α is stiffness factor. The default value for the
stiffness factor is 0.2
If segment belongs to a brick element:
1 A2
K = αB
2 V
Where
B: bulk modulus
A: segment area
V: element volume
α: stiffness factor (default is 0.2)
If a segment is a shell as well as the face of a brick element, the shell stiffness is used.
SLIDING WITH SEPARATION AND FRICTION
Surface-to-Surface
Used to simulate contact between two surfaces and works properly if the two surfaces are simply
convex
_______________________________________________________________________________________________
_______________________________________________________________________________________________
On the master side the segments must be connected to solid or shell elements
It is not allowed to put the same node in the two impact surfaces
Perform two times the algorithm for type node-to-surface contact
Symmetric computation slave and master surfaces play identical rule
First time solve penetration of nodes of surface 1 with respect to segments of surface 2
Second time solve penetration of nodes of surface 2 with respect to segments of surface 1
Is not effected by discontinuous (bad) mesh
Penalty stiffness is the minimum of the impacting segments
Precision is improved, however, cost is increased.
SINGLE SURFACE
• The single surface contact algorithm evolved from the surface to surface contact. Post contact
searching follows the procedure employed for the surface to surface contact.
• The algorithm loops through the contact segments and computes the normal segment vectors
and accumulates an area weighted average at the nodal points to determine normal vectors at
the nodal points.
• Then it loops through the bucket sorts and locates the node, which is nearest.
• It checks to see if nearest node is within a penetration tolerance.
• For shell elements, determine if the nearest node approaching the segment from positive or
negative side. Project both the node and the contact segment along the nodal line vectors.
• Check for interpenetrating nodes and if a node has penetrated apply a nodal point force.
CONTACT TYPES IN LSDYNA IMPLICIT
Contact interface types 3, m3, 5, m5, 10, m10, 13 are currently available:
SURFACE_TO_SURFACE
NODES_TO_SURFACE
ONE_WAY_SURFACE_TO_SURFACE
FORMING_SURFACE_TO_SURFACE
FORMING_NODES_TO_SURFACE
FORMING_ONE_WAY_SURFACE_TO_SURFACE
AUTOMATIC_SINGLE_SURFACE
2D_AUTOMATIC_SURFACE_TO_SURFACE
TIED_OFFSET
All implicit contact interfaces use the penalty method

• Disabling the shooting node logic is recommended
• SNLOG=1 on optional contact interface card "b"
• Oriented normal vectors are recommended
• Automatic contact types often fail for implicit
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Special parameter settings for implicit

• SHLTHK=1 or 2 on *CONTROL_CONTACT (required)
• ORIEN=2 on *CONTROL_CONTACT (recommended)
• SNLOG=1 on optional contact interface card "B" (recommended)
Contact Interface Normal Vectors

• Surface penetration determined using outward normal vector
• brick elements: unique outward direction for each face
• shell elements: sign of outward normal must be determined
• shell elements: right-hand rule on local connectivity gives outward normal
*CONTROL_CONTACT parameter ORIEN activates normal checking

• default "on" for input by part ID, "off" for input by segment list
Penetrating nodes are relocated during initialization

• severe deformation if normal vectors are incorrectly chosen
• observe model geometry at time=0 using postprocessor
• check "d3hsp" file for messages concerning reversed normal vectors
• This behavior is exactly the same for explicit and implicit simulations
Normal Vector Trouble
2D Example: penetration detected due to reversed normal vectors
slave slave
master master
Input geometry After penetration check
Automatic Normal Vectors
The LS-DYNA "automatic" contact interfaces allow randomly numbered segments

• approaching nodes are tracked, identifying and saving "outward" directions
• tracking algorithm assumes many small steps
• excellent for explicit analysis
_______________________________________________________________________________________________
_______________________________________________________________________________________________
• can fail for large, implicit steps
*CONTACT_AUTOMATIC_SINGLE_SURFACE is the only automatic type available for

implicit
• may be safe for some applications, if tracking is properly activated
• tracking is activated for all nodes that are initially in contact
• local relative velocities can be low even if global step size is large
• pillar collapse during roof crush
• be ware of shooting nodes, negative energy
TREATMENT OF SHELLS CONTACT- THICKNESS OFFESTS
• Not properly accounting for shell thickness In contact is responsible for most modeling and run
time difficulties such as shooting nodes, energy spikes, and nonphysical behavior.
• Contact surface or nodes must be offset to account for shell thicknesses. For single-surface
contact we must allow for contact from either side.
• Taking shell thicknesses into account in is optional for contact types 3 and 5.
• The thickness offsets are always included in contact types 13.
The following options may be specified:
EQ.0: thickness is not considered

EQ.1: thickness is considered but rigid bodies are excluded
EQ.2: thickness is considered including rigid bodies
Options 1 and 2 above activate the new contact algorithms in LS-DYNA.

• In metal forming and even crash analysis, membrane thinning changes the results by reducing
contact frictional forces and shell resultant forces.
• To include thinning in shells a flag must be set by global option on the shell related control
card:
Shell thickness change option, ISTUPD in *control_contact
EQ.0: no change
EQ.1: membrane straining causes thickness change
_______________________________________________________________________________________________
_______________________________________________________________________________________________
• By default membrane thickness changes are not accounted for in the single surface contact
algorithm. Thinning may be included by setting a flag on the contact control card.
Shell thickness changes considered in type 13 single-surface contact
EQ.0: no consideration
EQ.1: shell thickness changes are included
THICKNESS OFFSETS VIA NODAL NORMALS
The slave and master surfaces must be offset in the input by one-half the total shell thickness. This
also allows the segments to be oriented automatically.
THICKNESS OFFSETS VIA NODAL NORMALS
THICKNESS OFFSETS FOR NODES TO SURFACE CONTACT
PROJECTION OF CONTACT SURFACES BASED ON SEGMENTS
The two methods are of interest, shown here in 2 dimensions:
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Segment based projection is used in contact types a 3, a 5, a 10, a 13, and 15. All other contact
types used nodal normal projections if projections are used.
PROJECTION OF SURFACES
Advantages of nodal projection:

• Continuous contact surface.
Disadvantages of nodal projection:

• High cost due to nodal normal calculations (roughly 50% cost increase).
• T-intersections and other geometric complications create difficulties in projecting surface.
• Consistent orientations of contact surface segments are necessary.
REMARK
• Nodal projections are generally more accurate but are restricted to simple geometries.
• Segment based projection is MUCH cheaper.
SEGMENT BASED PROJECTION CREATES SOME PROBLEMS
The lack of a continuous contact surface requires special care in cases where nodes can
interpenetrate between contiguous segment.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
GUIDELINES:
• Contact type node-to-surface contact is simple and works almost all the time with if the master
surface is a closed surface.
• Contact type surface-to-surface is also reliable if both master and slave surfaces are closed.
• The advantages of these contact types is that contact forces can be monitored in “RCFORC”
with no need to specify a transducer.
• In general, fine mesh leads to better representation of the contact area and is more stable.
• Contact normals must point to the opposing surface in these contact types.
• Contact type Automatic Single Surface is the most efficient and reliable contact for general
surfaces if time steps are not too large so the tracking algorithm can work properly.
• Must use *contact_force_transducer to be able to obtain contact forces in “RCFORC”.
• A contact thickness must be defined in order to check for initial penetrations.
• In implicit analysis since time step is relatively large full searching my become necessary at
least every cycle.
• Bucket sorting is done every several cycles (user controlled) and it is an expensive part of the
contact algorithm. In some problems it may be necessary to reduce this to less frequency for
sorting.
• Distance to the nodes in the slave node bucket and 27 neighbor buckets are calculated. The cost
is proportional to N*(27N-1) where N is the number of nodes.
• Nodal based search may not be able to find the closest master segment, segment based search is
more reliable.
• Nodal based projection is more reliable than segment based projection. Segment based
projection may cause a node to interpenetrate between contiguous segments. This leads to
energy growth.
FRICTION
• Friction phenomena exist whenever contacts occur.

• Effect of friction may play very important role, for instance, in metal forming frictional effect
between the work piece and the die will strongly influence the forming process.
• Classical friction law (linear dry friction) and non-classical friction laws (like nonlinear friction
and elasto-plastic friction) exists for modeling friction.
• Classical friction law has physical deficiencies since it allows small relative motion of two
contacting bodies even if the friction force is less than µ s N .
• The Classical friction law ignores the dependence of friction coefficient on the relative sliding
velocity. This dependency is significant when the relative sliding velocity is large.
Dry Friction (Coulomb friction)

The actual coefficient of friction µ c = µ d + ( µ s − µ d )e − dc.v is assumed to depend on
µ s , µ d , and DC
Where
_______________________________________________________________________________________________
_______________________________________________________________________________________________
µs Coefficient of static fiction (defined as FS in manual)

µd Coefficient of dynamic friction (defined as FD in manual)
DC exponential decay coefficient
V relative velocity of the surface in contact
C1Co
V=
∆t
If exponential decay DC = 0 then µ c = µ s and f = µs N
If exponential decay DC ≈ 1.0e = ∞ then µ c = µ d
20
and f = µd N .
_______________________________________________________________________________________________
_______________________________________________________________________________________________
_______________________________________________________________________________________________
MATERIAL NONLINEARITY
_______________________________________________________________________________________________
ENGINEERING STRESS-STRAIN CURVE:
Nominal (engineering) stress σ n = Ρ / Α 0

Where p = applied load and A0 = undeformed cross-sectional area,
l − lo ∆l
Engineering strain: ε o = =
lo lo
p
True stress: σ =
A
Where A is the actual area
l
Logarithmic strain: ε = lu
lo
PLASTICITY
At low stress level, the stresses in elastoplastic material depend only on the state of strain;
however, above a certain stress level, called the yield stress, σ y , non-recoverable plastic
deformations are obtained.
Post- yielding behavior from uni-axial tension tests typically shows the following behaviors.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
• Hardening, like metals, concrete, rock with small deformations

• Ideal, crude idealization for steel, plastics, etc.
• Softening plasticity models, like dense sand, concrete with large deformations
ELASTIC-PLASTIC BEHAVIOR
Plasticity Theory:
• Yield criterion
(Von-Mises, typically)
• Flow rule
For metals use associated rule (flow rule and yield criterion are the same functions)
For granular materials like soil use non-associated rule (flow rule and yield criterion are
different functions
• Hardening rule
This is the rule that describes how the yield surface is changed by the history of the flow.
1. Kinematic hardening
2. Isotropic hardening
3. Mixed hardening
_______________________________________________________________________________________________
_______________________________________________________________________________________________
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Uniaxial Representation
σ
σyt
σy0
-σ~y
-σ y t -σ-y
Isotropic Hardening
Figure (a), yield surface expands uniformly.
Yield reappears when σ > σ B
Kinematic Hardening
Figure (b), yield surface translates in stress space but do not change size, shape, or orientation. The
elastic range of 2 oσ y is maintained.
t
σ + t σ~ =2o σ
y y y
Mixed Hardening
Figure (c), yield surface translate in stress space and change size.
t
2 oσ y < t σ y + σ y < 2 t σ y
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Converting Test Data to LSDYNA Material Data
Assumption: elastic deformation is small compared to total deformation and plastic flow occurs at
constant volume.
• Covert force (p) vs. elongation ( l − lo ) curve to engineering stress vs. engineering strain
σ n = Ρ / Α0
l − lo ∆l
εo = =
lo lo
• Convert engineering stress to Cauchy (true) stress and engineering strain to true strain
• Using incompresibility Al = Ao lo
p p l p l − lo
σ= = = (1 + ) = σ n (1 + ε o )
A Ao lo Ao lo
l l − lo
ε = lu ( ) = lu (1 + ) = lu (1 + ε o )
lo lo
• If one is interested in post necking strain then the following equation can be used
Ao
ε = lu ( )
A
_______________________________________________________________________________________________
_______________________________________________________________________________________________
The following figure depicts the applicability of the various strain and stress mesures:
_______________________________________________________________________________________________
DAMPING
_______________________________________________________________________________________________
DAMPING
• In general damping in structure is not viscous. It is due to hysteresis and usually modeled by
viscous damping.
• Treatment of damping:
1. Phenomenological damping methods:

Where actual physical dissipation is modeled (seldom used)
Elastic plastic hysteresis loss
Structural joint friction
Material microcracking
2. Spectral damping methods (Rayliegh or proportional damping)

Viscous damping is introduced by means of specified fraction of critical damping.
• Rayleigh damping is defined by
C = αM + βK
Where C, M, and K are damping, mass, and stiffness matrices.

α and β Are the mass and stiffness proportional constants.
• The keyword options are:
*DAMPING_GLOBAL
*DAMPING_PART_MASS
*DAMPING_PART_STIFFNESS
• For the first option you can either specify a load curve for damping or a system damping
constant “d”. With system damping the acceleration is computed as:
1
an = ( R ext − R int − R damp )
M
where R damp = dmv
• The best damping constant for the system is usually based on the critical damping factor for the
lowest natural frequency of interest.
• The last two keyword options assigns mass weighted damping and stiffness damping by part ID
• You can specify mass weighted damping by a load curve.
• Mass weighted damping damps all motion including rigid body motions. It is more effective for
low frequencies.
• Stiffness proportional damping can be input by specifying the value of β .
• Using the normal mode methods we can write
_______________________________________________________________________________________________
DAMPING
_______________________________________________________________________________________________
φiT Cφi = φiT (αM + βK )φi = α + βω i2 = 2ω iξ iδ ij
Where ω i is the ith frequency.

