You are on page 1of 274

Evaluation of Thermal-, Creep- and

Corrosion-Fatigue of Heat Recovery


Steam Generator Pressure Parts

SED
N
A L
LICE

R I

M AT E
Technical Report

Effective December 6, 2006, this report has been made publicly available in accordance
with Section 734.3(b)(3) and published in accordance with Section 734.7 of the U.S. Export
Administration Regulations. As a result of this publication, this report is subject to only
copyright protection and does not require any license agreement from EPRI. This notice
supersedes the export control restrictions and any proprietary licensed material notices
embedded in the document prior to publication.
Evaluation of Thermal-, Creep- and
Corrosion-Fatigue of Heat Recovery
Steam Generator Pressure Parts

1010440

Final Report, March 2006

EPRI Project Manager


R. B. Dooley

ELECTRIC POWER RESEARCH INSTITUTE


3420 Hillview Avenue, Palo Alto, California 94304-1395 PO Box 10412, Palo Alto, California 94303-0813 USA
800.313.3774 650.855.2121 askepri@epri.com www.epri.com
DISCLAIMER OF WARRANTIES AND LIMITATION OF LIABILITIES
THIS DOCUMENT WAS PREPARED BY THE ORGANIZATION(S) NAMED BELOW AS AN
ACCOUNT OF WORK SPONSORED OR COSPONSORED BY THE ELECTRIC POWER RESEARCH
INSTITUTE, INC. (EPRI). NEITHER EPRI, ANY MEMBER OF EPRI, ANY COSPONSOR, THE
ORGANIZATION(S) BELOW, NOR ANY PERSON ACTING ON BEHALF OF ANY OF THEM:

(A) MAKES ANY WARRANTY OR REPRESENTATION WHATSOEVER, EXPRESS OR IMPLIED, (I)


WITH RESPECT TO THE USE OF ANY INFORMATION, APPARATUS, METHOD, PROCESS, OR
SIMILAR ITEM DISCLOSED IN THIS DOCUMENT, INCLUDING MERCHANTABILITY AND FITNESS
FOR A PARTICULAR PURPOSE, OR (II) THAT SUCH USE DOES NOT INFRINGE ON OR
INTERFERE WITH PRIVATELY OWNED RIGHTS, INCLUDING ANY PARTY'S INTELLECTUAL
PROPERTY, OR (III) THAT THIS DOCUMENT IS SUITABLE TO ANY PARTICULAR USER'S
CIRCUMSTANCE; OR

(B) ASSUMES RESPONSIBILITY FOR ANY DAMAGES OR OTHER LIABILITY WHATSOEVER


(INCLUDING ANY CONSEQUENTIAL DAMAGES, EVEN IF EPRI OR ANY EPRI REPRESENTATIVE
HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES) RESULTING FROM YOUR
SELECTION OR USE OF THIS DOCUMENT OR ANY INFORMATION, APPARATUS, METHOD,
PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS DOCUMENT.

ORGANIZATION(S) THAT PREPARED THIS DOCUMENT

Electric Power Research Institute (EPRI)

Aptech Engineering Services, Inc.

J. Michael Pearson & Associates Co. Ltd.

NOTICE: THIS REPORT CONTAINS PROPRIETARY INFORMATION THAT IS THE


INTELLECTUAL PROPERTY OF EPRI. ACCORDINGLY, IT IS AVAILABLE
ONLY UNDER LICENSE FROM EPRI AND MAY NOT BE REPRODUCED
OR DISCLOSED, WHOLLY OR IN PART, BY ANY LICENSEE TO ANY
OTHER PERSON OR ORGANIZATION.

NOTE
For further information about EPRI, call the EPRI Customer Assistance Center at 800.313.3774 or
e-mail askepri@epri.com.

Electric Power Research Institute and EPRI are registered service marks of the Electric Power
Research Institute, Inc.

Copyright © 2006 Electric Power Research Institute, Inc. All rights reserved.
CITATIONS

This report was prepared by

Electric Power Research Institute (EPRI)


3420 Hillview Avenue
Palo Alto, CA 94304

Principal Investigator
R. B. Dooley

Aptech Engineering Services, Inc.


1253 Reamwood Avenue
Sunnyvale, CA 95030

Principal Investigators
J. Grover
S. Paterson

J. Michael Pearson & Associates Co. Ltd.

Principal Investigator
M. Pearson

This report describes research sponsored by EPRI.

The report is a corporate document that should be cited in the literature in the following manner:

Evaluation of Thermal-, Creep- and Corrosion-Fatigue of Heat Recovery Steam Generator


Pressure Parts. EPRI, Palo Alto, CA: 2006. 1010440.

iii
PRODUCT DESCRIPTION

The worldwide fleet of combined cycle units with heat recovery steam generators (HRSG) has
exhibited a disappointing track record with respect to reliability and availability in terms of
fatigue HRSG tube failures (HTF) which are thermal transient driven. This report, which forms
part of a series, will assist designer, owners, and operators with the technical basis to facilitate
specifying, designing, and operating HRSG in a manner to minimize fatigue damage.

Results and Findings


The report provides a set of steps that can be used in the specification and design phases of an
HRSG. It provides a set of life cycle actions to prevent fatigue HTF by the application of special
diagnostic instrumentation, which should be used during commissioning or early life to assess
the design. Details are provided on how an operator can develop a set of startup and shutdown
curves for the steam headers and drums. An extensive review, with examples, is provided of all
the major design codes around the world.

Challenges and Objective(s)


The most frequently occurring HTF damage mechanisms include thermal, corrosion and creep
fatigue. These are influenced by transiently high, thermally induced cyclic stresses caused by
inadequate condensate attemperator control, and startup and shutdown rates which are too fast or
do not allow the combustion turbine to match the capability of the HRSG tubing and headers.
The objective of the overall work is the provide a set of tools and guidelines which operators can
use to assess the thermal performance of the HRSG from design to operation.

Applications, Values, and Use


The techniques and processes described in this report are applicable to all types of HRSG.
Adoption of the necessary practices will put an organization on the road to world-class
performance.

EPRI Perspective
To address this major HTF issue in the HRSG, EPRI has already developed a series of
documents: HRSG Tube Failure Manual (EPRI Report 1004503), Delivering High Reliability
HRSG (1004240), and Diagnostic/Troubleshooting Monitoring (1008088). Organizations
needed an approach to recognize the thermal design deficiencies early in the life of an HRSG.
The current document now provides this. The next steps in EPRI’s HRSG Program are to
develop a startup/shutdown guideline and specific details on drains, attemperators, duct burners
and dampers. EPRI will also continue to test the efficacy of the overall approach delineated in
this series of reports/guidelines by conducting thermal monitoring at host HRSG plants.

v
Approach
The EPRI team first conducted a review of the major design codes around the world to identify
the relevant parts which address fatigue. They then assembled the 10 thermal mechanical loads
of concern and developed a road map approach for assessment of the susceptible HRSG circuits
and locations. They next developed on approach similar to that used in the HRSG cycle
chemistry world using a set of target and action levels for critical ramp rates. The final step was
to assemble the road map for assessment.

Keywords
Heat recovery steam generators (HRSG)
Combined cycle units
Tube failures
Thermal transients
Monitoring and diagnostics

vi
EPRI Proprietary Licensed Material

ABSTRACT

Fatigue failures in Heat Recovery Steam Generators (HRSGs) are a primary cause of availability
loss in combined cycle units. EPRI has already developed a series of reports to address thermal-,
creep-, and corrosion fatigue failures throughout the life of an HRSG. This current document
provides designers, owners and operators of HRSGs with the technical basis to facilitate
specifying, designing and operating HRSGs in a manner to minimize fatigue damage. The report
provides a set of steps that can be used in the specification and design phases of an HRSG. It also
includes the development of target and action limits for thermal mechanical loads and
temperature ramp rates. A detailed review of a number of design codes and standards used for
HRSG pressure part design is also included.

vii
EPRI Proprietary Licensed Material

ACKNOWLEDGMENTS

The authors of this manual wish to thank the following individuals for providing illustrations,
input, case studies or comments on the information in this report:

D. Agan APTECH
K. Koenig SCANA
L.M. Aljama Iberdrola
J. Biega Progress Energy
R. Anderson Progress Energy
I. Perrin Alstom
L. Douglas Nooter/Eriksen
G. Komora Nooter/Eriksen

ix
EPRI Proprietary Licensed Material

CONTENTS

1 INTRODUCTION....................................................................................................... 1-1
1.1 Purpose and Objectives ..................................................................................... 1-1
1.2 Avoidance of Off-Design Thermal Mechanical Transients ................................ 1-2
1.3 Roadmap Steps ................................................................................................ 1-3
1.4 Setting Target and Action Limits For Thermal Mechanical Loads ..................... 1-3
1.5 Design Details and Materials of Concern .......................................................... 1-4
1.6 Overview of Design Codes and Standards ....................................................... 1-5
1.7 Historical Perspective – Conventional Fossil-Fuel-Fired Units Lessons
Learned.................................................................................................................... 1-6
1.8 Historical Perspective – HRSGs, “Lessons Still to be Learned” ........................ 1-7
1.9 Organization of the Manual ............................................................................. 1-10

2 ROADMAP APPRAOCH .......................................................................................... 2-1


2.1 General Steps for Assessing and Correcting Pressure Part Fatigue
Damage Issues ........................................................................................................ 2-1
2.2 Life Cycle Actions to Prevent Pressure Part Fatigue Failures........................... 2-2
2.3 Assessment of HRSG Pressure Part Fatigue Failures...................................... 2-4

3 COMMON THERMAL-MECHANICAL DAMAGE MECHANISMS AND


INFLUENCING LOADS ............................................................................................... 3-1
3.1 HRSG Pressure Part Damage Mechanisms ..................................................... 3-1
3.2 Thermal-Mechanical Loads & Key Damage Influencing Parameters ................ 3-2

4 FATIGUE ASSESSMENTS – OWNER PROCUREMENT SPECIFICATION ........... 4-1


4.1 Specifying Definitions of Anticipated Operating Transients............................... 4-2
4.2 Specifying the Annual and Lifetime Operating Transient History ...................... 4-8
4.3 Specify What Design Calculations, Design Limits and Reports are
Required ................................................................................................................ 4-18

xi
EPRI Proprietary Licensed Material

4.4 Specify Special Diagnostic/Troubleshooting Monitoring Tests and


Acceptance Criteria to Be Performed During Commissioning................................ 4-26
4.5 Specify What Diagnostic/Troubleshooting Monitoring Instrumentation,
Controls, Data Acquisition Systems and Data Evaluation Algorithms are to Be
Used Throughout the Lifetime of the HRSG........................................................... 4-28

5 FATIGUE ASSESSMENTS – DESIGN..................................................................... 5-1


5.1 Owner to Specify that HRSG and Key Auxiliary System Designs Capable
of Causing Uncontrollable Transients with Large Tube Temperature
Differences During Startups Will Be Rejected.......................................................... 5-2
5.2 Prior to Contract Award, Owner to Review the Proposed Detailed Design
of HRSG and Key Auxiliary Systems to Verify that there are No Details and
Features Capable of Causing Uncontrollable Transients With Large Tube
Temperature Differences During Startups ............................................................... 5-4
5.3 Verify that the Assembly/Component/Feature/Construction Attributes and
Operating Parameters Meet Acceptable Design Limits............................................ 5-4
5.4 Set the Allowable Component Operating Limits................................................ 5-7
5.5 Prescribe the Operating Procedures and Permissibles................................... 5-14
5.6 Create a List of Important Design/Construction Damage Avoidance
Features and Required Construction QC/QA Tests and Limits.............................. 5-14
5.7 Specify the Design and Construction Requirements of Special Damage
Monitoring Instrumentation and Operating Tests to be Used during
Commissioning ...................................................................................................... 5-15
5.8 Specify Damage Monitoring Instrumentation, Controls, Data Acquisition
Systems and Data Evaluation Algorithms to be Used to Monitor Component
Damage Throughout the Lifetime of the HRSG ..................................................... 5-15

6 FATIGUE ASSESSMENTS – CONSTRUCTION...................................................... 6-1

7 FATIGUE ASSESSMENTS – COMMISSIONING..................................................... 7-1

8 FATIGUE ASSESSMENTS – OPERATION ............................................................. 8-1

9 FATIGUE ASSESSMENTS – MAINTENANCE ........................................................ 9-1

10 REFERENCES...................................................................................................... 10-1

A REVIEW OF FATIGUE ASSESSMENT FACTORS ................................................ A-1


Design by Rule/Design by Analysis .........................................................................A-1

xii
EPRI Proprietary Licensed Material

Exclusion Rules .......................................................................................................A-2


Fatigue Cycle ...........................................................................................................A-2
Loading Rate............................................................................................................A-4
S-N Curves ..............................................................................................................A-4
Endurance Limit .......................................................................................................A-5
Variable Amplitude Loading .....................................................................................A-6
Fatigue Curve or Fatigue Stress Correction Factors................................................A-6
Component Stresses .............................................................................................A-11
Stress Concentration Factors.................................................................................A-18
Stress Intensification Factors (for Analysis of Connections Acted on by
External Forces and Moments) ..............................................................................A-26
Equivalent Multiaxial Stress (Stress Intensity) .......................................................A-31
Sequence of Loading .............................................................................................A-36
Environment...........................................................................................................A-36
Temperature ..........................................................................................................A-38
Creep-Fatigue ........................................................................................................A-40
Fatigue Crack Growth ............................................................................................A-43
Creep .....................................................................................................................A-44
References.............................................................................................................A-46

B REVIEW OF FATIGUE DESIGN CODES AND STANDARDS ............................... B-1


American Codes and Standards ..............................................................................B-1
ASME Section I ...................................................................................................B-1
ASME Section III, Subsection NB ........................................................................B-1
ASME Code, Section III, Division 1, Subsection NH............................................B-3
ASME Code, Section VIII.....................................................................................B-7
ASME Section XI .................................................................................................B-8
API RP 579 ..............................................................................................................B-8
Other American Codes ..........................................................................................B-11
British Codes and Standards .................................................................................B-11
BS 1113.............................................................................................................B-11
BS PD 5500.......................................................................................................B-11

xiii
EPRI Proprietary Licensed Material

R5 and R6 .........................................................................................................B-14
BS PD 6493.......................................................................................................B-15
French Codes and Standards ................................................................................B-16
RCC-MR ............................................................................................................B-16
German Codes and Standards ..............................................................................B-17
TRD 301 ............................................................................................................B-17
TRD 508 ............................................................................................................B-18
Other German Codes and Standards ................................................................B-18
European Community Codes and Standards.........................................................B-19
EN 12952...........................................................................................................B-19
EN 13445...........................................................................................................B-23
Other European Codes and Standards..............................................................B-26
Comparison Of Fatigue Code Attributes ................................................................B-26
Exclusion Rules .................................................................................................B-26
Conclusions ...........................................................................................................B-30
References.............................................................................................................B-30

C ESTIMATING THE FLEXIBILITY/TUBE-TO-TUBE TEMPERATURE


DIFFERENCE INFLUENCED FORCES AND MOMENTS AT TUBE-TO-
HEADER CONNECTIONS.......................................................................................... C-1

D EXAMPLE FATIGUE ANALYSIS OF AN LP ECONOMIZER (PREHEATER)


WITH TUBE-TO-TUBE AND ROW-TO-ROW TEMPERATURE DIFFERENCES ...... D-1
Background Information .......................................................................................... D-1
Material Properties.................................................................................................. D-3
Nominal Elastic Stresses ........................................................................................ D-5
Structural Elastic Stresses ...................................................................................... D-5
Peak Maximum Principal Elastic Stresses .............................................................. D-6
Other Stress Concentration Factors........................................................................ D-7
Equivalent, Multiaxial Elastic Stress (Stress Intensity) Values ................................ D-7
Estimating the Fatigue Life Using the Markl Equation............................................. D-7
Estimating the Elastic Plastic Stress and Total Strain Values During the First
Loading Cycle ......................................................................................................... D-8

xiv
EPRI Proprietary Licensed Material

Estimating the Elastic Plastic Stress and Total Strain Values after Unloading
from the Peak Stress Loading Cycle....................................................................... D-9
Estimating the Cyclic Life with Mean Stress Corrections ...................................... D-11

E ESTIMATING THE BOREHOLE THERMAL SHOCK (LOADING MODE 1)


LIMITS FOR RISER/DOWNCOMER PIPE NOZZLE TO DRUM, NOZZLE PIPE
TO HEADER OR TUBE TO HEADER BOREHOLES .................................................E-1
Case Study 1 – HP Steam Drum Borehole Fatigue Assessment.............................E-2
Case Study 2 – HP Superheater Outlet Header Tube Borehole Fatigue
Assessment .............................................................................................................E-3
Consideration of Operating Limits for Borehole Cracking Prevention ......................E-5
Limitations of Current Design Code for Fatigue Assessment of Drum or
Header Boreholes ....................................................................................................E-7
References...............................................................................................................E-8

F MONOTONIC AND CYCLIC PROPERTIES OF HRSG PRESSURE PART


MATERIALS ................................................................................................................F-1
Cyclic Stress Strain Curve .......................................................................................F-3

xv
EPRI Proprietary Licensed Material

LIST OF FIGURES

Figure 2-1 Analysis steps to follow if a fatigue failure has already occurred .............................2-5
Figure 4-1 HP drum pressure decay curves for three relatively new triple-pressure
HRSGs with a design pressure of 2074 psig (143 barg). The times and minimum
HP drum pressure and water temperature values prior to restarts associated with
rapid hot, hot, warm, extended warm, cool and cold starts are shown. Based on
experience in similar units it is anticipated that the pressure decay response will
degrade (become faster) as the unit ages unless all the pressure blocking valves
are well maintained and/or are designed as sets of master/martyr valves.........................4-6
Figure 4-2 Example calculations of EN12953-3 allowable fluid temperature change rates
based on a conservative, bounding (thermal plus pressure stress) thermal-fatigue
assessment which considers loading mode Type 1 (Table 3-1). This loading mode
addresses the potential for HP drum to downcomer or riser nozzle borehole
cracking or HP superheater header borehole cracking. Meeting these limits based
on measured bulk fluid values does not ensure that individual boreholes with much
higher fluid ramp rates will not crack. Meeting these limits also does not ensure that
other loading modes described in Chapter 3 will not occur..............................................4-16
Figure 4-3 Allowable Transient Temperature Differences between Two Points in Close
Proximity ..........................................................................................................................4-17
Figure 4-4 Example of an allowable temperature differential chart. The permissible limits
shown are for an HP superheater manifold with specific dimensions and design
details...............................................................................................................................4-26
Figure 5-1 Estimates of the Stresses and Cycles to Cracking as a Function of Tube
Row-to-Row Temperature Difference.................................................................................5-6
Figure 5-2 Effect of Tube Offset Distance and Angle on Stresses in Tube/Header
Connection .........................................................................................................................5-7
Figure 5-3 Example control chart for an HP drum .....................................................................5-9
Figure 5-4 HP drum pressure/temperature ramp rate control chart with data from two
cold starts (CS), a warm start (WS), a hot start (HS) and a low load (LL) cycle
superimposed...................................................................................................................5-10
Figure 5-5 Example control chart for tube-to-tube temperatures in a secondary (high
temperature) reheater. The action level targets were derived from the analysis
shown in Figure 5-1 and engineering judgment. ..............................................................5-12
Figure 5-6 Measured bulk fluid temperatures into and out of a secondary reheater
measured during a warm start [4] ....................................................................................5-13
Figure 5-7 Plot of the secondary reheater tube metal temperatures at the timeslice
shown in Figure 5-6 that corresponded to the attemperator overspray quench event
[4] .....................................................................................................................................5-13

xvii
EPRI Proprietary Licensed Material

Figure A-1 Schematic of Fatigue Cycle ................................................................................... A-3


Figure A-2 Fatigue Design Curves for Unnotched Specimens ................................................ A-9
Figure A-3 Fatigue Design Curves for Weldments .................................................................. A-9
Figure A-4 Example of Local Variations in Weld Toe Angle that are Undersirable but
also Commonly Seen in Tube to Header Connections ................................................... A-20
Figure A-5 Comparison of Borehole Thermal Stress Correction Factors............................... A-21
Figure A-6 Borehole Pressure Stress Correction Factors (Note: dm= mean diameter of
tube, Dm= mean diameter of header and Tm=thickness of header) ................................ A-23
Figure A-7 Comparison of Various Tube to Header Stress Intensification Factors for a
Wide Range of HRSG Headers. (Note:Tnom=nominal thickness of header, Dm =
mean diameter of header)............................................................................................... A-24
Figure A-8 Effect of Header Flexibility Factor (Mean Radius to Nominal Thickness Ratio,
R/T) on the “Fatigue Test Derived” Stress Intensification Factor, SIF............................. A-27
Figure A-9 Effect of r/R Ratio on the Stress Intensification Factor ........................................ A-28
Figure A-10 Effect of t/T Ratio on the Stress Intensification Factor ....................................... A-29
Figure A-11 Comparison of Markl Data with SCF to Corrected* Fatigue Design Curves ...... A-30
Figure A-12 Example of Mohr’s Circle in the XY Plane ......................................................... A-32
Figure A-13 Example of Tube-Header Connection................................................................ A-33
Figure A-14 Three-Dimensional Mohr’s Circle for Biaxial Loading at the Outside Surface
of a Pressurized Tube. .................................................................................................... A-33
Figure A-15 Three-Dimensional Mohr’s Circle for Triaxial Loading at the Inside Surface
of a Pressurized Tube. .................................................................................................... A-34
Figure A-16 Comparison of Corrosion Fatigue Test Data for Carbon Steel Tubing
Tested in 135oC to 343oC (275oF to 650oF) Water.......................................................... A-38
Figure A-17 Temperature Based Cyclic Stress Correction Factors ....................................... A-39
Figure A-18 Fatigue Tests of 2-1/4Cr 1Mo Steel Performed at Room Temperature and
at 600oC (1110oF) with Continuous Cycling at 1 Cycle per Minute and with 30 and
300 Minute (5 hr) Tensile Hold Times. [A-40] ................................................................. A-40
Figure A-19 Outside Surface, Weld Toe Initiated Creep-Fatigue Crack ................................ A-41
Figure A-20 Outside Surface Weld Toe Initiated Creep-Fatigue Crack ................................. A-41
Figure A-21 Creep-Fatigue Interaction .................................................................................. A-42
Figure A-22 Typical Stress Rupture Curves (ASME Section III NH for 2¼Cr-1Mo Steel) ..... A-45
Figure A-23 Stress Rupture Curves of Figure A-22 Collapsed to a Single Line Using the
Larson-Miller Parameter.................................................................................................. A-46
Figure B-1 ASME Section III Subsection NB Fatigue Design Curves...................................... B-2
Figure B-2 Use of Neuber Relationship to Calculate Elastic-Plastic Strain ............................. B-5
Figure B-3 ASME Section III NH S-N Curves for 2¼Cr-1Mo Steel .......................................... B-6
Figure B-4 Multiplier on Cycles for Metal Temperature Differentials in ASME Section III
Subsection NB Exclusion Rules ........................................................................................ B-8
Figure B-5 ASME Section XI Fatigue Crack Growth Curves ................................................... B-9
Figure B-6 API 579 S-N Curves............................................................................................. B-10

xviii
EPRI Proprietary Licensed Material

Figure B-7 BS 5500 S-N Curve, Compared to ASME III NB.................................................. B-12
Figure B-8 BS PD 5500 S-N Curves...................................................................................... B-13
Figure B-9 TRD 301 Fatigue Design Curves ......................................................................... B-18
Figure B-10 Fatigue Design Curves of EN 12952-3 .............................................................. B-20
Figure B-11 EN 12952 SCF for Pressurized Nozzle.............................................................. B-21
Figure B-12 Stress Concentration Factor for Thermal Stresses ............................................ B-22
Figure B-13 Fatigue Design Curves of EN 13445.................................................................. B-24
Figure B-14 EN 13445 Design Curve for Unnotched Material ............................................... B-25
Figure C-1 Geometry Characteristics of Selected Types of Tube Bundles.............................. C-2
Figure C-2 Geometry of a single pass, two parallel row tube bundle with dogleg bends
at the bottom of the leading row and top of the training row. .......................................... C-15
Figure C-3 Example top and bottom tube metal temperatures for a 32 element tube
bundle with the configuration and geometry shown in Figure C-2. ................................. C-16
Figure C-4 Estimated nominal axial and bending stress at the tube to header
connections for the geometry shown in Figure C-2 and the tube metal temperatures
shown in Figure C-3. To estimate the fatigue life these nominal stresses would be
magnified by a stress concentration factor and the tube temperatures for the
remainder of the operating cycle would need to be measured to determine if the
stresses will just return to near zero or will reverse and produce a stress range that
exceeds these values...................................................................................................... C-17
Figure D-1 Feedwater heater tube bundle attributes ............................................................... D-2
Figure D-2 Design-basis tube and header temperatures. In addition to the nonuniform,
steady state, side to side temperature distribution associated with the three pass
design and header partition plates, the analysis assumed that a few tubes in Row 3
located near the inlet pipe nozzle and lower header partition plate were selectively
quenched with cold feedwater during a cold start to a value of approximately 106oF
(41oC).The tubes with the highest magnitude stresses were found to be the tubes
that were selectively quenched and located in the vicinity of the lower header inlet
pipe nozzle and lower header partition plate..................................................................... D-2
Figure D-3 Ramberg-Osgood cyclic stress strain curve .......................................................... D-4
Figure D-4 Coffin-Manson strain amplitude versus median reversals to failure curve for
SAE 1005 carbon steel at room temperature. SA 192 tubing is anticipated to have
similar fatigue behavior. .................................................................................................... D-4
Figure D-5 Schematic of the stress-strain history for Locations A and B during thermal
quench loading and subsequent shutdown unloading events......................................... D-17

xix
EPRI Proprietary Licensed Material

LIST OF TABLES

Table 3-1 Schematic of the Type 1 Thermal-Mechanical Loads................................................3-3


Table 3-2 Type 2 Thermal-Mechanical Loads ...........................................................................3-4
Table 3-3 Type 3 Thermal-Mechanical Loads ...........................................................................3-6
Table 3-4 Type 4A Thermal-Mechanical Loads .........................................................................3-7
Table 3-5 Type 4B Thermal-Mechanical Loads .........................................................................3-8
Table 3-6 Type 5A Thermal-Mechanical Loads .......................................................................3-10
Table 3-7 Type 5B Thermal-Mechanical Loads .......................................................................3-11
Table 3-8 Type 6 Thermal-Mechanical Loads .........................................................................3-12
Table 3-9 Type 7 Thermal-Mechanical Loads .........................................................................3-14
Table 3-10 Type 8 Thermal-Mechanical Loads .......................................................................3-15
Table 4-1 Suggested Characteristics of Selected HRSG Operating Transients ........................4-3
Table 4-2 Design-Basis HP Drum Pressures and Water Temperatures Prior to Various
Operating Startups .............................................................................................................4-6
Table 4-3 Sample Unit Number of Operating Cycles for Design Purposes ..............................4-9
Table 4-4 History of Actual Transients During the Early Years of Operation of Two Triple
Pressure HRSGs..............................................................................................................4-12
Table 4-5 Example Specified Design Basis Transients for the Two HRSGs Documented
in Table 4-4 ......................................................................................................................4-14
Table 4-6 Selected HRSG pressure part components/design features that should be
subjected to fatigue evaluations.......................................................................................4-21
Table A-1 Fatigue Curve Correction Factors in EN 12952-3 ................................................. A-12
Table A-2 Component Stresses For a Borehole Thermal Shock Fatigue Evaluation ............ A-14
Table A-3 Component Stresses For A Fatigue Evaluation of a Quench Event ..................... A-16
Table A-4 Micro-Notch Surface Roughness Stress Correction Factors for Non-Welded
Machined Surfaces ......................................................................................................... A-21
Table A-5 Sensitivity of cyclic life on stress range changes .................................................. A-22
Table B-1 Comparison of Fatigue Codes............................................................................... B-27
Table D-1 Cyclic Stress-Strain and Fatigue Evaluation ......................................................... D-13
Table D-2 Cyclic Stress-Strain and Fatigue Evaluation ......................................................... D-15
Table E-1 Fatigue analysis of a borehole region of a 1890-mm OD x 106-mm wall
(74.4-inch OD x 4.2-inch wall) SA209 drum with set-in nozzles. The shaded regions
show the required input values ......................................................................................... E-9

xxi
EPRI Proprietary Licensed Material

Table E-2 Example EN12952-3 Section 13 & Appendix B & C 2001 Edition Calculations
for a 323.8 mm OD x 28.6 mm Wall P91 HP Superheater Outletlet Header Borehole
Subjected to Pressure Changes and Thermal Shock Loadings...................................... E-15
Table F-1 Carbon Steel (e.g.,SA299, >1" thick) Material Properties Ref: ASME Section
II, Part D (2001)..................................................................................................................F-5
Table F-2 Grade 22 (2-1/4 Cr – 1 Mo steel) Material Properties Ref: ASME Section II,
Part D (2001.......................................................................................................................F-6
Table F-3 Grade 91 (9Cr-1Mo-V-Cb steel) Material Properties Ref: ASME Section II,
Part D (2001)......................................................................................................................F-7
Table F-4 Summary of Monotonic and Cyclic Properties..........................................................F-8
Table F-5 Materials Characterization Sheet .............................................................................F-9
Table F-6 Materials Characterization Sheet ...........................................................................F-10
Table F-7 Materials Characterization Sheet ...........................................................................F-11
Table F-8 Materials Characterization Sheet ...........................................................................F-12
Table F-9 Materials Characterization Sheet ...........................................................................F-13
Table F-10 Materials Characterization Sheet .........................................................................F-14
Table F-11 Materials Characterization Sheet .........................................................................F-15
Table F-12 Materials Characterization Sheet .........................................................................F-16
Table F-13 Materials Characterization Sheet .........................................................................F-17
Table F-14 Materials Characterization Sheet .........................................................................F-18
Table F-15 Materials Characterization Sheet .........................................................................F-19
Table F-16 Materials Characterization Sheet .........................................................................F-20
Table F-17 Materials Characterization Sheet .........................................................................F-21
Table F-18 Materials Characterization Sheet .........................................................................F-22
Table F-19 Materials Characterization Sheet .........................................................................F-23

xxii
EPRI Proprietary Licensed Material

1
INTRODUCTION

Although heat recovery steam generators (HRSGs) have existed since the 1950s, there has been a
major growth of the use of HRSGs in the last decade. With the increased popularity and capacity
of HRSGs have come increasingly more complex designs, operating at higher temperatures and
pressures than in the past, and with design features such as multi-pressure circuits and reheat.