• If only stiffness proportional damping is needed, then βω i2 = 2ω iξ i .
• If let say only 10% of critical damping is desired for the ω i frequency then
0.2
β=
ωi
• Stiffness proportional damping is effective for high frequencies and it is orthogonal to rigid
body motions.
• Stiffness proportional Rayleigh damping is introduced by replacing the incremental stress
computation by
α stiff
∆σ = C∆ε + C elas ∆ε
∆t
1 1 1
n+ n+ n+
2 2 2
elas
Where α stiff is the stiffness proportional damping coefficient, ∆t is the time step, and C is the
elastic consecutive tensor. The stiffness proportional damping coefficient is specified by material.
• Mass proportional damping is added with
α mass
Feint = ∫ BnT+1τ n +1 dV + M∆u e
Vn +1 ∆t
Where α mass is the proportional damping coefficient, M, is the element mass, matrix, and ∆u , is
the vector of nodal incremental displacements. In LSDYNA/IMPLICIT implementation, a unique
stiffness proportional damping coefficient may be specified for each part. However, one mass
proportional damping coefficient operates on the entire model mass. The damping force is
proportional to a midstep velocity approximation.
_______________________________________________________________________________________________
REDUCED INTEGRATION & HOURGLASSING
_______________________________________________________________________________________________
SELECTIVE REDUCED INTEGRATION

&
HOURGLASSING
SELECTIVE REDUCED INTEGRATION
Zero energy mode, hourglass mode, rank deficiency
A lower integration rule is desirable for the following reasons:

Lower computation cost.
Tendency to soften an element, thus countering the overly stiff behavior associated with an
assumed displacement field.
Reduce shear locking in elements with transverse shear deformations (in shell elements).
Softening comes about because certain higher order polynomial terms happen to vanish at
gauss points of a lower order rule, so that these terms make no contribution to strain energy.
However reduced integration rule produces zero energy deformation mode (or hourglass mode)
• (A pattern of nodal D.O.F. produces a strain that is zero at all integration points)
• If the integration points (gauss point) sense no strain under a certain deformation mode, the
resulting element stiffness matrix will have no resistant to that deformation mode.
• The stiffness matrix for bilinear four nodded elements has 8 eigenvalues in which lead to 8
independent displacement modes, they are:
Three rigid body displacement

Two translation and one rotation (will not contribute to the strain energy)
_______________________________________________________________________________________________
_______________________________________________________________________________________________
Three constant strain modes

Stretching, uniform extension, and shear
Two bending (flexural) modes
If exact or recommended integration is used the stiffness matrix will have
8-3=5 nonzero eigenvalues
(The 3 zero eigenvalues correspond to rigid body motion)
If reduced integration is used the stiffness matrix will have only 3 nonzero eigenvalues
So we will have two additional displacement modes that will not contribute to the strain energy,
they are the spurious zero-energy modes
Example:
Four-nodded plane elements integrated by one-integration point
u = cxy u=0
v=0 v = −cxy
_______________________________________________________________________________________________
_______________________________________________________________________________________________
The above displacement fields will lead to

ε xx = ε yy = γ xy = 0 At gauss points ξ = η = 0 (u ,ξ = u ,η = v,ξ = v,η = 0)
HOURGLASSING
Out-Of-Plane "W" Hourglass Mode in Shell Elements
undeformed deformed
The figure below depicts the effect of hourglass on deformation. The vehicle had the same initial
kinetic energy and both runs have the same termination time.
HOURGLASS CONTROL
Adding hourglass stiffness must provide stability to the element without stiffening the element to
legitimate modes that are already working well. To suppress the hourglass deformation modes,
hourglass viscosity stresses are added to physical stresses at the local element level.
Two methods are generally used:

1. Hourglass viscous forces
2. Hourglass stiffness forces
VISCOUS FORMS
• Standard LSDYNA (default)
_______________________________________________________________________________________________
_______________________________________________________________________________________________
• Flanagan-Beltyschko (2)
• Flanagan-Beltyschko with exact volume integration (3)
STIFFNESS FORMS
• Flanagan-Beltyschko (4)
• Flanagan-Beltyschko with exact volume integration (5)
Some Comments on the Hourglass Control Methods
• Viscous form sometimes works better and is the default.

• Stiffness form is more stable in many applications and is preferred for auto crash and sheet
stamping.
• The stiffness form may results in a stiffer response.
• Flanagan-Beltyschko hourglass control behaves better for large rotations.
• Fully integrated elements do not hourglass.
• If localized hourglass occurs in area far from design critical areas, then it is admissible.
• Hourglass modes are orthogonal to real deformation and rigid body motions.
• Hourglass energy vs. time appears in GLSTAT and MATSUM.
• Hourglass energy should be small relative to internal energy (0%-10% is good).
• The work done by hourglass stabiliztation forces dose not appear in energy balance equation
unless it is signaled in the CONTROL_ENERGY card.
The following cards are used to control hourglass parameters.
*CONTROL_ENERGY
• switch to have hourglass energy calculated
*CONTROL_HOURGLASS
• set hourglass type
• user can modify hourglass coefficient
*HOURGLASS
• set hourglass type and other parameters for use in a specific parts
*PART
• can be signaled to use a specific hourglass ID
_______________________________________________________________________________________________
QUASI STATIC SIMULATION
_______________________________________________________________________________________________
QUASI STATIC INITIALIZATIN

AND
QUASI STATIC INITIALIZATIN
In some cases it is desirable to have prestress loading of a structure before a transeint dynamic
analysis is performed. This situation occurs in many applications. To name few as follows:
• Submarine under hydrostatic pressure

• Gravity and body force loading
• Bird impact on rotating jet engine fan blades
• And many others.
It is possible to perform quasi-static initialization by three methods as follows:
1. Monotonic increment of pre loading

2. Dynamic relaxation
3. Implicit step calculation
These methods are described as follows:
1. Monotonic increment of pre loading
In here the applied load for prestress is applied in monotonic increments up to the final load before
transient analysis begins. For instance, if car crash is desired then, the gravity is applied with in a
load curve from time zero to some other time (say t1). Then this gravity loading is held constant for
the duration of the simulation. The impact velocity of the car can start with a second load curve
that has its starting time to be t1.
2. Dynamic relaxation
A load curve can be used for both prestressing and transient analysis. In the prestressing (or
dynamic relaxation) phase LSDYNA approximates a solution of the nonlinear quasi-static problem
by incorporating damping in the update of the displacement field.
3. Implicit step calculation
In here one can solve the prestress problem using an implicit solution phase. With in this phase one
needs to include the following card:
*INTERFACE_SPRINGBACK_DYNA3D
_______________________________________________________________________________________________
_______________________________________________________________________________________________
This will cause the code to generate a “dynain” file at the end of the simulation. This file contains
initial stresses imposed on the elements at the end of the simulation for the part defined. Then one
can take these initial-stresses and put into the input file for a second explicit phase with
prestressing.
WHY STATIC ANALYSIS WITH AN EXPLICIT CODE:
• Direct Solution May Fail When:

The material law is highly nonlinear. Extremely complicated material behavior is much
more easily accommodated with explicit methods
• Number of elements is extremely large

The IO’s are less important
The memory required is also less
Explicit method does not require matrix inversion
• Matrices must be re-evaluated at each time step and for most iterations
The CPU time of an explicit solution becomes competitive
• The problem includes several contact surfaces

Contact algorithms are very efficient in explicit codes
• Excessive rigid body motion

Explicit algorithm works very well
• Direct solution may fail when local buckling occur in large scale finite element models
Quasi-static analysis with large deformation and contact (like roof crash and door intrusion) is
expensive and some times impossible using implicit method. Explicit methods can be used with
time and mass scaling
Performing Static Simulations
• Dynamic FE codes account for inertia in formulation.

• Applied forces in static FE codes results in stress, however, in dynamic FE codes results in
stress waves.
• To model quasi-static events using explicit dynamic codes the events must be scaled in time
(from seconds to milliseconds).
• Need to apply forces or displacements in ramp curves. Sudden application of loads will results
in stress waves that could be much larger than stresses at the maximum applied forces when
performing static simulation using implicit LSDYNA.
_______________________________________________________________________________________________
_______________________________________________________________________________________________
1 2
• Note that to reduce kinetic energy E = mv we have two options. Either reduce velocity “v”
2
or reduce mass “m”.
• For instance to reduce E by ½ we can either reduce the velocity by 1 or the mass by ½.
2
Velocity reduction has more pronounced effect on kinetic energy reduction.
Three issues must be addressed to determine suitable loading rate when simulating quasi-static
tests using explicit LSDYNA:
• Strain rate effects (must not be included in material model).

• Amplitude of stress wave in material should be small ( F = σA = ρcvA ; c = E / ρ ).
Nonreflecting boundary can be used to reduce the effect of stress wave.
• Ratio of kinetic energy to strain energy should be small.
Note that when performing quasi-static simulations the higher the rate of loading the higher the
elongation at failure.
Quasi-Static Post-buckling Analysis Example
Time and Mass Scaling
Time scaling:
*CONTOL_TIMESTEP
• Minimum time step size can be specified. For shell elements and the following material:
*MAT_PLASTIC_KENEMATIC
*MAT_POWER_LAW_PLASTICITY
*MAT_STRAIN_RATE_DEPENDENT_PLASTICITY
_______________________________________________________________________________________________
_______________________________________________________________________________________________
*MAT_PIECEWISE_LINEAR_PLASTICITY
• The moduli in the element material properties will be modified to keep the time step as the
assigned value.
• You need to specify the parameter “TSLIMT” in the *CONTROL_TIMESTEP
• Dangerous exercise in FE simulation.
Mass scaling:
*CONTOL_TIMESTEP
• Minimum time step size can be specified, Preferred in FE simulation. For all elements and
materials;
• The mass in the element material properties will be modified to keep the time step as the
assigned value.
• You need to specify the parameter “DT2MS” in the *CONTROL_TIMESTEP.
• Positive values are for quasi-static analysis or time history analysis with insignificant inertial
effects.
• Negative values is the minimum time step size permitted and mass scaling is done if and only if
it is necessary to meet Courant criterion.
• You can bound mass increase by the parameter “ENDMASS” in the termination card.
LSDYNA EXPLICIT VS IMPLICIT COST
• Each increment in IMPLICIT analysis consists of at least one iteration, but usually more than
one.
• Each iteration in IMPLICIT requires the solution of a set of simultaneous equations
• The CPU cost per iteration is roughly proportional to the number of degrees of freedom in the
model squared.
• Each increment in EXPLICIT consists of one group of equations-there are no iterations.
• The CPU cost per increment is directly proportional to the number of degrees of freedom in the
model.
• As the size of the model increases, there is a point at which EXPLICIT may become more cost
effective than IMPLICIT for simulations that could be solved with either.
_______________________________________________________________________________________________
ADAPTIVITY
_______________________________________________________________________________________________
ADAPTIVITY
Theoretical and Practical Aspects of Mesh Adaptation
Objective
• Even with the development of newer and faster computers, it is desired to make simulation as
efficient as possible. Often times, the potential behavior of a model can only be guessed at,
requiring finer meshes and thus, greater amounts of processing time.
What is adaptive meshing?

• Adaptive meshing begins with the solution of an initial and usually coarse mesh.
• Based on error estimators, relative element discretization errors are quantified for the entire
mesh.
• In areas of large error, the mesh is refined or enriched
• The objective of the adaptation is to equally distribute error over the entire mesh.
What’s what?
• H-method alters the elements of the initial mesh by decreasing the size (refinement) or
increasing the size (de-refinement).
• P-method changes the order of the shape function polynomial while keeping the element
geometry constant.
• R-method moves the nodes of the elements while keeping their number and connectivity the
same.
• Can also have combinations of methods (h/p for example) or total remeshing.
Where It’s At -- H-Adaptivity
• H-Adaptivity - many major codes use this form of mesh adaptivity, including: LS-DYNA
Future Trends
More emphasis seems to being placed on the h/p method because of its’ ability to enrich and refine
the mesh.
LSDYNA ADAPTIVITY: H-adaptive refinement uses error norms to generate elements where
they are needed. Solution errors are minimized without a manual, trial and error approach. A one
side two neighbor rule is used for ease in implementation.
_______________________________________________________________________________________________
ADAPTIVITY
_______________________________________________________________________________________________
ADAPTIVE DATA STRUCTURE
Internal software and user databases are an issue in h-adaptivity.

• New element connectivities are generated for each adaptive remeshing. Requires resorting for
disjointness and subcycling. Connectivities must be available for post-processor also. Node
numbers do not change.
• In parallel computations the domain decomposition is repeated for load balancing
considerations (available soon).
• LS-TAURUS and LSPOST will read through and plot the adaptive database.
• Optimized mesh data is available for future run without adaptivity.
REFINEMENT INDICTOR
A deformation based indictor-using angles- no error indicator.

Check angular deformation between elements in plane and out-of-plane.
If ζ > ζ to1 the refinement, where ζ to1 is user determined. Limit for refinement has to be determined
by user as well as the number of refinements.
_______________________________________________________________________________________________
ADAPTIVITY
_______________________________________________________________________________________________
CONTINUE CALCULATION OR BACK UP?
After computing the changes in deformation one can:

• Refine the mesh and continue.
• Backup to an earlier state and repeat calculation.
Advantages and disadvantages of continuing:

• Speed but for accuracy the frequency of adaptive remeshing must be high and this may be
costly.
• Much easier implementation.
• Will be inefficient on MPP’s.
Advantages and disadvantages of repeating the calculation:

• Speed since only very intermittent error checks are made.
• Speed due to the fact that all the vector blocking of elements can be accomplished.
• Nearly all options work with adaptive remeshing.
• Domain decomposition on MPP’s.
• Much more difficulty in implementation.
_______________________________________________________________________________________________
ADAPTIVITY
_______________________________________________________________________________________________
_______________________________________________________________________________________________
SPRINGBACK
_______________________________________________________________________________________________
SPRINGBACK OPTIONS IN LS-DYNA
Springback analysis is important in metal forming, structural design

● Apply load, causing plastic deformation
● remove load, compute residual deformation and "springback" displacement
● static springback solution needed => explicit dynamics not good
● design problem: find tool shape which gives correct part geometry after springback
● two modeling strategies are available:
Implicit Method
● run entire simulation using static implicit method
● load completely removed at final time step
● control rigid body modes using BCs with late birth time (if necessary)
Seamless Springback method

● standard explicit simulation used to apply load
● automatic, seamless switch to implicit for static springback
● switch formulation at termination time
● springback for a subset of parts, additional BC’s optional
● all contact interfaces automatically killed
Minimal input required

● *INTERFACE_SPRINGBACK_SEAMLESS
➤ part set ID
➤ optional constraints to eliminate RB modes
● *CONTROL_IMPLICIT keywords are optional
➤ change number of steps (default: NSBS=1)
➤ change stepsize (default: DT0 = TERM)
➤ alternate "springback" defaults used: IAUTO=1, IAS=1 (see manual)
Implicit Springback
Single vs. multi-step springback

● many springback problems can be unloaded in one nonlinear step
● for large springback, geometric nonlinearity requires several steps
Artificial stabilization
● Allows multi-step springback
● artificial springs constrain motion of every node
● springs soften during simulation, vanish at termination time
● accurate springback only when stabilization is completely removed
_______________________________________________________________________________________________
SPRINGBACK
_______________________________________________________________________________________________
Automatic time step control

● modify new step size using iteration count of previous step
● repeat failed steps with smaller time step size
● works in conjunction with artificial stabilization for springback
Nonlinear, Multi-Step Springback Analysis
Single vs. multi-step springback

● many springback problems can be unloaded in one nonlinear step
● for large springback, geometric nonlinearity requires several steps
● seamless default: try single step, activate two features below just in case...
Automatic Time Step Control

● adjusts step size automatically to hold convergence "effort" constant
● repeats failed steps with exponentially smaller step size
● very, very persistent
Artificial Stabilization
● artificial springs added to model
● springback over several steps
● springs vanish at termination time
● stiffness multiplier option
● load curve option for k vs. t
10.0
Spring Stiffness
1.0
0.1
0.0
0.0 1.0
Simulation Time
_______________________________________________________________________________________________
SPRINGBACK
_______________________________________________________________________________________________
Seamless Springback
Restart
● small restart input deck may include:
➤ all *CONTROL_IMPLICIT commands
➤ *INTERFACE_SPRINGBACK
● binary d3dump file automatically generated after forming phase
● no need to repeat forming simulation if springback fails
Implicit Springback Simulation

Accuracy Issues
Accurate Thickness Stress Distribution is Critical A

B
● formability analysis requires accurate strain
➤ thinning, splitting, wrinkling, plastic set
➤ 3 thickness gauss points are often sufficient
● springback is driven by stress distribution
➤ 7 or 9 thickness gauss points may be needed
F
ζ ζ ζ
h/2 h/2 h/2
- σy σy - σy σy - σy σy
- h/2 - h/2 - h/2
Location A Location B Location B (unloaded)
_______________________________________________________________________________________________
SPRINGBACK
_______________________________________________________________________________________________
Implicit Springback Performance
• SGI R10000 machine

• fully incore stiffness matrix
• single precision default solver
Computing requirements for implicit springback
no trim with trim

elements 40,000 26,000
memory 400 Mb 368 Mb
CPU 53 min 13 min
Trimming improves CPU time (less flops), but

poor connectivity causes large fill-in of [K] -1,
hence large memory requirement
_______________________________________________________________________________________________
TROUBLE SHOOTING
_______________________________________________________________________________________________
TROUBLE SHOTING
Consider the simple elastic truss, one degree of freedom per node:
u1 u2
elastic stiffness K=EA/L, so static equilibrium gives

F = Kx =
EA
(u2 − u1 )
L
We cannot solve this problem unless a constraint is imposed on one degree of freedom. The model
has a rigid body mode - unconstrained rigid body translation is possible.
1-D Truss with Rigid Body Mode
One solution is to add a constraint to one node
FL
u1=0 u2 =
EA
The constraint could be zero displacement, or the node could be attached to another structure which
is "grounded".
Another solution is to consider the truss as a spring-mass system, and solve a transient dynamics
problem of the form F=Ma.
1-D Truss with Spring Support
Eliminate rigid body motion by adding spring support
k F
u1 u2
now static equilibrium requires
_______________________________________________________________________________________________
TROUBLE SHOOTING
_______________________________________________________________________________________________
F = ku1   k 0  u  F 

FL  ⇒  EA EA   1  =  
= u2 − u1  − L L  u 2  F 
EA 
and after inverting the stiffness matrix, the displacements are
 k1 0   F   u1 
1 L  
= 
k EA   F  u2 
in the limit where the spring becomes infinitely stiff, the previous solution is recovered
FL
u1 = 0, u2 =
EA
When the spring vanishes, the inverse stiffness is undefined and no solution can be found: the
stiffness matrix is singular
LS-DYNA will issue a "negative eigenvalue" warning message during stiffness matrix inversion.
Static Simulations Require Constraints
• Boundary conditions and rigid body modes

• static implicit simulation requires boundary constraints
• rigid body modes must be eliminated
• apply translational constraints to three nodes
a: reference node, dx=dy=dz=0

• eliminates all translational modes
b: node along X-axis, dy=dz=0

• eliminates rotations about y- and z-axis
c: node along Y-axis, dz=0

• eliminates rotation about x-axis
c
Y
a
X
Numerical Roundoff in the Linear System
Implicit FEA simulations are susceptible to numerical errors

• Force (residual) vector computation
_______________________________________________________________________________________________
TROUBLE SHOOTING
_______________________________________________________________________________________________
area, volume integration, stress evaluation

similar to explicit methods
single precision usually adequate
• Gaussian Elimination
roundoff errors during stiffness factorization
rule of thumb: numerical digits lost = c log(condition number)
In well conditioned model, equilibrium iteration corrects G.E. errors:
Example - Nodal Displacements in Beam Tower Model, mm.
linear nonlinear
double precision 127.7 127.7
single precision 124.2 127.8
In poorly conditioned models, increased sensitivity to errors in force vector compound the
problem.
Condition Number vs. Mesh Density
The condition number of a model is effected by mesh density.
Example: simple beam modeled with n beam elements

• Lowest eigenvalue remains virtually unchanged
• Highest eigenvalue (axial mode) increases with mesh refinement:
K K K K
M M M M
k k k
ω2 = ω2 = 2 ω2 = n
m m m
one mass two masses n - masses
L0
• for n-element lumped mass beam model, element length L = ,
n
 EA 
(ρAL ) = αn
k −1
λmax = ω 2 = n = n 3
m  L 
• so condition number varies with mesh density cubed
Condition Number vs. Boundary Conditions
The condition number is effected by boundary conditions
_______________________________________________________________________________________________
TROUBLE SHOOTING
_______________________________________________________________________________________________
Example: simple beam modeled with n beam elements

• Highest eigenvalue remains virtually unchanged
• Lowest eigenvalue (cantilever mode) increases with added constraint
λmin = ω λmin = 7.4ω λmin = 39.8ω
cantilever simple support double clamp
• boundary conditions improve conditioning, decrease roundoff error
_______________________________________________________________________________________________
TROUBLE SHOOTING
_______________________________________________________________________________________________
NONLINEAR CONVERGENCE PROBLEMS
Convergence trouble is the most common problem. Error messages displayed by LS-DYNA :
● iteration limit reached

displacement and energy tolerances were not satisfied, abandon step
● divergence
out-of-balance force R is growing, reform K and continue iterations
● energy explosion
energy norm grows beyond 1,000,000x initial value, abandon step
● negative eigenvalues
error from linear equation solver while computing K-1
input parameter gives procedure (default: ignore, try to continue)
● line search stepsize zero

new displacement ∆u does not reduce R
reform K if not new last iteration, otherwise abandon step
Nonlinear Convergence Problems
Several diagnostic tools are available

● NLPRINT print flag / sense switch for diagnostic information
● negative eigenvalue flag optionally repeats with smaller step immediately after encountering
negative eigenvalues, avoiding potential inverted element trouble (DEFAULT=ignore negative
EVs)
● “<ctrl-c> conv” overrides tolerances, forces convergence this step
● “<ctrl-c> iter” creates “d3iter” plot database every iteration
Procedures for solving convergence problems

● determine reason for termination (examine error messages)
● activate print flags to get more information
● view deformed geometry during iteration process using "d3iter" database
● carefully inspect input deck
Troubleshooting Non-Convergence
Iteration Limit Reached error message

● observe convergence progress of displacement, energy norms
(view d3hsp file or activate NLPRINT flag for screen output)
_______________________________________________________________________________________________
TROUBLE SHOOTING
_______________________________________________________________________________________________
● if tolerances nearly satisfied:

allow a few more iterations
relax tolerance slightly
interactively force convergence using "<ctrl-c> converge"
● if convergence progress is slow

switch to Full Newton method (ILIMIT=1)
check for incorrect material properties
increase step size
● if convergence progress is erratic

decrease contact penalty stiffness scale factor 10x (monitor penetration)
decrease step size
Divergence and Energy Explosion error messages:

● some part of model is likely deforming very rapidly
● generate d3iter database using "<ctrl-c> iter"
● observe model during iteration process, watch for locally large deformation
● check for unbounded applied loads
● check for incorrect yield stress or hardening (temporarily switch to elastic)
● check contact definitions for reversed normals or excessive penetration
● decrease time step size
Negative Eigenvalue error message - four possibilities
1) Non-physical material properties

● negative Young's modulus, negative slope of stress-strain curve, etc…
● check d3hsp file to be sure LS-DYNA read the input deck properly
2) Inverted elements
● negative volume elements in initial mesh (check in preprocessor)
● elements inverted when nodes relocated during contact initialization
check messages in d3hsp file, view mesh at time=0 with postprocessor
● elements inverted due to excessive deformation
activate plot state output to get many pictures just before trouble begins
3) Rigid body modes

● review boundary conditions and constraints
● watch out for beams that are free to spin about their axes
4) A feature in the model is not available for implicit analysis

● warning messages are not always issued when illegal feature is used
● check keywords against list in Appendix M of User's Manual
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Lab Exercise Problems
These example problems are designed with two objectives in mind:
• Demonstrate and teach LS-DYNA syntax

• Provide understanding of nonlinear implicit/explicit FE analysis
___________________________________________________________________________________
Problem #1: Tensile Test
___________________________________________________________________________________
Objectives
• Learn how to activate LS-DYNA’s implicit mode.
• Learn how to select linear or nonlinear analysis.
Problem Description
A static tensile test is simulated using shell elements. One end of the specimen is constrained,
while concentrated nodal loads are applied at the other end. Uniform stresses develop in the
narrowed center section.
Input Filename: tensile1.k
Procedure
Copy the input file to your local directory. Using an editor, view the input file and answer the
following questions:
1. Which shell element formulation is used? __________________________________

2. What hourglass control is used? __________________________________________
3. How many steps are used to apply the load? ________________________________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Run the simulation, and post-process using LS-POST. Record below the total applied load and
total tip displacement (HINT: the ascii database NODFOR includes data for each of the tip
nodes).
4. Applied load _________ Tip displacement _________
Modify the input deck to increase the applied load 100x, and rerun the simulation. Record the new
data:
6. Is the simulation linear or nonlinear ? (Does the tip displacement scale linearly with applied
load?)
7. What types of nonlinearity exist in this problem? ( ) material ( ) geometric ( ) contact
Modify the *CONTROL_IMPLICIT_NONLINEAR keyword to perform the “other” type of

simulation, and record the results below for the large applied load:
9. What is the purpose of the *CONTROL_IMPLICIT_LINEAR keyword? Does this keyword

select a linear solution? _______________________________________________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
*TITLE
implicit tensile test
$
$ test coupon: 200 mm long, 20 mm width, 2.67 mm thickness, 50.8 mm gauge length
$
$ units; mm, s, ton, N
$
$ A. Tabiei
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
*CONTROL_TERMINATION
1.0000
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
$ imflag dt0 iefs nstepsb igs
1 1.0 0 0 0
$
$ nlsolvr ilimit maxref dctol ectol rctol lstol
0 0 0 0.0 0.0 0 0
$ dnorm divflag inistif nlprint
0 0 0 0
$
$ lsolvr prntflg negeig
0 0 0
$
$ iauto iteopt itewin dtmin dtmax
0 0 0 0.0 0.0
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
*DATABASE_BINARY_D3PLOT
0.010
*DATABASE_NODFOR
0.010
*DATABASE_NODOUT
0.010
*DATABASE_NODAL_FORCE_GROUP
2
*SET_NODE_LIST
2
29,30,31,32,33,38,39
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
*MAT_elastic
$ MID RO E PR
1 0.780E-08 0.207E+06 0.280E+00
$
*SECTION_SHELL
1 0
2.670E-00 2.670E-00 2.670E-00 2.670E-00
$
*PART
shell tensile strip
1 1 1
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
*DEFINE_CURVE
1
0.00000000000000E+00 0.0000000000000E+00
1.00000000000000E+00 2.0000000000000E+02
$
*load_node_point
$ Node DOF LCID SF
204 1 1 1
205 1 1 1
206 1 1 1
207 1 1 1
212 1 1 0.5
213 1 1 0.5
$
*BOUNDARY_SPC_NODE
30 0 1 1 1 1 1 1
31 0 1 1 1 1 1 1
32 0 1 1 1 1 1 1
33 0 1 1 1 1 1 1
38 0 1 1 1 1 1 1
39 0 1 1 1 1 1 1
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
*NODE
14 1.000000000E+01 1.600000000E+01 0.000000000E+00
*ELEMENT_SHELL
1 1 38 30 14 40
*END
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
LS-DYNA Implicit Workshop
Problem #2: Elastic-Plastic Tensile Test
___________________________________________________________________________________
Objectives
• Learn how to observe convergence behavior of nonlinear equilibrium iterations.
• Learn to use automatic time step control for nonlinear problems.
• Learn the benefits of force vs. displacement controlled simulations.
Problem Description
A static tensile test is simulated using shell elements and a nonlinear, elastic-plastic material
model. One end of the specimen is constrained, while concentrated nodal loads are applied at the
other end. Uniform stresses develop in the narrowed center section.
Input Filename: tensile2.k
Procedure
1. Which material model is used? What is the yield stress? ______________________

2. How is load applied?___________________________________________________
3. How many steps are used to apply the load? ________________________________
Run the input deck. Does the job run to completion? ( ) normal termination ( ) error
termination
4. At what time does the solution begin to struggle? time = ____________________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Using the postprocessor, plot the X-displacement of an end node vs. time.
5. What is happening at the time shown above? _____________________________
6. max end displacement _____ max eff. stress _______ max eff. strain _____
Activate the nonlinear print flag to get more information about the nonlinear solution process.
7. What two methods are available for this? __________________________________
8. Repeat the simulation. Why does the simulation stop? _______________________
Switch from load control to displacement control, and repeat the simulation (Hint: helpful
keywords are commented out in the original input deck). Using the postprocessor, again plot the
X-displacement of an end node.
10. Why is this problem easier to solve? ______________________________________
Return to the original input deck, and activate automatic time step control (IAUTO=1 on the
keyword *CONTROL_IMPLICIT_AUTO). Repeat the simulation.
11. What happens? ______________________________________________________
Using the “ASCII” menu, load the GLSTAT database and plot the step size vs. time.
13. When does the step size begin to decrease? _________________________________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
*TITLE
implicit tensile test, elastic-plastic material
$
$ test coupon, 200 mm long, 20 mm width, 2.67 mm thickness, 50.8 mm gauge length
$
$
$ By A. Tabiei
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
1.0000
$
$========1=========2=========3=========4=========5=========6=========7=========8
$
$ imflag dt0 iefs nstepsb igso
1 0.01 0 0 0
$
0 0 0 0.0 0.0 0 0
0 0 0 0
$
0 0 0
$
0 0 0 0.0 0.0
$
$========1=========2=========3=========4=========5=========6=========7=========8
$
0.0100
*DATABASE_EXTENT_BINARY
$ neiph neips maxint strflg sigflg epsflg rltflg engflg
1 1
$ cmpflg ieverp beamip
$
$
*DATABASE_NODFOR
0.0001
2
*SET_NODE_LIST
2
29,30,31,32,33,38,39
$
$
*DATABASE_ELOUT
0.0001
*DATABASE_HISTORY_SHELL
71,72,73,74,75,76,77,78
79,80
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*CONTROL_SHELL
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
$ WRPANG ITRIST IRNXX ISTUPD THEORY BWC MITER PROJ

0 0 0 0 0 0 1 0
$
*SECTION_SHELL
$ ID elform
1 16
$ t1 t2 t3 t4
2.670E-00 2.670E-00 2.670E-00 2.670E-00
$
*PART
shell tensile strip
1 1 1
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
$ material 37, load curve gives hardening
*MAT_TRANSVERSELY_ANISOTROPIC_ELASTIC_PLASTIC
$ MID RO E PR SIGY ETAN R HLCID
1 0.780E-08 0.207E+06 0.280E+00 0.200E+03 0.572E+03 0.140E+01 13
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
$ material 37, constant hardening modulus
$*MAT_TRANSVERSELY_ANISOTROPIC_ELASTIC_PLASTIC
$ 1 0.780E-08 0.207E+06 0.280E+00 0.200E+03 0.572E+03 0.140E+01 0
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
$ hardening curve: effective stress vs. effective plastic strain
*DEFINE_CURVE
13
0.00000000 200.0
0.00006657 200.5
0.00013650 201.1
0.00020990 201.6
0.00028690 202.2
0.00036780 202.7
0.00045280 203.2
0.00054200 203.8
0.00091670 205.8
0.00136600 207.9
0.00190600 210.0
0.00255300 212.1
0.00472900 217.2
0.00654700 220.3
0.00891000 223.4
0.03259000 250.8
0.04616000 270.5
0.06516000 291.5
0.09176000 313.8
0.13830000 342.4
0.20810001 373.2
0.31279999 406.2
0.46990001 441.6
0.70560002 479.7
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
*LOAD_NODE_POINT
204 1 3 1
205 1 3 1
206 1 3 1
207 1 3 1
212 1 3 0.5
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
213 1 3 0.5
*DEFINE_CURVE
3
0.00 0.00
1.00 2000.00
2.00 2000.00
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
$*BOUNDARY_PRESCRIBED_MOTION_NODE
$ 204 1 2 4 1.00
$ 205 1 2 4 1.00
$ 206 1 2 4 1.00
$ 207 1 2 4 1.00
$ 212 1 2 4 1.00
$ 213 1 2 4 1.00
$*DEFINE_CURVE
$ 4
$ 0.00 0.00
$ 1.00 10.00
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
*BOUNDARY_SPC_NODE
30 0 1 1 1 1 1 1
31 0 1 1 1 1 1 1
32 0 1 1 1 1 1 1
33 0 1 1 1 1 1 1
38 0 1 1 1 1 1 1
39 0 1 1 1 1 1 1
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
*NODE
14 1.000000000E+01 1.600000000E+01 0.000000000E+00
*ELEMENT_SHELL
1 1 38 30 14 40
*end
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Problem #3: Ellipsoidal Dome
___________________________________________________________________________________
Objectives
• Learn the behavior of different element formulations in static implicit simulations.
• Learn how to minimize hourglass problems.
Problem Description
A static load is applied to the center of an ellipsoidal dome. Shell elements are used. Nodes at the
base of the dome are constrained, and included in a NODFOR output database.
Input Filename: ellipse.k
Procedure
1. What is the maximum value of the applied load? _____________________________
2. How many load steps are used? _________________________________________
3. What shell element formulation is used? __________________________________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
4. What hourglass control type is used? _____________________________________
Run the simulation, and postprocess the results.
5. What happens? _______________________________________________________
6. Applied load __________ Center displacement _______
Repeat the simulation using the S/R Hughes-Liu shell (ELFORM=6), and postprocess the results.
Repeat the simulation using the Fast Fully Integrated shell (ELFORM=16), and postprocess the
results.
9. What conclusion can you draw about the effect of hourglass deformation on overall structural
stiffness for this problem? _____________________________________________
Concentrated applied loads are often responsible for initiating hourglass deformation. Edit the
input deck and replace the single concentrated nodal load with pressure acting on the adjacent
segments (Hint: Helpful keywords are included in the comments inside the input deck). Repeat
the simulation using the default shell.
10. What happens? ______________________________________________________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
*TITLE
PINCHED ELLIPSE
$
$ A. Tabiei, March 99
$
1.0000
$
$========1=========2=========3=========4=========5=========6=========7=========8
$
1 0.01 0 0 0
$
0 0 0 0.0 0.0 0 0
0 0 0 0
$
0 0 0
$
1 0 0 0.0 0.0
$
$ imass gamma beta
0 0.0 0.0
$
$========1=========2=========3=========4=========5=========6=========7=========8
$
1 1
0.01
*DATABASE_GLSTAT
0.01
*DATABASE_NODFOR
0.01
2
*SET_NODE_LIST
2
1 7 13 19 25 31 37 48
54 60 66 72 78 84 91 97
103 109 115 121 127 206 212 218
224 230 236 242
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
*CONTROL_HOURGLASS
4
$
*MAT_elastic
$ MID RO E PR
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
1 7.890E-09 2.100E+05 3.000E-01