More complex design and operating conditions have led to premature failures of pressure parts,
in relatively young units. A previous EPRI report showed that the primary cause of tube failures
in HRSGs was corrosion-fatigue, followed by thermal-fatigue, and flow accelerated corrosion.
[1]

Low cycle fatigue caused by transient thermally induced stresses during startups and shutdowns
and other major unit transients is not a new mechanism, since it has been a major cause of
pressure part failures in utility steam generators and steam turbines. The fact that modern
HRSGs are seeing more fatigue failures than older designs of HRSG indicates that the evolution
to more modern designs specified for significantly higher flow rates, pressures and temperatures,
more challenging combustion turbine exhaust gas temperatures and more complex steam cycles
has focused on minimizing installed cost and maximizing thermal performance, to the detriment
of mechanical performance (reliability and operating flexibility). The impact of the substantial
increases in HRSG design conditions, physical size and in operating complexities of combined
cycle generating units on the arduous duties imposed on the pressure parts during startups and
shutdowns have not been adequately understood or optimized.

1.1 Purpose and Objectives

The purpose of this report is to provide designers, owners and operators of HRSGs with a
technical basis to facilitate specifying, designing, and operating HRSGs in a manner that
minimizes fatigue damage, and increases reliability. This report considers three types of fatigue:
thermal-fatigue, corrosion-fatigue, and creep-fatigue, and will not address other less frequently
experienced types of fatigue such as flow-induced vibration.

Other objectives of this report include providing methods that will be helpful with the following
fatigue avoidance actions:
• Determination by the HRSG designer of the compatibility of the design of pressure parts with
the project specific design lifetime numbers of the different types of startup–shutdown
cycles.

1-1
EPRI Proprietary Licensed Material

Introduction

• Determination by the HRSG designer of the recommended and permissible limits of the CT
exhaust gas temperature and flow and of HP steam temperature, pressure and flow, all versus
time, for startups from different HRSG initial HP drum pressure and for normal shutdowns
that are compatible with the project specific design lifetime numbers of the different types of
startup–shutdown cycles.
• Determination by the HRSG designer and possibly the EPC contractor of the operating
procedures for performing startups and shutdowns that will avoid damaging, off-design
transient thermal mechanical loading events and simultaneously achieve the dispatch,
environmental and pressure part reliability requirements for the unit.

1.2 Avoidance of Off-Design Thermal Mechanical Transients

It is not the intent of this report to provide the technical basis to accept off-design thermal
mechanical transients that can and should have been corrected during the design and construction
of the HRSG. For example, thermal transients caused by migration of undrained or reverse flow
of condensate or spraying steam down to or below the saturation temperature or slug feeding of
cold feedwater into hot tubes are all off-design transients that must be eliminated. These
transients should not occur in properly design units and should be eliminated through proper
design and construction of drain, attemperation and feedwater systems and controls.

Critical to the avoidance of premature failures in pressure parts is the elimination from the design
of HRSG and critical auxiliary systems, features such as improperly sloped steam piping or
poorly placed attemperators that introduce the potential for uncontrollable upset conditions
which to date are either overlooked or largely ignored by HRSG designers and purchasers.
Many very premature tube fatigue failures have occurred in HRSGs at header attachment welds
or tube bends which occurred within a few hundred starts and were caused by major thermal
transients during uncontrollable upset conditions that occur during startups or shutdowns for
which the operator can do little or nothing to prevent.

For example, the forward migration of undrainable condensate in HP superheaters and reheaters
during startups has been repeatedly highlighted as a severely damaging phenomenon for more
than 10 years and yet few, if any, horizontal gas path HRSGs have been designed and installed
with pressure parts and drains systems that are completely effective.

Overspraying down to saturation or below in HP superheaters and reheaters during load raising
at startup occurs on many HRSGs imposing very damaging thermal transients on downstream
headers and tubes, yet designs for new installations continue to largely ignore this serious design
deficiency.

Many HRSGs shutdown the CT from 50% load much too quickly, leaving the lower HP
superheater headers hot and subjected to severe quenching when condensate runs out of the tubes
during the CT deceleration or shortly after shutdown. Preheater/economizer inlet tubes are
subjected to large thermal mechanical loads when feedwater is first admitted during startups,
unless the HRSG includes an adequately sized feedwater recirculation system and appropriate

1-2
EPRI Proprietary Licensed Material

Introduction

controls to prevent large sudden changes in temperature of preheater/economizer inlet feedwater


temperature, yet this is often not done.

Until owners independently review and obtain effective modifications to proposed designs of
HRSGs and drains systems before HRSG OEM selection and contract award, these sources of
very premature fatigue damage will continue to be installed on new installations. Unless these
HRSG design deficiencies are successfully corrected on new installations fatigue failures within
a few hundred starup-shutdown cycles will continue to overwhelm the benefits of the broader
application of fatigue evaluations discussed in this report aimed at preventing more widespread
pressure parts damage in several thousands of startup-shutdown cycles.

To help owner/operators and HRSGs manufacturers obtain the appropriate design and operating
control features to eliminate the common off-design thermal mechanical transients, EPRI plans
to publish HRSG Startup and Shutdown Guidelines and a Guide on the Design and Optimization
of HRSG Drain and Feedwater Sytems.

1.3 Roadmap Steps

This report provides a set of steps that can be used to address the HRSG pressure part design
details that have experienced early service life fatigue failures (e.g., tube to header connection
failures, drain to header or manifold failures, high energy piping failures, etc). The evaluation
steps suggested have been developed from various international design codes and standards.
Rather than attempt to develop complex, extremely accurate, well validated fatigue assessment
steps the approach taken in this report was to use a combination of relatively simple analysis
steps and safety margins that will provide conservative estimates of fatigue life and accurate
estimates of the life improvement that will be obtained after changes to operation, maintenance
or design. The overall objective being to motivate fatigue resistant design details and
operational/maintenance procedures so that there is high assurance that HRSG pressure part
fatigue failures will not occur within the specified or desired life of the unit.

1.4 Setting Target and Action Limits For Thermal Mechanical Loads

It is envisaged that one of the primary uses of this document will be to help HRSG owner
operators determine safe limits for potentially damaging thermal-mechanical parameters such as
temperature ramp rates, pressure ramp rates, tube to header temperature differences, tube to tube
temperature differences, etc.). The long term vision, of which this document is one of the key
ingredients, is to develop a set of industry consensus limits for potentially damaging thermal-
mechanical parameters. This report provides some of the technical framework for such
guidelines. It is anticipated that EPRI’s first Startup and Shutdown Guideline and Interim
Consensus Thermal-Mechanical Limits Guideline will be published in 2006. The
Diagnostic/Troubleshooting Monitoring to Identify Damaging Cycle Chemistry or Thermal
Transients in Heat Recovery Steam Generator Pressure Parts, EPRI Palo Alto, CA:2005
(Product ID 1008088) provides guidance on special instrumentation that is required to identify
the severity and operating features associated with potentially damaging HRSG thermal
transients.

1-3
EPRI Proprietary Licensed Material

Introduction

Most HRSG manufacturers provide some minimal thermal mechanical limits for HRSG pressure
parts. The majority of these are steady state limits on temperature and pressure. For example the
maximum hot reheat steam outlet temperature may be limited to 566oC (1050oF) or the minimum
preheater feedwater inlet temperature may be limited (for natural gas fired units) to 63oC
(145oF). These limits are set to prevent damaging the pressure parts during steady state periods
of operation. For example, the maximum and minimum temperature limits described above are
intended to prevent longterm overheat-creep and dew point corrosion.

Limits on the thermal-mechanical parameters to ensure that fatigue damage does not result in
pressure part failures within the anticipated design life of the unit are often not adequately
addressed in the design through commissioning stages of the HRSG. Often, the only thermal-
mechanical parameter limit that is prescribed with the purpose of preventing fatigue failures is
the HP drum pressure gradient (bar/min or psi/min) and/or HP drum water temperature ramp rate
(oC/min or oF/min). These ramp rate limits are typically determined using calculations performed
in accordance with TRD 301, Annex 1 or EN 12952, Part 3, Section 13 rules. As will be shown
later modified versions of these rules are needed to ensure that the limits derived are
conservative for creep-fatigue or corrosion fatigue and to allow the benefit of hold periods and
engineered, nonlinear ramp rates to be included in assessments.

It is envisaged that these thermal mechanical target and action limits will be utilized in the same
fashion as EPRI’s cycle chemistry targets and action limits. They will be provided to the operator
and when they are exceeded they will lead to the requirement to take appropriate corrective
action. For example, if a reheater harp is experiencing tube-to-tube temperature differences with
a magnitude and frequency which will exceed the limits that have been determined to be
permissible then the operator would need to either ammeliorate or eliminate the transient by
taking appropriate operational, maintenance or design modification actions or would need to
alert dispatch and management personnel that the frequency of the operating transient producing
these events must be reduced on an annual basis in order to meet the longterm reliability goals
for the unit.

1.5 Design Details and Materials of Concern

The following HRSGs pressure boundary details have either experienced in-service fatigue
failures or are of significant concern with regard to fatigue:
• Tube-header connections
• Header boreholes
• Drain to header connections
• Riser pipe nozzle and downcomer pipe to drum connections and boreholes
• Pipe nozzle to header connections
• Header end cap welds
• Pipe-to-pipe circumferential welds, especially at alloy and/or wall thickness transitions
• Inner surface of pipe and pipe branches downstream of attemperators

1-4
EPRI Proprietary Licensed Material

Introduction

• Tube-to-tube circumferential welds connecting 180-degree return bends to straight sections


of tubing
• 180-degree return bends
• From the outside surface at the intrados of small-radius tube bends near headers

The primary materials of interest include carbon steel, Grade 22 (2¼Cr-1Mo), and Grade 91
(9Cr-1Mo-V-Cb) as well as the corresponding weld metals.

The technical discussions in this report are generally limited to very localized plastic strains in
pressure part components where the bulk of the pressure part material remains in an elastic (i.e.,
not permanently deformed) condition. The cyclic damage may occur with no visible deformation
to the tube, pipe or header etc. HRSG pressure parts are sometimes subjected to very severe
cyclic damage such as shakedown, ratcheting or thermal quench induced fractures which are not
addressed in this report. When tube bowing or other permanent, visible deformation is present
this may indicate that very severe thermal-mechanical loads have occurred and that further
investigation and diagnostic monitoring is needed to determine if the loading which caused the
permanent deformation is a recurring event that will lead to pressure part failures.

1.6 Overview of Design Codes and Standards

Unfortunately, there is no single code or standard that adequately addresses all of the fatigue
design issues for HRSG pressure parts. Appendix B provides a review of a number of codes or
standards that have been used for HRSG pressure part design or that have fatigue design rules
that could be used for HRSG pressure part design. Two key conclusions regarding these codes
and standards include:
• The most frequently used HRSG pressure part design code (ASME Section I) does not
include any fatigue rules or requirements
• When fatigue is considered, TRD 301, Annex 1, EN 12952 Part 3, EN 13480, Part 3 and
ASME Section VIII, Division 2 are the design codes that have most frequently been applied.
They provide the fundamental steps in fatigue evaluation but they all have a number of
deficiencies:
– They are non-conservative in respect to creep-fatigue
– They don’t fully address corrosion fatigue
– They each have a small, but incomplete, set of pressure and thermal stress solutions
relevant to HRSG pressure boundary components
– They generally assume that the rates of change of fluid temperatures are constant
throughout shutdowns and startups, thus neglecting the damaging effects of initial step
changes in temperature during startups and failing to advise on the beneficial use of
carefully designed nonlinear ramp rates in conjunction with planned hold periods for heat
soaking

1-5
EPRI Proprietary Licensed Material

Introduction

– They assume that each shutdown/startup operational cycle produces a single stress cycle,
whereas field measurements often indicate that multiple significant stress cycles are
experienced during shutdowns and startups

To achieve the objectives of providing a simple, conservative analysis relevant to the high
priority HRSG pressure part details an enhanced version of the fatigue rules in TRD 301, Annex
1, EN 12952 Part 3, and EN 13480, Part 3 was developed to overcome some of these
deficiencies.

1.7 Historical Perspective – Conventional Fossil-Fuel-Fired Units Lessons


Learned

It is useful to recall some of the evolution of fatigue issues and design rules, developed based on
service experience and research on steam turbines and pressure parts in fossil-fuel-fired power
plants. A summary of some of the salient events in the United Kingdom (UK) are(5):
• In the 1950s a program of monitored two-shifting and quick-start trials was undertaken on
each of the lead power stations in the UK to determine the optimal procedures for operating
the plant and to obtain data on the behavior of the boiler and turbine-generator units. These
trials continued into the 1970s as the 500 MW and 660 MW fleets became established
• Commencing in the 1960s and continuing at least into the 1990s, a program of longterm
creep and fatigue testing (with tests durations approaching 120,000 hrs) was undertaken
under the auspices of the CEGB, Parsons, and GEC. This work was focused on steam turbine
rotors and included a comparison of proposed life consumption algorithms in terms of their
ability to predict the endurance levels observed in tests.
• In the 1980s and 1990s, numerous steam turbine components were metallurgically examined
to determine the actual rate of life consumption. The fleet of 100 MW and 120 MW units that
were examined had typically completed between 150,000 and 200,000 running hours plus up
to 4500 stop/start cycles. The first signs of microcracking on some rotors allowed validation
of the creep-fatigue life consumption algorithms.

It is also useful to consider other findings summarized in Reference 5 derived from service
experience mainly from conventional fossil-fuel-fired plants:
• During startups the boiler parts such as tubes, headers, ducts and supporting structure would
move considerably, giving rise to expansion-induced strains. A boiler furnace structure
would typically expand by 250 mm (10-inches) vertically and 150 mm (6-inches)
horizontally.
• Header borehole ligament cracking after 300 to 5500 stop/start cycles was a well documented
problem and was often associated with the borehole surfaces being rapidly quenched from a
temperature near 500oC (932oF) to temperatures in the range of 350oC (662oF) as a result of
forward flow of condensate and improper operation of desuperheaters in futile attempts to
prevent occasional over temperature excursions during startups.
• The CEGB, as “informed buyers”, specified the design requirements for plants to include
being capable of 5000 starts after an 8 hr shutdown, 1000 starts after a 36 hr shutdown and

1-6
EPRI Proprietary Licensed Material

Introduction

200 cold starts for duty towards the end of the unit life. This requirement lead to radical
design changes to eliminate excessive forces and moments at the tube to header connections
and a significant reduction in header thickness. At that time, it was believed that a header
thickness less than 40 mm (1.6-inch) was required to avoid ligament cracking due to the
pendant superheater condensate quench events. More recent evaluations have suggested that
even thinner header thicknesses may be required to avoid cracking.
• Fatigue cracking in steam turbines that had been designed for 5000 hot starts, 1000 warm
starts, and approximately 200 cold starts was rare, as care was taken to protect these
expensive components from severe thermal transients. One expert estimated that plants
pressed into two-shift duty might experience three to seven years of additional service before
thermal cycling problems would be seen.
• A significant problem in radiant and convection superheater coils was steam flow blockage
by condensate in the lower return bends (leading to tube overheating) followed by quenching
of the tube and header borehole when the condensate suddenly flushed through. This led to
header ligament temperature differentials far in excess of the “old rule” of not permitting
more than a 40oC (72oF) temperature differential to develop across the header ligaments.
During the initial portion of the startup, prior to the quench events, the final superheater
outlet temperature rise rates were approximately 1.7oC/min (3oF/min) though overshoots of
temperature to 640oC (1184oF) could occur on the superheater platens. The temperature rise
rates at outlet of platen superheaters was significantly higher.
• Condition monitoring programs (with extensive use of thermocouples and rare uses of strain
gages) required a staff detail of at least three persons, 100 stop/starts, and three months to one
year to identify, correct, and verify the correction of damaging stop/start thermal transients.
These efforts had the benefit of allowing faster stop/start transients without incurring
excessive damage. For example, steam turbine startups in a 680MW unit after 6 to 15 hrs
offline could be achieved within 1 hr of commencing the boiler purge and 45 minutes after
firing the boiler. Steam turbine cold starts in a 400 MW unit were improved from 8 hrs to 2.5
hrs. A 400MW steam turbine which previously was cooled for rotor inspection in 200 hrs by
natural cooling was safely force cooled in 60 hrs by the use of an online fatigue life
monitoring system.

1.8 Historical Perspective – HRSGs, “Lessons Still to be Learned”

The historical perspective given in Section 1.7 for conventional plants is illuminating. It would
appear that modern HRSGs are experiencing pressure part thermal transient issues as were
already experienced, studied and, in many cases, corrected in conventional fossil-fuel-fired
plants. There are certainly significant differences between the design and operation of HRSGs
compared with conventional plants but the fundamental thermal-mechanical damage issues and
solutions are annoyingly similar. Below is a summary of some of the salient service experiences
for HRSGs(1):
• A 20 m (65-ft) high by 8 m (25-1/2-ft) wide tube module will expand approximately 150 mm
(6-inch) vertically and 60 mm (2-1/2-inches) horizontally during heating from 20oC (68oF) to
540oC (1004oF).

1-7
EPRI Proprietary Licensed Material

Introduction

• Header borehole ligament cracking hasn’t been observed in HRSGs but very few inspections
of HP superheater header or HP drum boreholes have been performed for HRSGs that have
experienced 300+ stop/start cycles and upwards of about 50,000 fired hours. On the other
hand numerous examples of HRSG HP superheater and reheater headers being rapidly
quenched by condensate up to 200oC (360oF) below the header temperature by forward flow
of undrainable condensate during startups and also by condensation during CT shutdown at
shutdowns have been documented. Many HRSGs have HP superheater headers with wall
thickness greater than 40 mm (1.6-inch). Whereas 40mm was considered to be an acceptable
header thickness for cycling operation on conventional boilers, unlike HRSG lower headers,
conventional boilers are not subjected to condensate run off from pendant tube loops during
shutdowns, which are potentially more damaging than forward migration of undrained
condensate during startups. Thus for cycling service the higher temperature headers of
HRSG HP superheaters and reheaters need to be limited to not more than about 25mm (1.0
inch). Because the majority of installed HRSGs have headers significantly thicker than
25mm (1.0 inch), there is every reason to believe that borehole cracking will become a
serious issue in headers that experience creep conditions when on load after these units
accumulate 300 to 1000+ stop/start cycles.
• Whereas superheater and reheater tubes in most conventional boilers designed after the 1960s
have relatively flexible tube arrangements capable of accommodating large tube temperature
variations, the tube and header arrangements of large horizontal gas path HRSGs are much
stiffer and are particularly vulnerable to fatigue damage from cyclic occurrence of even
moderate tube-to-tube temperature differences, as evidenced by the very premature tube
failures at attachment welds to headers on many HRSGs within 2 to 300 starts.
• It is still rare for HRSG procurement specifications to include requirements for the plant to
be capable of a set of specified stop/start cycles. As a result many HRSGs have been
designed to ASME Section I which has no fatigue design requirements. Some HRSGs
supplied for installations specified for cycling service are little different than the HRSGs the
same supplier provides for base load installations. Many HRSGs have HP superheater outlet
headers greater than 50-mm (2-inch) thick. More recently a few HRSGs that have been
designed for cycling have radical design changes to eliminate excessive forces and moments
at the tube to header connections and a significant reduction in header thickness to about
25mm or less. These will accommodate both the compressive stresses developed by the
significantly faster steam temperature ramp rates required during startups of HRSGs than are
required on conventional boilers and limit the magnitude of the peak tensile thermal stress
during the condensate quenching part of the thermal cycle at shutdowns. Comparative data
on the influence of HP steam temperature ramp rates and header thickness on HP superheater
outlet header cyclic life expenditure is provided in Reference 1.
• Heat up rates for HRSGs is a subject of much controversy. Very often, the methods used
have been improperly applied and/or are non-conservative. HRSG startup design curves
often indicate that, except for any required steam turbine hold, hot, warm and cold starts can
be achieved in less than 1 hour. Such aggressive starts from cold and even warm initial
HRSG conditions may lead to premature creep-fatigue cracking at the intersection of the
tubeholes and bore of HP superheater headers, particularly when the header thickness is
significantly greater than 25-mm (1-inch). Commissioned plants tend to have 2 hr hot starts,
3 hr warm starts, and 4+ hr cold starts. Some combined cycle plants with added thermal

1-8
EPRI Proprietary Licensed Material

Introduction

transient monitoring instrumentation on the HP drum, HP superheater outlet header and


steam turbine components use startup rates that are somewhat shorter than these values, but
have not operated with enough stop/start cycles to know if the HRSG pressure parts are
experiencing undesirable rates of fatigue damage.
• On-line diagnostic/troubleshooting monitoring programs, with extensive use of
thermocouples of HRSG pressure parts are beginning to become more common. The more
extensive monitoring programs have installed 500 or more tube/header/downcomer/drum
metal temperature measuring thermocouples. For these extensively monitored HRSGs it has
required monitoring of approximately 100 stop/starts, and three months to more than a year
of service to identify and correct damaging stop/start thermal transients. The benefits of these
efforts are just being realized.
• Although a few studies have reported a benefit of allowing faster stop/start transients without
incurring excessive damage, the primary benefit of diagnostic/troubleshooting monitoring
has been to identify and allow correction of very severe and undesirable pressure part thermal
transients. Furthermore the methods used to predict damage expenditure rates may be
unsound and/or non-conservative.
• The opportunities to optimize the overall startup and shutdown rates while simultaneously
avoiding excessive fatigue damage have been studied and in selected cases have been
implemented but have yet to be proven with longterm field experience.
– For example, in a triple pressure, horizontal gas flow, vertical tube combined cycle unit
with a 150 MW gas turbine, cyclic damage problems were identified within the first
60,000 hr/200 starts/100 trips of service. After initial assessment of the damage
susceptible HRSG components, thermocouples were added to selected HP superheater
outlet and LP economizer inlet tubes. Initial monitoring was performed for 4 months and
6 stop/start cycles. Subsequent analysis identified ramp rates as high as 80oC/min
(144oF/min) during startups. Many examples of temperature imbalance among
components and several quenching events of -240oC/min (-432oF/min) were discovered.
High cooldown rates associated with use of nitrogen purging were also identified.
Analyses of these transients performed in accordance to TRD 301, Annex 1 resulted in
changes to operating practices and some design alterations.
• Corrosion fatigue associated with low pH, oxygen saturated water has led to failures in a few
hundred stop/start cycles or less in many of the water-touched HRSG components
• Creep-fatigue associated with severe tube to header and/or tube to tube temperature
differences has led to numerous tube to header connection failures in both primary (lower
temperature) and secondary (higher temperature) sections of HP superheaters and reheaters
• Thermal-fatigue failures have occurred at tube to header or drain to header connections in all
the water and steam touched tube bundles
• Creep- and thermal-fatigue failures have occurred in the steam-touched high energy piping
due to poor flexibility, attemperator water quenching or poor design details.
• Valve spindle distortion (due to restrained thermal expansion) and valve seat debris buildup
and erosion have been experienced in HRSGs that cycle frequently.

1-9
EPRI Proprietary Licensed Material

Introduction

It should be apparent from a comparison of the conventional, fossil-fuel-fired and combined


cycle HRSG experiences that very similar fatigue problems are being experienced with closely
related root causes, except that premature fatigue failures are occurring even earlier on horizontal
gas path HRSGs than they did on conventional fossil-fired boilers. As HRSGs accumulate more
stop/start cycles it is anticipated that fatigue failures will become more of a problem unless new
units are designed or modified where practical and operated within appropriate fatigue
prevention design limits, maintained to ensure that operational capabilities are not diminished
due to leaking valves or similar issues, and/or enhanced using input from
diagnostic/troubleshooting monitoring measurements.

1.9 Organization of the Manual

The following topics are covered in this manual:

Chapter 2 – “Roadmap Approach” provides a step by step process for the assessment and
correction of HRSG pressure part fatigue issues.

Chapter 3 – “Common Thermal-Mechanical Damage Mechanisms and Influencing Loads”


provide a catalog of the types of HRSG pressure part thermal mechanical loads that are of
concern, most which have produced multiple failures in modern triple pressure HRSGs.

This chapter is followed by five chapters that discuss the use of fatigue assessments for the
primary life cycle phases of HRSG:

Chapter 4 – “Fatigue Assessments – Owner Procurement Specification”

Chapter 5 – “Fatigue Assessments – Design”

Chapter 6 – “Fatigue Assessments – Commissioning”

Chapter 7 - “Fatigue Assessments – Operation”

Chapter 8 – “Fatigue Assessments – Maintenance”

The manual also includes six appendixes. These address:

Appendix A – “Review of Fatigue Assessment Factors”, which lists the numerous steps and
approaches to fatigue assessment and their relevance to the evaluation of HRSG pressure parts.

Appendix B – “Review of Fatigue Design Codes and Standards”, which provides a summary and
comparison of the design codes and standards that are relevant to HRSG pressure part fatigue
assessments.

Appendix C – “Estimating Flexibility- Tube to Tube Temperature Difference Influenced Forces


and Moments”, which provides a detailed set of closed form hand calculations for estimating the

1-10
EPRI Proprietary Licensed Material

Introduction

tube to header forces, moments and nominal axial and bending stresses resulting from row to row
or tube to tube temperature differences within individual tube bundles (harps).

Appendix D – “Example Fatigue Analysis of an LP Economizer (Preheater) Harp with Row to


Row and Tube to Tube Temperature Differences”, which provides a detailed step by step fatigue
assessment for off-design conditions and then compares the estimated life with the design basis
analysis.

Appendix E – “Estimating Borehole Thermal Shock Limits”, which provides a detailed example
of a thermal shock analysis performed using the TRD 301, Annex 1 rules.

Appendix F – “Monotonic and Cyclic Properties of Common HRSG Pressure Part Materials”,
which provides a catalog of properties for carbon steel, Grade 22 (2-1/4 Cr-1Mo) steel and Grade
91 (9Cr-1Mo-V-Cb) steel.

1-11
EPRI Proprietary Licensed Material

2
ROADMAP APPRAOCH

2.1 General Steps for Assessing and Correcting Pressure Part Fatigue
Damage Issues

Fatigue assessment of HRSG pressure parts is performed in the following steps:


1. Owner to specify that HRSG and key auxiliary system designs capable of causing
uncontrollable transients with large tube temperature differences during startups will be
rejected.
2. Owner to review proposed detailed design of HRSG and key auxiliary systems prior to
contract award to ensure that all sources of uncontrollable transients capable of causing large
temperature differences have been eliminated.
3. Identify components and design features of concern
4. Identify the potential loads of concern
5. Identify the thermal-mechanical parameters that control the loads
6. Measure or estimate the load-controlling parameters for all anticipated operating stop/start
cycles
7. Perform thermal-mechanical stress analyses for the each component/design feature/operating
cycles of concern
8. Make judgments on the most probable potential fatigue damage mechanism (i.e., corrosion
fatigue, thermal-fatigue, creep-fatigue)
9. Estimate the annual and lifetime fatigue life consumption for each component/design
feature/operating cycle of concern
10. Use this knowledge to alter operating procedures/thermal-mechanical parameter limits or
design features to reduce calculated peak thermal-mechanical stresses such that the fatigue
life consumption, unit dispatch and environmental restrictions are, as closely as practical, all
achieved. The objective is to produce optimal dispatch capabilities, if feasible, while
simultaneously meeting emissions requirements and achieving fatigue life consumption
levels that will ensure that the HRSG pressure parts will perform reliably and safely
throughout the lifetime of the unit. Often there are conflicts between ideal dispatch
requirements, permissible air permit out-of-compliance times and fatigue life expenditure
rates that require one or more of these ideals to be compromised.
11. Verify, through additional diagnostic/troubleshooting monitoring and unit operation, that the
alterations or corrective actions have been successful

2-1
EPRI Proprietary Licensed Material

Roadmap Appraoch

2.2 Life Cycle Actions to Prevent Pressure Part Fatigue Failures

Thermal-mechanical fatigue damage is irreversible and its expenditure must be minimized from
first firing of the CT/HRSG to conserve life expenditure to meet the longer term planned service
life of the unit. Implementation of modifications to correct design features that cause high life
expenditure rates is often either impractical or very costly in time and material costs once the
HRSG has been fabricated.

Avoiding thermal-mechanical damage in HRSG pressure parts requires diligence throughout all
the life cycle stages of the unit. Some of the key actions required to prevent fatigue damage are
documented in the following roadmaps. The owner specification and design roadmap steps are
discussed in Sections 4 and 5 respectively.