$
*SECTION_SHELL
$ SID ELFORM
1 0
3.000E-00 3.000E-00 3.000E-00 3.000E-00
$
*PART
SHELL
$ PID SID MID ADPOPT
1 1 1
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
*LOAD_NODE_POINT
$ NID DOF LCID SF
164 3 1 -1.0
$
*DEFINE_CURVE
1
0.00000000000000E+00 00.000000000000E+00
1.00000000000000E+00 500000.00000000E+00
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
$*LOAD_SEGMENT
$$ LCID SF N1 N2 N3 N4
$ 2 -1.0 154 155 164 163
$ 2 -1.0 155 156 165 164
$ 2 -1.0 163 164 173 172
$ 2 -1.0 164 165 174 173
$
$*DEFINE_CURVE
$ 2
$0.00000000000000E+00 00.000000000000E+00
$1.00000000000000E+00 7500.0000000000E+00
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*NODE
1 22.627417 -16.97056299999 0.0 7 7
*ELEMENT_SHELL
1 1 1 2 8 7
*END
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Problem #4: Adaptive Ellipsoidal Dome
___________________________________________________________________________________
Objectives
• Learn to activate mesh adaptivity in an implicit simulation.
• Learn how to minimize hourglass problems.
Problem Description
A static load is applied to the center of an ellipsoidal dome. Shell elements are used. Nodes at the
base of the dome are constrained, and included in a NODFOR output database. Adaptivity is used
to automatically refine the mesh in areas of high curvature.
Input Filename: aellipse.k
Procedure
1. How frequently will the mesh be evaluated for refinement? ____________________
2. How many times can each element be subdivided? __________________________
3. How do you indicate which parts will be adapted? ___________________________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
4. Which element formulation is used? _____________________________________
5. How is load applied? __________________________________________________
Execute the simulation, and view the results with the postprocessor.
7. How many (auto) time steps were used? __________________________________
8. Does the adaptive mesh improve the hourglassing problem ___________________
Switch to pressure driven load application, and repeat the simulation. Postprocess the results.
Using the NODFOR database, verify that the load is applied correctly as the mesh is refined.
10. How many (auto) time steps were used? __________________________________
11. Does the pressure load improve nonlinear convergence? ______________________
Experiment with shell element formulation #16.
12. Does shell type #16 improve hourglassing? _________________________________
13. Does shell type #16 improve convergence behavior (number of auto steps)? ______
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
*TITLE
PINCHED ELLIPSE
$
$ A. Tabiei, March 99
$
1.0000
$
*CONTROL_ADAPTIVE
$ adpfreq adptol adpopt maxlvl tbirth tdeath lcadp ioflag
0.10 5.000 2 3 0.0 0.0 0 0
$ adpsize adpass ireflg adpene
0.0000000 1 0 1.0
$
$========1=========2=========3=========4=========5=========6=========7=========8
$
1 0.01 0 0 0
$
0 0 0 0.0 0.0 0 0
0 0 0 0
$
0 0 0
$
1 0 0 0.0 0.0
$
$ imass gamma beta
0 0.0 0.0
$
$========1=========2=========3=========4=========5=========6=========7=========8
$
1 1
0.01
*DATABASE_GLSTAT
0.01
*DATABASE_NODFOR
0.01
2
*SET_NODE_LIST
2
1 7 13 19 25 31 37 48
54 60 66 72 78 84 91 97
103 109 115 121 127 206 212 218
224 230 236 242
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
$
*CONTROL_HOURGLASS
4
$
*MAT_elastic
$ MID RO E PR
1 7.890E-09 2.100E+05 3.000E-01
$
*SECTION_SHELL
$ SID ELFORM
1 2
3.000E-00 3.000E-00 3.000E-00 3.000E-00
$
*PART
SHELL
$ PID SID MID ADPOPT
1 1 1 0
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
*LOAD_NODE_POINT
$ NID DOF LCID SF
164 3 1 -1.0
$
*DEFINE_CURVE
1
0.00000000000000E+00 00.000000000000E+00
1.00000000000000E+00 500000.00000000E+00
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
$*LOAD_SEGMENT
$$ LCID SF N1 N2 N3 N4
$ 2 -1.0 154 155 164 163
$ 2 -1.0 155 156 165 164
$ 2 -1.0 163 164 173 172
$ 2 -1.0 164 165 174 173
$
$*DEFINE_CURVE
$ 2
$0.00000000000000E+00 00.000000000000E+00
$1.00000000000000E+00 7500.0000000000E+00
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*NODE
1 22.627417 -16.97056299999 0.0 7 7
*ELEMENT_SHELL
1 1 1 2 8 7
*END
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Problem #5: Ball Pressed Into Plate
___________________________________________________________________________________
Objectives
• Learn when implicit methods can run faster than explicit methods.
• Learn to troubleshoot contact and convergence problems.
Problem Description
A hemispherical ball of brick elements is displaced into a plate of brick elements. The plate is
supported around its edges.
Input Filename: hemi.k
Procedure
Copy the input file to your local directory. The simulation is initially set up to run explicitly. Start
the job and estimate how long it will take to finish. Stop the job using “<ctrl-c> stop” when your
patience runs thin.
1. How long will the explicit job take to run?__________________________________
Edit the input deck and activate the implicit method. Run the simulation, postprocess the results,
and examine the contact condition (HINT: cut a section plane normal to the Y-direction using
SPLANE buttons).
2. Why is the ball flattening rather than deforming the plate?
Deactivate the “shooting node logic” by setting SNLOG=1 on the Optional Card B of
*CONTACT_SURFACE_TO_SURFACE. Repeat the simulation.
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
3. How deeply does the ball penetrate the plate?
To eliminate penetration, use the penalty scale factor to increase the contact interface penalty
stiffness by 10x (SLSFAC=1.0 on *CONTROL_CONTACT). Repeat the simulation.
4. Is the penetration problem improved? ____________________________________
5. Why does the ball now jump around in the X-Y plane? _______________________
6. Does the problem run to completion? _____________________________________
Use *BOUNDARY_SPC_SET to add appropriate constraints to the ball (use the existing node set
ID #1.) Repeat the simulation.
7. How long does the simulation take to run using implicit method? _______________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
*TITLE
half ball impact problem
$
$ Units: m, s, kg, N
$
$ A. Tabiei
$
$ ENDTIM
1.00
$
$========1=========2=========3=========4=========5=========6=========7=========8
$
0 0.010 0 0 0
$
0 0 0 0.0 0.0 0 0
0 0 0 0
$
0 0 0
$
1 30 0 0.0 0.1
$
$ imass gamma beta
0 0.0 0.0
$
$========1=========2=========3=========4=========5=========6=========7=========8
$
*CONTROL_CONTACT
$ SLSFAC RWPNAL ISLCHK SHLTHK PENOPT THKCHG ORIEN
0.0000 0 1 0 0 0
$ USRSTR USRFAC NSBCS INTERM XPENEN
0 0 0 0 0.00
$
*CONTACT_SURFACE_TO_SURFACE
$ SSID MSID SSTYP MSTYP SBOXID MBOXID SPR MPR
1 2 3 3 0 0 0 0
$ FS FD DC V VDC PENCHK BT DT
0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.000E+00 0 0.000E+00 0.000E+00
$ SFS SFM SST MST SFST SFMT FSF VSF
0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.000E+00
$ optional card A
$ soft
0
$ optional card B
$ penmax thkopt shlthk snlog
0.0 0 0 0
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
5.0000E-02
$
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*DATABASE_RCFORC
1.0000E-06
$
*DATABASE_GLSTAT
1.0000E-06
$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
*MAT_ELASTIC
1 7.850E+3 210.0E+9 0.3
*MAT_ELASTIC
2 7.850E+3 210.0E+9 0.3
*PART
elastic plate
1 1 1
*PART
elastic ball
2 2 2
*SECTION_SOLID
1 0
*SECTION_SOLID
2 0
*CONTROL_HOURGLASS
5
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*BOUNDARY_PRESCRIBED_MOTION_SET
1 3 2 1 1.0
*DEFINE_CURVE
1
0.00000000E+00 0.00000000E+00
1.00000000E+00 5.00000000E-01
2.00000000E+00 5.00000000E-01
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*SET_NODE_LIST
1
515 518 521 524 527 530 533 536
539 560 563 566 569 572 575 590
593 596 599 602 605 631 633 635
637 667 669 671 673 691 693 695
697 704 707 710 713 716 719 745
747 749 751 781 783 785 787 805
807 809 811 817 819 821 823 829
831
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$$
$*BOUNDARY_SPC_SET
$$ NSID CID DOFX DOFY DOFZ DOFRX DOFRY DOFRZ
$ 1 0 1 1 0 0 0 0
$$
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*NODE
1-2.500000000E+00-2.500000000E+00 1.000000000E+00 7 7
*ELEMENT_SOLID
1 1 1 33 35 3 2 34 36 4
*END
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________

Problem #6: Truck Bumper
___________________________________________________________________________________
Objectives
• Learn the behavior of different contact interfaces in static implicit simulations.
• Learn how to set input parameters for implicit contact problems.
Problem Description
The rear bumper of a truck is modeled using shell elements. A solid, rigid bar is displaced into the
bumper, causing plastic buckling of the support. Post-buckling response is determined.
Input Filename: bumper.k
Procedure
1. What type of contact interface is used?_____________________________________
2. Is the thickness of the shell elements considered? ___________________________
Run the simulation and postprocess the results.
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
3. Why do the first few steps converge so easily? _____________________________
4. When does convergence trouble begin? ___________________________________
5. Why does the simulation stop? __________________________________________
Activate thickness offsets using *CONTROL_CONTACT, and repeat the simulation.

Convergence will be difficult, so temporarily increase the default displacement and energy
convergence tolerances to DCTOL=0.005 and ECTOL=0.05 using keyword
*CONTROL_IMPLICIT_NONLINEAR. Watch the values of the convergence norms displayed
on the screen.
6. Which norm is decreasing most slowly? ( ) displacement ( ) energy
Stop the job after about 10 steps using the interactive command “<ctrl-c> stop”. Postprocess the
results, and magnify the displacements 10x.
7. What is wrong with the contact interface? _________________________________
Switch to contact interface *CONTACT_AUTOMATIC_SURFACE_TO_SURFACE. The

AUTOMATIC… contact types are superior when gaps are opening. Don’t forget to activate the
consistency flag (CONST=1) on *CONTROL_PARALLEL, since this is currently required for
implicit mode. Repeat the simulation, stop after about 10 steps, and check the contact results
using the postprocessor.
8. Is the contact interface now separating properly? ___________________________
9. What is wrong? ______________________________________________________
Disable the “shooting node logic” using SNLOG=1 on the fifth contact card, “Optional Card B.”
Change the convergence tolerances back to their default values, and run the entire simulation
10. Does the force/displacement curve look reasonable? __________________________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
*TITLE
truck rear bumper
$
$ units: in,lbf,sec
$
0.0500000
$
$========1=========2=========3=========4=========5=========6=========7=========8
$
1 0.0001 0 0 0
$
0 0 0 0.0 0.0 0 0.0
$ 0 0 0 0.005 0.050 0 0.0
0 0 0 1
$ arcctl arcdir arclen arcmth arcdmp
0 0 0.0 0 0
$
0 0 2
$
1 100 0 0.0001 0.001
$
$ imass gamma beta
0 0.0 0.0
$
$ ias scale tstart tend
0 0.0 0 0
$
$========1=========2=========3=========4=========5=========6=========7=========8
$
$ displace bumper 5 inches
$
*BOUNDARY_PRESCRIBED_MOTION_RIGID
4 2 2 1 1.0000000 0 0.0000000 0.0000000
*DEFINE_CURVE
1 0 1.0000000 1.0000000 0.0000000 0.0000000
0.00000000E+00 0.00000000E+00
0.50000000E-01 5.00000000E+00
2.00000000E-01 5.00000000E+00
$
$========1=========2=========3=========4=========5=========6=========7=========8
$
*CONTROL_CONTACT
$ SLSFAC RWPNAL ISLCHK SHLTHK PENOPT THKCHG ORIEN
0 0 0 0 0 0 0
$ (second card blank)
$
*CONTROL_PARALLEL
$ NCPU NUMRHS CONST PARA
0 0 0 0
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
$
4 3 3 3
0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0 0.0000000 0.0000000
0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
$ optional card A (blank)
$ optional card B
$ PENMAX THKOPT SHLTHK SNLOG
0 0 0 0
$
$========1=========2=========3=========4=========5=========6=========7=========8
$
*DATABASE_GLSTAT
1e-4
*DATABASE_RCFORC
1e-4
1e-4
$
$========1=========2=========3=========4=========5=========6=========7=========8
$
$ material properties
$
*MAT_PLASTIC_KINEMATIC
$ 1020 steel bilinear material model
1 7.30000-4 3.00000+7 0.3000000 30000.000 1.25000+5 0.0000000
0.0000000 0.0000000 0.0000000
*MAT_RIGID
2 7.30000-4 3.00000+7 0.3000000 0.0000000 0.0000000 0.0000000
$ constrain all dof of frame
1.0000000 7.0000000 7.0000000
0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
*MAT_PLASTIC_KINEMATIC
$ 1020 steel bilinear material model
3 7.30000-4 3.00000+7 0.3000000 30000.000 1.25000+5 0.0000000
0.0000000 0.0000000 0.0000000
*MAT_RIGID
4 7.30000-4 3.00000+7 0.3000000 0.0000000 0.0000000 0.0000000
$ allow only y dof of impactor
1.0000000 6.0000000 7.0000000
0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
*MAT_POWER_LAW_PLASTICITY
$ alternate 1020 steel material model
5 7.30000-4 3.00000+7 0.3000000 91353 .1911
0 0
*SECTION_SHELL
1 6 0.0000000 5.0000000 0.0000000 0.0000000 0
0.2500000 0.2500000 0.2500000 0.2500000
*SECTION_SHELL
2 6 0.0000000 5.0000000 0.0000000 0.0000000 0
0.2500000 0.2500000 0.2500000 0.2500000
*SECTION_SHELL
3 6 0.0000000 5.0000000 0.0000000 0.0000000 0
0.1875000 0.1875000 0.1875000 0.1875000
*SECTION_SOLID
4
$*CONTROL_SHELL
$0,0,0,0,0,0,2
$
*PART
support
1 1 1
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*PART
frame
2 2 2
*PART
bumper
3 3 3
*PART
impactor
4 4 4
$
$ ---------------------------------------------------------------
$
*NODE
1 2.400000000E+01 0.000000000E+00 0.000000000E+00 0 0
*ELEMENT_SOLID
1 4 1813 1849 1855 1819 1814 1850 1856 1820
*ELEMENT_SHELL
1 1 1 13 14 2
*END
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Problem #7: Door Beam
___________________________________________________________________________________
Objective
• Learn to perform a springback simulation using an entirely static implicit analysis.
• Learn the benefits of Full Newton method for contact problems.
• Learn about the available displacement convergence norm options.
Problem Description
A doorbeam subassembly is deformed by a rigid pole. Shell elements are used throughout, and
nodal rigid bodies are used to spotweld the components of the doorbeam. The pole is displaced to
deform the doorbeam, then retracted to evaluate springback.
Input Filename: doorbeam.k
Procedure
1. What type of contact interface is selected? Why? ____________________________
2. Why is the doorbeam chosen as the slave side? _____________________________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Run the simulation and observe the convergence behavior.

Since this is a contact dominated problem, frequent stiffness reformations should help improve
convergence. Modify the nonlinear solution strategy to use more stiffness reformations and fewer
equilibrium iterations. Reform the stiffness matrix every five iterations, and allow twenty
reformations before giving up on the step. Also, relax the displacement convergence tolerance to
0.010.
Repeat the simulation. Postprocess the results and observe the force-deflection curve.
3. Why is the force-deflection behavior not smooth? ___________________________
Select the alternate displacement tolerance scheme using DNORM=1 on

*CONTROL_IMPLICIT_NONLINEAR, and repeat the simulation.
4. Does the alternative displacement tolerance become ( ) more or ( ) less strict when total
displacements are large, as they are at the end of this problem?
5. Is the force-deflection curve more smooth? Why? __________________________
6. What is the springback deflection at the center of the doorbeam? _______________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
*TITLE
static implicit door beam, 150 mm
$
1.10000
$
$ ---------------------------------------------------------------
$
1 0.010 0 0 1
$
0 0 0 0.000 0.00 0 0
$ dnorm divflag inistif nlprt
0 0 0 0
$
0 0 0
$
1 200 0 0.0 0.05
$
$ ---------------------------------------------------------------
$
0.0001000
$
*DATABASE_RCFORC
0.0001000
$
$ ---------------------------------------------------------------
$
2 7.83500-6 200.00000 0.2800000 0.2067000 0.0000000 5.00000+9 0.0000000
0.1700000 4.0000000 0 0
0.0000000 0.0230000 0.0940000 0.1380000 0.1510000 0.3010000 0.7010000 0.9010000
0.2070000 0.2580000 0.3330000 0.3610000 0.3910000 0.4390000 0.5060000 0.5280000
*MAT_RIGID
4 7.83500-6 200.00000 0.2800000 0.0000000 0.0000000 0.0000000
1.0000000 6.0000000 7.0000000
0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
*SECTION_SHELL
8 16 0.0000000 5.0000000 0.0000000 0.0000000 0
3.0000000 3.0000000 3.0000000 3.0000000 0.0000000
*SECTION_SHELL
4 16 0.0000000 5.0000000 0.0000000 0.0000000 0
3.0000000 3.0000000 3.0000000 3.0000000 0.0000000
*PART
PSHELL BBRKT
48 8 2 0 0 0 0 0
*PART
PSHELL BBRKT2
52 8 2 0 0 0 0 0
*PART
PSHELL DOORBEAM
56 8 2 0 0 0 0 0
*PART
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
4 4 4 0 0 0 0 0
$
$ ---------------------------------------------------------------
$
*BOUNDARY_PRESCRIBED_MOTION_RIGID
4 2 2 1 1.0000000 0 0.0000000 0.0000000
$
*DEFINE_CURVE
1
0.00000000E+00 0.00000000E+00
1.00000000E+00 1.50000000E+02
2.00000000E+00 0.00000000E+00
$
$ ---------------------------------------------------------------
$
*CONTROL_CONTACT
0.1000000 0.0000000 0 1 0 0 0
0 0 0 0 0.0000000 0 0 0
$
*CONTACT_NODES_TO_SURFACE
2 1 2 2 0 0 1 1
0.2000000 0.2000000 0.0000000 0.0000000 0.0000000 1 0.0000000 1.0100000
0.0000000 0.0000000 0.0000000 0.0000000 1.0000000 1.0000000
*SET_PART
1
4
*SET_PART
2
56
$
$ ---------------------------------------------------------------
$
*CONSTRAINED_NODAL_RIGID_BODY
3 0
*SET_NODE_LIST
3
77041 77045 77049 77230 77231 77232
4 0
*SET_NODE_LIST
4
77039 77043 77047 77204 77205 77207 77208 77213
5 0
*SET_NODE_LIST
5
76945 76946 76961 77261 77262 77263
6 0
*SET_NODE_LIST
6
76937 76940 76955 77243 77244 77245 77246 77247
$
$ ---------------------------------------------------------------
$
*NODE
1 3.745967041E+03-1.019032959E+03 5.000000000E+02 0 0
*ELEMENT_SHELL
1 4 1 6 7 2
*END
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Problem #8: Seamless Springback
___________________________________________________________________________________
Objective
• Learn to run a combined explicit/implicit simulation for dynamic analysis followed by static
springback.
Problem Description
A tensile strip of shell elements is slowly loaded using the explicit dynamic method. At the end of
the simulation, the analysis type is automatically switched to static implicit, the load is removed,
and a static unloaded solution is obtained.
Input Filename: springback.k
Procedure
1. How is load applied to the strip? _________________________________________
2. How is load released for springback? _____________________________________
Add the keywords *INTERFACE_SPRINGBACK_SEAMLESS and *SET_PART_LIST to

activate seamless springback. Run the simulation.
3. What is the termination time for the explicit solution? _______________________
4. How many implicit steps were taken during springback? _____________________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
5. What was the implicit time step size for springback? _________________________
6. How was this step size chosen? Does it’s value matter? _____________________
Postprocess the results and verify the springback solution analytically by extracting the following
information:
Elastic strain calculation:

7. Maximum axial stress in tensile test = _____________________________________
8. Young’s Modulus for this material = ______________________________________
9. Elastic strain = sigma/E = ______________________________________________
Springback displacement:
10. Specimen length after loading = _________________________________________
11. Elastic springback = strain * L = _________________________________________
Validation:
12. Computed displacement of tip node after loading ____________________________
13. Computed displacement of tip node after springback _________________________
14. Computed displacement due to springback _________________________________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
*TITLE
1D tensile test with seamless springback
$ This is a simple uniaxial tensile test with seamless springback.
$ Prescribed velocity is applied to the nodes at the end of the
$ strip, and removed using a death time of 0.005 to allow springback.
$ so springback (elastic) strain should be 2.98e-3
$ Length (loaded) = 44.38
$ so springback tip displacement should be 0.132 (at node 303)
$ units: mm, ton, N, N/mm^2
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ ENDTIM
0.005
*CONTROL_TIMESTEP
$ DTINIT SCFT ISDO TSLIMT DTMS
0.000 0.900 0 -3.0E-07
$ DT/CYCL
5.00E-04
*DATABASE_BINARY_RUNRSF
$ DT/CYCL
2000
*DATABASE_GLSTAT
1.0e-5
*DATABASE_ELOUT
1.0e-5
*DATABASE_NODOUT
1.0e-5
*DATABASE_HISTORY_SHELL
1,2,199,200
*DATABASE_history_node
303
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ NEIPH NEIPS MAXINT STRFLG SIGFLG EPSFLG RLTFLG ENGFLG
1 1
$ CMPFLG IEVERP BEAMIP DCOMP SHGE STSSZ
1 2
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ The following *CONTROL_IMPLICIT cards are not necessary, since
$ all of the default settings are used. They are included for
$ convenience in case you wish to try changing something.
$
$
$ imflag dt0 iefs
0 0.50 0 2
0 0 0 0.00 0.00 0 0
$ dnorm divflag inistif
0 0 0
0 0 0
0 0 0 0 0
$
$ ias scale tstart tend
0 0.0 0 0
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*INTERFACE_SPRINGBACK_SEAMLESS
982
6 7 0
1 6 0
303 5 0
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*SET_PART_LIST
982
1
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*CONTROL_ADAPTIVE
$ ADPFREQ ADPTOL ADPOPT MAXLVL TBIRTH TDEATH LCADP IOFLAG
0.100E-01 0 2 1 4.900E-03 1.0000000 0 0
$ ADPSIZE ADPASS IREFLG ADPENE
0.0000000 0 0 5.0000
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*PART
PART PID = 1
$ PID SID MID EOSID HGID GRAVTY ADOPTY
1 1 1 1
*SECTION_SHELL
$ SID ELFORM SHRF NIP
1 2 1.0 5.0
$ T1 T2 T3 T4
1.00 1.00 1.00 1.00
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*MAT_TRANSVERSELY_ANISOTROPIC_ELASTIC_PLASTIC
1 0.787E-08 0.207E+06 0.33 1.00 4
$ stress-strain curve for the sheet metal
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$--------LOAD CURVE DATA FOR STRAIN STRESS ------*
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*DEFINE_CURVE
$ LCID SIDR SCLA SCLO
4 0
$ A1 O1
0.0000E+00 0.3282E+03
0.1000E-01 0.4245E+03
0.3000E-01 0.5061E+03
0.5000E-01 0.5492E+03
0.7000E-01 0.5796E+03
0.9000E-01 0.6034E+03
0.1100E+00 0.6231E+03
0.1300E+00 0.6400E+03
0.1500E+00 0.6548E+03
0.1700E+00 0.6680E+03
0.1900E+00 0.6800E+03
0.2100E+00 0.6910E+03
0.2300E+00 0.7011E+03
0.2500E+00 0.7105E+03
0.2700E+00 0.7194E+03
0.2900E+00 0.7276E+03
0.9990E+01 0.7316E+02
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$-----------LOAD CURVE DATA FOR BINDER TRAVEL------*
$--------1---------2---------3---------4---------5---------6---------7---------8
*DEFINE_CURVE
$ LCID SIDR SCLA SCLO
1 0
$ A1 O1
0.0 0.0
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
1.0E-03 1000.0
5.0E-03 1000.0
$--------1---------2---------3---------4---------5---------6---------7---------8
*BOUNDARY_PRESCRIBED_MOTION_NODE
$ NID DOF VAD LCID SF VID DEATH
301 1 0 1 1.0 0.0050
302 1 0 1 1.0 0.0050
303 1 0 1 1.0 0.0050
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*NODE
$ NODE X Y Z TC RC
1 0.976562500E-03 0.100019989E+02 0.000000000E+00 6
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*ELEMENT_SHELL
$ EID PID N1 N2 N3 N4
1 1 1 2 3 4
*END
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Problem #9: Beam and GNL
Beam in bending
Objective:
• Get familiar with some of the syntax associated with implicit analysis.
• Learn to perform nonlinear geometrical analysis.
• Learn to the effect of NLSOLVR=1 or 2.
Introduction:
This workshop models a simple cantilever beam. The beam is 100 mm long and 2 mm by 2 mm
cross-section. The material is elastic with steel properties.
Input Filename: beam.k
Exercise:
Open the file beam.k with your favorite text editor. Understand the content of the file. Run the file
and post process. Generate the nodal force and displacement plots using Taurus. Record the end
displacement and the fixed end nodal forces. Run the problem with the given parameters below and
fill in the table (please note that the termination time is 1.0, however, the time in the *define curve
is 2.0).
Tip deflection Nodal force at fixed end