Owner Specification

1. Define each type of anticipated operational transient


2. Define the yearly and total lifetime number of each type of operational transient
3. Specify what design calculations and methods, design limits and reports are required
4. Specify for transient conditions that may occur during startups and shutdowns the tube
to tube temperature differences to be used as inputs to specified design verification
calculations
5. Specify what special damage monitoring instrumentation is required
6. Specify special damage monitoring tests and acceptance criteria to be performed
during commissioning
7. Specify what damage monitoring instrumentation, controls, data acquisition systems
and data evaluation algorithms are to be used throughout the lifetime of the HRSG

Design

1. Owner to specify that HRSG and key auxiliary system designs with the potential to
cause uncontrollable transients with large tube-to-tube or row-to-row temperature
differences during startups or shutdowns will be rejected.
2. Prior to contract award, Owner to review in detail the proposed design of HRSG and
key critical auxiliary systems and verify that there are no features or details that may
cause large tube-to-tube and/or tube row-to-row temperature differences during upset
transients at startups and/or shutdowns.
3. Verify that the assembly/components/feature/construction attributes and operating
parameters meet acceptable design limits (documented with design calculations and
reports)

2-2
EPRI Proprietary Licensed Material

Roadmap Appraoch

4. Set the allowable component operating limits compatible with the specified cyclic life
and calculated life expenditure rates in 1. above (e.g., fluid temperature/pressure ramp
rates and holds during shutdowns and startups, tube-to-header or tube-to-tube
temperature differences, inner to mid wall temperature difference in drums and headers)
5. Prescribe the operating procedures and permissibles to ensure that the component
operating limits are not exceeded
6. Create a list of important damage avoidance features (e.g., weld preparations, fit-ups,
tolerances, alignments, surface finish, weld procedures, heat treatments, weld profile,
weld quality, base metal quality, bend ovality limits, etc.) and the required QC/QA tests
and limits to be implemented during manufacture/fabrication to ensure that these features
are obtained
7. Specify the design and manufacture/fabrication requirements of special damage
monitoring instrumentation and operating tests to be performed during commissioning
8. Specify the design and construction requirements of damage monitoring
instrumentation, controls, data acquisition systems and data evaluation algorithms to be
used to monitor component damage throughout the lifetime of the HRSG

Construction

1. Verify that key, life limiting design/construction features (e.g., weld preparations, fit-
ups, tolerances, alignments, surface finish, weld procedures, pre-heat, post weld heat
treatment, other heat treatments, weld profile, weld quality, base metal quality, bend
ovality limits, etc.) are within their acceptable limits
2. Install special diagnostic monitoring instrumentation

Commissioning

1. Measure and evaluate (using the special diagnostic instrumentation) the key damage-
influencing thermal mechanical parameters (e.g., pressures, temperatures, pressure and
temperature ramp rates, key temperature gradients such as tube-to-tube, tube-to-header,
inner to mid wall temperature difference in drums and headers) throughout the
anticipated range of operating transients and operating procedures
2. Modify or tune the design and/or operating procedures as required to meet the limits of
the key damage-influencing thermal mechanical parameters

2-3
EPRI Proprietary Licensed Material

Roadmap Appraoch

Operation

1. Operate within the allowable limits of the damage-influencing parameters


2. Modify and tune operational practices as required to meet the damage-influencing
parameter limits

Maintenance

1. Maintain equipment (e.g., drain valves, start up vent valves, attemperator spraywater
control and block valves, HP bypass pressure control and spray water control and
isolation valves, all HP steam and feedwater isolation valves and other control/stop
valves) to ensure that unanticipated operational transients caused by malfunctioning or
worn equipment/control systems are prevented
2. Monitor and analyze the key damage-influencing factors throughout the lifetime of the
HRSG and set inspection intervals based on estimated life consumption values
3. Perform visual and nondestructive evaluations to verify that the severity of anticipated
damage has not been exceeded and to detect unanticipated damage (e.g., bowed tubes,
damage associated with water hammers or inadequate clearances, corrosion, broken or
misaligned gas baffles, etc.)
4. Repair or replace components or component features prior to failure
5. Identify opportunities to optimize the life of the HRSG components

2.3 Assessment of HRSG Pressure Part Fatigue Failures

If the fatigue prevention actions have not been successful and a pressure part failure has occurred
then a detailed assessment to determine its cause and develop engineered corrective action
should be performed. Figure 2-1 provides an example of some of the steps that might be included
in the assessment. Each situation is different and parameters that influence failures may be
readily available in some cases or may require considerable effort to obtain in other cases.
Experience has indicated that corrective actions made prior to fully understanding the root cause
and influence factors that produced the fatigue failure are frequently unsuccessful and often lead
to repeat failures or failures in other components, often after shorter service periods than the
initial failure.

2-4
EPRI Proprietary Licensed Material

Roadmap Appraoch

Figure 2-1
Analysis steps to follow if a fatigue failure has already occurred

2-5
EPRI Proprietary Licensed Material

3
COMMON THERMAL-MECHANICAL DAMAGE
MECHANISMS AND INFLUENCING LOADS

3.1 HRSG Pressure Part Damage Mechanisms

The pressure part failure modes that are of primary concern and are addressed in various design
codes are:
• Ductile rupture
• Creep rupture
• Brittle fracture
• Excessive elastic deformation including elastic instability
• Excessive plastic deformation including plastic instability and incremental collapse
• Creep buckling
• Fatigue (Thermal-, Corrosion-, Creep-, Vibration-Induced)
• Stress Corrosion Cracking
• Metal wastage or corrosion

The key damage mechanisms that are predominantly controlled by transient thermal mechanical
loads are:
• Thermal quench induced fracture (ductile rupture)
• Thermal-fatigue
• Thermal heating and cooling shocks; e.g., at attemperator outlets
• Creep-fatigue
• Corrosion fatigue

These damage mechanisms have commonly occurred in HRSGs at or near:


• Tube and pipe to header or drum connections
• Nozzle to header or drum
• Pipe tees and nozzle to header connections downstream of attemperators
• Drain line to header connections

3-1
EPRI Proprietary Licensed Material

Common Thermal-Mechanical Damage Mechanisms and Influencing Loads

• Tube to header upstream and downstream of attemperators


• At 180-degree return bends welds
• High transient stress locations within high energy piping systems, manifolds, or nozzles
• Fatigue cracking within the drum or header boreholes has been predicted will occur as units
accumulate more stop/start cycles.

3.2 Thermal-Mechanical Loads & Key Damage Influencing Parameters

The thermal-mechanical loads that have been identified as being responsible for HRSG pressure
part fatigue failures include:
• Pressure changes, especially the consequential changes in saturation temperature. Heat
transfer to steam-containing pressure parts by either condensation heating during initial
heating, and pressure-raising or condensate quenching by condensate migration forward
during startups or by condensation run off from tubes after CT shutdown. These all develop
very high temperature gradients and thermal stresses at the inner surfaces that are transiently
“wetted” by condensate. Quenching at tube holes by sub-cooled condensate is nucleate
boiling with potentially very high heat transfer rates.
• Fluid temperature changes leading to local through-wall or axial temperature gradients
• Nonuniform temperatures (e.g., between adjacent tubes attached to the same header, between
tubes and headers, etc.) in combination with restriction to the thermal expansions
• Load transfer between cold and hot conditions leading to unloading of some tube supports
and transfer of their load to other tubes
• Changes in external loads (e.g., piping thermal expansion, dead weight loads applied to
headers/manifolds or humping of pipes/headers/drums caused by condensate pooled along
the bottom of hotter, almost horizontal pipes)
• Transient temperature differences in evaporators during initial startup between tubes and
downcomers with relatively stiff interconnecting pipe arrangements
• Steam or water hammer, often associated with undrained condensate pooled along the bottom
of long, inadequately drained pipes, which consolidates into large plugs of water when steam
flow is initiated
• Forced vibrations

Some of the common damage influencing parameters and loading modes that have produced
through-wall HRSG pressure part failures are summarized in Tables 3.1 through 3.10. There are
undoubtedly other loading modes responsible for pressure part damage and numerous subtleties
for each of the loading modes depicted here, however these simplified, somewhat generic
depictions should provide the reader with a good starting point with regard to understanding the
attributes that must be considered and controlled by more enlightened design and operation to
avoid fatigue damage.

3-2
EPRI Proprietary Licensed Material

Common Thermal-Mechanical Damage Mechanisms and Influencing Loads

Table 3-1
Schematic of the Type 1 Thermal-Mechanical Loads

Type of Loading: Type 1 - Header/ drum borehole stress


Description of Stress: Local through-wall and/or across ligament header or drum temperature gradients

Damage Mechanisms: Thermal-, creep- or corrosion- fatigue

Locations:
1. Drum or header tube/borehole surface or adjacent inside diameter surface

Key factors influencing damage:


1. Step changes in fluid temperature across ligaments between individual tubes
2. Fluid temperature ramp rates in individual tubes
3. Through-wall temperature gradient
4. Saturation temperature changes caused by changes in operating pressure
5. Header thickness and diameter
6. Tube hole through header wall - configuration/spacing/ chamfer
7. Type of tube to header connection design

Comments:
1. Transients that cause through-wall temperature gradients are often referred to as thermal cooling
or heating shocks
2. Local thermal shock of header or drum penetrations result when fluid temperatures are hotter or
cooler than the header or drum mid-thickness (mean wall) temperature. Large step changes in
temperature and very high heat transfer rates (nucleate boiling) in tube holes and header/drum
bore during condensate migration or introduction of subcooled attemperator spray water are
possible
3. Both header thickness and header diameter significantly influence peak wall temperature
gradients because heating and cooling of header wall is partly conduction from tubes and the
surface of tubeholes that have higher heat transfer from fluid than at the bore of the header

3-3
EPRI Proprietary Licensed Material

Common Thermal-Mechanical Damage Mechanisms and Influencing Loads

Table 3-2
Type 2 Thermal-Mechanical Loads

Type of Loading: Type 2 - Tube/drain line to header discontinuity stress


Description of Stress: Tube/drain line to header temperature differences. Local discontinuity stresses
are caused by differential heating or cooling of the tube relative to the header. For example, when cooler
fluid such as condensate or attemperator spraywater is introduced into a hot tube/header the tube will cool
and contract far more rapidly than the header, producing a local tensile stress that will be highest at the
toe of the tube to header reinforcing fillet weld. There is also a discontinuity pressure stress that results
from the difference in section properties between the tube and header.

Damage Mechanisms: Thermal-, creep- or corrosion- fatigue

Location:
1. Weld toe on tube side of tube to header connection.
2. With water-touched components and poorly controlled water chemistry multiple parallel corrosion
fatigue cracks may initiate at the ID surface in same approximate location
Key factors influencing damage:
1. Fluid or tube metal temperature ramp rates or step changes in individual tubes
2. Difference between the tube and header mid-wall temperatures
3. Operating pressure
4. Tube and header thickness
5. Connection design
6. Weld toe angle
Comments:
1. The primary discontinuity stress is produced by thermal expansion mismatches caused by a
combination of transient differences between the tube wall and bulk mean header wall
temperature.
2. Local discontinuity stresses are caused by differential heating or cooling of the tube relative to the
header. For example, when cooler fluid such as condensate or unflashed attemperator spraywater
is introduced into a hot tube/header the tube will cool and contract more rapidly than the header,
producing a local tensile stress that will be highest at the toe of the tube to header reinforcing fillet
weld.
3. There is also a discontinuity pressure stress that results from the difference in section properties
between the tube and header.
4. Finite element thermal stress analysis of a generic set-in tube to header connection (right hand
figures) subjected to a rapid thermal downshock and high heat transfer coefficients indicated that
with this connection design the thermal gradient in the vicinity of the tube-to-header weld is
minimal and produces an insignificant weld toe stress even though the end of the tube contracts
away from the header tubehole surface during the thermal quenching event. Under this simulated

3-4
EPRI Proprietary Licensed Material

Common Thermal-Mechanical Damage Mechanisms and Influencing Loads

condition very high tubehole surface stresses are calculated to occur. The calculated tube-to-
header weld toe stresses are very low. The weld toe stresses are low because very rapid
conduction of heat occurs across the tube to header weld region, thus maintaining the local
differential expansion of this portion of the tube and header and minimizing the weld toe stresses.

3-5
EPRI Proprietary Licensed Material

Common Thermal-Mechanical Damage Mechanisms and Influencing Loads

Table 3-3
Type 3 Thermal-Mechanical Loads

Type of Loading: Header/drum humping (bowing) and downcomer/riser pipe to drum or tube-to-header
discontinuity stress
Description of source of stress:
1. Top to bottom header or drum temperature differences
2. Pipe to drum temperature differences
Damage Mechanisms: Thermal-, creep- or corrosion- fatigue
Location:
1. Weld toe on the pipe side of branch to drum shell connection
2. With poorly controlled water chemistry, multiple parallel corrosion fatigue cracks may initiate at the
ID surface in the same approximate location
Key factors influencing damage:
1. Differences in the heat transfer rate and drum metal temperature changes on the water-touched
and saturated steam-touched surfaces
2. Amount of offset of top headers and risers from drum center-line and resulting bending moment at
individual riser connections to drum
3. Fluid or riser pipe metal temperature ramp rates in individual risers
4. Difference between the riser pipe and drum mid wall temperatures
5. Operating pressure
6. Riser pipe and drum thickness
7. Connection design, and surface roughness
8. Weld toe angle
Comments:
1. This loading mode involves drum humping (bowing) due to differences in heat transfer rates within
the watertouched and saturated steam touched surfaces of the drum and local discontinuity stress
at the reinforced downcomer or riser pipe nozzle to drum connections.
2. The intensification of the local pressure and thermal stresses at these connections are controlled
by the design (e.g., magnitude of offset pipe bending moment, flanged/set-on/set-in/set-through
branch configuration, hole diameter, wall thicknesses, wall thickness transition angles, borehole
chamfer, weld toe radius, complete joint penetration v. full strength-partial joint penetration weld
configuration, etc.) and fabrication (i.e., weld discontinuities, out-of-roundness, surface roughness
machined condition, etc.) characteristics. The magnitudes of the local stresses are controlled by
the operating characteristics (i.e., temperature changes, temperature gradients, temperature ramp
rates, relationship between fluid pressure and the thermal transients, etc.).

3-6
EPRI Proprietary Licensed Material

Common Thermal-Mechanical Damage Mechanisms and Influencing Loads

Table 3-4
Type 4A Thermal-Mechanical Loads

Type of Loading: Tube row-to-row temperature differences

Description of Stress: Constrained thermal expansion of tube rows connected to common lower and upper
headers

Damage Mechanisms: Thermal-fatigue, creep-fatigue or thermal quench induced fracture in superheaters


and reheaters. Thermal-fatigue or corrosion fatigue in evaporators, economizers or preheaters

Location:
1. Weld toe on tube side of the tube to header connection.
2. With water-touched components and poorly controlled water chemistry multiple parallel corrosion
fatigue cracks may initiate at the ID surface in same approximate location or a short distance from
the inlet of the tube where corrodents deposit.
Key factors influencing damage:
1. Difference in tube row metal temperatures
2. Assembly design features (i.e., multiple tube rows connected to the same top and bottom
headers)
3. Tube design features (i.e., dogleg bends, tube dimensions)
4. Tube to header connection design features (i.e., set-on/set in, complete joint penetration [CJP] v.
partial joint penetration [PJP] welds, weld toe profile)

Comments:
1. This loading mode is caused by global tube row-to-row thermal expansion differences
2. Tube row temperature differences can be large during startups and/or shutdowns
3. This loading mode is unique to “pant leg” tube design (i.e., two or more parallel rows connected to
common upper and lower headers in a single pass harp configuration)
4. The tube to header forces/moments/stresses are intensified by poor tube-to-header connection
design/fabrication and/or “dog leg” bends at either header.

3-7
EPRI Proprietary Licensed Material

Common Thermal-Mechanical Damage Mechanisms and Influencing Loads

Table 3-5
Type 4B Thermal-Mechanical Loads

Type of Loading: Tube-to-tube temperature differences within any row of tubes


Description of Stress: Constrained thermal expansion of tube rows connected to common lower and upper
headers
Damage Mechanisms: Thermal-fatigue, creep-fatigue or thermal quench induced fracture in superheaters
and reheaters. Thermal-fatigue or corrosion fatigue in evaporators, economizers or preheaters
Location:
1. Weld toe on tube side of the tube to header connection.
2. With water-touched components and poorly controlled water chemistry multiple parallel corrosion
fatigue cracks may initiate at the ID surface in same approximate location

Key factors influencing damage:


1. Difference in tube row metal temperatures
2. Assembly design features (i.e., multiple tube rows connected to the same top and bottom
headers)
3. Tube design features (i.e., dogleg bends, tube dimensions)
4. Tube to header connection design features (i.e., set-on/set in, CJP/PJP welds, weld toe profile)
5. Large transient differences in individual tube-to tube temperatures caused by condensate
migration during startups or overspraying at attemperators
6. Efficacy of the HP superheater and reheater drains system design and operation
Comments:
1. This loading mode is caused by global tube-to-tube thermal expansion differences within any row
of tubes or by row to row tube temperature differences
2. Transient tube temperature differences can be large during startups and/or shutdowns
3. This loading mode can occur in “pant leg” tube design (i.e., two or more parallel rows connected to
common upper and lower headers in a single pass harp configuration) or in single row, single or
multiple pass configurations
4. In harps with pant leg configurations a combination of Type 4A and 4B loading is likely to occur
5. The tube to header forces/moments/stresses are intensified by poor tube-to-header connection
design/fabrication and/or “dog leg” bends at either header

3-8
EPRI Proprietary Licensed Material

Common Thermal-Mechanical Damage Mechanisms and Influencing Loads

6. In superheaters and reheaters (left hand schematic) severe tube-to-tube temperature variations
have been associated with introduction of unflashed attemperator spraywater in the tube element
in the projected area of the inlet/outlet nozzles, forward flow of undrained condensate or from
backflow of drain condensate
7. In economizers (right hand schematic) severe tube-to-tube temperature variations have been
associated with introduction of cold feedwater at low flow rates into tubes near the inlet nozzle,
especially with top fed harp designs.
8. Designs with multipass configurations with header partition plates tend to exacerbate the
likelihood of cracking associated with these thermal transients

3-9
EPRI Proprietary Licensed Material

Common Thermal-Mechanical Damage Mechanisms and Influencing Loads

Table 3-6
Type 5A Thermal-Mechanical Loads

Type of Loading: Tube row-to-row temperature differences in modules comprising multiple pass tube row
assemblies

Description of Stress: Differential thermal expansion of economizer/ preheater, superheater or reheater


tube rows between cold and normal hot conditions that in some types of economizer/preheater and
superheater design transfers the tube support gravity load from the hotter to cooler tubes

Damage Mechanisms:
1. Thermal- or corrosion-fatigue in economizers or preheaters.
2. Thermal- or creep-fatigue in some designs of superheaters and reheaters with fixed supports

Location:
1. Weld toe on tube side of tube to header connection
2. With poorly controlled water chemistry multiple parallel corrosion fatigue cracks may initiate at the
ID surface in same approximate location
3. At return bend butt weld connections
4. On the intrados of the return bends
Key factors influencing damage:
1. Difference in tube row metal temperatures
2. Harp design features (i.e., multiple tube rows connected to a common header, 180-degree upper
return bends, rigid or flexible header supports)
3. Tube design features (i.e., dogleg bends, tube dimensions)
4. Tube to header connection design features (i.e., set-on/set in, CJP/PJP welds, weld toe angle)

Comments:
1. Differential thermal expansion of economizer/preheater tube rows between cold and hot normal
operating conditions that in this type of economizer/preheater design with rigid supports transfers
the tube support gravity loads from hotter to cooler tubes.
2. Deformation of tubes in the bundle occurs when hot if constraints prevent the outlet header from
lifting when the tubes are hot or prevent the bottom of the tube bundle from deflecting towards the
stack end.
3. It is likely that a combination of Type 5A and 5B loads will be active.
4. This loading mode has also produced stress corrosion cracks at the intrados of upper return
bends

3-10
EPRI Proprietary Licensed Material

Common Thermal-Mechanical Damage Mechanisms and Influencing Loads

Table 3-7
Type 5B Thermal-Mechanical Loads

Type of Loading: Tube-to-tube temperature differences in modules comprising multiple tube row
assemblies

Description of Stress: Differential thermal expansion of economizer/ preheater, superheater or reheater


tube rows between cold and normal hot conditions that in some types of economizer/preheater design
transfers the tube support gravity load from the hotter to cooler tubes

Damage Mechanisms:
1. Thermal- or corrosion-fatigue in economizers or preheaters.
2. Thermal-, or creep-fatigue in some designs of superheaters and reheaters with fixed supports

Location:
1. Weld toe on tube side of tube to header connection
2. With poorly controlled water chemistry multiple parallel corrosion fatigue cracks may initiate at the
ID surface in same approximate location
3. At return bend butt weld connections
4. On the intrados of the return bends

Key factors influencing damage:


1. Difference in temperature of tubes in the same row
2. Assembly design features (i.e., multiple tube rows connected to a common header)
3. Tube design features (i.e., dogleg bends, tube dimensions)
4. Tube to header connection design features (i.e., set-on/set in, CJP/PJP welds, weld toe angle)

Comments:
1. Differential thermal expansion between cold and hot normal operating conditions in some designs
of HP superheater and economizer/preheater of tubes in the same row that in this type of design
economizer/preheater or HP superheater design lifts the hotter tubes off the support beam/hanger
tube beneath the intrados of the 180-degree return bends and transfers their share of the weight
of the tubes/water/lower headers to the coolest tubes in the same row.
2. It is likely that a combination of Type 5A and 5B loads will be active.
3. This loading mode has also produced stress corrosion cracks at the intrados of upper return
bends

3-11
EPRI Proprietary Licensed Material

Common Thermal-Mechanical Damage Mechanisms and Influencing Loads

Table 3-8
Type 6 Thermal-Mechanical Loads

Type of Loading: Manifold to header pipe nozzle temperature differences or pipe nozzle to header
gradients

Description of Stress: Pipe nozzle to header temperature differences and/or constrained thermal
expansion of multiple pipe nozzles connected to a common manifold and header

Damage Mechanisms:
1. Thermal-, or creep-fatigue in superheaters and reheaters
2. Thermal- or corrosion-fatigue in evaporators, economizers or preheaters

Location:
1. At the root of pipe nozzle thickness transitions
2. OD surface of the dogleg pipe nozzle to lower manifold connection
3. Inside surface or outside surface of the pipe nozzle to header connection
Key factors influencing damage:
1. Design:
a) Flanged/set-on/set-in/set-through branch configuration
b) Hole diameter
c) Wall thicknesses
d) Wall thickness transition configuration
e) Borehole chamfer
f) Weld toe profile
g) CJP v. full strength PJP weld configuration, etc.
2. Fabrication:
a) Weld discontinuities
b) Out-of-roundness
c) Surface roughness
d) Machined condition
3. Operating characteristics:
a) Temperature changes
b) Temperature gradients
c) Temperature ramp rates
d) Relationship between fluid pressure and the thermal transient

3-12
EPRI Proprietary Licensed Material

Common Thermal-Mechanical Damage Mechanisms and Influencing Loads

Comments:
1. Through-wall nozzle pipe failures have occurred in HRSG pipe nozzles when the nozzles had poor
thickness transitions (i.e., at Grade 22 to Grade 91 alloy transitions) and were repeatedly
subjected to forward flow of condensate and/or interstage attemperator overspray quench events.
2. Humping of the manifold caused by condensate pooled along the bottom of the hot pipe has
probably contributed to the loading mechanism
3. The FEA images above do not have thickness or alloy transitions in the nozzles.
4. In absence of the thickness transitions the most likely locations for fatigue cracking to occur within
the pipe nozzles under repeated quenching events are: (i) the OD surface of the dogleg pipe
nozzle to lower manifold connection, (ii) the inside surface of the side of the pipe nozzle to header
connection, or (iii) the outside surface of the corner of the pipe nozzle to header connection.
5. If attemperator spraywater is poorly distributed within a group of manifold-to-header nozzles then
the nozzle-to-manifold connections in those nozzles which are most severely quenched and which
have dogleg bends will probably have the highest risk of cracking.

3-13
EPRI Proprietary Licensed Material

Common Thermal-Mechanical Damage Mechanisms and Influencing Loads

Table 3-9
Type 7 Thermal-Mechanical Loads

Type of Loading: Tube row-to-row alloy differences

Description of Stress: Constrained thermal expansion due to row-to-row tube alloy differences

Damage Mechanisms: Thermal-fatigue, creep-fatigue in superheaters and reheaters

Location:
1. Weld toe on tube side of the tube to header connection
Key factors influencing damage:
1. Different alloys and thermal expansion coefficients of tube rows within a harp
2. Difference in tube row metal temperatures
3. Harp design features (i.e., configuration of tube rows, header supports)
4. Tube design features (i.e., dogleg bends, tube dimensions)
5. Tube to header connection design features (i.e., set-on/set in, CJP/PJP welds, weld toe profile)

Comments:
1. The thermal conductivity of T22 (2-1/4Cr-1Mo) is approximately 5 to 50 percent higher than T91
(9Cr-1Mo-V-Nb) with the largest differences occurring at low temperatures
2. The thermal expansion coefficient of T22 (2-1/4Cr-1Mo) is approximately 10 to 18 percent higher
than T91 (9Cr-1Mo-V-Nb)
3. T22 tubing will heat up more rapidly (especially during initial heating) and expand to a greater
extent, for a given temperature change, than T91 tubing
4. T22 to T91 row-to-row alloy dependent differential thermal expansion produces tube-to-header
stress/strain magnitudes during cold starts that are similar to what would occur if both rows were
T91 material and row-to-row temperature differences of approximately 90oC (160oF) occurred

3-14
EPRI Proprietary Licensed Material

Common Thermal-Mechanical Damage Mechanisms and Influencing Loads

Table 3-10
Type 8 Thermal-Mechanical Loads

Type of Loading: Piping flexibility and/or severe thermal quenching of piping

Description of Stress: Constrained thermal expansion of piping, or local thermal shock of piping
Damage Mechanisms:
1. Thermal-, or creep-fatigue in superheaters and reheaters
2. Thermal- or corrosion-fatigue in evaporators, economizers or preheaters

Location:
1. Piping loads may cause tube-to-header connection cracking
2. At welded butt or tee joint pipe to pipe or pipe to header connections
3. Pipe parent material at a locations away from weld connections (near attemperators)
4. Rotation of headers causes bending stresses at tube connections to headers
Key factors influencing damage:
1. Inadequate piping flexibility
2. Weld connection design/fabrication
3. Severe quenching events

Comments:
1. Steam or water hammer piping loads may also cause unanticipated loads to be placed on
headers/tubes/pipes
2. For the case shown in the schematic the pipe expansion loads resulted in tube-to-header
connection cracking.
3. Cracking incidents have been experienced within the piping runs themselves.
4. HRSG pipe fatigue cracking has been experienced mainly at welded butt or tee joint connections
but in a least one incident produced through-wall cracks in the pipe parent material at a location
well away from any welds.

3-15
EPRI Proprietary Licensed Material

4
FATIGUE ASSESSMENTS – OWNER PROCUREMENT
SPECIFICATION

The owner specification roadmap includes the following steps:

1. Define each type of anticipated operational transient

2. Define the yearly and total lifetime number of each type of operational transient

3. Specify what design calculations and methods, design limits and reports are required

4. Specify for transient conditions that may occur during startups and shutdowns the tube to
tube temperature differences to be used as inputs to specified design verification calculations

5. Specify what special damage monitoring instrumentation are required

6. Specify special damage monitoring tests and acceptance criteria to be performed during
commissioning

7. Specify what damage monitoring instrumentation, controls, data acquisition systems and data
evaluation algorithms are to be used throughout the lifetime of the HRSGThese steps will
each be discussed in the following sections.

There are a wide variety of objectives that lead to the requirement of performing fatigue
calculations of HRSG pressure parts. The primary objective of HRSG designers’ use of design
fatigue calculations should be to verify by a conservative calculation method acceptable to the
owner that each proposed HRSG pressure part’s design is compatible with the cyclic life
specified by the owner when acted on by the anticipated design-basis thermal/mechanical loads.

In addition to the requirement of verifying that the design meets some minimum code
requirements, HRSG owners should also want to utilize the fatigue analyses to:

1. Determine the tolerance of the design to transient, off-design operating conditions that have
been identified in various technical papers and guidelines as typical for certain HRSG design
features.

2. Develop optimal operating practices for shutdowns and startups from different initial HRSG
conditions that minimize the startup times and emissions. This needs to be done while
ensuring that HRSG pressure part fatigue failures will not occur during the anticipated life of

4-1
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

the unit while operating within practical limits minimizing startup durations and enabling the
unit to meet air permit out-of-compliance time limits.

3. Develop control charts for the fatigue damage influencing parameters (e.g., fluid pressure
and temperature ramp rates, controllable tube-to-tube temperature differences) within limits
that will ensure that fatigue failures will not occur during the anticipated life of the unit.
Examples of target and action limit control charts are presented in Figures 5-4 and 5-5

4. Set maintenance and condition assessment examination schedules

5. Determine the high priority locations which should be included in diagnostic monitoring
efforts and scheduled condition assessments

6. Make run/repair/replace decisions with regard to thermal-mechanical, corrosion-, or creep-


fatigue damage within the HRSG pressure parts.

It should be apparent that the design basis objectives and the HRSG owner/operators objectives
are generally not fully compatible, nor are all the listed owners’ objectives obtained from a
normal code fatigue assessment. To achieve the owners’ objective requires the preparation of a
detailed HRSG procurement specification with special items included that address what the
owner has determined will be the design-basis duty cycle of the HRSG throughout its lifetime
and what special fatigue calculations are required from the designer. These procurement
specification requirements are addressed in the remainder of this chapter.

4.1 Specifying Definitions of Anticipated Operating Transients

The first step in the owner specification roadmap is the specification of the characteristics of
anticipated major operating transients. Owners sometimes balk at specifying the planned lifetime
numbers of starts from different operating conditions because the operating regime beyond a few
years is uncertain. The probability for most fossil generating plants is that the economic benefits
for being able to cycle will increase substantially above current expectations in the medium and
long term particularly on gas fired installations where escalation of fuel gas cost is likely to
exceed current expectations. Thus, it is prudent that owners’ specify more frequent cycling starts
than current conservative estimates indicate when establishing the lifetime number of starts from
different initial conditions to be specified for the design of new installations.

There is no industry consensus regarding the definition of cold starts, warm starts, hot starts, load
changes, normal cooldowns, maintenance or spin cooldowns, startup trips, on-load trips/rapid
restarts, etc. Although the condition of a steam turbine or of a conventional fossil fired boiler
(which have substantial stored energy in the water and metal of the evaporator system and can be
boxed up tight on the gas side) following a normal cycling service shutdown can be predicted
reasonably closely as a function of the off-load hours, this is not the case with HRSGs. They
have less stored energy and lose heat from the stack when the the CT is at crank speed and from
leakage through many isolation points on the HP section, especially through large startup vent
and HP bypass control valves. On cycled units, leakage from eroded HP superheater and
reheater drain valves, blowdown isolation valves, etc. increase as seat damage increases each

4-2
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

startup-shutdown cycle. For example, a new HRSG that was capable of maintaining a few bars
of pressure in the HP drum after more than 50 hours off load decayed to atmospheric pressure in
less than 18 hours after less than 100 starts.

A rational method of setting definitions of the operating transients is to first determine what
hours are likely to correspond to each of the anticipated operating transients. An example is
given in Table 4-1. For the HRSG, the classification of type of start should be based on the HP
drum pressure immediately prior to commencing the start, since metal temperatures in the HP
superheater headers cool close to the HP saturation temperature. Although the importance of
maintaining the HP drum pressure as high as possible during the shutdown is widely recognized,
in practice the rate of decay of drum pressure during the off-load period varies widely. Critical to
the preservation of drum pressure is the maintenance of tight shutoff at all points of HP steam
and feedwater isolation, including HP drains, vents, blowdowns, steam and feedwater isolations,
HP bypass, etc. It is also important to avoid intermittent drum topping throughout the off-load
period, which accelerates pressure decay and is also damaging to the preheater.

Table 4-1
Suggested Characteristics of Selected HRSG Operating Transients

Cold starts: made from an initial pressure of 0 barg (0 psig) and drum contents or drum wall
temperature less than 75oC (167oF). Likely to follow an off-load time following a normal
shutdown of more than 56 hrs. An initial drum water temperature of 5oC (41oF) should be used
in design life calculations for cold start-normal shutdown cycles.

Cool starts: made from initial conditions of drum water temperature above about 75oC (167oF)
up to an initial drum pressure of about 4 barg (58 psig) made from 0 barg (0 psig) pressure, but
where the HP drum water temperature remains. Cool starts are likely following a recommended
normal shutdown and weekend off-load from Friday evening until Monday morning. An initial
drum temperature of 75oC (167oF) is recommended for use in design life calculations for cool
start-normal shutdown cycles.

Warm starts: made from an initial HP drum pressure prior to the startup of between
approximately 4 barg (58 psig) to 32 barg (464 psig) , (drum saturation temperature between
152oC (306oF) and 239oC (462oF). Provided the HRSG is boxed up after the shutdown with high
HP drum pressure between 75% and 80% of design pressure and all HP section steam and
feedwater isolation valves are maintained leak tight, warm starts are likely following a
recommended normal shutdown and off-load from Saturday evening until until Monday morning
(about 36 hours off-load). However some HRSGs if shutdown with lower HP pressure when
boxed up, or when HP valves have been poorly maintained, may have difficulty maintaining HP
drum pressure above the lower threshold for warm starts of about 4 barg after 12 hours off-load.
For design verification purposes, an initial drum and HP superheater outlet header temperature
of 152oC (306oF) is recommended for design life calculations for warm start-normal shutdown
cycles. However, if operation experience with the unit indicates that consistently higher
pressures than 4 barg can be obtained for all warm starts, then a higher pressure can be used
for recalculation of design life expenditure per warm start-normal shutdown cycle

Hot starts: made from an initial pressure in excess of 32 barg (464 psig); HP drum saturation
temperature in excess of 239oC (462oF). Provided the HRSG is boxed up after the shutdown
with high HP drum pressure between 75% and 80% of design pressure and all HP section
steam and feedwater isolation valves are maintained leak tight, hot starts are likely following
normal shutdowns for off-load periods of up to 12 hours, possibly longer. However, if HP drum
pressure is significantly lower when boxed up after shutdown, or HP steam and feedwater
isolation is poor, then even 8 hour shutdowns may not maintain HP pressure above 32 barg. In

4-3
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

which case overnight shutdowns have to be followed by warm starts. For design verification
purposes, an initial drum and HP superheater outlet header temperature of 239oC (462oF) is
recommended for design life calculations for hot start -normal shutdown cycles.

Trip (Combustion turbine (CT), Steam turbine (ST), or HRSG induced) followed by restart within
2 hours: The trip shutdown should assume that condensation occurs in HP superheater tubes
during the CT coastdown and substantial quantities of condensate at or below saturation
temperature quench cool the tubeholes in lower headers with very high heat transfer rates.
Trips cause substantially higher thermal mechanical stress at the intersection of tubeholes and
the bore of lower headers than is the case for shutdowns performed in accordance with the
Recommended Shutdown Procedure described below.

Combustion turbine load changes. Combustion turbine characteristics must be clearly defined
and agreed by the owner and the combustion turbine supplier before contract, including:
1. Specifying the lowest possible combustion turbine exhaust gas temperature with the
turbine at its minimum load
2. Specifying combustion turbine controls to incorporate the means for the operator/DCS
auto controls to select exhaust gas temperature versus time characteristics and
deloading prior to shutdowns; also during startups.

Supplementary burner firing. The specific loading conditions must be clearly defined by the
owner and manufacturer. It is especially important a) to specify the minimum CT load for
continuous stable operation of the unit together with the maximum duct firing heat input at the
minimum load and b) to ensure by heat balances that the attemperators can control HP
superheater and reheater outlet steam temperatures with not less than 28°C (50°F) of superheat
at the attemperator outlet. Two stages of desuperheating may be required to avoid overspraying
close to or below saturation temperature at attemperator outlets.
“Maintenance” shutdown: It is recommended that prior to maintenance, during deloading of the
CT/HRSG from baseload that HP pressure is lowered in accordance with the maximum rate of
change of HP saturation temperature recommended by the HRSG OEM. A somewhat higher
rate of change of HP saturation temperature may be permissible but thermal mechanical
analysis is required to calculate the cyclic life expenditure per maintenance shutdown-cold start
cycle. The specific loading conditions must be clearly defined by the owner and manufacturer.
This may involve controlled lowering of HP pressure during the deloading and shutdown of the
CT followed by natural cooling (with the combustion turbine restricted only to crank speed). It
should be noted that rapid depressurization of HRSGs is potentially very damaging. Forced
cooling of the HRSG by spin cooling the combustion turbine at purge speed is not
recommended since it is potentially very damaging to the HP drum and HP superheater
headers.
Recommended Shutdown Procedure: First, the HP steam pressure should be maintained at
between 75% and 80% of the design maximum operating pressure prior to the combustion
turbine shutdown. Secondly, the DCS should be programmed to provide a linearly reducing CT
exhaust gas temperature set point to the CT combustion controls that reduces the combustion
turbine exhaust gas temperature at a predetermined rate. This will gently cool the HP
superheater headers at a safe rate close to the HP saturation temperature to ensure
condensation during or after the combustion turbine shutdown can’t quench cool the lower
headers. The specific loading conditions must be clearly defined by the owner and
manufacturer.

Alternative shutdown procedure usually used at the plant. The specific loading conditions must
be clearly defined by the owner and manufacturer. At some plants this may be similar to the
Recommended Shutdown Procedure, above. In other plants this may shutdown the unit with low
HP pressure so that the HRSG restart will be a cold, cool or warm start which takes significantly
longer and is still more damaging to the HRSG than a hot start. Some plants also deload the

4-4
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

combustion turbine from about 50% load rapidly prior to immediate shutdown of the combustion
turbine. This leaves the HP superheater and reheater headers at high temperature and
vulnerable to severe quench cooling when condensation commences at some point during or
shortly after the combustion turbine shutdown or when attemperation spraywater leakage occurs
past block valves whenever the feedwater pump is run.

It is important that the assumed pressure, temperature and fluid ramp rate values be realistic or
somewhat conservative. For example Table 4-2 lists the lowest anticipated HP drum water
temperature and pressure for four types of startup transients. Figure 4-1 graphically presents
these time/HP drum pressure/HP drum water temperature relationship along with some measured
pressure decay values in three relatively new triple pressure HRSGs. However, HP pressure
decay rates vary substantially from installation to installation and over time in a given unit as
they are strongly influenced by the operating and maintenance practices at the plant. HRSGs
capable of achieving the relatively slow HP pressure decay rates similar to those in Figure 4-1,
after less than 12 months with significantly less than 100 starts could not hold HP pressure above
atmospheric pressure longer than 16 hours due to leakage past many HP steam and feedwater
section isolating valves.

The owner must predict the lifetime number of CT/HRSG startups following scheduled
shutdowns overnight, after weekend shutdowns and for discrete longer periods that are to be
used in fatigue evaluations. It is then crucially important that the method of combined cycle
shutdown is established, since this will determine the temperature and pressure conditions in the
HRSG at the end of the natural cooldown period offload. If not performed correctly as
recommended above, the shutdown will cause severe condensate quenching damage in lower
headers of the HP superheater. Despite the seemingly obvious importance of establishing an
appropriate shutdown procedure for each of the combined cycle components, few installations go
into operation with a thoughtfully engineered normal shutdown procedure.

To determine the conditions in the HRSG after each of the different offload periods to be
assumed for fatigue evaluations, it is necessary to specify the HP pressure versus time
characteristic to be assumed for determining the HP pressure prior to each category of startup.
The pressure decay rate specified for the analysis should conservatively allow for degradation of
isolation in the HP and feedwater isolation valves over the period of cycling service between
planned outage opportunities to replace or restore tight shutoff at all HP isolation points. It is rare
that any of the foregoing logical and essential steps towards establishing the basis for
determining the initial conditions at startups after different anticipated shutdowns and offload
periods are followed.

To simplify analysis and err to the side of conservatism, it may be desirable to specify that the
hot start conditions be used for the design analysis of all hot and rapid starts. The extended warm
start condition could be specified for all anticipated warm and extended warm starts. The cold
start conditions could be specified for all anticipated cool and cold starts.

4-5
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

Figure 4-1
HP drum pressure decay curves for three relatively new triple-pressure HRSGs with a
design pressure of 2074 psig (143 barg). The times and minimum HP drum pressure and
water temperature values prior to restarts associated with rapid hot, hot, warm, extended
warm, cool and cold starts are shown. Based on experience in similar units it is
anticipated that the pressure decay response will degrade (become faster) as the unit ages
unless all the pressure blocking valves are well maintained and/or are designed as sets of
master/martyr valves.

Table 4-2
Design-Basis HP Drum Pressures and Water Temperatures Prior to Various Operating
Startups

Type of Startup Off-line Lowest HP drum pressure Lowest HP drum water


time, hrs at beginning of start temperature at beginning of
start

Hot < 12+ 32 barg (464 psig) 239oC (462oF)

Warm 8 to 36 4 barg (58 psig) 152oC (306oF)

Cool 24 to 60+ 0 barg (0 psig) 75oC (167oF)

Cold >72 0 barg (0 psig) 5oC (41oF)

4-6
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

Another aspect of the specification of the operating transient forecast that must be carefully
prepared is the proper coupling together of shutdown and startup “transients” to define complete
operating “cycles”. These are two very important points: firstly the crucial importance of
“designing” and specifying the method of shutdown, (since this determines the HRSG conditions
prior to the next startup), and secondly that unless correctly performed, shutdowns are potentially
significantly more damaging than the startup part of each thermal cycle. Despite this emphasis it
is rare for owners’ specifications to require “engineered shutdown” methods even in those owner
specifications that do define different types of startup.

A common and substantially non-conservative error that is often observed in design-basis fatigue
assessments is that of neglecting to model the complete operating cycles. For some critical
components such as the thicker high temperature headers, the cooling part of the cycle develops
similar or even greater stresses of opposite sign to those developed in the same critical pressure
part and location during prudently cautious startups. Thus the practice of basing fatigue damage
evaluations on just the heating part of the thermal cycle during startup underestimates the fatigue
damage by a factor approaching 10 or more.

The correct practice to avoid this seemingly perpetual and serious fatigue assesment error is
firstly to specify the requirement for designing the normal method of combined cycle shutdown.
Secondly there is a need to define each type of operating transients forecast to be evaluated by
always referring to complete operating cycles. For example: 500 low load/load return + 100
normal shutdown/hot starts + 200 normal shutdown/warm starts + 10 normal shutdown/cold start
+ 5 trip/rapid restart + 2 maintenance shutdown/cold start cycles per year.

One final important point that requires emphasizing seems to be the confusion or controversy
within the industry as to what thermal-mechanical loads have to be taken into account during
design verification of fatigue analyses. In the first place, as discussed in the introduction, the
owner and HRSG design/construct team should specifically exclude HRSG design features
which introduce the risk of uncontrolled upsets during startups with the potential for large
uncontrolled thermal transients. For example, the severe thermal downshock events caused by
migration of undrainable condensate or by gross overspraying or leakage of spraywater at
desuperheater outlets that have been measured in numerous HP superheater and reheater tube
bundles and have caused many tube and piping failures.

On horizontal gas path HRSGs, the consequential tube-to-tube temperature differences caused by
these upsets during startups will probably exceed the maximum compatible with the required
cyclic life. Regarding steam pressure and temperature ramp rates and holds, the HRSG supplier
should provide, with his tender proposal, prelimnary proposals for startup curves from different
initial HRSG conditions and for normal shutdowns. These should be based on his preliminary
fatigue assessment and should be compatible with the specified number of startup-shutdown
cycles. The owner should utilize the preliminary HRSG startup and shutdown requirement
conditions to develop coordinated combined cycle startup and shutdown curves that integrate the
preliminary requirements of HRSG, steam and gas turbines and also air permit out-of-
compliance limits. The later curves should then be used by the HRSG designer for the
comprehensive fatigue evaluation of all HRSG pressure part components judged to have
potentially limited cyclic life.

4-7
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

It is recommended that the owner specifies in detail the scope of required HRSG fatigue
evaluations, including:

1. Verification by the HRSG designer that HP drum and the top and bottom headers of the HP
superheater and reheater have acceptable fatigue life when operated with the owner’s
intended startup curves

2. Determination by the HRSG designer of maximum permissible tube-to-tube and tube-to-


header temperature differences for each section of the HRSG, based on reasonable
assumptions such as that two quench events caused by attemperator spray will occur during
each stop/start cycle.

3. The owner could also require that the designer provide design and operating procedures that
will meet the estimated tube-to-tube and tube-to-header temperature difference limits. If the
HRSG designer’s fatigue analysis is based on tube-to-tube temperature differences that are
unrealistically small, there is often no low-cost action that can be taken to correct the error
once the unit has been installed.

4. The owner should require the HRSG designer to provide data measured on HRSGs of similar
design to demonstrate that the design-basis predictions of transient tube temperature
anomalies are realistic before the design is accepted by the owner and fabrication
commenced.

5. Similar assessment requirements could be specified for each tube bundle after the likely
loading modes and underlying damage influencing factors that have produced service failures
of pressure parts are compiled. EPRI report “Delivering High Reliability Heat Recovery
Steam Generators” [2] provides numerous examples of the thermal-mechanical factors that
have produced HRSG pressure part fatigue failures.

4.2 Specifying the Annual and Lifetime Operating Transient History

The 2nd step in the owner specification roadmap is the specification of the number and types of
transient events such as stop/starts or load changes to be experienced over the design life of the
HRSG. Table 4-3 provides alternative examples of operating transient forecasts that might be
appropriate, one for a unit anticipated to be predominantly base loaded, the other for a unit
anticipated to be cycled for an average of 8 months per year. It is strongly recommended that
owners predict and specify numbers of operating cycles that contain a prudent margin to account
for the impact of continuing rapid rate of increase in gas price relative to other fuels used for
power generation.

4-8
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

Table 4-3
Sample Unit Number of Operating Cycles for Design Purposes

Predicted Operating Regime Predominantly Baseload with Regular Cycling Service for
continuous operation until an average of 8 months per
later years of life year

Transient Type Average Cycles in 30 Average Cycles in 30


Annual Year Life Annual Year Life

Cold start from ambient temperature 5 150 5 150


after normal shutdown

“Cool” HRSG start following 6 to 7 200 33 1,000


recommended normal shutdown,
and off-load from Friday evening
until Monday morning
Warm start after normal shutdown, 8 to 9 250 10 300
and off-load for 36 hours

Hot starts following normal 16 to 17 500 167 5,000


shutdown, and offload overnight

Trips followed by restart within 2 8 to 9 250 8 to 9 250


hours

Forced cool shutdowns followed by a 5 150 5 150


cold start from ambient

Total shutdown/startup cycles for 1,500 6,850


design

If the owner neglects to specify conservatively predicted numbers of each type of transient then
only in respect to some of the fatigue failure load mechanism do the design codes provide
guidance, albeit with inherent conservatism for components that operate below about 400°C
(752°F). However, for components that operate at higher temperature the design codes may be
non-conservative in respect to both creep-fatigue as well as to evaluation of some thermal
mechanical loading mechanisms, such as tube row-to-row and tube-to-tube transient temperature
differences, pipe humping, etc, that have been the root cause loading mechanisms for most of the
many tube and pipe attachment weld failures experienced in HRSGs to date.

For example, if the owner doesn’t specify the design numbers of cycles, then EN12952-3,
para.13.1.2 requires the designer to assume that 2,000 shutdown + cold startup cycles will occur
and in this circumstance also sets the maximum allowable fatigue design usage factor, Df, to 0.4.
The fatigue design usage factor, Df, is defined as follows:

4-9
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

n
⎛n⎞
Df = ∑⎜ ⎟ Equation 4-1
i =1 ⎝ N ⎠i

where:
Df = fatigue usage factor or life consumption fraction
n = total number of design cycles of a given type of transient such as a normal shutdown
followed by a cold startup
N = design allowable number of cycles for the effective stress range and temperature
estimated for the cycle

Thus a fatigue life consumption by 2,000 shutdown + cold starts of 0.4 at end of design life is
equivalent to a fatigue life expenditure of 1.0 after 5,000 shutdown + cold start. For those lower
temperature components and thermal mechanical loading mechanisms assessed this is
conservative because shutdowns + warm or (especially) hot starts are significantly less damaging
than a normal shutdown + cold start cycle. Therefore the HRSG should be able to perform many
more than 5000 starts when the majority will be hot or warm without experiencing fatigue
failures.

However, for higher temperature components which are subjected to creep-fatigue, and for all
those transient loading mechanisms caused by upset conditions, (which are the predominant
source of the many HRSG tube and pipe failures experienced to date), neither EN12952-3, nor
any other code or guideline discussed in this report provide adequate guidance. Consequently
they are almost invariably excluded from fatigue assessments by designers and owners.

When the design operating types and numbers of transients are specified by the owner, then
EN12952-3 allows the maximum allowable fatigue design usage factor, Df , to be increased to a
value of less than or equal to 1.0.

TRD 301, Annex 1 indicates that if the owner only specifies the number of shutdown/ cold
startup transients then the total number of cycles shall be assumed to be 5 times the number of
cold starts to “ensure adequate reserves for startups from warm.” This may or may not be a
conservative assumption. It is far better that the owner develops and specifies a somewhat
conservative forecast of the types and number of all anticipated transients expected annually and
for the plant lifetime.

From the discussion above it should be obvious that it is desirable for the owner to specify more
than just shutdown/cold startup cycles. There are however some additional important issues to be
addressed by the owner when specifying the design-basis cyclic history.

The owner should also be aware of and understand the ramifications and potentially non-
conservative limitations of the fatigue exemption rules that are included in the codes or
standards. These may be used by the designer to justify that no detailed fatigue assessment is
necessary. For example, EN12952-3, para. 5.5 states that “boiler components are deemed to be
exposed to cyclic loading if the boiler was designed for more than 500 cold startups.”

4-10
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

Many of the numerous tube and pipe failures at attachment welds in HRSGs have occurred in
significantly fewer shutdown + cold startup cycles (many of them hot and warm starts that are
less damaging than cold starts) than 500. This highlights one of the inadequacy of such
simplified exclusion rules. Unfortunately the definition of cold starts as it relates to this section
of this standard is not clearly defined. It would be worthwhile for the owner to include in their
specification that “With respect to EN12952-3, para. 5.5 all starts and load changes, where the
pressure change exceeds 50% of the maximum allowable working pressure (MAWP), shall be
assumed to be cold starts.” Better still, the owner should specify that a detailed fatigue analysis
be performed irrespective of any fatigue exemption rules.

Additional fatigue analysis exemption rules are included in EN12952-3, para 13.3.4. In this
section of the standard, the designer is exempted from performing a fatigue analysis if all of six
conditions are met. The two conditions that relate to the design basis operating transient forecast
are: (i) the number of load changes where the pressure exceeds 50% of the maximum operational
pressure must be less than 3,000 and (ii) the number of operational transients where the pressure
changes less than 50% of the maximum operational pressure must be less than 10,000.

Defining fatigue analysis exemption rules by reference to pressure changes is fundamentally


unsound. It is likely to lead to omission of major thermal mechanical loading mechanisms from
fatigue life evaluations because the dominant source of inelastic stresses that cause localized
yielding and fatigue damage are thermal stresses. These are caused by through wall temperature
gradients in thicker walled pressure parts, or by tube-to-tube or tube-to-header temperature
differences or by pipe reactions and moments caused by top to bottom temperature differences in
pipes.

Pressure stresses are generally limited to below the elastic limit by design codes and are
significant contributors to stress range when peak thermal stresses are also tensile. If the owner
wants to be certain that a reliable fatigue analysis is performed then it is important to either
specify more than 3000 shutdown/startup design cycles that are clearly specified to have pressure
changes that are greater than 50% of the MAWP or, even better, to state that the exclusion rules
shall not be used.

It is recommended that owners specify that, regardless of exemptions in codes and guidelines
adopted for fatigue analysis, the HRSG designer is required to perform a fatigue assessment for
each of the specified numbers and types of shutdown-startup operational cycles. It is strongly
recommended that before the HRSG supply contract is awarded that the detailed scope of fatigue
analysis and life assessments should be agreed and incorporated in the contract documents. This
should include identification of which components will be assessed, the methodology to be used
for the assessment, the input assumptions to be used for boundary conditions during startups and
shutdowns including ramp rates, hold points, and transient temperature differences during
predictable upset conditions. Furthermore the specification should require the fatigue analysis to
be completed by the designer and accepted by the owner before materials are procured or
fabrication commences.

Some of the design fatigue assessments specified in the codes may be overly conservative (in
which case they overly prohibit the designers’ choices or overly restrict ramp rates during

4-11
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

shutdowns and/or startups). Other code assessments may be non-conservative when used
beyond the circumstances envisaged when they were first empirically formulated (in which case
the design fatigue analysis may significantly underestimate the fatigue life expenditure damage
rates or lead to aggressive, damaging startup and/or shutdown methods).

By insisting on a fatigue analysis that evaluates the elastic stress and total strain range for each
cycle, the owner is then able to optimize the startup and shutdown durations. For example, by
trading off a higher damage per cycle for the comparatively small number of cold starts in order
to comply with air permit requirements out-of-compliance time limits, or by utilizing a slightly
different hot start method to reduce the damage increment per hot start/shutdown procedure.
Furthermore the owner could use a longer, less damaging shutdown procedure to permit a
somewhat more rapid startup.

Optimization of startup and shutdown procedures to meet air permit limits or to minimize
operating or maintenance costs cannot be performed without data, which defines the influence of
alternative methods for startups from different initial conditions or different methods of
shutdown on the cyclic life expenditure rates for the critical HP superheater outlet header and
manifold and HP drum parts.

The following example will illustrate these issues and a meaningful way of addressing the
fatigue exemption rule issues. Table 4-4 provides an example of an operating history for an HP
drum with a maximum allowable operational pressure of 15.3 MPa (2219 psig). The total
number of anticipated cold starts over the forecast 30 year life is 90. If this table were provided
in the owner specification with no further guidance then the designer might invoke EN12952-3,
para. 5.5, and conclude that since there are less than 500 cold starts no fatigue evaluation is
required.

Table 4-4
History of Actual Transients During the Early Years of Operation of Two Triple Pressure
HRSGs

Transient Shutdown + Cold Shutdown + Shutdown + Hot Load Change


Start Warm Start Start

Number per year 3 40 15 300

Maximum drum 13.25 MPa (1921 13.25 MPa (1921 13.25 MPa (1921 13.25 MPa (1921
pressure during psig) psig) psig) psig)
transient

Minimum drum 0 0.4 MPa 3.8 MPa 8.9 MPa


pressure during
transient (58 psig) (551 psig) (1291 psig)

4-12
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

A more complete and somewhat, but not overly conservative, operating cycle forecast will
provide better input to the designer and will result in a more meaningful fatigue assessment.
Table 4-5 provides an example of the operating cycle forecast that might be specified for units
expected to operate in a similar fashion to the unit shown in Table 4-4. In the development of
Table 4-5 it was assumed that as the plant ages that it may become more difficult to maintain
pressure above ambient, thus the unit will experience cool starts instead of hot starts. It was also
assumed that the units may be required to load swing more than twice a day frequently and may
experience far more hot starts than the historical trend from a similar but relatively new unit
illustrates. The objective in developing this design-basis operating cycle forecast is not to
overburden the designer with excessive cycling requirements but to provide a realistic but
somewhat conservative forecast of the anticipated duty of the unit.

If on the other hand, the owner clarifies the definition of cold starts as described above then all
the transients with a pressure change greater than 7.65 MPa (1109 psig) would be counted as
cold starts. For the information in Table 4-5 all the operational transients except the load changes
would be counted as design basis cold starts (i.e., the design basis cold starts = 75 +1000 + 375 =
1450). Since this exceeds 500 the designer would assume that the HP drum is “exposed to cyclic
loading.”

A quick check of the required HP drum water temperature ramp rate, vt, at startup can be made
using the conservative lower bound allowable ramp rate estimation given in EN12952-3:2001,
paragraph 5.5:

⎧ ⎡α x d ⎤⎫ Z
vt = ⎨550 N / mm 2 − po ⎢ m m − 0.5⎥ ⎬ 2 Equation 4-2
⎩ ⎣ 2 x ems ⎦ ⎭ ems

where:
vt = drum water temperature ramp rate, oC/sec. If the estimated value of vt is greater than the
expected ramp rate and is positive then no further fatigue evaluation is required.
po = maximum allowable operating pressure, MPa
αm = upper bound stress concentration factor for the pressure stress (assumed to be 4 or may be
calculated for specific cylindrical or spherical shells stress concentration factor solutions)
dm = mean diameter of the drum, mm
ems = minimum wall thickness of the shell (drum), mm
Z= thermal stress factor including an upper bound stress concentration factor for the thermal
stress
550 N/mm2 = a worst case minimum allowable alternating stress range.

4-13
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

Table 4-5
Example Specified Design Basis Transients for the Two HRSGs Documented in Table 4-4

Initial HP Drum Pressure and Number of Cycles for Regular


Water Temperature Cycling Service for an
average of 8 months per year

Transient Type Minimum Minimum Average Cycles in 30


pressure temperature Annual Year Life

Cold start from ambient temperature 0 MPa 5oC 5 150


after normal shutdown
(0 psig) (41oF)

“Cool” HRSG start following 0 MPa 75oC (167oF) 35 1050


recommended normal shutdown,
and off-load from Friday evening (0 psig)
until Monday morning

Warm start after normal shutdown, 0.05 MPa (7 111oC 10 300


and off-load for 36 hours psig) (232oF)

Hot starts following normal 3.5 MPa 243oC 167 5,010


shutdown, and off-load overnight (500 psig) (470oF)

Load change 8.9 MPa 303oC 720 21,600


(1291 psig) (577oF)

Trips followed by restart within 2 8.3 MPa 289oC 9 270


hours (1200 psig) (569oF)

Forced cool shutdowns followed by a 0 MPa 5oC (41oF) 5 150


cold start from ambient
(0 psig)

Total number of cycles for design 28,530


Notes:
1 MPa = 1 N/mm2 = 10 bar
MAWP = 15.3 MPa (2219 psig)
Maximum HP drum pressure during transient = 13.25 MPa (1921 psig)
Maximum HP drum water temperature during transient = 333oC (631oF)
Fatigue exemption rules shall not be used
Designer to provide:
Allowable HP drum pressure/water temperature ramp rate limits for shutdown and startup portions of
each operating cycle including the duration and combustion turbine exhaust conditions and HP drum
water conditions for required holds
Allowable HP superheater headers pressure/steam temperature ramp rate limits for shutdown and
startup portions of each operating cycle including the duration and combustion turbine exhaust
conditions and HP superheater steam conditions for required holds
Maximum allowable tube-to-tube in same row, tube row to tube row, and tube-to-header temperature
differences for every tube bundle (assuming two events of equal magnitude but opposite sign occur
during every stop/start cycle) and a description of the key operating variables that must be controlled
to ensure these limits are achieved.

4-14
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

For the HP drum example the value of vt is estimated to be:

⎧ ⎡ 4 x1784mm ⎤ ⎫ 2o C mm 4 / Ns
vt = ⎨550 N / mm 2 − 15.3 N / mm2 ⎢ − 0.5⎥ ⎬ 2
= 0.0076o C / s (0.46o C / min)
⎩ ⎣ 2 x106mm ⎦ ⎭ 106mm

To meet these conservative requirements would require approximately 713 minutes (~12 hrs) to
heat the HP drum from 5oC to 333oC. This is unlikely to be an acceptable time for cold starts.
This illustrates the importance of the design/fabrication details used since this simplified model
assumes worst case design/fabrication details. If these poor design/fabrication details were used
then the very slow startup times estimated would be required to avoid fatigue damage.

Graphical results obtained using this relationship for HP drums with a range of maximum
allowable working pressure (MAWP), mean diameters and wall thickness and for Grade 91 and
Grade 22 superheater outlet manifolds are presented in Figure 4-2. In the case of the HP drum
described above the designer would probably conclude that a more detailed, less conservative
fatigue assessment that takes credit for the more fatigue resistant design/fabrication details used
is required since anticipated drum water temperature ramp rates are likely to be in the range of
2oC/min to 3oC/min (3.6o to 5.4°F/min) to achieve cold startup times in not less than 3 hrs.

One aspect of startups that is often not adequately considered and evaluated is the early part of
the transient when very rapid fluid touched metal temperature changes occur that are not readily
estimated from the bulk pressure or temperature values. For example, when starting from
ambient conditions, the steam wetted surfaces of the HP drum and HP superheater are raised
very rapidly by condensation heating to 212°F when boiling commences, thereby developing an
almost instantaneous step change in the inner surface temperature before the bulk mean wall
temperature begins to increase. The ramp rates calculated by design codes assume quasi steady
state ramps. When step increases in surface temperatures occur during the inirial stage of
heating then it is important that HP pressure is held constant for for a predetermined period at the
lowest controllable pressure to allow time for conduction to raise the bulk mean wall temperature
and establish the desired quasi steady state temperature difference between inner suraface and
mean wall temperatures before continuing the pressure raising at the predetermined ramp rate.