NLSOLVR=2, tip load=10 N
Exact solution for load=25 N
Q-1:
Why there is a difference between exact solution and LSDYNA/IMPLICIT?
Q-2:
Is equilibrium satisfied? What happens when NLSOLVR is equal to one?
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
*TITLE
GNL BEAM
$
$ This file is to demonstrate beam element and GNL
$ 100 mm long beam with 2x2 mm rectangular cross-section
$ A. Tabiei,
$
$ mm, s, N, ton
$
$ ENDTIM
1.000E+00
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*DATABASE_MATSUM
0.100E-05
*DATABASE_NODOUT
0.100E-05
*DATABASE_GLSTAT
0.100E-05
*DATABASE_NODFOR
0.100E-05
*DATABASE_BNDOUT
0.100E-05
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ DT/CYCL
0.100E-03
*DATABASE_HISTORY_NODE
$ id1
11
*DATABASE_nodal_force_group
5
*SET_NODE_LIST
5
1 11
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
1 0.01 0 0 0
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
2 0 0 0.000 0.00 0 0
$ dnorm divflag inistif nlprt
0 0 0 0
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*PART
BEAM
$ PID SID MID
3 4 2
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*MAT_ELASTIC
$ MID RO E PR
2 7.830E-09 2.070E+05 2.800E-01
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*SECTION_BEAM
$ SID ELFORM QR CST
4 1
$ TS1 TS2 TT1 TT2
2.0 2.0 2.0 2.0
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*LOAD_NODE_POINT
$ NODE DOF LCID SF
11 1 1 1.0
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*DEFINE_CURVE
1
0.00000000000000E+00 0.0000000000000E+00
2.00000000000000E+00 20.000000000000E+00
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*NODE
$ NODE X Y Z
1 0.500000000E+02 0.500000000E+02 0.000000000E+00 7 7
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*ELEMENT_BEAM
$ EID PID N1 N2 N3
1 3 1 2 35
*END
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Problem #10: Axial Buckling of Cylindrical Shell
Objective:
Get familiar with some of the syntax associated with implicit analysis.
Learn to perform stability analysis.
Learn the difference in an implicit simulation and explicit simulation for a quasi-static problem.
Introduction:
This workshop models an imperfect cylindrical shell. The initial geometric imperfections where
obtained from performing eigenvalue analysis and superimposing fractions of the eigenmodes on
the perfect shell. The material is elastic with Steel properties. Stability problems in general can be
modeled with implicit and some times with explicit. This example will show how to do that.
Input Filename: cylind.k
Exercise:
Open the file cylind.k and understand its content. The problem is set up to perform an implicit
analysis. Change the problem to an explicit analysis with least minimum change to the file (please
note that you do not need to remove the implicit analysis control cards). Run the problem and post
process. Look at the resultant nodal forces using the ASCII file NODFOR. Now change the file
back to perform implicit analysis. Run the file and compare the resultant nodal forces. Change the
termination time to 0.001 s and the analysis to explicit analysis. Run the file and compare the nodal
resultant forces to the two cases before.
Q-1: What do you think? Can we perform explicit analysis for this problem?
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
*TITLE
Post Buckling Analysis
$ Imperfect shell
$ Steel, thickness=2.0 mm, length=101 mm, diamter=202 mm.
$ By A. Tabiei, March 1, 1999
$ units: mm, s, ton, N
0.10000
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
0.01
*DATABASE_BINARY_D3THDT
0.01
*DATABASE_GLSTAT
1.0e-5
*DATABASE_ELOUT
1.0e-5
*DATABASE_NODFOR
1.0e-5
*DATABASE_NODOUT
1.0e-5
*DATABASE_SECFORC
1.0e-5
1
21 356 347
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*control_implicit_dynamics
$ IMASS GAMA BETA
1 0.5 0.25
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*control_implicit_general
$ IMFLAG DT0
1 0.01
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*control_implicit_nonlinear
$ NSOLVR ILIMIT MAXREF DCTOL ECTOL LSTOL
2 0.001 0.01 0.9
$ DNORM DIVERG ISTIF LPRINT
0 0 0 0
$ ARCCTL ARCDIR ARCLEN ARCMTH ARCDMP
$ 103 3
1 0 0 0.00 0
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*MAT_elastic
$ MID RO E PR
1 7.890E-09 2.100E+05 3.000E-01
*SECTION_SHELL
1 6
2.0000000 2.0000000 2.0000000 2.0000000
*PART
MATERIAL FOR PART : P1
1 1 1
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*SET_NODE_LIST
3
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
1 2 3 4 5 6 7 8
9 10 11 12 13 14 15 16
17 18 19 20 21 233 234 235
236 237 238 239 240 241 242 243
244 245 246 247 248 249 250 251
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*DATABASE_CROSS_SECTION_SET
1 0 0 2 0 0 0 0
*SET_NODES_LIST
1
211 212 213 214 215 216 217 218
219 220 221 222 223 224 225 226
227 228 229 230 231 443 444 445
446 447 448 449 450 451 452 453
454 455 456 457 458 459 460 461
*SET_SHELL_LIST
2
181 182 183 184 185 186 187 188
189 190 191 192 193 194 195 196
197 198 199 200 381 382 383 384
385 386 387 388 389 390 391 392
393 394 395 396 397 398 399 400
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*DEFINE_CURVE
1
0.00000000E+00 0.00000000E+00
0.10000000E+00 5.00000000E+00
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
3 1 2 1 -1.00
$*CONSTRAINED_NODAL_RIGID_BODY
$ 3
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*NODE
1 1.015973969E+02-8.854999542E+00-1.012129974E+02 5 0
*ELEMENT_SHELL
1 1 1 22 23 2
*END
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Problem #11: Implicit/Explicit Analysis

___________________________________________________________________________________
Objective:
• Learn to run a combined explicit/implicit simulation.
Problem Description:
A steel plate is modeled with solid elements. The plate is constrained on all edges and loaded by
uniform pressure. The uniform pressure loading is to be analyzed using an implicit step. Then the
loaded plate is to be impacted by half a sphere. The impact scenario is to be analyzed using an
explicit step.
Input Filename: im-ex-plate.k

im-ex-ball.k
Procedure:
Copy the input files to your local directory. Using an editor, view the input file and answer the
1. How is load applied to the plate? _________________________________________
2. What kind of element formulation is used for the solid element? ________________
Add the keywords *INTERFACE_SPRINGBACK_DYNA3D for the plate part (you do not need
to add constrains). This control card will force LSDYNA to generate an ascii file called “dynain” at
the end of the simulation. This file contains the internal stresses at the element integration points
and the deformed geometry at the termination time.
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Run the simulation.
3. What is the maximum effective stress (Von Misses) at the last state? ____________
4. Now examine the “dynain”. Can you explain what is in it? _____________________
Open the file im-ex-ball.k with your favorite text editor. This file contains both the nodal point for
the plate and half sphere. Need to cut the initial stresses and the new deformed nodal coordinate
from “dynain” file and put in this file (in-ex-ball.k).
5. What kind of load is applied in this file, and what is the load curve? _____________
6. What is the given initial velocity for the sphere? _____________________________
Comment the initial velocity card and run the simulation. Do an animation of the effect stress.
7. How is the effective stress change as a function of time, is it constant? ___________
Now you can apply the initial stress and run the problem. Post processes.
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
*TITLE
PLATE UNDER PRESSURE IMPLICIT
$ A. Tabiei
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ IMASS GAMA BETA
1 0.5 0.25
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ IMFLAG DT0
1 0.0001
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
2 0.001 0.01 0.9
2 2 1 1
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*control_implicit_auto
$ IAUTO ITEOPT ITEWIN DTMIN DTMAX
1 100 1.0E-6 1.0e-3
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*INTERFACE_SPRINGBACK_DYNA3D
123
*SET_PART_LIST
123
1
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ ENDTIM
1.000E+00
*DATABASE_GLSTAT
0.001E+00
$ DT/CYCL
0.100E+00
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*CONTROL_ACCURACY
1,1
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*PART
PART PID = 1
$ PID SID MID
1 1 1
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ MID RO E PR SIG ETAN FAIL
1 7.890E+03 2.100E+11 3.000E-01 3.150E+08 0.000E+00 0.000
$ C P LCSS
0.0 0 0
$ (EPS) effective plastic strain vs efftive stress (ES)
0.0 0.019 0.05 0.165 0.33 0.495 0.625 0.0
315E+06 315E+06 427.8E+06 500.8E+06 504.3E+06 506.5E+06 400E+06 0.0
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*SECTION_SOLID
$ SID ELFORM
1 2
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*NODE
1-2.500000000E+00-2.500000000E+00 1.000000000E+00 7 7
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*ELEMENT_SOLID
$ EID PID N1 N2 N3 N4 N5 N6 N7 N8
1 1 1 33 35 3 2 34 36 4
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*DEFINE_CURVE
1
0.00000000E+00 0.00000000E+00
1.00000000E+00 1.00000000E+08
2.00000000E+00 1.00000000E+08
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*LOAD_SEGMENT
1 1 33 35 3
*END
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
*TITLE
half ball impact problem using explicit/implicit formulation
$ 1.414 m radius half ball impacting plate 5x5x1 m (Steel).
$ A. Tabiei
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
0.01
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*CONTROL_ENERGY
$ HGEN RWEN SLNTEN RYLEN
2 1 2 1
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
1.0000E-02
*DATABASE_RCFORC
1.0000E-06
*DATABASE_GLSTAT
1.0000E-06
*DATABASE_NODOUT
0.100E-05
*DATABASE_ELOUT
0.100E-05
*DATABASE_HISTORY_NODE_SET
$ id1
10
*SET_NODE
10
240,242,272,274
*DATABASE_HISTORY_SOLID
$ id1
113
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ DT/CYCL
0.100E-02
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*CONTROL_ACCURACY
1,1
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
1 7.890E+03 2.100E+11 3.000E-01 3.150E+08 0.000E+00 0.000
$ C P LCSS
0.0 0 0
0.0 0.019 0.05 0.165 0.33 0.495 0.625 0.0
315E+06 315E+06 427.8E+06 500.8E+06 504.3E+06 506.5E+06 400E+06 0.0
2 7.890E+03 2.100E+11 3.000E-01 3.150E+08 0.000E+00 0.000
$ C P LCSS
0.0 0 0
0.0 0.019 0.05 0.165 0.33 0.495 0.625 0.0
315E+06 315E+06 427.8E+06 500.8E+06 504.3E+06 506.5E+06 400E+06 0.0
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*PART
elastic plate
1 1 1
*PART
elastic ball
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
2 2 2
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*SECTION_SOLID
1 2
*SECTION_SOLID
2 2
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*INITIAL_VELOCITY
1
0,0,250
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*SET_NODE_LIST
1
515 518 521 524 527 530 533 536
539 540 541 542 543 544 545 546
547 548 549 550 551 552 553 554
555 556 557 560 563 566 569 572
575 576 577 578 579 580 581 582
583 584 585 586 587 590 593 596
599 602 605 606 607 608 609 610
611 612 613 614 615 616 617 631
633 635 637 638 639 640 641 642
643 644 645 646 647 648 649 650
651 652 653 667 669 671 673 674
675 676 677 678 679 680 681 682
683 684 685 686 687 688 689 691
693 695 697 698 699 700 701 704
707 710 713 716 719 720 721 722
723 724 725 726 727 728 729 730
731 745 747 749 751 752 753 754
755 756 757 758 759 760 761 762
763 764 765 766 767 781 783 785
787 788 789 790 791 792 793 794
795 796 797 798 799 800 801 802
803 805 807 809 811 812 813 814
815 817 819 821 823 824 825 826
827 829 831 832 833
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*NODE
$ NODE X Y Z TC RC
515-0.707106829E+00-0.707106829E+00-0.460000396E+00
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*ELEMENT_SOLID
1 1 1 33 35 3 2 34 36 4
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*CONTACT_SURFACE_TO_SURFACE_TITLE
$ CID IF. NAME
1 IF000001
1 2 0 0 0 0 0 0
0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.000E+00 0 0.000E+00 0.000E+00
0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.100E+01 0.100E+01
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*SET_SEGMENT
1
515 540 542 518
540 541 543 542
518 542 544 521
542 543 545 544
521 544 576 560
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
544 545 577 576

560 576 578 563
576 577 579 578
515 590 606 540
540 606 607 541
590 599 612 606
606 612 613 607
599 704 720 612
612 720 721 613
704 713 726 720
720 726 727 721
541 607 647 543
543 647 649 545
545 649 683 577
577 683 685 579
607 613 651 647
647 651 653 649
649 653 687 683
683 687 689 685
613 721 761 651
651 761 763 653
653 763 797 687
687 797 799 689
721 727 765 761
761 765 767 763
763 767 801 797
797 801 803 799
563 578 699 693
578 579 685 699
693 699 701 697
699 685 689 701
697 701 813 807
701 689 799 813
807 813 815 811
813 799 803 815
713 821 826 726
726 826 765 727
821 823 827 826
826 827 767 765
823 831 833 827
827 833 801 767
831 811 815 833
833 815 803 801
*SET_SEGMENT
2
1 3 35 33
3 5 37 35
5 7 39 37
7 9 41 39
9 11 43 41
11 13 45 43
13 15 47 45
15 17 49 47
17 19 51 49
19 21 53 51
21 23 55 53
23 25 57 55
25 27 59 57
27 29 61 59
29 31 63 61
33 35 67 65
35 37 69 67
37 39 71 69
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
39 41 73 71
41 43 75 73
43 45 77 75
45 47 79 77
47 49 81 79
49 51 83 81
51 53 85 83
53 55 87 85
55 57 89 87
57 59 91 89
59 61 93 91
61 63 95 93
65 67 99 97
67 69 101 99
69 71 103 101
71 73 105 103
73 75 107 105
75 77 109 107
77 79 111 109
79 81 113 111
81 83 115 113
83 85 117 115
85 87 119 117
87 89 121 119
89 91 123 121
91 93 125 123
93 95 127 125
97 99 131 129
99 101 133 131
101 103 135 133
103 105 137 135
105 107 139 137
107 109 141 139
109 111 143 141
111 113 145 143
113 115 147 145
115 117 149 147
117 119 151 149
119 121 153 151
121 123 155 153
123 125 157 155
125 127 159 157
129 131 163 161
131 133 165 163
133 135 167 165
135 137 169 167
137 139 171 169
139 141 173 171
141 143 175 173
143 145 177 175
145 147 179 177
147 149 181 179
149 151 183 181
151 153 185 183
153 155 187 185
155 157 189 187
157 159 191 189
161 163 195 193
163 165 197 195
165 167 199 197
167 169 201 199
169 171 203 201
171 173 205 203
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
173 175 207 205

175 177 209 207
177 179 211 209
179 181 213 211
181 183 215 213
183 185 217 215
185 187 219 217
187 189 221 219
189 191 223 221
193 195 227 225
195 197 229 227
197 199 231 229
199 201 233 231
201 203 235 233
203 205 237 235
205 207 239 237
207 209 241 239
209 211 243 241
211 213 245 243
213 215 247 245
215 217 249 247
217 219 251 249
219 221 253 251
221 223 255 253
225 227 259 257
227 229 261 259
229 231 263 261
231 233 265 263
233 235 267 265
235 237 269 267
237 239 271 269
239 241 273 271
241 243 275 273
243 245 277 275
245 247 279 277
247 249 281 279
249 251 283 281
251 253 285 283
253 255 287 285
257 259 291 289
259 261 293 291
261 263 295 293
263 265 297 295
265 267 299 297
267 269 301 299
269 271 303 301
271 273 305 303
273 275 307 305
275 277 309 307
277 279 311 309
279 281 313 311
281 283 315 313
283 285 317 315
285 287 319 317
289 291 323 321
291 293 325 323
293 295 327 325
295 297 329 327
297 299 331 329
299 301 333 331
301 303 335 333
303 305 337 335
305 307 339 337
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
307 309 341 339

309 311 343 341
311 313 345 343
313 315 347 345
315 317 349 347
317 319 351 349
321 323 355 353
323 325 357 355
325 327 359 357
327 329 361 359
329 331 363 361
331 333 365 363
333 335 367 365
335 337 369 367
337 339 371 369
339 341 373 371
341 343 375 373
343 345 377 375
345 347 379 377
347 349 381 379
349 351 383 381
353 355 387 385
355 357 389 387
357 359 391 389
359 361 393 391
361 363 395 393
363 365 397 395
365 367 399 397
367 369 401 399
369 371 403 401
371 373 405 403
373 375 407 405
375 377 409 407
377 379 411 409
379 381 413 411
381 383 415 413
385 387 419 417
387 389 421 419
389 391 423 421
391 393 425 423
393 395 427 425
395 397 429 427
397 399 431 429
399 401 433 431
401 403 435 433
403 405 437 435
405 407 439 437
407 409 441 439
409 411 443 441
411 413 445 443
413 415 447 445
417 419 451 449
419 421 453 451
421 423 455 453
423 425 457 455
425 427 459 457
427 429 461 459
429 431 463 461
431 433 465 463
433 435 467 465
435 437 469 467
437 439 471 469
439 441 473 471
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
441 443 475 473

443 445 477 475
445 447 479 477
449 451 483 481
451 453 485 483
453 455 487 485
455 457 489 487
457 459 491 489
459 461 493 491
461 463 495 493
463 465 497 495
465 467 499 497
467 469 501 499
469 471 503 501
471 473 505 503
473 475 507 505
475 477 509 507
477 479 511 509
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*NODE
1-0.250000000E+01-0.250000000E+01 0.100000000E+01 7 7
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*INITIAL_STRESS_SOLID
1 8
1.264E+07 1.264E+07 1.545E+08-1.433E+08-5.396E+07-5.396E+07 1.085E-04
9.326E+07-5.625E+07 1.428E+08-1.291E+08-7.521E+07-4.939E+04 2.427E-03
1.194E+07 1.194E+07 1.559E+08-1.617E+08 3.029E+05 3.029E+05 3.449E-03
-5.625E+07 9.326E+07 1.428E+08-1.291E+08-4.939E+04-7.521E+07 2.427E-03
1.083E+08 1.083E+08-3.672E+07 1.437E+08-5.201E+07-5.201E+07 3.402E-04
3.096E+07 1.755E+08-2.662E+07 1.255E+08-8.016E+07-5.402E+06 3.038E-03
1.156E+08 1.156E+08-5.140E+07 1.519E+08-1.755E+07-1.755E+07 4.352E-03
1.755E+08 3.096E+07-2.662E+07 1.255E+08-5.402E+06-8.016E+07 3.038E-03
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*DEFINE_CURVE
1
0.00000000E+00 1.00000000E+08
2.00000000E+00 1.00000000E+08
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*LOAD_SEGMENT
1 1 33 35 3
*END
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Problem #12: Bolt Analysis