4-15
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

Figure 4-2
Example calculations of EN12953-3 allowable fluid temperature change rates based on a
conservative, bounding (thermal plus pressure stress) thermal-fatigue assessment which
considers loading mode Type 1 (Table 3-1). This loading mode addresses the potential for
HP drum to downcomer or riser nozzle borehole cracking or HP superheater header
borehole cracking. Meeting these limits based on measured bulk fluid values does not
ensure that individual boreholes with much higher fluid ramp rates will not crack. Meeting
these limits also does not ensure that other loading modes described in Chapter 3 will not
occur.

The designer may determine if the fatigue analysis exemption rules in Section 13 of EN12952-3
could be met. Using the specified design basis operational transients (e.g., Table 4-5) the
designer would determine how many transients involve a pressure change greater than 50% of
the MAWP (i.e., 6,660 transients) and how many involve lesser pressure changes (i.e., 21,870).
These are more than the limiting values of 3,000 and 10,000 respectively so the HP drum would
not be exempt from a fatigue analysis. If on the other hand, less cycles were specified then an
additional four criteria would also need to be met to exempt the drum from a fatigue analysis.
The other exemption rules in EN 12952-3 that must all be met for the component to be exempt
from a fatigue assessment include:

1. The component has been designed for sustained pressure in accordance to the EN 12952-3
rules. This will most likely be true. If the sustained pressure design was performed with
ASME Section 1 then a comparison with the EN 12952-3 would be required, however the
ASME Section 1 sustained pressure design rules are generally more conservative than those
in EN 12952-3, which results in thicker components that are significantly more susceptible to
fatigue damage

4-16
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

2. The mechanical loadings on branches (e.g., downcomer and riser nozzles), are below a
limiting level. The mechanical loading on branches will be strongly influenced by the
assumptions on the severity of temperature differences between risers and downcomers
(especially external downcomers which are more susceptible to subcooling while offload) or
between tube rows or elements. Also by the severity of humping caused by drum mean wall
top-to-bottom temperature differences, which are significantly influenced by the rate of
saturation temperature change in the HP drum during pressure raising and also during
pressure lowering when flashing of steam from the water causes high heat transfer to the
water-wetted surface of the drum. Humping of the drum causes significant reactions and
bending moments at the connections of riser pipes to the drum, especially when the pipes are
offset from the bottom of the drum. Design basis assumptions on these temperature
differences have frequently been found to be non-conservative when compared to measured
values on operating units.

3. The transient temperature differences between two points within close proximity (see Figure
4-3) are less than two different limiting levels

Figure 4-3
Allowable Transient Temperature Differences between Two Points in Close Proximity

4-17
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

In addition to determining the maximum permissible through-wall, radial temperature gradient


adjacent to the header or bore borehole, most design codes also require that the axial, midwall
temperature gradient not exceed the same temperature difference over a characteristic distance
defined by the wall thickness and mean radius dimensions of the shell and branch portions of
connections.

Rather than leaving judgments regarding these exemption criteria to the designer, the owner is
advised to specify that fatigue exemption rules shall not be invoked. Specifying what analyses
shall be performed, what assumptions should be made and what information shall be provided
from the analyses is also recommended.

4.3 Specify What Design Calculations, Design Limits and Reports are
Required

All HRSG pressure parts will be subjected to sustained load design calculations. These
calculations are based on plastic instability and creep rupture limits. In addition, fatigue
calculations may be made of various components and features. There are numerous design codes
and standards available that can and are being used to check the fatigue design of HRSG pressure
parts (see Appendix A). Some of these codes and standards will provide some assurance that
those aspects of the design that are addressed by the code will be tolerant to the anticipated
operational transients. However, the owner will not receive assurances on those apects of the
design that are not addressed by the code unless the appropriate evaluations are precisely
specified in detail in the procurement specification.

Since no single code or standard adequately addresses all the relevant fatigue design issues it is
currently recommended that, with regard to fatigue assessments, the items in Table 4-6 be
included in procurement specifications for the HRSG pressure parts. It is suggested that the
HRSG purchaser use this table to develop a list of components/design features that they will
include in the HRSG procurement specification with regard to required fatigue analyses and
analysis reports.

It is also recommended that the purchaser specify what information is required in the fatigue
analysis reports. Often only partial results of analyses are included in these stress/fatigue analysis
reports. For example, in a flexibility analysis it is common practice to only report the locations
and magnitudes of the combined code stresses in the few locations with the highest stresses. It is
important to require that the reports include the input parameters including all assumed boundary
conditions and the estimated forces, moments, displacements, and individual pressure, thermal
expansion stresses and combined code stress values for all node points in the flexibility model.
This will provide more comprehensive information for selecting high priority condition
assessment locations and will provide a useful basis for comparing in-service as-found
conditions with the design-basis assumptions and estimations.

A common source of underestimation of peak thermal-mechanical stresses and gross


overestimation of fatigue life is the use of incorrect boundary conditions. No significant extra
effort is required by the designer to provide this more comprehensive information.

4-18
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

It is common practice for design analyses to be performed with simplified models. For example,
flexibility analyses to determine the loads applied to the headers for a three row module may be
modeled with 1 row of tubes with no bends and with the modulus of elasticity of the tubes
increased three fold to simulate the stiffness of three tube rows. Alternatively for an analysis of
tube stresses in a bundle with multiple passes and partitioned headers the designer may choose to
only include the tube elements at the ends of the headers and on both sides of the partition plates
in the model.

Although this and other similar design stress analysis practices might be acceptable for
determining if the design meets minimum requirements of design codes, they provide very little
useful additional information to the purchaser. Such features might include the peak localized
thermal-mechanical stresses required for determination of fatigue life at tube and pipe
attachments to headers. To overcome this deficiency the purchaser should require that all tube
rows and elements be included in the model.

With some HRSG tube and header arrangements it is neceassry to include more than one tube
harp and the interconnecting pipes in an elastic model. This extra modeling effort will require
only minimal extra time for the designer and will provide an excellent as-designed baseline
model that can be readily modified and run again if off-design tube temperature differences are
identified during commissioning or commercial operation.

It is not sufficient to merely ensure that the designer performs fatigue analysis for predictable
conditions associated with major transients. Such events might include startups from different
initial conditions and normal controlled shutdowns, during which the gas and fluid conditions
imposed on the pressure parts can be controlled to reliably reproduce closely similar fluid
conditions and thus thermal gradients in thicker pressure parts for each of these types of event.
Most of the many very premature fatigue-related failures experienced on horizontal gas path
HRSGs have been caused by more severe thermal transients. These have been caused by upset
conditions that have been completely overlooked by the designers and are difficult to quantify
for the purposes of fatigue analyses without installing thermocouples on many tubes.

More recently, some enlightened owners have also required that the designer demonstrate with
appropriately sophisticated fatigue analysis that the design is tolerant to temperature differences.
These are additional to those that can be predicted from proposed startup and shutdown
characteristics, but have been detected and quantified using diagnostic thermocouples attached to
many tubes in HRSGs that have similar design features. Examples of the type of events that the
HRSG specification could require an assessment of by the designer (to verify that the fatigue life
and planned cyclic design life are compatible) include, but should not be limited to:
• Tube row-to-row temperatures where panels have more than one row of tubes with the same
headers. Published tube temperature measurements indicate row-to-row tube temperature
differences of up to 80oC (150oF) in gas inlet panel of the HRSG during the CT prestart
purge;
• Absence of any of the essential constituent features of good design practice for HP
superheater and reheater drains systems is likely to result in forward migration of undrained
condensate during startups. Published data indicates that when forward migration occurs at

4-19
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

startups it causes temperature differences between tubes in the same row of final HP
superheaters of typically 80oC (150oF), and in the final reheater of 160 oC (300 oF) or even
greater.
• Tube-to-tube temperature differences of the order of 80oC (150oF) occur between tubes in the
feedwater inlet row of LP preheater and other economizer bundles during every startup.

Some purchasers have adopted the approach of requiring the designer to calculate the maximum
allowable combinations of tube-to-tube in individual row and row-to-row tube temperature
differences for the specified design number of startups, Then subsequently verifying these are
achieved with diagnostic thermocouples attached to carefully selected tubes during
commissioning. The problem with this approach is that if there is any deficiency in the design of
the drains system, or interconnecting pipes which do not have a positive slope to the condensate
drain pots, then transiently large tube-to-tube temperatures will occur. This results from the
forward migration of condensate when steam flow is initiated, or if when load ramping from
lower loads during startups overspraying occurs at HP superheater or reheater desuperheater
outlet. These are usually practically impossible to eliminate by design modifications because of
space restrictions and high cost.

For HPSH headers/manifolds, HP drums and greater than 25mm (1 inch) thick reheater
headers/manifolds the purchaser should require that an allowable temperature differential
diagram similar to the illustrative example shown in Figure 4-4 be included as part of the
analysis report.

4-20
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

Table 4-6
Selected HRSG pressure part components/design features that should be subjected to fatigue evaluations

Component Design features Loading Modes Fatigue analysis Comments


HP drum Riser and (Type 1) Borehole thermal The Type 1 loading can If the waterside oxide cracking check required
shock. be assessed with by EN 12952, Part 3, Section 13, Paragraph
downcomer to
EN 12952, Part 3, 13.4.3 indicates that the oxide may crack
drum connections (Type 2) Riser & downcomer
Section 13. A finite during any of the cycles then an additional
and boreholes pipe to drum temperature
element (FE) model of safety factor of 10 on cycles to cracking
differences.
the complete should be used to account for corrosion
(Type 3) Downcomer & riser evaporator circuit used fatigue. The results of the analysis should be
HP drum
reactions and moments applied in conjunction with the recommended drum pressure/water
feedwater inlet
by downcomer and riser pipes EN12952-3 will be temperature ramp rate curves for each of the
nozzle borehole
induced by humping of the required to assess the specified types of stop/start and low load
drum due to large top-to-bottom Type 2 and 3 loading cycles. If pressure, temperature holds during
temperature differences during modes. The FE model specific operating cycles are to be
pressure raising (caused by could also be used to recommended then the limits on the pressure,
higher heat transfer of the remove some of the water temperature ramp rates before and after
steam washed surfaces during conservatism inherent the hold and the time, temperature and
condensation heating) should in the EN12952-3 pressure of the hold should be provided.
be included in the fatigue borehole thermal shock
assessment stress estimates
Final HP Tube-to-header (Type 1) Borehole thermal The Type 1 loading can The results of the analysis should be
superheater boreholes, shock be assessed with recommended steam pressure, temperature
and final nozzle-to-header EN 12952, Part 3, ramp rate curves for each of the specified
(Type 2) Tube to header or
reheater inlet or manifold pipe Section 13. types of stop/start and low load cycles. If
drain line to manifold midwall
and outlet borehole, drain pressure/temperature holds during specific
temperature differences A finite element (FE)
headers and line to lower operating cycles are to be recommended then
model used in
for all lower manifold (Type 3) Header top-to-bottom the limits on the pressure/water temperature
conjunction with the
return headers boreholes temperature differences ramp rates before and after the hold and the
EN12952-3 will be
where these time/temperature/pressure of the hold should
(Type 6) Nozzle pipe to required to assess the
exceed 25 mm be provided. It is critically important to ensure
manifold or header midwall Type 2, 3 and 6 loading
(1-inch) that the analysis evaluates the maximum
temperature differences modes. The FE model
thickness stresses developed by the recommended
could also be used to
steam pressure/temperature ramp rate curves
remove some of the
for both the shutdown and startup phase of
conservatism inherent
each thermal cycle for each of the specified
in the EN12952-3
types of start/stop operating transients. The
borehole thermal shock

4-21
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

Component Design features Loading Modes Fatigue analysis Comments


stress estimates stress or strain range and mean stress values
used in the fatigue calculations must include
the combined shutdown and startup portions
of the each of the operating cycles.
For superheater or reheater bundles with
outlet temperatures above 426oC (800oF) an
additional safety margin of 10 on cycles to
cracking should be used to account for creep-
fatigue interaction effects.
Peak stresses caused by tube to header
temperature differences often occur
simultaneously with direct and bending
stresses caused by transient tube-to-tube and
row-to-row temperature differences. The
stresses are additive at some locations
Temperature gradient across the ligament
between tubeholes when tube-to-tube temp
differences are present for several minutes
may produce transiently higher stresses which
should be considered
HP Pipe to manifold (Type 1) Borehole thermal The Type 1 loading can HP superheater and reheater outlet manifolds
superheater bores intersection shock be assessed with are usually thicker and larger diameter then
and reheater EN 12952, Part 3, their respective outlet headers. Thus the
outlet manifold Section 13. outlet manifolds will probably establish the
pipe branches steam pressure and temperature ramp rate
curves. The results of the analysis should be
recommended steam pressure, temperature
ramp rate curves for each of the specified
types of stop/start and low load cycles. If
pressure/temperature holds during specific
operating cycles are to be recommended then
the limits on the pressure/water temperature
ramp rates before and after the hold and the
time/temperature/ pressure of the hold should
be provided. It is critically important to ensure
that the analysis evaluates the maximum

4-22
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

Component Design features Loading Modes Fatigue analysis Comments


stresses developed by the recommended
steam pressure/temperature ramp rate curves
for both the shutdown and startup phase of
each thermal cycle for each of the specified
types of start/stop operating transients. The
stress or strain range and mean stress values
used in the fatigue calculations must include
the combined shutdown and startup portions
of each of the operating cycles.
For superheater or reheater bundles with
o o
outlet temperatures above 426 C (800 F) an
additional safety margin of 10 on cycles to
cracking should be used to account for creep-
fatigue interaction effects.
HP Tube-to-header (Type 1) Borehole thermal The Type 1 loading can On many HRSGs, in addition to heating (or
superheater and pipe-to- shock be assessed with cooling) during the first part of each startup (or
and reheater header (Type 2) Tube to header EN 12952, Part 3, shutdown) the pipes and headers downstream
headers boreholes, midwall temperature differences Section 13. A finite of HPSH and reheater desuperheaters are
downstream of element (FE) model subjected to large temperature downshocks
de- used in conjunction with and upshocks at fast rates when substantial
superheater the EN12952-3 will be desuperheating is suddenly required and then
outlets required to assess the a little later in the loading is rapidly reduced or
Type 2 loading mode. curtailed. Slow response of desuperheater
The FE model could temperature control loops often causes 2 or 3
also be used to remove thermal downshock/upshock cycles during
some of the each startup and again during each shutdown.
conservatism inherent
in the EN12952-3
borehole thermal shock
stress estimates
All headers Tube-to-header (Type 4A) Row-to-row Global pressure + These flexibility analyses should assess the
connections temperature differences thermal expansion + fatigue life of the tube-to-header connections
dead weight load resulting from the worst case tube-to-tube
Header surface (Type 4B) Tube-to-tube within
transfer flexibility temperature differences within the tube
(stresses individual row temperature
analysis modeled with bundle during each operating cycle
produced by differences
the tubes, headers,
tubing/pipe
piping and supports all

4-23
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

Component Design features Loading Modes Fatigue analysis Comments


nozzle loads) (Type 5A) Row-to-row included in the model. Anticipated tube bowing/header bending or
temperature differences in (EN13480-3 or B31.1 rotation from row-to-row and tube-to-tube
multipass assemblies flexibility analyses) temperature differences should be included,
especially for bundles with header partition
(Type 5B) Tube-to-tube within
plates
individual row temperature
differences in multipass For superheater or reheater bundles with
o o
assemblies outlet temperatures above 426 C (800 F) an
additional safety margin of 10 on cycles to
(Type 7) Tube row alloy
cracking should be used to account for creep-
differences
fatigue interaction affects.
The tube to tube temperature differences
used in the analysis should be obtained by the
HRSG designer from field measurements on
closely similar designs that have been heavily
instrumented. If dynamic simulations are used
to derive these temperatures then the
purchaser should require that these values be
verified by measurement during
commissioning since experience has been
that the dynamic simulation derived tube
temperature differences are often significantly
optimistic
Internal and Girth welds, (Type 8) Piping flexibility Global pressure + For superheater or reheater bundles with
o o
external piping especially at thermal expansion outlet temperatures above 426 C (800 F) an
(Type 6) Midwall temperature
fittings and flexibility analysis and additional safety margin of 10 on cycles to
differences between nozzle
thickness and/or local axial and through- cracking should be used to account for creep-
pipes or between nozzle pipes
alloy transitions thickness (thermal fatigue interaction effects.
and headers or manifolds
shock) analysis
Pipe spools near
(Type 2) Pipe throughwall (EN13480-3 or B31.1)
attemperators
temperature gradients (near
attemperators)
HP Endplates Pressure stress + through- Stress analysis used in For superheater or reheater bundles with
o o
superheater thickness and axial temperature combination with outlet temperatures above 426 C (800 F) an
and RH gradient induced thermal shock fatigue curves in additional safety margin of 10 on cycles to
headers and EN12952-3 cracking should be used to account for creep-
manifolds fatigue interaction effects.

4-24
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

Component Design features Loading Modes Fatigue analysis Comments


Piping, Supports Thermal expansion flexibility EN13480-3 or B31.1 For superheater or reheater bundles with
header, drums analysis, local axial and outlet temperatures above 426oC (800oF) an
through-thickness temperature additional safety margin of 10 on cycles to
or alloy transition induced cracking should be used to account for creep-
thermal shock fatigue interaction effects

4-25
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

Figure 4-4
Example of an allowable temperature differential chart. The permissible limits shown are
for an HP superheater manifold with specific dimensions and design details.

The curves in Figure 4-4 document the permissible combination of top-to-bottom and midwall to
inside surface temperature differentials for a specific header, manifold or drum. This combines
the Mode 1 (borehole thermal shock) and Mode 3 (header top-to-bottom temperature difference)
loadings. Permanent midwall thermocouples must be installed in the top and bottom of the
header, manifold or drum at a few selected locations along its length to use this type of control
chart. This type of control chart should be provided for the HP drum, HP superheater outlet
header, HP superheater lower headers and manifolds, reheater headers with greater than 25 mm
(1-inch) wall thickness and the manifolds and headers at the LP economizer (preheater) inlet.
This type of control chart has been used, in combination with appropriate thermocouples, in large
fossil-fuel-fired plants as the technical basis for shortening startup and shutdown schedules.

4.4 Specify Special Diagnostic/Troubleshooting Monitoring Tests and


Acceptance Criteria to Be Performed During Commissioning

The companion document to this report, “Diagnostic/Troubleshooting Monitoring to Identify


Damaging Cycle Chemistry or Thermal Transients in Heat Recovery Steam Generator Pressure
Parts” [3] provides detailed guidance on the selection of locations where tube temperature
measuring thermocouples should be attached to selected tubes. Ideally these
diagnostic/troubleshooting thermocouples should be installed prior to commissioning so that the
HRSG manufacturer and owner/operator can verify that the limits described in the last section

4-26
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

are not being exceeded or are corrected before extensive damage occurs. These
diagnostic/troubleshooting thermocouples will also provide the knowledge required to alter or
fine-tune the shutdown and startup procedures. The objective is to provide the greatest operating
flexibility without exceeding the thermal-mechanical fatigue limits.

It is important that the owner includes a requirement for this monitoring in the plant procurement
specification. This diagnostic/troubleshooting monitoring is not a standard item. As a minimum,
it is recommended that the following monitoring instrumentation be required:

Permanently installed midwall thermocouples on the top and bottom of the HP steam drum at
the same circumference at a few locations along the length of the drum. These will be used to
verify that the permissible top-to-bottom temperatures and midwall and saturation
temperature (estimated from the measured drum pressure) are not exceeded. This system of
monitoring has provided the technical basis for increasing the startup rates in large
conventional plants with no ill effects on the steam drum. This was done by controlling the
operation during startup to come close to but not above the temperature differential
permissibles then readjusting the operational procedures or control logic to duplicate the
successful operating procedures on all subsequent startups. Much care is required when
considering permitting an increase in pressure raising rate to ensure that drum top-to-bottom
temperature differences are limited to a safe value that does not develop excessive bending
stresses at the attachment of the shorter risers from the furnace headers

Thermocouples installed midwall on the top and bottom of a selected final HP superheater
header or manifold and on the outside diameter surface of few selected outlet tubes (e.g., near
the ends of the header/manifold, in the projected area of the nozzles and midway between the
header/manifold ends and nozzle pipe). These will be used to verify that the permissible
temperature ramp rate, tube outside surface-to-header midwall temperature difference and
header top-to-bottom temperature difference are not exceeded.

A selected number of tube temperature measuring thermocouples be installed prior to


commissioning. These will be used to verify that the row-to-row and tube-to-tube within
individual row temperature difference limits are not exceeded. The selection of locations for
these thermocouples should be made using the roadmap approach in
“Diagnostic/Troubleshooting Monitoring to Identify Damaging Cycle Chemistry or Thermal
Transients in Heat Recovery Steam Generator Pressure Parts” [3]

These diagnostic thermocouples should be used during commissioning to verify or tune the
DCS controlled combustion turbine exhaust gas conditions, startup vent/ bypass control
valves, drain and vent, attemperator controls, etc. to maintain acceptable temperature ramp
rate and temperature differences between the temperatures measured.

4-27
EPRI Proprietary Licensed Material

Fatigue Assessments – Owner Procurement Specification

4.5 Specify What Diagnostic/Troubleshooting Monitoring Instrumentation,


Controls, Data Acquisition Systems and Data Evaluation Algorithms are to
Be Used Throughout the Lifetime of the HRSG

The owner/operator should consider specifying the HRSG pressure part monitoring requirements
to be used throughout the life of the unit. As a minimum it is recommended that software
algorithms be added to the plant historian system to automatically count and characterize
operating cycles. The algorithms should ideally count each stop/start cycle and provide output on
time/date of the beginning and end of the cycle, the minimum (negative) HP drum water
temperature ramp rate while reducing load, the off line or low load hours, the minimum HP drum
water temperature prior to restarting and the maximum (positive) HP drum water temperature
ramp rate while increasing load. If midwall HP drum and HP superheater header thermocouples
are included in the design then these measured values should be included in the cycle
characterizations.

More sophisticated algorithms could be developed to also provide the incremental and
cumulative fatigue damage life consumption on a cycle by cycle basis for the downcomer/riser
nozzle to HP drum connections and/or for other critical pressure parts/features such as the HP
superheater tube-to-header connections. To utilize the monitors throughout the life of the plant
will require the installation of permanent thermocouples on the components/features being
assessed.

Other HRSG fatigue monitoring systems that should be included in the procurement
specification are:
• HP superheater/ reheater attemperator spraywater monitoring systems to alert the operators to
potentially damaging overspray conditions. Thermocouples should be attached to the outside
surface of bare sections of tubes carefully selected based on measurements in closely similar
designs. They should be located within the projected image of the inlet nozzle pipes and in a
few tubes well away from the nozzles that are immediately downstream of attemperators
used in combination with thermocouples in pipe thermowells placed in accordance with the
attemperator designers recommendations.
• LP economizer/preheater feedwater tube temperature monitoring to identify maldistribution
of low flow rates of cold feedwater between feedwater inlet row tubes. As a minimum, a few
carefully selected tubes in the projected image of the feedwater inlet nozzles and away from
the nozzles should be instrumented
• HP superheater and reheater tube-to-tube temperature difference monitoring to identify the
occurrence of very damaging forward migration of undrained condensate during startups.
This would require a more robust version of the diagnostic/troubleshooting monitoring
thermocouples. Unless data from detailed measurements on closely similar designs are
available it will be necessary to install diagnostic thermocouples on a significant number of
tubes to determine which tubes become selectively quenched or overheated. After this has
been determined then a small number of permanent thermocouples could be installed.
• HP steam to outlet header midwall temperature differential monitoring.

4-28
EPRI Proprietary Licensed Material

5
FATIGUE ASSESSMENTS – DESIGN

The fatigue design roadmap includes the following steps:

1. Owner to specify that HRSG and key auxiliary system designs will be rejected which have
the potential to cause uncontrollable transients with large tube-to-tube or row-to-row
temperature differences during startups or shutdowns.

2. Prior to contract award, Owner to review in detail the proposed design of the HRSG and key
critical auxiliary systems and verify that there are no features or details that may cause large
tube-to-tube and/or tube row-to-row temperature differences during upset transients at
startups and/or shutdowns.

3. Verify that the assembly/component/ feature/construction attributes and operating parameters


meet acceptable design limits (documented with design calculations and reports)

4. Set the allowable component operating limits compatible with the specified cyclic life and
calculated life expenditure rates in 1. above (e.g., fluid temperature/pressure ramp rates
during stops and starts, tube to header temperature differences, inner wall to midwall
temperature gradients in drums and headers, tube to tube temperature differences)

5. Prescribe the operating procedures and permissibles to ensure that the component operating
limits are not exceeded

6. Create a list of important design/construction damage avoidance features and required


construction QC/QA tests and limits to ensure that these features are obtained

7. Specify the design and construction requirements of special damage monitoring


instrumentation and operating tests to be used during commissioning

8. Specify the design and construction requirements of damage monitoring instrumentation,


controls, data acquisition systems and data evaluation algorithms to be used to monitor
component damage throughout the lifetime of the HRSG

These steps will each be discussed in the following sections.

5-1
EPRI Proprietary Licensed Material

Fatigue Assessments – Design

5.1 Owner to Specify that HRSG and Key Auxiliary System Designs
Capable of Causing Uncontrollable Transients with Large Tube
Temperature Differences During Startups Will Be Rejected

In horizontal gas path (HGP) HRSGs, the magnitude of tube-to-tube and/or row-to-row
temperature differences measured on several different designs of HRSG HP superheater or
reheater sections can be very large. This can be caused by forward migration of undrained
condensate when steam flow is initiated each startup or by overspraying down to or close to
saturation temperature at desuperheater outlet during CT load increasing from lower loads. This
is likely to lead to premature fatigue cracking at tube to header attachment welds, especially in
designs that have tubes offset from the centerline. The design details and features that contribute
to forward migration of condensate, to overspraying at desuperheater outlets and to large
transient temperature differences between the feedwater inlet row and other rows of LP
economizer/preheaters must be eliminated from the HRSG and key auxiliary systems design for
new installations

Most installed HGP HRSGs have one or more design deficiencies in HP superheater and reheater
drainage arrangements. These include inadequate forward slope in the steam flow direction to
enlarged condensate drain pots of all HRSG interconnecting pipes and in the cold reheat pipe,
and in the detailed design and arrangement of drains systems that make it impossible to remove
all condensate before steam flow is initiated during startups. The Owner’s specification should
reject HRSG and key auxiliary system designs that have not incorporated all of the detailed
elements identified in Reference 2 and other publications. These collectively ensure that
condensate will, with certainty, be quickly and completely removed from the HRSG at every
startup to a blowdown vessel before forward steam flow commences. It is important that the
HRSG designer be made responsible for the detailed design of the entire HRSG drains removal
and disposal system, since deficiencies in some of the installed units stem from splitting
responsibilities between HRSG supplier and the EPC contractor

Many HRSG operators overspray too close to saturation temperature at the HP superheater and
reheater desuperheater outlet. This often occurs during loading of the CT through the lower loads
when the CT exhaust gas temperature is higher than when the CT is at baseload. The steam flow
rates are also low because pressure is usually being simultaneously raised so that a portion of
evaporation is used to pressurize steam spaces in the HP drum and superheater. Slow response
of conventional desuperheater temperature control loops leads to sudden demand for excessive
high spraywater flow rate to attempt to prevent a high steam temperature trip or an automatic CT
load runback. This is followed by rapid shutoff when steam temperature quickly falls below the
desuperheater temperature control setpoint. Often this damaging thermal cycle is repeated two
or three times during a single startup.

The same phenomenon occurs on some units during deloading to lower loads prior to unit
shutdown. This unacceptable, very damaging phenomenon, which is particularly prominent on
units with the GE 7FA+e CT. These machines produce maximum exhaust gas temperatures of
1200°F (649°C) as low as about 30% CT output. This probably occurs because the selection by
the HRSG designer of the desuperheater position within the heating surface is based on steady
state heat balances at part load operating conditions. During startups and reloading through

5-2
EPRI Proprietary Licensed Material

Fatigue Assessments – Design

lower load range, when steam conditions are far from stable and CT exhaust gas temperature is
increasing rapidly, desuperheaters have great difficulty in preventing steam temperature
overshoots. This is because of the slow response of desuperheater steam temperature feedback
control loops at the same time as when gas temperature is increasing rapidly and steam flow rate
is also low.

To prevent high steam temperature trips or automatic CT load runbacks, operators tend to lower
the desuperheater HP steam temperature setpoint. This considerably increases the spraywater
demand and exacerbates the magnitude of overspraying. In some cases the extent of
overspraying is because the single stage of desuperheating is too early in the steam superheating
process. But even on HRSGs with about 65% of the HP superheater surface preceding the
desuperheater and 35% of the surface downstream of the desuperheater, gross overspraying
occurs.

To minimize cost, HRSGs currently employ a single stage of desuperheating in the HP


superheater and similarly in the reheater. Resolution of this design deficiency, which is the root
cause of potentially severe fatigue damage to downstream components, may require the adoption
of two stages of desuperheater with part of the heat transfer surface located between the two
staged desuperheaters. Such a change was long ago adopted on conventional fossil-fired boilers
to eliminate overspraying at desuperheater outlet and to improve control of HP steam
temperature.

The Owner’s specification should reject HRSG designs that cannot demonstate with certainty
that under the dynamic conditions developed during initial loading up to 60% load during
startups there will always be a minimum of 28°C (50°F), preferably more, superheat at HP
superheater and reheater desuperheater outlets. This can be demonstrated either by DCS trend
plots from units with identical CT and HRSG design, or by dynamic analysis of steam flow,
pressure and temperature at the HP superheater and reheater desuperheater inlet and outlet and at
the HP outlet during startup and loading from different initial HRSG conditions. It should also be
confirmed that HP superheater and reheater outlet steam temperatures can be controlled to design
setpoint temperatures without significant steam temperature overshoot during all startups.

The inlet row of LP economizer/preheaters is vulnerable to large transient downshocks when


cool feedwater is first admitted into the hot stagnant economizer/preheater at startups. The
downshock develops a potentially damaging negative temperature difference between the inlet
pass of tubes and all subsequent passes in the economizer/preheater. To reduce this temperature
difference below 80°C (150°F) a condensate recirculation system that returns an adjustable
controlled flow rate of LP economizer/preheater outlet feedwater to temper the cold feedwater
inlet flow is required. The design of the condensate recirculation system, including its flow
capacity and complex controls to cater for many upset operating conditions that will arise, is
critical to the efficacy of this system in eliminating fatigue damage to the LP
economizer/preheater.

EPRI intends to publish a guideline for the design and operation of HP superheater and reheater
drains and feedwater control systems in 2007

5-3
EPRI Proprietary Licensed Material

Fatigue Assessments – Design

5.2 Prior to Contract Award, Owner to Review the Proposed Detailed


Design of HRSG and Key Auxiliary Systems to Verify that there are No
Details and Features Capable of Causing Uncontrollable Transients With
Large Tube Temperature Differences During Startups

Modifications to the proposed design of HRSG and key auxiliary systems essential to ensuring
that there will be no condensate migration and no overspraying during startups must be agreed
prior to contract award. Otherwise the changes will impact on many different design attributes
and the control systems.

5.3 Verify that the Assembly/Component/Feature/Construction Attributes


and Operating Parameters Meet Acceptable Design Limits

Verifying that the assembly/ component/ feature/ construction attributes and operating
parameters meet acceptable design limits is a primary objective of fatigue design
codes/standards. Once the pressure parts have been designed to meet the sustained load
conditions (using ASME Section I or EN 12952-3 for tubing, ASME B31.1 or EN 13480-3 for
piping and other relevant codes, standards or code cases) a check of their anticipated fatigue
performance is undertaken if required by the procurement specification and/or anticipated
operating service forecast.

The most common fatigue check that is performed for HRSG pressure parts is an assessment of
the HP drum borehole thermal shock loading (Type 1 loading, Table 3-1). Appendix D provides
and example of this type of analysis performed in accordance with EN12952-3, Section 13 rules.
As discussed in Appendix D these simplified code analyses have limitations that are both overly
conservative in some respect and optimistic in other respects. Perhaps the most significant
deficiencies of the current code approaches are the inability to evaluate the combined effect of
top to bottom header or drum temperatures and humping (bowing) that occur at the same time as
borehole thermal shock events and the inability to easily evaluate the benefit of temperature
holds. For new or cycling designs it is recommended that a more comprehensive finite element
cyclic stress evaluation be performed to determine the acceptable limits of heating and cooling
rates and borehole surface to midwall and top to bottom drum/manifold/header temperature
gradients.

Figure 4-4 show an example of a useful control chart developed from such an analysis. This type
of analysis should be performed for the HP drum and other selected HRSG headers/manifolds
such as the HP superheater headers and lower manifolds, reheater headers with thickness greater
that 25mm (1-inch), and economizer inlet headers that are suspected to be prone to borehole
quenching events associated with introduction of cold feedwater. A thorough determination of
the impact of pressure ramp rate and holds on drum top to bottom mean wall temps must use
different heat transfer coefficients for the steam washed surface (condensation heating). The
water wetted surface heat transfer coefficient will vary depending on vapor content, velocity and
turbulence of mixture. To determine the startup and shutdown pressure or fluid temperature
ramp rate limits both the top to bottom mean temp difference and the near borehole midwall to
inside surface temperature gradient should be evaluated.

5-4
EPRI Proprietary Licensed Material

Fatigue Assessments – Design

For units that are anticipated to be extensively cycled, or when required by the owner, additional
fatigue analyses will need to be performed. Analysis of the HP remote steam drum nozzles and
HP superheater headers under the Type 1 loading (Table 3-1) is used to set the permissible drum
water and superheater temperature ramp rates. Few HP drums and few, if any, HP superheater
outlet headers have been inspected internally and thus there is no evidence available to date
whether cracks have initiated at the drum/manifold/header borehole locations. It is quite likely
that inspections will find borehole crack initiation in thicker HP superheater headers that have
accumulated a few hundred starts.

The most common recorded pressure part fatigue failures have occurred at tube-to-header
connections under a combination of loading mode Types 2 through 5 (Table 3-1). The stresses at
the tube, pipe nozzle or drain line-to-header attachment are controlled by the temperature
difference between the tubes connected to common headers and in some designs to the
temperature difference between the tube and header midwall. These transient, often very short
lived, but nevertheless potentially very damaging loads are generally unrelated to the bulk fluid
ramp rates and thus are not prevented by setting and controlling the rate of change of bulk fluid
temperature. Instead, the severe temperature differences and thermal loads typically occur when
there is an upset condition that results in maldistribution of fluid temperatures within the tube
bundle due to a variety of causes such as:
• Forward flow of undrained condensate or excessive attemperator spraywater in superheaters
and reheaters
• Reverse or stagnant flow of fluid in some evaporator tubes during the early part of startups
before natural circulation is fully established
• Recirculation of stratified, subcooled water in HP evaporators following long weekend
shutdowns
• Steaming or blockage by unvented air pockets at the top of economizers and preheaters
without reliable venting arrangements
• Slug feeding or leakage of low flow rates past damaged feedwater valve seat of cold
feedwater into LP economizers/preheaters

The loading produced by these events is not addressed in the stress solutions included in
TRD301, Annex 1 or EN 12952-3. The constrained global thermal expansion of the tubes within
the tube bundle will produce the largest portion of the loads applied to nozzles or tube-to-header
connections. This loading is mentioned in EN 12952-3 Sections 7.5 and 11.4.2 and suggests that
the flexibility analysis methods of the European Norm piping standard (EN 13480-3, Section 12)
should be used to assess this loading. The forces and moments at the tube-to-header connections
can be estimated using commercially available piping flexibility codes or can often be easily first
estimated using hand solutions to assess whether more sophisticated analysis is justified. An
example hand calculation is shown in Appendix C.

Figure 5-1 illustrates estimated forces and moments for a single pass tube bundle with two
parallel rows of tubes with dogleg bends at the top and bottom of the harp. The header to header
spacing was approximately 15.6 meters (50ft) and the tubes were 57.2 mm OD x 3.6 mm wall
thickness (2.25-inch OD x 0.14-inch wall thickness) with a 28-degree dogleg bend. This analysis

5-5
EPRI Proprietary Licensed Material

Fatigue Assessments – Design

indicates that tube row temperature differences less than approximately 28oC (50oF) can
probably be tolerated with this reheater bundle design, whereas tube row temperature differences
greater than approximately 56oC (100oF) may lead to tube-to-header fatigue failures within the
anticipated cyclic life of the plant. During startups, migration of undrainable condensate caused
tube to tube temperature differences varying between 139 and 178°C (250 and 320°F), which
may cause crack initiation in of the order of 100 cycles.

Figure 5-1 highlights the crucial importance, even for HRSGs intended to perform few starts, of
cutting no corners in the design of drainage arrangements of HRSG tubes, including ensuring all
interconnecting pipes drain in the direction of steam flow to condensate drain pots so that there is
nowhere that condensate can collect. The detailed design of the drain system between drains
points and drains vessel is also critical to the reliable removal of all condensate before steam
flow commences at every startup.

Figure 5-2 presents a sensitivity study of the stress magnitude in a similar harp design with a
range of dogleg bend angle and tube offset distances. The allowable tube temperature differences
are very sensitive to the tube bundle configuration and tube-to-header connection design. A
sensitivity study of a single pass, two row harp indicated that the magnitude of the nominal,
combined axial + bending stress was a strong function of the dogleg tube bend angle and the
distance between the tube rows.

Figure 5-1
Estimates of the Stresses and Cycles to Cracking as a Function of Tube Row-to-Row
Temperature Difference

5-6
EPRI Proprietary Licensed Material

Fatigue Assessments – Design

Figure 5-2
Effect of Tube Offset Distance and Angle on Stresses in Tube/Header Connection

Given the common requirement of minimizing the distance between tube rows, designs with
small tube offset angles and minimized tube offset distances from the header centerline provide
the greatest tolerance to tube temperature differences. The magnitude of these calculated elastic
stress values scale in a linear fashion with the magnitude of the tube to tube temperature
difference. The intensified peak elastic stress at the tube-side toe of the tube to header weld
would be expected to be 3 to 5 times higher than the values shown. This illustrates the benefit of
keeping offset distances either very small or very large, especially for larger dogleg bend angles.

5.4 Set the Allowable Component Operating Limits

The previous section discussed two stress analysis/fatigue life models that provide limits for key
parameters that must be maintained below safe limits in order to avoid HRSG pressure part
fatigue cracking within the anticipated life of the unit. Key parameters that must be controlled to
avoid fatigue cracking include:
• HP drum pressure/water temperature ramp rate during the shutdown and startup portions of
each operating cycle
• HP superheater headers pressure/steam temperature ramp rate during the shutdown and
startup portions of each operating cycle
• Tube-to-tube and tube-to-header temperature differences

5-7
EPRI Proprietary Licensed Material

Fatigue Assessments – Design

It is important that the HRSG designer ensures that the design includes no features or details that
would require operating limits on these key parameters that are not practically achievable. In
this respect it is the designer who is primarily responsible for ensuring that the design of HRSG,
interconnecting pipes and drains removal system can reliably remove all condensate from the HP
superheater and reheater to the drains vessels.

The HRSG designer is also entirely responsible that the location of HP superheater and reheater
desuperheaters within their respective heating surfaces is such that under all operating conditions
the desuperheater controls are capable of controlling HP superheater and reheater outlet steam
temperatures safely below the respective outlet header and pipes design temperature whilst also
maintaining a minimum of 28 °C (50°F at desuperheater discharge. This includes the dynamic
conditions during loading of the CT from lower loads at startups.

It is also important that the safe limits for those key parameters associated with controllable
steam pressure and temperature ramps and holds are determined by conservative analysis of the
design. These must be communicated to the owner/operator in ways that make it clear what
controllable operating conditions can be tolerated by specific components/design features. This
overcomes the concern that the designers can’t fully control the way the unit is operated. The
owner/operator needs to know the limits of the operating parameters that must be controlled to
avoid fatigue cracking. Ideally these limits would be provided as targets and control limits and
action levels in a fashion similar to EPRI’s cycle chemistry guidelines. Examples of some useful
control charts are described below.

Figure 5-3 illustrates the commonly used pressure/temperature ramp rate control curves that are
produced as part of a TRD 301, Annex 1 or EN 12952-3 fatigue assessment for drum or header
borehole regions. This example diagram presents the combination of HP drum pressure and HP
drum water temperature ramp rate limits required for the shutdown and startup portions of each
design basis operating cycle to prevent fatigue cracking of the HP drum nozzle boreholes.

Control curves of this type should be used to track the HP evaporator drum and HP superheater
pressure/water temperature ramp rate during all shutdown and startups. In the case of the HP
superheater, the temperature ramp rates should be determined using tube temperature measuring
thermocouples. An example of this tracking for an HP drum is shown in Figure 5-4. The HP
drum water temperature ramp rate during the cold start made on the lead HRSG (without the
steam turbine running) and during the warm start exceeded the values required to ensure that
fatigue cracking is prevented within the anticipated operating life of the plant. To prevent these
potentially damaging transients from being repeated, a simple monitoring system was added to
the control room to alert the operators when they were approaching the ramp rate limits.

These types of control curves can be developed for each specific HRSG, and need to include
when temperature/pressure holds are used to minimize the magnitude of thermal stress.

5-8
EPRI Proprietary Licensed Material

Fatigue Assessments – Design

Figure 5-3
Example control chart for an HP drum

5-9
EPRI Proprietary Licensed Material

Fatigue Assessments – Design

Figure 5-4
HP drum pressure/temperature ramp rate control chart with data from two cold starts (CS),
a warm start (WS), a hot start (HS) and a low load (LL) cycle superimposed.

There is the tendency to conclude from these pressure/temperature ramp rate control charts that
pressures should be decreased quickly during shutdowns to increase the allowable thermal stress
and retain the same total stress. This logic is seriously flawed. Although dropping pressure
reduces the pressure induced component of total stress, the consequential rapid reduction in
saturation temperature results in larger temperature changes, turbulent boiling of evaporator
water and thus significantly increased heat transfer and non-linear temperature gradients at the
inner surfaces of the drum wall and borehole surfaces.

Although in HP drums the pressure induced stresses are a significant part of the borehole
stresses, the thermal stresses caused by rapid pressure decreases significantly exceed the the
reduction obtained in pressure stress. Furthermore, by lowering the HP pressure during
shutdowns, the restart after an overnight shutdown becomes a cool start for the HRSG. This
requires much longer and may still be more damaging than a hot start, while the steam turbine is
hot and requires hot steam for its restart.

In higher temperature HPSH headers the peak thermal stresses are significantly greater than the
pressure stresses because the permissible stress under mechanical, sustained loadings are
significantly reduced for the higher design temperatures. For HP superheater outlet headers,
deliberately lowering the pressure doesn’t immediately cool the header because the header is still
heated by superheated steam. However at some stage after shutdown, condensation will occur in
the HP superheater tubes and then run down and quench the lower headers and manifolds. Later
in the shutdown the upper headers will also cool to the saturation temperature.

5-10
EPRI Proprietary Licensed Material

Fatigue Assessments – Design

If pressure is preserved to the extent possible, then during the next start after an overnight
shutdown, the HP superheater will heat up by approximately 230oC (450oF) to 565oC (1050oF)
predominantly by convection (thus near linear throughwall temperature gradients) with little or
no condensation heating (unless the HP pressure is ramped up aggressively fast). If, on the other
hand, pressure is purposely reduced then this restart would start with an initial temperature closer
to120oC to 150oC (250oF to 300oF), and initial heating will be by condensation, with an order of
magnitude higher heat transfer rate and more severe and non-linear throughwall temperature
gradient.

The other type of fatigue damage prevention control chart that is needed is a tube-to-tube and
tube-to-header temperature difference limit chart. Figure 5-5 provides an example of such a
chart. This was derived from a constrained thermal expansion/flexibility/fatigue analysis of the
tube bundle and tube-to-header connections. The warm start data shown in Figure 5-6 illustrates
how this controls chart and action limits are used.

Note the two severe thermal quench events. The first was attributed to condensate migration and
resulted in a very rapid -139oC (-250oF) temperature drop. The second quench event was
attributed to overspraying of the interstage attemperator due to manually lowering the reheater
outlet steam setpoint to a value of 510oC (950oF). This was to prevent high hot reheat steam
temperature trips experienced when the set point was left at the design setting of 565°C
(1050°F). This resulted in a rapid drop in fluid temperature of -222oC (-400oF) at desuperheater
outlet. Unfortunately it was impossible to know if these quench events (which were measured in
bulk fluid measurement made downstream of the attemperator in the primary to secondary
reheater crossover pipe) persisted and were as severe within the reheater tube bundle and if they
produced significant maldistributions of the reheater tubes. To overcome this deficiency,
diagnostic/ troubleshooting thermocouples were added to the top and bottom of selected tubes on
each row of the single pass, two parallel row reheater bundles.

During this warm start two severe thermal quench events were recorded in the bulk fluid
downstream of the attemperator in the crossover pipe between the primary and secondary
reheater bundles. To understand the extent to which these quench events produced temperature
maldistributions within the secondary reheater tube bundle approximately 70 tube temperature
measuring thermocouples were installed on the top and bottom of both rows of selected tube
elements.

Figure 5-7 presents an example of the tube temperature pattern across the two side-by-side
reheater tube bundles at 02:35 hours during the second severe thermal quench event, which was
caused by reheater attemperator overspray to below the saturation temperature. Each of the two
side by side reheater bundles has 33 tube elements. As is shown, all the tubes near the secondary
reheater inlet pipe nozzle closest to the blind end of the reheater crossover pipe and furthest from
the interstage attemperator were approximately 139oC (250oF) colder than the hottest tubes
within the bundle. A few tubes near each of the other three lower, inlet pipe nozzles also had
very low temperatures. Most of the unflashed attemperator water was transported to the blind
end of the secondary reheater manifold where its momentum was destroyed and it was blown by
steam up the closet inlet pipe nozzle. Some smaller droplets of unflashed water were transported
by steam into the earlier three inlet pipe nozzles and then into the tubes closest to these nozzles.

5-11
EPRI Proprietary Licensed Material

Fatigue Assessments – Design

At the outlet of the reheater the coldest tubes were in line with the inlet nozzle furthest from the
attemperator and were approximately 78oC (140oF) colder than the hottest tubes in the bundle.

During the overspraying event the steam temperature recorded in the manifold pipe downstream
of the attemperator was about 177oC (350oF). The nonuniform distribution of unflashed
spraywater at the inlet to the tubes of the secondary reheater caused a tube temperature
difference at the bottom (inlet) of the reheater of approximately 139oC (250oF). At the top
(outlet) of the secondary reheater 78oC (140oF) tube temperature differences were recorded.
Using the “cycling unit” limits shown in Figure 5-5 leads to Action Level 3 since the average
tube temperature difference in the tube bundle is between 83oC (150oF) and 111oC (200oF).

According to the control chart shown in Figure 5-5 this severity of tube-to-tube temperature
difference should only be permitted twice per year. The corrective action to prevent these very
damaging quenching events requires a redesign of the reheater crossover piping to provide more
straight length between upstream pipe bends and the attemperator and also between the
attemperator and the inlet to the secondary reheater. In addition, a change to the operating
procedures limiting the manual setting of the reheat outlet setpoint that controls the interstage
attemperator to a value of not less than 560oC (1040oF). To avoid frequent high hot reheat steam
temperature trips, the owner had to re-engineer the desuperheater steam temperature control
logic to incorporate an anticipatory control action.

Figure 5-5
Example control chart for tube-to-tube temperatures in a secondary (high temperature)
reheater. The action level targets were derived from the analysis shown in Figure 5-1 and
engineering judgment.

5-12
EPRI Proprietary Licensed Material

Fatigue Assessments – Design

Figure 5-6
Measured bulk fluid temperatures into and out of a secondary reheater measured during a
warm start [4]

Figure 5-7
Plot of the secondary reheater tube metal temperatures at the timeslice shown in Figure
5-6 that corresponded to the attemperator overspray quench event [4]

5-13
EPRI Proprietary Licensed Material

Fatigue Assessments – Design

5.5 Prescribe the Operating Procedures and Permissibles

The previous section described a couple of useful control charts that could be used to track and
set action limits for potentially damaging thermal-mechanical transients. Setting the operating
procedures and permissibles could be done using this information but is more often done with
consideration of many operating requirements. For example an HP steam drum
temperature/pressure hold may be needed to allow the chemistry of the saturated steam to meet
the requirement of the steam turbine. These hold times would be engineered to also reduce the
through-wall, nozzle to drum and tube-to-tube temperature differences and may permit more
rapid heating prior to and after these hold periods. The benefit of pressure/temperature hold
periods must be assessed on a case by case basis and then used to define operating procedures or
programmed operating permissibles.

5.6 Create a List of Important Design/Construction Damage Avoidance


Features and Required Construction QC/QA Tests and Limits

The designer should be encouraged to provide quality control/quality assurance (QC/QA)


checklists listing the key construction features that must be controlled to ensure that the as-built
components and design features have the required fatigue damage resistance. These may include
items such as:
• Weld size, geometry (e.g., weld to radius), fit up and quality
• Surface finish, chamfers
• Out-of-roundness

Communication of the importance of and required limits for these key construction parameters to
the fabricator and erector along with a diligent QC/QA program will go a long way towards
ensuring that the as-built HRSG is free of construction flaws that will fail prematurely.

Since tube-to-header welds are the most common feature that has produced fatigue failures these
should be given special attention. Fatigue failures of tube-to-header joints have in some cases
been accelerated by a variety of weld fabrication attributes such as:
• Undersized welds or welds with flaws such as incomplete penetration or incomplete fusion
that were designed as either full strength partial joint penetration welds in set-in branch
configurations or complete joint penetration welds in set-on branch configurations
• Fillet weld toe radii much smaller than 3.2 mm (1/8”)
• Tube to header connection alloy transitions such as a T91 (modified 9% chromium alloy)
tube connecting to a P22 (2-1/4% chromium 1% molybdenum) header
• Poor tube to header fit-up of shop-welded panels due to header distortions during welding of
tubes

Designs with tube to header alloy transitions should be avoided. The remainder of these
construction quality fatigue influence factors must be controlled with communication of

5-14
EPRI Proprietary Licensed Material

Fatigue Assessments – Design

requirements to the fabricators and craftspersons, and a thorough quality control and quality
assurance program.

5.7 Specify the Design and Construction Requirements of Special Damage


Monitoring Instrumentation and Operating Tests to be Used during
Commissioning

During commissioning much effort is spent verifying that the plant can operate within the
engineered design procedure limits and can meet numerous performance limits. Commissioning
is also an ideal time to verify that the fatigue damage influencing cyclic thermal and pressure
loads are well within the safe limits defined during the design for the entire range of anticipated
operating cycles. To perform this fatigue cracking avoidance verification will require installing
carefully selected tube and header temperature measuring thermocouples as is outlined in EPRI’s
“Diagnostic/ Troubleshooting Monitoring to Identify Damaging Cycle Chemistry or Thermal
Transients in Heat Recovery Steam Generator Pressure Parts” [3].

If the control charts discussed previously were prepared as part of the design then the acceptable
limits on HP drum and HP superheater temperature ramps rates and tube-to-tube and tube-to-
header temperature limits will have already been determined. If operating events are identified
that are outside these limits then alterations to the operating procedures or limits or required
design alteration can be made before extensive fatigue damage has accumulated.

5.8 Specify Damage Monitoring Instrumentation, Controls, Data


Acquisition Systems and Data Evaluation Algorithms to be Used to Monitor
Component Damage Throughout the Lifetime of the HRSG

Most modern HRSGs have been designed and built with very little thought to in-service damage
monitoring. Periodic examinations of crack susceptible locations are possible in the accessible
locations but is often limited due to lack of access. This elevates the importance of
diagnostic/troubleshooting monitoring to track and control the fatigue damage influencing
thermal-mechanical factors such as fluid ramp rates in individual tubes in the HP superheater or
tube-to-tube and tube-to-header temperature differences in all but the most flexible or fatigue
damage tolerant tube bundles.

Ideally the control charts described above will be developed and the action limits followed. If on
the other hand, the owner/operator wants more freedom to operate with less regard for the
potential fatigue damage incurred then it would be wise to develop a fatigue damage monitoring
and life consumption evaluation system. Fatigue damage monitoring systems have been installed
on a number of HRSGs for various purposes.

By combining the EN 12952-3/EN 13480/PD5500 models discussed above with a cycle counting
algorithm such as a rainflow cycle counter and linking this with measurements of appropriate
tube temperature and fluid pressure measurements it is possible to keep track of the potential
fatigue damage occurring and forecast when detailed examinations or component
repair/replacements should be scheduled.

5-15
EPRI Proprietary Licensed Material

6
FATIGUE ASSESSMENTS – CONSTRUCTION

The construction roadmap includes the following steps:

1. Verify that key, life limiting design/construction features (e.g., weld preparations, fit-ups,
tolerances, alignments, slopes of steam piping and the associated position of drains and drain
pots, surface finish, weld procedures, pre-heat, post weld heat treatment, other heat
treatments, weld profile, weld quality, base metal quality, bend ovality limits, etc.) are within
their acceptable

2. Install special diagnostic monitoring instrumentation

These steps have been discussed extensively in References 2 and 3.

6-1
EPRI Proprietary Licensed Material

7
FATIGUE ASSESSMENTS – COMMISSIONING

The commissioning roadmap includes the following steps:

1. Measure and evaluate (using the special diagnostic instrumentation) the key damage-
influencing thermal mechanical parameters (e.g., pressures, temperatures, pressure and
temperature ramp rates, key temperature gradients such as tube-to-tube, tube-to-header, inner
to mid wall temperature difference in drums and headers) throughout the anticipated range of
operating transients and operating procedures

2. Modify or tune the design and/or operating procedures as required to meet the limits of the
key damage-influencing thermal mechanical parameters

These steps have been discussed extensively in References 2 and 3. A separate companion report
to this report which addresses optimization of startup and shutdowns to minimize HRSG
pressure part damage will be published in 2006.

The key issue with regard to fatigue assessment during the commissioning phase is to thoroughly
review the design basis assumptions and compare the design assumptions regarding temperature
differences, temperature gradients and temperature ramp rates to the values obtained from
thermocouple measurements. Where the measured temperatures exceed the design-basis values
detailed fatigue analysis should be performed to determine if the measured values are acceptable
or if operational or design changes are necessary to achieve the specified cyclic life of the unit.

7-1
EPRI Proprietary Licensed Material

8
FATIGUE ASSESSMENTS – OPERATION

The operation roadmap includes the following steps:

1. Operate within the allowable limits of the damage-influencing parameters

2. Modify and tune operational practices as required to meet the damage-influencing parameter
limits

These steps have been discussed extensively in References 2 and 3. A separate companion report
to this report which addresses optimization of startup and shutdowns to minimize HRSG
pressure part damage will be published in 2006.

The key issue with regard to fatigue assessment during the operational phase of unit life is to
periodically review the design basis assumptions and compare the design assumptions regarding
temperature differences, temperature gradients and temperature ramp rates to the values obtained
from thermocouple measurements. Where the measured temperatures exceed the design-basis
values detailed, fatigue analysis should be performed to determine if the measured values are
acceptable or if operational or design changes are necessary to achieve the specified cyclic life of
the unit.

8-1
EPRI Proprietary Licensed Material

9
FATIGUE ASSESSMENTS – MAINTENANCE

The maintenance roadmap includes the following steps:


1. Maintain equipment (e.g., drain valves, startup vent valves, attemperator spraywater control
and block valves, HP bypass pressure control and spray water control and isolation valves, all
HP steam and feedwater isolation valves and other control/stop valves) to ensure that
unanticipated operational transients caused by malfunctioning or worn equipment/control
systems are prevented
2. Monitor and analyze the key damage-influencing factors throughout the lifetime of the
HRSG and set inspection intervals based on estimated life consumption values
3. Perform visual and nondestructive evaluations to verify that the severity of anticipated
damage has not been exceeded and to detect unanticipated damage (e.g., bowed tubes,
damage associated with water hammers or inadequate clearances, corrosion, broken or
misaligned gas baffles, etc.)
4. Repair or replace components or component features prior to failure
5. Identify opportunities to optimize the life of the HRSG components

Fatigue assessments can be extremely useful for prescribing what, where and when to perform
HRSG pressure part condition assessments. It is common practice to perform examinations of
those HRSG pressure part details that have exceeded 50% of their predicted minimum fatigue
life or when failures, visual examinations or upset operational events suggest that accelerated
damage may have occurred. High priority locations for examination should be identified from a
combination of experience on similarly designed and operated units, expert opinions, as-is stress
and fatigue analysis, and visual examinations.

The stress and fatigue analyses required for this effort are often significantly different to design-
basis stress and fatigue assessments. In the former case the question being asked is which
locations are most prone to cracking and what is the anticipated difference in life (i.e., relative
life) between similar design features such as tube to header connections through and across a
given tube bundle. Design-basis stress and fatigue analyses are often simplified and do not
provide knowledge of the absolute or relative life of HRSG pressure part features. They simply
determine if the design features meet certain requirements that are perceived by the Code
committees to produce safe, acceptable design when evaluated with design basis rather than as-is
design and operational conditions.

9-1
EPRI Proprietary Licensed Material

10
REFERENCES

[1] Heat Recovery Steam Generator Tube Failure Manual. EPRI, Palo Alto, CA: 2002.
1004503.

[2] Delivering High Reliability Heat Recovery Steam Generators. EPRI, Palo Alto, CA: 2003.
TR-1004240.

[3] Diagnostic/Troubleshooting Monitoring to Identify Damaging Cycle Chemistry or Thermal


Transients in Heat Recovery Steam Generator Pressure Parts. EPRI, Palo Alto, CA: 2005.
1008088.

[4] Anderson, R. and M. Pearson, “Quantifying Thermal Transients in Heat Recovery Steam
Generators,” EPRI Boiler Tube and HRSG Tube Failures and Inspections International
Conference, San Diego, CA. , Nov 2004

[5] Extracts from a letter received from Mr. G. Horseman, Siemens Power Generation,
Materials at High Temperatures, Vol. 18, No. 4, 2001, pp. 193-209

10-1
EPRI Proprietary Licensed Material

A
REVIEW OF FATIGUE ASSESSMENT FACTORS

This appendix provides an overview of the general attributes of fatigue design rules and analysis
procedures incorporated into the Codes and Standards that are commonly used in the design of
HRSG pressure parts. The various Codes and Standards do not always base their design criteria
on the same technical criteria. As a result, it is sometimes difficult to compare one code with
another. A brief review of the concepts and terminology used in commonly used Codes and
Standards will be presented below, so that the reader can understand the general approaches
used. This is followed in Appendix B by summary overviews of the design codes and standards
that have been used in the design of HRSG pressure parts. The design codes and standards that
have been reviewed include [A-1] through [A-25]:
1. ASME Section I
2. ASME Section III, Subsection NB
3. ASME Code, Section III, Division 1, Subsection NH
4. ASME Code, Section VIII
5. ASME Section XI
6. ASME B31.1
7. API RP 579
8. BS 1113
9. BS PD 5500
10. R5, PD6539
11. R6, PD6493, BS7910
12. RCC-MR
13. TRD 301, Annex 1
14. TRD 508
15. AD-Merkblatt S2
16. EN 12952-3
17. EN 13445
18. EN 13480-3

Design by Rule/Design by Analysis

Some of the older design codes are known as “design by rule” codes. This means that these
codes established basic rules for the design of pressure vessels that covered all of the anticipated
aspects of the pressure vessel operation. There was no need to perform detailed stress analysis
for example, or to consider specific damage mechanisms such as fatigue. The effect of this
approach was to require relatively high factors of safety. These factors of safety incorporated a
long history of successful plant operation.

A-1
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

“Design by analysis” codes were developed to allow higher design allowable stresses. In
exchange for allowing use of the higher allowable stresses (and correspondingly lower factors of
safety) the code-writing bodies required detailed stress analysis and consideration of various
damage mechanisms, including fatigue. More stringent minimum inspection standards and
greater condition monitoring standards may also be required.