___________________________________________________________________________________
Objective:
• Learn to rigid body, prescribed motion, and equilibrium check.
A steel beam section is bolted as shown. The bolt in modeled by both rigid and deformable
material. The bolt is given prescribed displacement so that the bolt is pulled out of the slotted hole.
Input Filename: bolt.k
Procedure:
following question:
1. What is the allowed direction for motion of the rigid material (*mat_rigid)? _______
(hint: look at the third card of *mat_rigid).
2. Put the *boundary_prescribed_motion_rigid with the approperiate parameters. Use the defined
curve for the motion.
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Run the simulation and look at the contact forces.
3. Switch the material of the bolt to deformable. You can do that by changing the material ID of
the deformable material. Put the *boundary_prescribed_motion_set with the appropriate
parameters. Use the defined curve and the defined node set for the motion (do not forget to
remove the *boundary_prescribed_motion_rigid from the file).
Run the simulation and look at the contact forces.
4. Plot the x-component of contact force. Plot sum of the nodal internal forces of the nodal set
specified (hint: look at the NODFOR).
5. Is equilibrium satisfied? ________________________________________________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
*TITLE
Bolt Analysis By A. Tabiei
0.0100000
*CONTROL_CONTACT
$ SLSFAC RWPNAL ISLCHK SHLTHK
0.100 2
$ USRSTR USRFAC NSBCS INTERM
*CONTROL_ENERGY
2
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ IMASS GAMA BETA
0 0.5 0.25
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ IMFLAG DT0
1 0.00001
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
2 20 5 0.001 0.01 0.9
2 2 1 1
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
1 100 1.0E-7 1.0e-4
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
1.00000-3
*DATABASE_NODFOR
1.00000-5
*DATABASE_SLEOUT
1.00000-3
*DATABASE_RCFORC
1.00000-5
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
1 7.83000-9 2.00000+5 0.3300000 345.00000 0.0000000 0.0000000 0.0000000
0.0000000 0.0000000 0 0
0.0000000 0.0200000 0.0300000 0.1500000 0.3000000 0.4000000 10.000000 0.0000000
345.00000 515.00000 600.00000 690.00000 740.00000 760.00000 700.00000 0.0000000
*MAT_RIGID
22 7.85000-9 2.10000+5 0.3000000 0.0000000 0.0000000 0.0000000
1.0000000 6.0000000 7.0000000
0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
2 7.83000-9 2.20000+5 0.3300000 445.00000 0.0000000 0.0000000 0.0000000
0.0000000 0.0000000 0 0
0.0000000 0.0200000 0.0300000 0.1500000 0.3000000 0.4000000 10.000000 0.0000000
445.00000 615.00000 700.00000 790.00000 840.00000 860.00000 800.00000 0.0000000
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*SECTION_SHELL
1 16 0.0000000 0.0000000 0.0000000 0.0000000 0
2.0000000 2.0000000 2.0000000 2.0000000 0.0000000
*SECTION_SHELL
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
2 16 0.0000000 0.0000000 0.0000000 0.0000000 0

2.0000000 2.0000000 2.0000000 2.0000000 0.0000000
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*PART
Mat.-Type: 24
27 1 1
*PART
Mat.-Type: 20 ** Rigid **
28 2 2
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*DEFINE_CURVE
1 0
0.00000000E+00 0.00000000E+00
0.01000000E+00 1.50000000E+01
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$*BOUNDARY_PRESCRIBED_MOTION_rigid
$ 28 2 2 1 1.0000000 0 0.0000000 0.0000000
1 2 2 1 1.0000000 0 0.0000000 0.0000000
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
1
*SET_NODE_LIST
1
810 811 812 813 814 815 816 817
818 819 820 821 822 823 824 825
826 827 828
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*NODE
$ NODE X Y Z TC RC
1 0.406250000E+02-0.762940005E-07 0.240000000E+02
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*ELEMENT_SHELL
1 27 220 278 372 280
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
2 3 0 0 0 0 0 0
0.2200000 0.2000000 0.0000000 0.0000000 0.0000000 0 0.0000000 0.0000000
1.0000000 1.0000000 0.0000000 0.0000000 0.0000000 0.0000000
$ optional card A
$ soft
0
$ optional card B
0.0 0 0 1
*SET_SEGMENT
$ Slave segments, surface: 1
$
2
220 278 372 280
280 372 284 221
372 279 224 284
278 219 279 372
221 284 373 283
283 373 78 14
373 288 15 78
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
284 224 288 373

228 282 378 292
292 378 285 229
378 280 221 285
282 220 280 378
229 285 379 295
295 379 79 22
379 283 14 79
285 221 283 379
12 72 164 70
72 13 76 164
164 76 16 71
70 164 71 11
13 75 165 76
75 14 78 165
165 78 15 80
76 165 80 16
20 84 170 74
84 21 77 170
170 77 13 72
74 170 72 12
21 87 171 77
87 22 79 171
171 79 14 75
77 171 75 13
625 674 766 672
674 626 678 766
766 678 627 673
672 766 673 624
626 677 767 678
677 429 493 767
767 493 430 682
678 767 682 627
629 686 772 676
686 630 679 772
772 679 626 674
676 772 674 625
630 689 773 679
689 437 494 773
773 494 429 677
679 773 677 626
427 485 577 487
487 577 491 428
577 486 431 491
485 426 486 577
428 491 578 490
490 578 493 429
578 495 430 493
491 431 495 578
435 489 583 499
499 583 492 436
583 487 428 492
489 427 487 583
436 492 584 502
502 584 494 437
584 490 429 494
492 428 490 584
*SET_SEGMENT
$ Master segments, surface: 1
$
3
297 83 424 432
291 297 432 434
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
287 291 434 448

286 287 448 449
264 286 449 475
263 264 475 477
262 263 477 478
260 262 478 498
233 260 498 525
230 233 525 303
223 230 303 354
222 223 354 355
218 222 355 356
418 424 83 207
215 218 356 357
214 215 357 358
213 214 358 360
212 213 360 362
211 212 362 364
364 418 207 211
355 354 662 664
303 658 662 354
303 525 758 658
525 498 756 758
498 478 754 756
478 477 752 754
477 475 751 752
448 434 748 749
434 432 714 748
432 424 713 714
424 418 712 713
475 449 750 751
713 712 819 809
418 364 697 712
364 362 691 697
362 360 685 691
360 358 681 685
358 357 680 681
357 356 665 680
356 355 664 665
712 697 818 819
697 691 817 818
691 685 816 817
685 681 815 816
681 680 814 815
680 665 813 814
665 664 812 813
664 662 811 812
662 658 810 811
809 820 714 713
820 821 748 714
821 822 749 748
822 823 750 749
823 824 751 750
824 825 752 751
825 826 754 752
826 827 756 754
827 828 758 756
828 810 658 758
449 448 749 750
*END
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Problem #13: Crash Analysis

___________________________________________________________________________________
Objective:
• Learn to difference between explicit and dynamic implicit.
A steel front rail is constrained on one end and displacement control is applied on the other end.
The analysis to be run using explicit and implicit formulation.
Input Filename: crash-implicit.k
Procedure:
following question:
1. There are two boundary conditions specified.

What are these boundary conditions ? ________________________________________
2. How is motion prescribed and what out put is requested for post processing? ______
Run the simulation and post processes.
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
3. Plot reaction forces.
Change the simulation to an explicit simulation and run it.
4. Plot reaction forces. Is there a difference in the forces between the two runs. _______
5. Comment on the simulations. ____________________________________________
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
*TITLE
Symmetric Short Crush Tube
$
$ Symmetric model - 1/4 of the tube
$
$ Units: mm, kg, ms, kN, GPa, kN-mm
$
$
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$
$$$$ Control Ouput
$
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$
$
$...>....1....>....2....>....3....>....4....>....5....>....6....>....7....>....8
$ f i f f f
$ endtim endcyc dtmin endneg endmas
1.00 0 0.0 0.0 0.0
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ IMASS GAMA BETA
1 0.5 0.25
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ IMFLAG DT0
1 0.0001
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
2 20 5 0.005 0.01 0.9
2 2 1 1
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
1 10 1.0E-6 1.0e-4
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$
*SET_NODE_LIST
$ SID DA1 DA2 DA3 DA4
1
$ NID1 NID2 NID3 NID4 NID5 NID6 NID7 NID8
99999 253 254 255 256 257 258 259
486 487 712 713 714 715 716
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*SET_NODE_LIST
2
1 2 3 4 5 6 7 414
415 532 533 534 535 536
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*BOUNDARY_SPC_SET
1,0,1,1,0,0,0,0
1,3,2,99,-1.0
*DEFINE_CURVE
99
0.0,0.0
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
1.00,50.0
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*CONTROL_ENERGY
$ i i i i
$ hgen rwen slnten rylen
2 2
$
$ f i
$ DT/CYCL LCDT - time interval between state dumps (D3PLOT)
0.1
$
$ i i i i i i i i
$ i i i
1
$
*DATABASE_BINARY_D3THDT
$ DT/CYCL LCDT - time interval between time history data dumps (D3THDT)
999999
$
2
*DATABASE_NODFOR
$ DT/CYCL
0.0001
$
$ Define nodes that output into NODOUT
$...>....1....>....2....>....3....>....4....>....5....>....6....>....7....>....8
$ id1 id2 id4 id4 id5 id6 id7 id8
99999 414 486
$
*DATABASE_NODOUT
$ DT/CYCL
0.001
*DATABASE_GLSTAT
$ DT/CYCL
0.001
*DATABASE_MATSUM
$ DT/CYCL
0.001
*DATABASE_SLEOUT
$ DT/CYCL
0.001
$
$
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$
$$$$ Define Contacts - Sliding Interfaces
$
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$
$
*CONTACT_AUTOMATIC_SINGLE_SURFACE
$...>....1....>....2....>....3....>....4....>....5....>....6....>....7....>....8
$ ssid msid sstyp mstyp sboxid mboxid spr mpr
0
$ Equating ssid to zero means that all segments are included in the contact
$
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
$ fs fd dc vc vdc penchk bt dt
0.08 0.08
$
$ sfs sfm sst mst sfst sfmt fsf vsf
$
$
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$
$$$$ Define Materials and Parts
$
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$
$
$...>....1....>....2....>....3....>....4....>....5....>....6....>....7....>....8
$ mid ro e pr sigy etan eppf tdel
1 7.830E-06 200.0 0.3 0.207 10.0
$
$ c p lcss lcsr
0 0
$ PLASTIC STRESS/STRAIN CURVES
$ eps1 eps2 eps3 eps4 eps5 eps6 eps7 eps8
$ es1 es2 es3 es4 es5 es6 es7 es8
0.000E+00 8.000E-02 1.600E-01 4.000E-01 9.900E+01
2.070E-01 2.500E-01 2.750E-01 2.899E-01 3.000E-01
$
*SECTION_SHELL
$ SHELL ELEMENT CROSS-SECTIONAL PROPERTIES
$...>....1....>....2....>....3....>....4....>....5....>....6....>....7....>....8
$ sid elform shrf nip propt qr/irid icomp
1 16 3.0000
$ t1 t2 t3 t4 nloc
2.000E+00 2.000E+00 2.000E+00 2.000E+00
$
*PART
$ HEADING
corner1
$...>....1....>....2....>....3....>....4....>....5....>....6....>....7....>....8
$ pid sid mid eosid hgid igrav adpopt
1 1 1
$
$
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$
$$$$ Define Nodes and Elements
$
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$
$
$ NODAL POINT CARDS
*NODE
$
$...>....1....>....2....>....3....>....4....>....5....>....6....>....7....>....8
$ NID X Y Z TC RC
99999 0.0 0.0 274.0 0 0
*ELEMENT_SHELL
$...>....1....>....2....>....3....>....4....>....5....>....6....>....7....>....8
752 1 547 552 553 553
*BOUNDARY_SPC_NODE
$...>....1....>....2....>....3....>....4....>....5....>....6....>....7....>....8
$ NID CID X Y Z RX RY RZ
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
1, 0,1,1,1, 1, 1, 1
2, 0,1,1,1, 1, 1, 1
3, 0,1,1,1, 1, 1, 1
4, 0,1,1,1, 1, 1, 1
5, 0,1,1,1, 1, 1, 1
6, 0,1,1,1, 1, 1, 1
7, 0,1,1,1, 1, 1, 1
414, 0,1,1,1, 1, 1, 1
415, 0,1,1,1, 1, 1, 1
532, 0,1,1,1, 1, 1, 1
533, 0,1,1,1, 1, 1, 1
534, 0,1,1,1, 1, 1, 1
535, 0,1,1,1, 1, 1, 1
536, 0,1,1,1, 1, 1, 1
537, 0,0,1,0, 1, 0, 1
542, 0,0,1,0, 1, 0, 1
547, 0,0,1,0, 1, 0, 1
552, 0,0,1,0, 1, 0, 1
557, 0,0,1,0, 1, 0, 1
562, 0,0,1,0, 1, 0, 1
567, 0,0,1,0, 1, 0, 1
572, 0,0,1,0, 1, 0, 1
577, 0,0,1,0, 1, 0, 1
582, 0,0,1,0, 1, 0, 1
587, 0,0,1,0, 1, 0, 1
592, 0,0,1,0, 1, 0, 1
597, 0,0,1,0, 1, 0, 1
602, 0,0,1,0, 1, 0, 1
607, 0,0,1,0, 1, 0, 1
612, 0,0,1,0, 1, 0, 1
617, 0,0,1,0, 1, 0, 1
622, 0,0,1,0, 1, 0, 1
627, 0,0,1,0, 1, 0, 1
632, 0,0,1,0, 1, 0, 1
637, 0,0,1,0, 1, 0, 1
642, 0,0,1,0, 1, 0, 1
647, 0,0,1,0, 1, 0, 1
652, 0,0,1,0, 1, 0, 1
657, 0,0,1,0, 1, 0, 1
662, 0,0,1,0, 1, 0, 1
667, 0,0,1,0, 1, 0, 1
672, 0,0,1,0, 1, 0, 1
677, 0,0,1,0, 1, 0, 1
682, 0,0,1,0, 1, 0, 1
687, 0,0,1,0, 1, 0, 1
692, 0,0,1,0, 1, 0, 1
697, 0,0,1,0, 1, 0, 1
702, 0,0,1,0, 1, 0, 1
707, 0,0,1,0, 1, 0, 1
712, 0,0,1,0, 1, 0, 1
14, 0,1,0,0, 0, 1, 1
21, 0,1,0,0, 0, 1, 1
28, 0,1,0,0, 0, 1, 1
35, 0,1,0,0, 0, 1, 1
42, 0,1,0,0, 0, 1, 1
49, 0,1,0,0, 0, 1, 1
56, 0,1,0,0, 0, 1, 1
63, 0,1,0,0, 0, 1, 1
70, 0,1,0,0, 0, 1, 1
77, 0,1,0,0, 0, 1, 1
84, 0,1,0,0, 0, 1, 1
91, 0,1,0,0, 0, 1, 1
98, 0,1,0,0, 0, 1, 1
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
105, 0,1,0,0, 0, 1, 1
112, 0,1,0,0, 0, 1, 1
119, 0,1,0,0, 0, 1, 1
126, 0,1,0,0, 0, 1, 1
133, 0,1,0,0, 0, 1, 1
140, 0,1,0,0, 0, 1, 1
147, 0,1,0,0, 0, 1, 1
154, 0,1,0,0, 0, 1, 1
161, 0,1,0,0, 0, 1, 1
168, 0,1,0,0, 0, 1, 1
175, 0,1,0,0, 0, 1, 1
182, 0,1,0,0, 0, 1, 1
189, 0,1,0,0, 0, 1, 1
196, 0,1,0,0, 0, 1, 1
203, 0,1,0,0, 0, 1, 1
210, 0,1,0,0, 0, 1, 1
217, 0,1,0,0, 0, 1, 1
224, 0,1,0,0, 0, 1, 1
231, 0,1,0,0, 0, 1, 1
238, 0,1,0,0, 0, 1, 1
245, 0,1,0,0, 0, 1, 1
252, 0,1,0,0, 0, 1, 1
259, 0,1,0,0, 0, 1, 1
$
$
*END
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
Problem #14: Bumper Analysis