Exclusion Rules

Exclusion rules are criteria that in effect state that fatigue is not considered to be a concern for a
given component design and anticipated operating transient forecast. These criteria are often
based on the predicted number of startup-shutdown cycles, or based on a local temperature
difference. These provisions are typically based on conservative evaluations using allowable
stresses and lower bound fatigue design curves. The exclusion limit defined by a temperature
difference is usually based on the temperature difference across the wall of the vessel, or at two
locations within 2 Rt along the surface of a cylindrical vessel.

For HRSGs, it may be appropriate to also include an assessment of the transient, as well as
steady state temperature difference between adjacent tubes connected to common lower and
upper headers, or between the tubes and header.

Fatigue Cycle

Fatigue failures are the result of cyclic stresses and strains, where the magnitude varies with
time. The simplest case is constant amplitude loading, where the stress oscillates between two
values. There are five terms that can be used to describe the stress variation with time, as shown
in Figure A-1. The stress cycle can be completely described by any two of these terms. The two
most obvious terms are the maximum and minimum stress levels (σmax and σmin) in the cycle (the
maximum stress is also often called the peak stress). The mean, or average, stress is 0.5*(σmax +
σmin) . The stress range is the peak-to-peak distance, defined by σmax - σmin. The stress range,
Δσ, should not be confused with the stress amplitude, which is one-half the stress range.

The final term used to describe the stress cycle is the R-ratio, which is the ratio of σmin/σmax. A
stress cycle that varies between +σ and –σ has an R-ratio of -1. A stress cycle that varies
between zero and σ has a value of R = 0. As R approaches a value of 1 (i.e., the minimum and
maximum stress values converge) the value of the stress range, Δσ, by definition becomes very
small. When Δσ is small, there is normally little fatigue damage, except when the mean stress is
high.

A-2
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Figure A-1
Schematic of Fatigue Cycle

The single most important aspect of fatigue loading is the stress range, Δσ. The R-ratio can also
be important because, in general, the higher the R-ratio, the greater the fatigue damage for a
given stress range.

The R-ratio is important in welded structures because welds contain residual stresses. As the
weld metal cools and solidifies, it wants to shrink. This shrinkage puts the last part of the weld
to cool into tension (other parts of the weld will be in compression, so that the net force across
the section is zero). The magnitude of the tensile residual stress may be as high as the yield
stress of the material.

When a cyclic load is superimposed on top of the weld residual stress, local plasticity limits the
peak stress to the yield stress of the material. The cyclic stress then varies between σyield and
(σyield – Δσ). Thus, the R-ratio is relatively high, but is also a function of the stress range. As a
result, fatigue design curves for weldments do not explicitly mention the R-ratio. It is assumed
that the peak stress is equal to the yield stress, and the R-ratio is a function of the stress range.

However it is important to understand that if there are no residual stresses, the R-ratio will be
lower and the fatigue damage will be therefore be less for a given stress range. As a result, it is
possible to make a distinction between as-welded and stress relieved welds. Most codes do not
make this distinction, based on a belief that it is not always possible to ensure that the stress
relief is effective (i.e., that the weld is fully stress relieved).

Most fatigue cycles in HRSGs are more complex than shown in Figure A-1. The pressure cycle,
for example, increases from zero to the operating pressure over some period of time, known as
the rise time. The pressure is then held nominally at the operating pressure for some period of
time, known as the hold time. The time for the pressure to decrease to zero at shutdown is
known as the fall time.

A-3
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

The rise time and hold time are important when time dependent plasticity or environmental
effects (i.e., creep fatigue, corrosion fatigue) become important. For watertouched components
that are strained beyond the limits of the protective surface oxide, the rise time or strain rate will
strongly influence the damage accumulation per cycle. For a constant strain range and mean
stress, slower strain rates (longer rise times) will produce increased corrosion fatigue damage.
This would suggest that increased loading rates would be beneficial. This is only true if the
increased loading rate does not increase the strain range and mean strain of the cycle, which is
seldom the case.

Loading Rate

The reader should be aware that there are two dominant causes of damaging cyclic plasticity at
high local stress points. The first is the transient temperature difference or quasi steady state
through-wall temperature gradient developed during heating and cooling by steam or water
during startups and shutdowns, respectively. The second is large, very rapid changes in
temperature caused by condensation heating of HPSH headers and the steam drum during cold
starts, by rapid increases and decreases in steam temperature. These often involve overspraying
at HP superheater and reheater desuperheater outlets during startups as attemperation water
demand changes rapidly, and sudden quench cooling by forward migration of condensate in HP
superheaters and reheaters during startups. In this case the loading rates and minimum and
maximum temperatures are controlled by the distribution and gradients of metal temperature.

The high stress regions such as the header boreholes or weld toes of the tube to header
connections in components such as primary reheaters or HP superheaters, that are often designed
to operate at temperatures where the rate of creep deformation is relatively slow, may be
plastically deformed under short term transient thermal loading events such as forward flow of
condensate or unflashed attemperator water. After the transient condition has ended, the region
that yielded will have a residual stress that is opposite in sign to the transient stress and a residual
strain that is the same sign but of lower magnitude than the transient strain. During subsequent
sustained load operation these local regions with residual stress will attempt to relax. The
relaxation may occur by time dependent plastic deformation of the material or for the more creep
resistant materials such as T91 may relax by grain boundary cavitation. In either case additional
strain will accumulate leading to significant increases in the damage accumulation for each of
these thermal transient followed by steady state load stress relation events. At higher
temperatures oxide growth and cracking will participate and exacerbate the fatigue damage.

S-N Curves

Fatigue design curves are typically based on test results from small, smooth specimens, typically
loaded in uniaxial tension or rotating bending. The number of cycles to failure is determined
experimentally for a number of specimens tested under varying conditions. The simplest test is
to perform constant amplitude loading at a given R-ratio, for varying stress ranges. The test is
continued until the part fails, or failure is imminent. By plotting the stress range against the
number of cycles to failure, a stress range versus cycles to cracking curve can be generated. This

A-4
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

is usually expressed as an S-N curve, where S is the stress range and N is the number of cycles to
failure. In some cases, the fatigue curve is presented in terms of strain range rather than stress
range.

The S-N curve generated in this manner represents the mean failure curve. In other words, if one
used this curve in design, one would expect one-half of the parts to fail in N cycles with some of
the failures occurring in significantly fewer than N cycles. To convert the mean failure curve
into a design curve, the design curve is lowered by a correction factor. This correction factor can
be statistically based (e.g., a two-standard deviation from the mean that represents a 95% lower
bound) or it can be based strictly on factors of safety (either on the stress range or the number of
cycle to failure).

It should also be noted that any S-N curve is valid only for the specific conditions of the
underlying tests. Some of the variables that could lead to different S-N results are listed below:
• Test specimen v. component geometry
• Minimum to mean data scatter
• Test specimen v. component size
• Test specimen v. component surface finish
• Type of loading
• Mean stress
• Environment (i.e., temperature, water touched)

Endurance Limit

Many metallic materials, including steel, show an endurance limit for unnotched specimens.
This is a horizontal threshold in the S-N curve, below which cyclic loading causes no fatigue
damage.

However, under variable amplitude loading, the concept of an endurance limit is questionable.
High cyclic stresses early in life may cause damage (in the form of microscopic crack growth)
that causes fatigue damage at stress levels below the nominal endurance limit. In these cases, the
order of loading is important.

Weldments may not experience a threshold at all, due to the assumed crack-like defects present
in the weld, which grow under even very small cyclic stresses.

Since pressure boundary parts are not typically loaded with high-cycle, low amplitude loading,
the concept of the endurance limit may not be relevant to the design of most HRSG components.
The key exception is flow-induced vibrations, which is not considered in this report.

A-5
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Variable Amplitude Loading

The basic S-N curve assumes that all of the cycles of loading are identical, i.e., the same load
cycle is repeated over and over again until the part fails. In the real world, there are numerous
different types of magnitudes of stress cycles. All of the design codes use the concept of linear
summation of damage (also known as cumulative damage or Miner’s rule [A-26]).

Under linear summation of damage, the stress history is broken into a number of discrete
segments, with n1 cycles of stress range Δσ1, n2 cycles of stress range Δσ2, and so on. For each
segment, the fatigue damage ratio is calculated as di = ni/Ni, where ni is the number of applied
cycles at stress range Δσi, and Ni is the number of cycles to failure from the S-N curve at that
stress range. Fatigue failure is assumed to occur when the sum of all the di values equals 1.0.
This value of the sum of the di is often called the fatigue usage factor or the fatigue damage ratio:

ni
d =∑ Equation A-1
Ni

Many of the design codes have added adjustment factors that limit the permitted fatigue usage
factor when the anticipated loading spectrum is not well defined. Most of the standards allow a
fatigue usage of 1.0 because the design curves themselves are perceived to have sufficient built
in conservativism. A number of codes limit the estimated fatigue usage factor. For example:
• BS PD 5500 has a simplified fatigue design approach that conservatively limits the fatigue
usage to 0.6
• TRD 301 restricts the number of cold starts such that the fatigue usage from cold starts alone
must be less than 0.2. For variable amplitude loading, TRD 301 restricts the fatigue usage
factor to 0.5.
• EN 12952 allows a usage factor of 1.0, but for the simplified analysis when the fatigue loads
have not been specified and 2000 cold starts have been assumed, the usage factor is limited
to 0.4.

Fatigue Curve or Fatigue Stress Correction Factors

Adjustments or correction factors must be applied to empirical S-N curves in order to develop
design fatigue curves that account for the differences between a laboratory fatigue test and the
range of anticipated field conditions.

For the most part, the design curves in different design codes adopt different approaches to these
correction factors. For example in the ASME Section VIII, Div.2 fatigue curve a worst case
mean stress correction is included in the fatigue curve and no further mean stress correction is
required. In contrast, the TRD 301 and EN 12952-3 fatigue rules require a mean stress correction
to be estimated and used to adjust the estimated stress range and no mean stress correction has
been included in the fatigue curves.

A-6
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

In addition to the mean stress correction, the ASME Section VIII, Division 2 fatigue curves are
corrected for the following other factors:
• A correction of 2 on cycles for experimental scatter in the fatigue data from the mean to
minimum values
• A correction factor of 2.5 on cycles to compensate for the possible increased size of the
component/feature being assessed compared with the laboratory fatigue specimens
• A correction factor of 4 on cycles to account for rougher surfaces and the effect of the
atmosphere of the component/feature being assessed compared to the smooth, laboratory
atmosphere conditions used to develop the fatigue curve

These individual factors were the basis for the total correction factor of 20 on cycles used in the
development of the ASME fatigue design curve. These factors have frequently been called safety
factors by HRSG designers whereas this was never the intent of the ASME Code writers. The
factors were judgments made during the development of the code to provide realistic estimates of
the fatigue life of real components/features based on the fatigue data from tests that were far
more benign than what most real components will experience. It is incorrect to assume that the
actual cycles to cracking will be twenty times longer than predicted by the corrected fatigue
curves. For example, if corrosion-fatigue or creep-fatigue are active damage mechanisms then
larger correction factors are needed to predict the cycles to cracking with no extra safety
margins.

The term “correction factor” as used here is not a factor of safety per se, but represents the
difference between the mean fatigue line from the supporting test data and the design curve
defined by the code or standard. These correction factors can be defined on stress or on cycles
(life).

The ASME Codes incorporate a correction factor of either 2 on stress or a factor of 20 on life.
The original BS 5500 standard incorporates a correction factor of 2.2 on stress or a factor of 15
on life. TRD 301 and EN 12945 include a correction factor of 1.5 on stress or a factor of 10 on
life. In each of these codes, whichever correction factor produces the lowest life is used.

Some of the S-N curves are linear (on a log-log basis), or at least segmented linear. Others,
including ASME, TRD and EN design codes, use a curved S-N relationship that is approximately
tangent to the straight line S-N curves in the mid-cycle region and is concave upward.

For the linear S-N curves, the relationship between a correction factor on stress and a correction
factor on life is fixed. However, applying the two correction factors to curved S-N relationships
is not as straight forward. At low cycles, the correction factor on life provides a more
conservative lower bound than the correction factor on stress. Conversely, at high cycles the
correction factor on stress provides a more conservative lower bound than the correction factor
on life. The transition between these two correction factors may not be smooth. As an example,
the ASME Code has a design curve that contains a cusp at the transition between the two criteria
at about 1.2x104 cycles.

A-7
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Another way of correcting laboratory fatigue test data into the design fatigue curves is to specify
the number of standard deviations the design curve is drawn below the mean or median data.
API 579 claims a design curve 2 standard deviations below the mean, EN 13445 uses 3 standard
deviations, and the original BS 5500 used 4 standard deviations below the mean. This method of
specifying the degree of conservatism lends itself to a statistical definition of the probability of
failure. The standard deviation, and thus the degree of conservatism, depends on the number of
data points that were used to develop the mean or median fatigue curve.

Design curves for weldments fall into two general categories, based on the type of specimen used
to generate the underlying data. The ASME Codes, TRD 301, and EN 12952 use S-N curves
based on tests of unnotched specimens. BS PD 5500 and AD-Merkblatt S2 use S-N curves
based on tests of weldments. API 579 and EN 13445 provide both types of S-N curves, although
the API 579 curve is the same as the ASME Section VIII curve. The original BS 5500 used a
compromise solution, using unnotched test specimens derived from ground flush weldments. In
some cases the design curves for weldments are broken into two further subsets – as-welded and
stress relieved – but as noted previously this is rarely done because it is difficult to ensure an
adequate stress relief.

Figure A-2 shows a comparison of the various S-N curves for unnotched specimens. The ASME
curves and the BS 5500 curve are based on stress amplitude rather than stress range, so the
curves were adjusted by a factor of two for comparison with the other curves. In the 1000-10000
cycle range (the expected number of cycles for HRSG components that operate below 750oF
(400oC)) the curves are similar, except for the BS 5500 curve. However, the BS 5500 has a
higher combined set of correction factors to account for statistical variation, surface roughness,
size effects, etc. than the other curves (see the discussion on fatigue curve correction factors
above).

The S-N curves for weldments are shown in Figure A-3. Two sets of curves are shown, to
compare similar weld details. The fillet welds represent API Class 100, BS PD 5500 Class F2,
and EN 13445 Class 100. The butt welds represent API Class 40, BS PD 5500 Class W, and EN
13445 Class 32. The EN 13445 curves provide the most conservative design basis.

A-8
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Figure A-2
Fatigue Design Curves for Unnotched Specimens

Figure A-3
Fatigue Design Curves for Weldments

A-9
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

The complexity and significant differences in methods used among various design
codes/standards is based on the numerous correction factors that are employed to relate fatigue
curves based on laboratory data to the characteristics of real pressure part design features. Fifteen
correction factors identified below must be determined as part of the assessment and these
correction factors must be appropriately linked with the stress definitions (e.g., nominal stress,
structural stress, hot spot stress, peak stress, etc) used.

Fatigue assessments are based on simple relationships between the stress range, Δσ, and the
number of cycles to cracking, N. In some cases, the relationship is a straight line that can be
described by an equation of the form:

N = C Δσ n Equation A-2

where:

N= cycles to cracking

C, n = empirical values that are a function of the material, test specimen type, test loading, test
temperature and test environment. The value of n is usually in the range of –1.8 to -5.0. For
welds the value of n is usually close to -3

In other cases, the basic relationship is a curve, either defined by an equation or by a figure in the
Code. In those cases, there is something equivalent to the value C in Eq. A-2 that adjusts the
basic curve to account for differences between the test conditions and the real world conditions.

C =falloy (ftest specimen)(fsize)(ftest loading)(ftemperature)(ftest environment)(fstatistical) Equation A-3

falloy = material coefficient


ftest specimen = test specimen correction factor
fsize= test specimen to component size correction factor
ftest loading = test loading correction factor
ftemperature = service temperature correction factor
fenvironment = service environment correction factor
fstatistical = correction factor to adjust from mean or median fatigue curve to lower bound
fatigue curve
One of the primary differences between the common fatigue rules is the stress definition and the
way the correction factors are grouped and estimated.

Δσstructural = [σthermal (fthermal)+σpressure(fpressure)+σexternal(fexternal)]max Equation A-4


- [σthermal (fthermal)+σpressure(fpressure)+σexternal(fexternal)]min
fthermal= stress concentration factor (SCF) for thermal component of stress
fpressure = SCF for pressure component of stress
fexternal = SCF for component of stress resulting from externally applied forces and
moments

A-10
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Δσ = notch or peak stress range


= Δσstructural (fsf)(fround)(fconfig)(fweld)(fmean stress)(fe-p) Equation A-5
fsf = surface finish/mill scale correction factor
fround = out-of-roundness correction factor
fconfig= connection configuration correction factor
fweld= weld configuration/quality correction factor
fmean stress= mean stress correction factor
fe-p= elastic-plastic correction factor
Table A-1 below illustrates which correction factors are used for the fatigue assessment of
header or drum borehole thermal shocks performed in accordance to EN12952-3. It is
recommended that the table be used to verify that key correction factors have been included in
the analysis for all HRSG pressure part fatigue analyses. For EN12952 this review revealed that
the most significant missing correction factor is the test loading correction factor. The very
significant detrimental influence of slow loading rates and long hold times for components
subjected to creep fatigue is not included in EN12952-3.
The labaoratory testing that the was the basis of the fatgue design curves was done with rapid
loadings and not tensile or compressive hold periods. To overcome this deficiency it is
recommended that the total fatigue life consumption (i.e., usage factor) estimated using
EN12952-3 for components operating at steady load temperatures above 430oC (800oF) be
limited to 0.10. Alternatively an analysis such as ASME Code, Section III, Division 1,
Subsection NH that incorporates the detrimental influences of elevated temperature hold times
should be performed.

Component Stresses
For a thorough fatigue evaluation of a specific location it is important to ensure that each of the
component stresses 1 have been considered. These component stress values must be elevated
appropriately to account for stress concentrations. Concentrated component stresses acting in the
same direction are then summed up to determine the principal stress value in that direction and
combined with the other principal stresses and converted into an equivalent multiaxial stress (i.e.,
sometimes referred to as a stress intensity value).
It is recommended that a table of component stresses be prepared such as Table A-2, which
considers the evaluation of header or drum borehole pressure and thermal shock stresses at the
header ID to borehole intersection and Table A-3 which considers the tube side weld toe of a
tube to header connection being acted on by external forces and moments, pressure and local
temperature gradients. This check table will verify that the nominal stress calculations and
associated stress concentration factors are being combined appropriately. There are a variety of
stress concentration factors used in the design codes. A few examples based on the EN12952-3
code are included in Tables A-2 and A-3.

1
The term “component stress” as used here means a stress for a given component of loading such as pressure
loading, thermal loading, etc. It should not be confused with the components of multiaxial stresses that act in
different directions, which are used to calculate the principal stresses. Multiaxial and principal stresses are
discussed in a later section.

A-11
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Table A-1
Fatigue Curve Correction Factors in EN 12952-3
N = CΔσn EN12952-3 Evaluation of Borehole Thermal Shock
N = cycles to cracking
C, n = empirical values that are a function of the material, test C = f (room temperature tensile strength)
specimen type, test loading, test temperature and test n = -1.828
environment.
C = falloy (ftest specimen)(fcsize) (ftest loading) (ftemperature) (ftest environment)
(fstatistical)
falloy = material coefficient C = f (room temperature tensile strength)
ftest specimen = test specimen correction factor Not included
fsize= test specimen to component size correction factor Not included
ftest loading = test loading correction factor Not included. No hold time or rise time corrections are included, therefore
creep fatigue damage is not considered and a separate correction factor
needs to be used for conditions where creep fatigue damage may be
active. To overcome this deficiency it is recommended that the total fatigue
life consumption (i.e., usage factor) estimated with EN12952-3 for
components operating at steady load temperatures above 430oC (800oF)
be limited to 0.10.
ftemperature = service temperature correction factor Included as a factor Ct*. The final estimated peak stress range is divided
by Ct*
fenvironment = service environment correction factor A check is included to ensure that the critical tensile and compressive
strain limits of the oxide scale are not exceeded
fstatistical = correction factor to adjust from mean or median Factors SL = 10 and SS =1.5 are used to reduce the median fatigue curves
fatigue curve to lower bound fatigue curve by a factor of 10 on cycles or 1.5 on stress range, whichever produces the
shortest life estimation. An additional limit of 0.4 on the fatigue usage
factor is required when the fatigue loading is not well defined and 2000
cold starts have been assumed
Δσstructural = [σthermal (fthermal)+σpressure(fpressure)+σexternal(fexternal)]max
- [σthermal (fthermal)+σpressure(fpressure)+σexternal(fexternal)]min
fthermal= stress concentration factor for thermal component of Included as αt which has values between 0.8 and 2.0 which depends on
stress the heat transfer coefficient and the ratio of the tube and header mean
diameters
fpressure = stress concentration factor for pressure component of Included as αm which has values from 2 to > 10 depending on the ratio of
stress the tube and header nominal thickness and mean diameter values and the
header mean radius to nominal thickness. In certain cases the value of αm
may be increased or reduced by 10% depending on the tube to header
connection and weld detail design and quality

A-12
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

fexternal = stress concentration factor for externally applied forces Limits are included for external forces
and moments component of stress
Δσ = notch or peak stress range
= Δσstructural(fsf)(fround)(fconfig)(fweld)(fmean stress)(fe-p)
fsf = surface finish/mill scale correction factor Included as Cko which is a function of the surface roughness and room
temperature tensile strength
fround = out-of-roundness correction factor Not included
fconfig= connection configuration correction factor Included as Ck1, Ck2, Ck3 which are a function of the tube to header
connection and weld detail design and the estimated cycle to failure.
These are not applied for analysis of borehole thermal shock evaluations
fweld= weld configuration correction factor Included as Ck1, Ck2, Ck3 which are a function of the tube to header
connection and weld detail design and the estimated cycle to failure.
These are not applied for analysis of borehole thermal shock evaluations
fmean stress= mean stress correction factor For elastic conditions a Gerber mean stress correction is used. For
partially plastic conditions the Gerber mean stress correction is used but
the mean stress value is assumed to be the cycle temperature based yield
stress minus half the peak elastic stress range
fe-p= elastic-plastic correction factor For fully plastic conditions, when the cyclic stress exceeds twice the yield
stress, an elastic-plastic correction is made based on the Neuber rule and
an elastic-fully plastic stress assumption. The total elastic plastic strain
range determined with the Neuber/elastic perfectly plastic behavior:
Δε = Δσ2/ [(2 x Ret*)(E)]
is then converted back to a virtual elastic stress value.
Δσvirtual, elastic = Δε (E)
The combined relationship which is used in EN12952-3 is:
Δσvirtual, elastic = Δσ2/ [(2 x Ret*)]

A-13
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Table A-2
Component Stresses For a Borehole Thermal Shock Fatigue Evaluation
Component stresses for a borehole thermal shock fatigue evaluation
for Location A shown to the right

Location Load Stress Orientation Stress equation Anticipated Comments


(with respect to the range of
drum or header) SCF
Drum or Throughwall Tangential (hoop), σt (αt)(βLt*)(E)(Tmidwall – Tid)/(1-ν) 0.8 to 2 See αt , EN12952-3, Fig. 13.4.8). The
header temperature nominal stress is based on a flat plate
inside gradient thermal stress solution
diameter to Axial, σax (αt)(βLt*)(E)(Tmidwall – Tid)/(1-ν) 0.8 to 2 Not required to be determined since it is the
borehole same as the maximum principal stress (hoop
intersection stress)
in Radial (through thickness), 0 on surface NA
longitudinal σr
plane Internal Tangential (hoop), σt (αm)(P)(dms)/(2ems) 2.2 to 5 See αm , EN12952-3, Fig. 13.4.5). The
(Location A) pressure nominal stress is based on the header mean
diameter stress without the borehole present

A-14
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Axial, σax ~½ the hoop stress Not required to be detemined since it is


between the maximum principal stress (the
hoop stress) and the minimum principal
stress (the radial stress)
Radial (through thickness), -P
σr

Notes:
αt = stress concentration factor for thermal stress
βLt*= linear thermal expansion coefficient
E = modulus of elasticity
Tmidwall, Tid = midwall and inside surface temperatures
ν = Poisson’s ratio
αm = stress concentration factor on pressure stress
P = gauge pressure
dms = mean diameter of header
ems = nonminal thickness of header

A-15
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Table A-3
Component Stresses For A Fatigue Evaluation of a Quench Event

Component stresses for a fatigue evaluation of the


tension (during quench event) sideof a tube to header
connection with external loads, internal pressure, and
local temperature gradients (Location A shown in the
figure to the right)

Location Load Stress Nominal stress Anticipate Comments


Orientation equation d range of
(with respect to SCF
the tube)
Fillet weld External forces and moments Tangential Insignificant
toe on tube resulting from (i) constrained thermal (hoop), σt
side of tube expansion tube loads due to tube Axial, σax F/Atube + Moop/Ztube 2 to 5 SCF strongly influenced by
to header temperature differences or tube alloy amount of tube offset as well
connection transitions within the harp, (ii) piping as weld design, etc
on the side loads on the header, (iii) redistribution Radial (through Zero on surface
of the tube of dead weight due to tube thickness), σr
loaded in out temperature differences
of plane Internal pressure Tangential (αm)(P)(dms)/(2ems) Can usually be ignored since
bending (hoop), σt the combined hoop stress due
when the to external forces plus
tube is pressure will be between the
colder than maximum axial and minimum
its neighbors radial principlal stresses due to
Location A) these loads
Axial, σax P (Aitube)/(Atube) 2 to 5
Radial (through Zero on surface
thickness), σr

A-16
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Local temperature gradients Tangential Usually insignificant See EN12952-3, Para. 13.3.5
(hoop), σt for temperature gradient limits
Axial, σax Usually insignificant
Radial (through Zero on surface
thickness), σr
Notes:
F = axial force
Atube = cross-sectional area of tube
Moop = out of plane bending moment
Ztube = section moduls of tube
P = gauge pressure
Aitube = cross-section area of tube bore

A-17
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Stress Concentration Factors

Stress concentration factors (SCFs) are perhaps the most difficult and confusing aspect of fatigue
analysis of HRSG pressure parts. The main source of confusion with regard to stress
concentration factors is the numerous terms used to describe SCFs. These generally fall into two
categories, the first being factors that are used to estimate the increase in stress resulting from a
gross structural discontinuity such as a header borehole or tube to header connection and the
second being factors used to estimate the increase in stress associated with local structural
discontinuities such as surface roughness, cracked mill scale, or weld characteristics. Some
examples of the terms used are:
• Stress intensification factors, SIF. These are experimentally derived factors in EN 13480-3
and B31.1 and ASME Section III which are relevant to connections acted on by external
loading. SIF factors account for the gross structural discontinuity associated with a tube to
header or nozzle to header connection acted on by external bending moments but they only
partially account for local weld characteristics. The basis of ASME SIF values was a
comparison of fatigue stress values of a specific connection to the performance of a welded
butt joint in 100 mm (4”) size pipes [A-27]. After developing the SIF values in this way it
was discovered that the SIF values for connections with carefully prepared and flush ground
welds was in the range of 0.5 to 0.64, thus all the SIF values are a factor of 1.6 to 2 lower
than would be anticipated for stress concentration factors for welds with characterisitics that
are typical of industry practice.
• Stress indices, i. These are incorporated in ASME Section VIII, Division 2 and are used to
estimate each of the three orthogonal pressure stress components for specific locations such
as the longitudinal plane, inside corner of a header borehole for nozzles and laterals based on
a nominal header hoop stress made with the assumption that the borehole was not present.
The stress indices are used to increase the value of nominal internal pressure hoop stress
(calculated without the gross structural discontinuity present). They are not relevant to
stresses produced by loads other than internal pressure. The range of validity of the stress
indices often does not make them useful for certain HRSG pressure part nozzles or laterals.
In EN12952-3 the factor that is used to estimate the borehole pressure stress from the
nominal header mean diameter hoop stress is referred to as a stress concentration factor.
• Fatigue strength reduction factors, FSRF. This is a factor in ASME Section VIII, Division
2 that is multiplied to the equivalent stress values which accounts for local structural
discontinuities. For example for attachments to pressure vessels made with fillet welds a
fatigue strength reduction factor of 4 is required [A-28]. Similar factors are included in
EN12952-3 but they are referred to as “micro-notch effect” surface finish and weld
correction factors. Ref. [A-29] provides a review of FSRFs for pressure vessel components.

SCFs are sometimes determined analytically and sometimes are determined experimentally.
These two approaches are often produce significantly different SCF values. Analytically derived
stress concentration factors are also prone to signicant errors. Stress concentration factors
derived with finite element models or other analytical methods may vary widely depending on
the weld toe angle assumed and the mesh size and characteristics of the model. Significant
errors may also occur if overly simplistic or inappropriate models are used to calculate stress

A-18
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

concentration factors. For example The analytical models for “Local Stresses in Spherical and
Cylindrical Shells due to External Loadings” published in Welding Research Council bulletin
WRC 107 [A-30] were based on branch connections that were modeled as solid plugs rather than
hollow cylinders. The stress concentrations factors derived from from these models are probably
conservative for the evaluation of the header side of a tube to header connection but may be
significantly nonconservative for tube side of the tube to header connection.

The work summarized in WRC 297 [A-31] attempted to overcome this deficiency by updating
the WRC 107 approach for cylinder to branch connections modeled with hollow cylinders.
WRC 297 is useful for some HRSG tube to header and nozzle pipe to header connections but
many common header configurations are outside the range of validity the WRC 297 models. In
many cases, further experimental testing or analytical modeling is required to determine an
appropriate gross and local structural stress concentration factors for tube to header or nozzle
pipe to header connections that are acted on by external forces and moments.