___________________________________________________________________________________
Objective:
• Learn to rigid body, prescribed motion, difference in implicit static and dynamic.
A steel beam section is bolted as shown. The bolt in modeled by both rigid and deformable
material. The bolt is given prescribed displacement so that the bolt is pulled out of the slotted hole.
Input Filename: bumper.k
Procedure:
1. What materials are specified, what is constrained. ____________________________
2. How is motion prescribed. ______________________________________________
Run these simulation with static and dynamic implicit and compare contact forces.
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
*KEYWORD
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*TITLE
BARRIER-BUMPER IMPACT
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ ENDTIM
1.000E+00
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ IMASS GAMA BETA
0 0.5 0.25
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ IMFLAG DT0
1 0.0001
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
2 20 5 0.005 0.01 0.9
2 2 1 1
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
1 100 1.0E-7 1.0e-3
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*CONTROL_CONTACT
$ SLSFAC RWPNAL ISLCHK SHLTHK THKCHG ORIEN
0.100 2 2
$
*CONTROL_ENERGY
$ HGEN RWEN SLNTEN RYLEN
2 2 1 1
*DATABASE_GLSTAT
.100E-02
*DATABASE_RCFORC
.100E-02
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ DT/CYCL
.100E-00
0 0 3 0 1 1 1 1
0 0 0 0 0 0
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*PART
$HEADING
PART PID = 1 PART NAME :BUMP.S
$ PID SID MID
1 1 2
*PART
$HEADING
PART PID = 2 PART NAME :RAILS.H
$ PID SID MID
2 1 2
*PART
$HEADING
PART PID = 3 PART NAME :BRACKS.H
$ PID SID MID
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
3 3 2
*PART
$HEADING
PART PID = 4 PART NAME :WALL.LIN
$ PID SID MID
4 1 1
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
$ MID RO E PR SIGY ETAN EPPF TDEL
2 7.830E-09 2.070E+05 2.800E-01 2.100E+02 0.000E+00 0.000E+00 0.000E+00
$ C P LCSS LCSR
.000E+00 .000E+00 .000E+00 .000E+00
$ EPS1 EPS2 EPS3 EPS4 EPS5 EPS6 EPS7 EPS8
0.000E+00 3.090E-02 4.090E-02 5.000E-02 1.510E-01 3.010E-01 7.010E-01 9.100E-01
$ ES1 ES2 ES3 ES4 ES5 ES6 ES7 ES8
2.100E+02 3.000E+02 3.140E+02 3.250E+02 3.900E+02 4.380E+02 5.050E+02 5.270E+02
*MAT_RIGID
$ MID RO E PR N COUPLE M ALIAS
1 7.830E-09 2.070E+05 2.800E-01 0.000E+00 0.000E+00 0.000E+00
$ CMO CON1 CON2
1.0 7.0 7.0
$LCO or A1 A2 A3 V1 V2 V3
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*SECTION_SHELL
$ SID ELFORM SHRF NIP PROPT QR/IRID ICOMP
1 16 .100E+01 3.0 1.0 .0
$ T1 T2 T3 T4 NLOC
1.000E+00 1.000E+00 1.000E+00 1.000E+00
*SECTION_SHELL
$ SID ELFORM SHRF NIP PROPT QR/IRID ICOMP
3 16 .100E+01 3.0 1.0 .0
$ T1 T2 T3 T4 NLOC
1.000E+00 1.000E+00 1.000E+00 1.000E+00
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*SET_NODE_LIST
$ SID
1
253 254 255 267 289 290 294 457
458 472 490 500
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*BOUNDARY_SPC_SET
1,0,1,0,1,0,0,0
1,2,2,99,-1.0
*DEFINE_CURVE
99
0.0,0.0
1.00,50.0
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*NODE
$ NODE X Y Z TC RC
1-0.500000000E+03-0.152500000E+03 0.313752014E+03
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*ELEMENT_SHELL
1 4 1 10 11 2
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
1 2 0 0 0 0
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
.100E+00 .000E+00 .000E+00 .000E+00 .000E+00 0 .000E+00 .000E+00

.000E+00 .000E+00 .000E+00 .000E+00 .000E+00 .000E+00 .100E+01 .100E+01
$ optional card A
$ soft
0
$ optional card B
0.0 0 0 1
*SET_SEGMENT
$ SID DA1 DA2 DA3
1
$ N1 N2 N3 N4
100 101 344 115
101 102 345 344
102 103 346 345
103 104 347 346
104 105 348 347
105 106 107 348
115 344 113 114
344 345 112 113
345 346 111 112
346 347 110 111
347 348 109 110
348 107 108 109
116 117 101 100
117 118 102 101
118 119 103 102
119 120 104 103
120 121 105 104
121 122 106 105
123 124 135 136
124 125 134 135
125 126 133 134
126 127 132 133
127 128 131 132
128 129 130 131
114 113 142 143
113 112 141 142
112 111 140 141
111 110 139 140
110 109 138 139
109 108 137 138
136 135 349 152
135 134 350 349
134 133 351 350
133 132 352 351
132 131 353 352
131 130 144 353
152 349 150 151
349 350 149 150
350 351 148 149
351 352 147 148
352 353 146 147
353 144 145 146
151 150 158 159
150 149 157 158
149 148 156 157
148 147 155 156
147 146 154 155
146 145 153 154
143 142 354 161
142 141 355 354
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
141 140 356 355

140 139 357 356
139 138 358 357
138 137 160 358
161 354 124 123
354 355 125 124
355 356 126 125
356 357 127 126
357 358 128 127
358 160 129 128
106 162 359 107
162 163 164 359
107 359 166 108
359 164 165 166
129 167 170 130
167 168 169 170
145 144 360 173
144 130 170 360
173 360 171 172
360 170 169 171
172 174 175 173
173 175 153 145
176 177 166 165
177 137 108 166
168 167 361 178
167 129 160 361
178 361 177 176
361 160 137 177
122 179 162 106
179 180 163 162
181 182 362 185
182 172 171 362
185 362 183 184
362 171 169 183
181 186 187 182
182 187 174 172
188 189 190 191
189 163 180 190
192 193 363 194
193 165 164 363
194 363 189 188
363 164 163 189
195 196 193 192
196 176 165 193
184 183 197 198
183 169 168 197
168 178 364 197
178 176 196 364
197 364 199 198
364 196 195 199
181 200 209 186
200 201 208 209
201 202 207 208
202 203 206 207
203 204 205 206
192 210 219 195
210 211 218 219
211 212 217 218
212 213 216 217
213 214 215 216
188 220 365 194
220 221 366 365
221 222 367 366
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
222 223 368 367

223 224 225 368
194 365 210 192
365 366 211 210
366 367 212 211
367 368 213 212
368 225 214 213
184 226 369 185
226 227 370 369
227 228 371 370
228 229 372 371
229 230 231 372
185 369 200 181
369 370 201 200
370 371 202 201
371 372 203 202
372 231 204 203
224 223 235 236
223 222 234 235
222 221 233 234
221 220 232 233
220 188 191 232
195 219 373 199
219 218 374 373
218 217 375 374
217 216 376 375
216 215 237 376
199 373 242 198
373 374 241 242
374 375 240 241
375 376 239 240
376 237 238 239
198 242 226 184
242 241 227 226
241 240 228 227
240 239 229 228
239 238 230 229
304 305 408 308
305 204 231 408
308 408 306 307
408 231 230 306
309 310 305 304
310 205 204 305
311 312 313 314
312 224 236 313
315 316 409 317
316 214 225 409
317 409 312 311
409 225 224 312
238 237 407 300
237 215 301 407
300 407 303 299
407 301 302 303
302 301 316 315
301 215 214 316
307 306 300 299
306 230 238 300
304 318 321 309
318 319 320 321
315 322 325 302
322 323 324 325
315 317 411 322
317 311 326 411
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
322 411 328 323

411 326 327 328
329 330 412 331
330 299 303 412
331 412 325 324
412 303 302 325
327 326 332 333
326 311 314 332
334 335 330 329
335 307 299 330
304 308 413 318
308 307 335 413
318 413 336 319
413 335 334 336
320 319 336 338
338 336 334 337
339 323 328 340
340 328 327 333
341 324 323 339
342 329 331 343
343 331 324 341
337 334 329 342
*SET_SEGMENT
2
$ N1 N2 N3 N4 A1 A2 A3 A4
1 2 11 10
2 3 12 11
3 4 13 12
4 5 14 13
5 6 15 14
6 7 16 15
7 8 17 16
8 9 18 17
10 11 20 19
11 12 21 20
12 13 22 21
13 14 23 22
14 15 24 23
15 16 25 24
16 17 26 25
17 18 27 26
19 20 29 28
20 21 30 29
21 22 31 30
22 23 32 31
23 24 33 32
24 25 34 33
25 26 35 34
26 27 36 35
28 29 38 37
29 30 39 38
30 31 40 39
31 32 41 40
32 33 42 41
33 34 43 42
34 35 44 43
35 36 45 44
37 38 47 46
38 39 48 47
39 40 49 48
40 41 50 49
41 42 51 50
_______________________________________________________________________________________________
WORKSHOP
_______________________________________________________________________________________________
42 43 52 51
43 44 53 52
44 45 54 53
46 47 56 55
47 48 57 56
48 49 58 57
49 50 59 58
50 51 60 59
51 52 61 60
52 53 62 61
53 54 63 62
55 56 65 64
56 57 66 65
57 58 67 66
58 59 68 67
59 60 69 68
60 61 70 69
61 62 71 70
62 63 72 71
64 65 74 73
65 66 75 74
66 67 76 75
67 68 77 76
68 69 78 77
69 70 79 78
70 71 80 79
71 72 81 80
73 74 83 82
74 75 84 83
75 76 85 84
76 77 86 85
77 78 87 86
78 79 88 87
79 80 89 88
80 81 90 89
82 83 92 91
83 84 93 92
84 85 94 93
85 86 95 94
86 87 96 95
87 88 97 96
88 89 98 97
89 90 99 98
$*CONTROL_PARALLEL
$1,0,1
$*CONTACT_AUTOMATIC_SINGLE_SURFACE
$ CID CONTACT INTERFACE TITLE
$ 2 IF02
$ 1 2 3 3 0 0
$ .100E+00 .000E+00 .000E+00 .000E+00 .000E+00 0 .000E+00 .000E+00
$ .000E+00 .000E+00 .000E+00 .000E+00 .000E+00 .000E+00 .100E+01 .100E+01
$ optional card A
$ soft
$ 0
$ optional card B
$ 0.0 0 0 1
$---+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
*END
_______________________________________________________________________________________________

ImplicitNotes

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ImplicitNotes

Uploaded by

Copyright:

Available Formats

LS-DYNA IMPLICIT NOTES

Livermore Software Technology Corporation

Ala (Al) Tabiei, Ph.D.

7374 Las Positas Road

Tel: (925) 449-2500

Finite Element Equation & Time Integration 2-13

Equilibrium and Nonlinearity 3-21

Incremental Equations and Linearization 4-25

Finite Element Modeling Techniques 5-49

Contact and Friction 6-77

Material Nonlinearity 7-88

Reduce Integration & Hourglass Phenomenon 9-96

Quasi-Static Initialization and Quasi-Static Simulations 10-100

Adaptive Meshing 11-104

Trouble Shooting 13-112

LS-DYNA OFFERS NEW IMPLICIT ANALYSIS CAPABILITY

keyword input and arbitrary ID numbering are available in implicit mode

FEATURES AVAILABLE FOR IMPLICIT MODE

LS-DYNA FEATURES CURRENTLY AVAILABLE FOR IMPLICIT ANALYSIS

solid element formulations 1, 2

beam element formulation 1

shell element formulations 2,6,10,16

2-D solid element formulations 12,13,15

contact interface types 3, m3, 5, m5, 10, m10, 13:

ACTIVATING IMPLICIT MODE

Activating Implicit Analysis

Three types of analyses can be performed

New keywords have been added

● not all features are available in implicit mode

*CONTROL_IMPLICIT_GENERAL (required for implicit)

*CONTROL_IMPLICIT_STABILIZATION (optional, metal forming only)

User's Manual contains helpful notes on each input parameter

A Note on Implicit Numerical Methods

Implicit governing equations contain two problems to solve

Nonlinear Problem: *CONTROL_IMPLICIT_NONLINEAR

Linear Problem: *CONTROL_IMPLICIT_LINEAR

Implicit Eigenvalue Analysis

Extract "n" eigenvalues by subspace iteration

Two additional databases output

• eigout: text file summarizing frequencies found

Simple input parameters

Implicit Boundary Conditions

Kinematic Boundary Conditions

Implicit Output Databases

Binary Output Databases

ASCII Output Databases

IN DYNAMIC SIMULATION---TIME STEPS

FE EQUATIONS AND TIME INTEGRATION

[M ]{u}+ [C ]{u}+ [K ]{u}= {R ext (t )} (1)

Equation (1) can be rewritten in the form

{FI (t )}+ {FD (t )}+ {R int (t )}= {R ext (t )} (2)

Direct Time Integration

[M ] = [M ([M ], [C ], ∆t )] is the effective mass matrix, and

EQUATION OF MOTION FOR TRANSLATIONAL VELOCITIES:

V and S are the actual volume and external surface

Flowchart for Explicit Time Integration

Compute internal nodal forces

Loop over elements: E = 1 to NELE

Implicit Governing Incremental Equations

M u n+1 + K t ( x )∆u 0 = R ( x ) n+1 − R ( x )

M: lumped mass matrix

Using Newmark method lead to the following equation

The iteration for equilibrium now begins using:

keyword input and arbitrary ID numbering are available in implicit mode

[M ]{u}+ [C ]{u}+ [K ]{u}= {R ext (t )} (1)

Dependence of Rexternal on current displacements

Shallow spherical caps

The FEM equation corresponding to linearized equilibrium is a matrix equation in N dof

Re-write the equilibrium equations as a function r of u :

r (u ) are the residuals (out-of-balance forces) at u . r (u ) is nonlinear!

Assume you are not in equilibrium at displaced position u so that r (u ) ≠ 0

Mx(t ) + F (x, x ) − P(x, t ) = 0

x, x , x = n dimensional acceleration, velocity, and geometry vectors

Fe (x, x ) = Cx + K∆x

Q = Mx + Cx + K∆x − P = 0