SCFs are intended to be used with specific loading modes, locations, stress components and
nominal stress formulas (see Tables A-2 and A-3). SCF values used to correct for local “micro-
notch” discontinuities rather than gross structural discontinuities are usually applied to the
estimated equivalent stress (stress intensity value). Stress values used in fatigue assessments
must include adjustments for both the gross and micro-notch effects. Many of the design codes
begin by estimating the nominal stress that would exist in the absence of both the gross and
micro-notch structural discontinuities. Examples include:
• The flat plat thermal shock equation σthermalshock = βLt*(Et*)(Tmidwall – Tid)/(1-ν)
• The mean diameter pressure stress equation (σ = p(Rm/t)
• The thick walled cylinder bending stress equation σb = M/{π(Ro4 – Ri4)/32/Ro}

These simplified nominal stress values must be corrected to account for gross structural
discontinuities (e.g., flat plate to header curvature corrections, borehole corrections, nozzle pipe
to header or tube to header connection stiffness corrections) and local structural discontinuities
(e.g., cracked mill scale, surface roughness, local variations in weld toe angle, or other weld
discontinuities). Figure A-4 shows an example of a tube to header connection that had been cut
out after a few years of service due to a throughwall creep-fatigue crack that initiated at tube side
weld toe at a circumferential location that did not have the highest bending stress but had the
most severe fillet weld toe angle. It was concluded that the severe weld toe radius significantly
increased the localized stresses at the weld toe. No cracking occurred during the short (few year)
service period at locations where the weld toe radius was 1/8-inch (3.2 mm).

A-19
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Figure A-4
Example of Local Variations in Weld Toe Angle that are Undersirable but also Commonly
Seen in Tube to Header Connections

To estimate the cyclic thermal stress at the inside corner of a header borehole the minimum and
maximum flat plate thermal shock stress values must be first corrected to account for the
curvature of the header and the presence of a borehole. Figure A-5 compares the borehole
thermal stress correction factors from TRD 301, Annex 1 and EN12952-3. For a typical range of
HRSG tube to header mean diameter ratios the anticipated value of this borehole thermal stress
gross structural discontinuity correction factor is in the range of approximately 1.4 to 2.0.

A-20
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Figure A-5
Comparison of Borehole Thermal Stress Correction Factors

This corrected thermal stress must be corrected by an additional amount to account for very
localized surface discontinuities such as cracked mill scale or surface roughness at the
header/borehole intersection. Unless welds or weld repairs are present in the vicinity of the
borehole to header inside diameter corner no micro-notch weld correction is needed. For smooth
hot finished boreholes with mill scale present the micro-notch stress correction factor will be in
the range of 1.1 to 1.6 for the commonly used HRSG materials. For machined surfaces the
surface roughness factors included in EN12952-3 are shown in the Table A-4 below. A number
of HRSG pressure part fatigue failures have initiated from within machining groove that had
peak to valley heights that were at the upper end or exceeded the levels shown in this table.

Table A-4
Micro-Notch Surface Roughness Stress Correction Factors for Non-Welded Machined
Surfaces

Peak to valley height of surface Room temperature tensile Stress correction


roughness, Rz strength factor

< 6 μm (240 μinch) Any 1.0

6 μm to 50 μm (1,970 μinch) 400 MPa (58 ksi) 1.1

6 μm to 50 μm 1000 MPa (145 ksi) 1.3

A-21
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

It is useful to consider the importance of these values on the possible error in fatigue life
assessments. Recall that the cycles to failure are related to stress range as follows:

Nf = C (Δσ)n

The value of n determines the stress sensitivity of the cyclic life. Values of n reported in
literature for tests relevant to HRSG components range from approximately -1.8 (for the fatigue
curve in EN 12952-3) to -5 (for the Markl fatigue curve [A-27]). The table below summarizes
the relative reduction in cycles to failure for increased stress range values. For each 10% (factor
of 1.1) increase in stress range the estimated life will be reduced 16% to 38%. For every factor of
1.2 to 1.5 increase in stress range the life will be reduced by half. There will be an order of
magnitude reduction in life for stress range increases of a factor of 1.6 to 3.6.

Table A-5
Sensitivity of cyclic life on stress range changes

For n = -1.8 For n = -5

Δσ2/Δσ1 Nf2/Nf1 Nf2/Nf1

1 100.0% 100.0%
1.1 84.2% 62.1%
1.2 72.0% 40.2%
1.3 62.4% 26.9%
1.4 54.6% 18.6%
1.5 48.2% 13.2%
1.6 42.9% 9.5%
1.7 38.5% 7.0%
1.8 34.7% 5.3%
1.9 31.5% 4.0%
2 28.7% 3.1%
2.1 26.3% 2.4%
2.2 24.2% 1.9%
2.3 22.3% 1.6%

A rough machined surface or a surface with thick mill scale will be expected to fail in
approximately half the number of cycles as a smoothly machined surface, free of mill scale given
the same loading history. It is recommended that fatigue calculations be performed with
somewhat conservative assumptions with regard to the micro-notch (local structural
discontinuity) correction factors. In addition it should be apparent that even relatively minor
underestimations of the magnitude of the gross structural discontinuities stress correction factors
will introduce significant nonconservative errors in the estimated cycles to failure.

A-22
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

To estimate the internal pressure component of the borehole cyclic stress the header mean
diameter stress must be corrected to account for the presence of the borehole and then corrected
an additional amount to account for mills scale or surface roughness micro-notches. Figure A-6
provides a comparison of borehole gross structural discontinuity correction factors. Note that
there are significant differences in the recommended correction factors which could produce
pessimistic or optimistic errors in fatigue life estimations. If credit is to be taken for the use of a
machined borehole surface then strict limits on and verification of the final surface roughness
need to be achieved. The EN12952-3 values shown in Figure A-6 do not include any mill scale
or surface roughness correction factors, so these corrections must be made in addition to the
initial borehole corrections are completed. For the typical range of tube to header mean diameter
ratios and header mean diameter to thickness ratios the anticipated value of this factor is in the
range of approximately 2.8 to 4.3.

Figure A-6
Borehole Pressure Stress Correction Factors (Note: dm= mean diameter of tube, Dm= mean
diameter of header and Tm=thickness of header)

To estimate the local stress at the tube side weld toe of a tube to header connection acted on by
internal pressure and external forces and moments the nominal bending stress must be corrected
to account for the increased stress associated with the tube to header or nozzle pipe to header
stiffness discontinuity. Figure A-7 compares three alternative estimations of the gross structural

A-23
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

correction factor for tube to header or nozzle pipe to header connections made with unreinforced
fabricated (welded) tees, UFT.

The stress intensification factor (SIF) shown in Figure A-7 is the ratio of the tube to header
bending stress (excluding any stress factors for local structural discontinuities associated with the
toe of the fillet weld) to the nominal tube bending stress resulting from external forces based on
the assumption that the gross structural discontinuity associated with the tube to header
connection did not exist (i.e., nominal stress = bending moment/ section modulus of the tube).
The red circles that are filled in had characteristics that were outside the range of the WRC 297
parameters.

Figure A-7
Comparison of Various Tube to Header Stress Intensification Factors for a Wide Range of
HRSG Headers. (Note:Tnom=nominal thickness of header, Dm = mean diameter of header)

The ASME B31.1, ASME Section III and EN 13480-3 codes have similar solutions for these
stress intensification factors. They allow an UFT to be considered as a “branch connection” if
certain condition are met such as the spacing between adjacent branch connection having a
center to center spacing that is larger than three times their inside radii in the longitudinal
direction and two times the inside radii in the circumferential direction. Five other factors must
also be met to use the branch connection SIF. These include fillet weld toe radius limits,
reinforcement limits, radial entry angle being 90-degrees. ASME Section III places a lower limit
of 2.1 on the SIF for a branch whereas ASME B31.1 doesn’t truncate the value of estimated SIF.

A-24
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

The estimated values of SIF using the B31.1 equation for branch connection for tube to header
connections leads to SIF values below one. It is likely that these values are well outside the range
of the empirical and analytical information used to derive the branch connection SIF values. Due
to the importance of and the confusion with SIF values used for fatigue modeling of HRSG
pressure parts a separate discussion of SIFs is included on the next page.

As an interim recommendation it is suggested that the SIF values obtained using the relationships
in WRC Bulletin 297 [A-31] be used for the fatigue assessment of tube side of tube to header
connections where the tube and header characteristics are within the range of applicability of the
tabulated correction factors. For nozzle or tube to header connections that fall outside the WRC
297 limits, full size fatigue tests should be performed of the connection using out of plane
bending cyclic loads to determine an appropriate value of SIF. Alternatively, analytical estimates
could be used to develop appropriate SIF values for the connection.

This corrected local tube to header or nozzle pipe to header stress must be further corrected for
anticipated small, localized notches associated with the toe of the fillet weld. This second
correction is made to account for local structural discontinuities such as surface roughness,
cracked mill scale, or local weld discontinuities such as a very local variation in weld toe angle.
Figure A-4 provides an illustration of a significant local variation in fillet weld toe angle. In a
single tube the weld toe radius varied around the periphery of the connection from 1/8” (3.2 mm)
to 1/64” (0.40 mm). After a few years of service cracking occurred in the tube. The cracking
initiated at the location with the sharpest weld toe angle even though this was not the position of
highest bending stress. No cracking had occurred where the weld toe angle was more rounded.

For a wide range of common HRSG tube to header connection dimensions ASME B31.1 limits
the the tube side weld toe radius to values in the range of 3mm to 4 mm to meet the branch
connection limitations [A-32]. ASME Section VIII, Division 2 limits this radius to values in the
range of 6.5 mm to 8 mm for nozzle connections that are analyzed with the stress indices for
pressure stress [A-33]. AWS D1.1 recommends a toe radius of 8mm for enhanced fatigue
performance or alternatively recommends using a 5 mm burr radius when toe dressing with a
burr grinder is used to enhance fatigue performance for components with wall thickness values
less than 20mm (0.79-inch) [A-34].

A 1/8” (3.2 mm) weld toe radius is readily achieved for most tube tube header configurations and
may be acceptable when the cyclic stresses are not overly severe. When enhanced fatigue
performance is required, a weld toe radius of 5/16-inch (8 mm) should be specified and verified
using the “coin” test.

In addition to the recommendations above regarding specification and control of tube side weld
toe radius, it is recommended that the weld correction factors in EN12952-3, Section B.5 be used
for the micro-notch (local structural discontinuity) stress correction factor when assessing tube to
header or nozzle pipe to header connections.

A-25
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Stress Intensification Factors (for Analysis of Connections Acted on by


External Forces and Moments)

For fatigue analysis of tube to header, nozzle pipe to header or riser or downcomer pipe to drum
connections subjected to external (e.g., flexibility and global constrained thermal expansion)
forces resulting from piping loads and temperature differences between tubes connected to
common headers the intensified stresses at the tube connection are commonly determined using
stress intensification factors, SIFs, applied to the estimated nominal bending stress. These
intensified stress values are then further corrected using micro-notch (local structural
discontinuity) stress concentration factors to account for surface roughness or anticipated
variations in weld discontinuity characterisitics.

Stress intensification factor formulas for pipe to cylindrical branch connections are incorporated
into ASME B31.1, ASME Section III and EN13480-3. These stress intensification factors have
been used by HRSG designers to evaluate the flexibility and fatigue design of tube bundles and
tube to header connections, especially for tube bundles that include header partition plates and
multiple passes or that are anticipated to have transiently high row to row or tube to tube
temperature differences within an individual row. It is useful to review the basis of the stress
intensification factors and how they are applied in design calculations.

The pipe to cylinder branch connection stress intensification factors incorporated in ASME
B31.1 were originally derived by A.R.C. Markl in the 1940s [A-27] based on fatigue tests
performed on 4-inch to 4-inch pipe assemblies. Markl found that the fatigue lives of butt welded
pipe to pipe connections were well represented with the following relationship:

(ΔS/2) = S = C x N0.2 Equation A-6

where:
C = 245,000 psi (1690 MPa) for median fatigue life
ΔS = nominal stress range (total cyclic moment divided by the section modulus of the pipe)
S = nominal stress amplitude = ΔS/2
N = Number of cycles to failure

For other more complex welded connections Markl found that equation A-6 could be modified
by an experimentally determined stress intensification factor, SIF.

SIF = C x N0.2/(ΔS/2) Equation A-7

Further evaluations led to the conclusion that the SIFs could be related to geometric, “flexibility
characteristics” of the components that were fatigue tested. For example, for a header to branch
tee with a welded-in, welded-on or extruded nozzle with no extra reinforcement the SIF could be
related to the nominal thickness, tn and mean radius, Rm of the header with the following
relationship:

SIF = 0.9/(tn/Rm)2/3 Equation A-8

A-26
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

The allowable nominal stress range, ΔS or the cycles to failure, N, for the 4-inch (106.6 mm)
pipe butt weld connections could be determined by combining Equations A-7 and A-8.

(ΔS/2) 0.9/(tn/Rm)2/3 = C x N0.2 Equation A-9

There is considerable debate regarding the accuracy and range of applicability of existing SIF
and flexibility characteristic relationships. It is not clear if the existing relationships are
appropriate for these tube-to-header connections. The technical basis for the B31 approach
includes work by Markl [A-27] and Rodabaugh [A-35, A-36].

Figure A-8 shows the stress intensification factor (SIF) data for the entire unreinforced,
fabricated tee tests reported in [A-27, A-35, A-36] plotted as a function of the radius/thickness
ratio, which is an important part of the B31 approach to calculating the SIF.

Figure A-8
Effect of Header Flexibility Factor (Mean Radius to Nominal Thickness Ratio, R/T) on the
“Fatigue Test Derived” Stress Intensification Factor, SIF

The in-plane bending (IPB) tests are shown as dark blue diamonds and the out-of-plane bending
(OPB) tests are shown as magenta squares. The points indicated by an X are the values of the
SIF predicted by the B31 equation for the geometry corresponding to each of the test points.

The SIF relationship for unreinforced fabricated tees (Equation A-8) is estimated based only on
the header R/T ratio and is independent of the nozzle or lateral dimensions. This stress
intensification factor is probably most relevant to the header side of a tube to header or nozzle

A-27
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

pipe to header connection. It is unlikely to provide useful estimates of the gross structural stress
correction factor for the tube side of tube to header connections.

The typical range of R/t values of tube-to-header connections in a typical HRSG is shown in
Figure A-8. The range of R/T values are shown as a bar along the bottom of the graph, with the
minimum, mean, and maximum R/t values shown by diamonds. It is clear from this graph that
none of the existing experimental data cover the range of R/T values appropriate to tube-to-
header connections.

About half of the data shown were developed for in-plane bending (where the loads applied to
the branch were oriented in the longitudinal direction of the header). The out of plane bending
loading is of primary interest with respect to tube to header analyses.

In tube-to-header connections there is often a large difference in the size of the tube and the
header. Figure A-9 shows the effect of the r/R ratio and Figure A-10 shows the effect of the t/T
ratio on the SIF, where the lower case letter refers to the tube and the capital letter refers to the
header.

Figure A-9
Effect of r/R Ratio on the Stress Intensification Factor

A-28
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Figure A-10
Effect of t/T Ratio on the Stress Intensification Factor

These plots demonstrate that the data used as the basis for the B31 approach are not
representative of the geometries of typical HRSG tubes and headers.

EN13480-3, ASME B31.1 and ASME Section III all have stress intensification factors for
“branch connections.” All three of these codes use the same relationship for the branch end SIF
which includes R/T, r/R, t/T, and a thickness transition geometry factors to attempt to account for
the inherent flexibility of both the header and branch.

To use these branch connection relationships the connection must meet a variety of conditions,
none of which would readily exclude the application of these SIF for HRSG tube to header
connections made with complete joint penetration welds and controlled weld toe angles.
Unfortunately these branch connection solutions do not produce meaningful results for common
HRSG tube and header dimensions. For a wide range of common HRSG tube to header
configurations and fillet weld sizes the SIF values determined with the branch connection
relationship were in the range of 0.10 to 0.50. Clearly these are not sensible.

Finite element modeling of HRSG tube to header connections indicates that the tube side stress
magnitude in a wide region near the weld toe is amplified by at least a factor of two and in some
cases will be greater than three. Of the three codes mentioned, only ASME Section III has a
lower limit on the branch side SIF. Section III limits this SIF to a value of 2.1. It is recommended
that code branch connection SIF relationships not be used for the fatigue assessment of HRSG
tube to header connections.

A-29
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Recall that Markl’s SIF approach was benchmarked against the fatigue performance of pipe to
pipe butt weld connections. Markl [A-27] stated that his tests “proved a stress-raising factor of
the order of 1.5 in relation to plain pipe”. Rodabaugh [A-35] later commented that “Markl’s
representative girth butt welds (for which if = 1.00) had a local stress concentration factor of
about 2.0”. Others have suggested that the baseline SCF for pipe butt welds is closer to 1.5.

Figure A-11 below compares the Markl fatigue data to fatigue design curves for a number of
design codes. The Markl data have been adjusted by a factor of 2 to convert the reported stress
amplitudes into stress ranges, and then increased by an additional factor of 2 to represent the
weld micro-notch (local discontinuity) weld stress concentration factor (based on Rodabaugh’s
conclusion).

Figure A-11
Comparison of Markl Data with SCF to Corrected* Fatigue Design Curves

* Note: All curves have been corrected to represent mean failure data without stress
concentration and other correction factors. The curves need to be adjusted for use in
lower bound life assessments.

The Markl data appear to fall within the scatter of the various design curves, each which are
presented with any correction factor removed (i.e., all the curves shown are median rather than
lower bound design fatigue life curves). This appears to validate the use of the SCF=2 approach
to fatigue, at least for the Markl data.

A-30
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Equivalent Multiaxial Stress (Stress Intensity)

Most laboratory fatigue tests are performed under uniaxial loading (i.e., the specimen is loaded
in one direction only). However, pressure boundary components are loaded in at least two
directions. Pressure loading inside a cylinder creates circumferential (hoop) stresses that are
approximately twice the value of stresses in the axial direction. On the inside surface, the
cylinder is also subjected to the internal pressure, so the inside surface is subjected to a triaxial
stress field, whereas the outside surface, lacking the applied pressure acting on it, is subjected to
a biaxial stress field.

To convert a situation described by multiaxial loading into one that can utilize the uniaxial S-N
curves, one needs to understand the concept of equivalent stress. An equivalent stress is the
uniaxial stress that produces the same fatigue damage as the applied multiaxial stresses.

To determine the equivalent stress, one also needs to understand the concept of principal stresses.
At any point within a structure, an arbitrarily oriented coordinate system can be defined to
describe the state of stress at that point. There are three stresses acting along the axes of the
coordinate system, (σx, σy, and σz) and three shear stresses (τxy, τxz, τyz) acting in each of the
three planes (i.e., the x-y plane, the y-z plane, and the z-x plane). There are only three stress
components on each plane (e.g., σx, τxy, and σy).

The magnitudes of the various stress components depend on the orientation of the coordinate
system. By rotating the coordinate system about each of the three axes, a unique condition can be
generated where the shear stresses all become zero.

A common way of representing this is Mohr’s circle, shown in Figure A-12. For each of the
three planes, the stresses in that plane are plotted along the abscissa (x-axis). The shear stress is
plotted on the ordinate (y-axis). A circle can be drawn through the two points on that plane (e.g.,
[σx, τxy] and [σy, -τxy]). The points at which the circle crosses the x-axis are the principal
stresses. The principal stresses are usually denoted as σ1, σ2, and σ3. The radius of the circle is
equal to the maximum shear stress, τmax. In pressure boundary parts remote from discontinuities,
the principal stress directions correspond with the hoop, axial, and radial directions.

A-31
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Header Tubes p = 2000 psi


OD = 12.750” OD = 2.00”
t = 1.969” t = 0.150”
Figure A-12
Example of Mohr’s Circle in the XY Plane

Under uniaxial loading, σ2 = σ3 = 0 and the state of stress can be described by a single circle of
radius σ1/2, centered at [σ1/2, 0]. As noted previously, pressure boundary parts away from any
discontinuities are loaded biaxially at the outer surface, and loaded triaxially on the inner surface.

Under multiaxial loading, there are three Mohr’s circles of different diameters. Each circle
represents a stress state in a given plane.

Some design codes use the maximum principal stress as the basis for the S-N diagram. Other
Codes use an equivalent stress (sometimes referred to as the stress intensity) such as the
maximum octahedral shear stress (von Mises or distortion energy criterion) or the maximum
shear stress (Tresca), with the Tresca criterion the one that is used most often. The Tresca
criterion is the diameter of the largest of the three Mohr’s circles, defined as the difference
between the maximum and minimum principal stress.

The differences between principal stresses and equivalent Tresca stresses are illustrated
considering the tube to header connection shown in Figure A-13. Because there are no shear
stresses at this location, the principal stresses are aligned with the hoop and axial direction.
Figure A-14 shows the Mohr’s circle for the outside surface. The principal stresses are σ1 =
12,333 psi, σ2 = 6,167 psi, and σ3 = 0. The loading is biaxial, as the radial stress is zero. The
equivalent multiaxial Tresca stress is 12,333 psi.

A-32
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Figure A-13
Example of Tube-Header Connection

Figure A-14
Three-Dimensional Mohr’s Circle for Biaxial Loading at the Outside Surface of a
Pressurized Tube.

A-33
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Figure A-15 shows the Mohr’s circle for the inside surface. The principal stresses are
σ1 = 12,333 psi, σ2 = 6,167 psi, and σ3 = -2000 psi. The loading is triaxial, as the radial stress is
non-zero. The equivalent multiaxial Tresca stress is 14,333 psi.

Figure A-15
Three-Dimensional Mohr’s Circle for Triaxial Loading at the Inside Surface of a
Pressurized Tube.

By comparing Figure A-14 and Figure A-15, it is clear that there is no difference in the
equivalent stress between the two locations if the equivalent stress is based on the maximum
principal stress. However, there is a difference in the equivalent stress between the two locations
if the equivalent stress is based on the Tresca, or maximum shear stress, (in this case, σ1-σ3).
This difference will translate into a difference in the predicted fatigue life, so it is important to
ensure that the correct stress criterion is used.

Some codes use the von Mises definition of the equivalent stress. The von Mises stress is
calculated by:

1
σ vm = (σ 1 − σ 2 ) 2 + (σ 2 − σ 3 ) 2 + (σ 3 − σ 1 ) 2 Equation A-10
2

The Tresca and von Mises criteria are equivalent only under uniaxial conditions.

A-34
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

For the pressurized tube example in Figure A-13 and using the thin-wall equations for pressure
stress, the stresses in tube T2 remote from any discontinuities are:

Inside surface Outside Surface

σhoop = 12,333 psi σhoop = 12,333 psi

σaxial = 6,167 psi σaxial = 6,167 psi

σradial = -2000 psi σradial = 0 psi

σmax principal = 12,333 psi σmax principal = 12,333 psi

σTresca = 14,333 psi σTresca = 12,333 psi

σvon Mises = 12,453 psi vvon Mises = 10,680 psi

Most of the standards use principal stresses to convert multiaxial stresses to equivalent stresses.
When the direction of loading changes during the cycle, ASME and BS PD 5500 use the Tresca,
or maximum shear stress, theory.

For the toe of a weld on the outside of a component, there is no difference between the maximum
principal stress and the Tresca stress, because the third principal stress is zero (no loading normal
to the surface). For the inside surface the pressure is non-zero (it is equal and opposite the
internal pressure).

For a high-pressure component, the difference between a maximum principal stress and a Tresca
stress can be significant. This is also true for locations within the wall of the pressure vessel.
Fatigue normally does not initiate within the thickness of the part (unless a defect is present), but
creep can initiate mid-wall.

EN 13445 provides fatigue design curves for both maximum principal and equivalent (Tresca)
stresses.

For consideration of creep, ASME Section III NH uses the von Mises criterion to calculate the
equivalent stress because of the plasticity effects. The French RCC-MR code applies a similar
methodology, but uses the Tresca stress instead.

Due to its simplicity and conservatism the maximum shear (Tresca) equivalent stress is more
widely used. However when following specific design code rules careful review of which
equivalent stress formula is required so that unnecessarily optimistic equivalent stress values are
used. Table B-1 in Appendix B summarizes the type of stress to be used for multiaxial loading
in each Code.

A-35
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Sequence of Loading

Fatigue cycles in pressure boundary parts are rarely the same magnitude and vary widely for
each type of operating transient cycle (e.g. shutdown followed by cold start, trip followed by
rapid restart, etc) and underlying mechanism producing the cyclic loading (e.g., forward flow of
undrained condensate, introduction of unflashed attemperator water, etc.)

The most common method of dealing with variable amplitude loading is the linear summation of
damage approach, or the cumulative fatigue damage rule discussed previously. Using this
approach, there is no significance to the order of loading.

However, this is not strictly true, especially for weldments. For the linear elastic stress based
fatigue assessments that are the basis of most of the design codes the sequence of loading used in
the analysis will not influence the outcome. On the other hand, a more detailed inelastic strain or
energy based fatigue model will produce fatigue life estimates that will be significantly
influenced by the order in which the load cycles are modeled. For the strain range based models,
such as illustrated in Appendix D, the estimated stress strain hysteres is history, mean stress and
resulting cyclic life estimations will all be altered for different sequences or varying magnitude
loadings. When using this type of model the sequence of loading should be considered and
modeled in as realistic a fashion as possible.

Crack growth will be discussed in more detail below, but for now it is only important to realize
that the rate of crack growth depends on the size of the crack. Larger cracks cause more crack
growth than small cracks, for the same applied stress range. As a result, the order of loading is
important for fatigue models that incorporate crack growth. In such conditions, it is conservative
to assume that the cycles are ordered from highest stress to lowest. At least one code makes this
recommendation, even for initiation-based fatigue design (e.g., S-N), even though the order of
loading makes no difference in a linear summation of damage analysis. It is recommended that
when fatigue crack growth analyses are performed the sequence of loading should be considered
and modeled in as realistic a fashion as possible.

Environment

Only a few of the standards address the effect of water or steam environment on fatigue. ASME
Section XI and PD 6493 have fatigue crack growth curves for water environments, but the PD
6493 fatigue crack growth curves are for seawater (for offshore platforms) and should not be
used for the assessment of watertouched HRSG pressure parts.

The primary method used by the Codes and Standards to ensure that the environment doesn’t
significantly reduce the cycles to cracking is a check on the integrity of the surface oxide scales
for components that are water-touched. For watertouched components, if the oxide isn’t
mechanically disrupted by exceeding its compressive or tensile strain tolerance then the fatigue
life is anticipated to be adequately estimated using the cycles to cracking estimates that were
derived from tests in air environments.

A-36
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

TRD 301 Annex 1, EN 12952-3 and EN 13345-3 place limits on the estimated change in internal
surface stress levels for watertouched components to ensure that the adherent, protective
magnetite oxide scale will not crack. These oxide stress limits assume that the oxide grows when
the pressure part is at normal steady load stress levels. To ensure that the oxide stress or strain
limits are not exceeded the maximum stress determined under all steady and transient loading
conditions is restricted to 200 MPa (29 ksi, 0.1% strain) above and 600 MPa (87 ksi, 0.3% strain)
below the value determined under steady load conditions. The total stress range is thus restricted
to 800 MPa (116 ksi, 0.4% strain).

The success of this approach has been demonstrated over the years by the significantly lower
number of corrosion fatigue failures that have been experienced in conventional fossil-fuel fired
plants designed using these rules compared with units that were designed using codes that did
not include this oxide scale stress/strain tolerance limit.

The effects of the environment can really only be addressed by performing additional testing in
the environment of concern, and generating S-N curves for that environment or an environment
correction factor for use with existing air test based S-N curves.

Figure A-16 compares some corrosion-fatigue test data performed on carbon steel tubing in hot,
oxygenated water with the ASME fatigue curves. Note that only one of the test results produced
a life above the ASME mean fatigue curve and 10 of the 25 test results fell below the ASME
design curve. The data with the lowest cycles to cracking generally were the samples exposed to
low pH, high dissolved oxygen water and were subjected to the slowest loading rates. For more
information on the effects of water on fatigue, refer to Refs. [A-37 to A-39].

Based on the test data in Figure A-16 it is recommended that a correction factor of 10 on cycles
to cracking be used for water-touched HRSG pressure parts to account for corrosion-fatigue
unless the EN 12952-3 oxide scale cracking check indicates that the critical compressive or
tensile oxide strains will not be exceeded under all loading conditions.

A-37
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Figure A-16
Comparison of Corrosion Fatigue Test Data for Carbon Steel Tubing Tested in 135oC to
343oC (275oF to 650oF) Water

Temperature

Many of the fatigue design codes are not relevant to high temperature applications. Only ASME
III NB, R5, and TRD 301 provide S-N curves for elevated temperature (Merkblatt uses the same
temperature curves as TRD 301). These curves can be used to develop a temperature correction
factor for use with other standards. The ASME III NB procedure only presents S-N data at
800°F (426°C) and 1000°F (537°C), so it is difficult to build an extrapolation down to lower
temperature design curves.

EN 12952 provides a correction factor for S-N curves for elevated temperature, shown in Figure
A-17. The correct factor (ft) is equal to the virtual stress amplitude for X cycles to failure at the
service temperature divided by the virtual stress amplitude to produce the same number of cycles
to failure at room temperature.

A-38
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Figure A-17
Temperature Based Cyclic Stress Correction Factors

EN 13445 uses the same temperature correction factor. API 579 has a temperature correction
factor, but it falls short of the creep regime. Figure A-17 also shows the temperature correction
factors derived from TRD 301, along with the PD 6493 relationship and the EN 12952 correction
factor for ferritic material.

Note that the design code correction factors are generally conservative for the representation of
rapidly loaded and unloaded tests with no hold times. With even moderate 30 minute to 5 hr
hold times these correction curves are very optimistic. They do not account for creep fatigue
damage associated with elevated temperature steady state operating (hold) periods.

Creep fatigue or oxidation-enhanced fatigue of elevated temperature steam-touched components


is known to significantly reduce their fatigue life but this not directly addressed in any of the
codes or standards reviewed. Figure A-18 provides an example of the reduction in fatigue life for
2-1/4Cr-1Mo steel specimens resulting from elevated temperature and elevated temperature plus
a very moderate (5 hour) hold at the peak tensile strain value. With even moderate hold times
the cyclic stress amplitude to produce a specific number of of cycles to cracking fall well below
the elevated temperature test data with no hold time. Longer hold times will further reduce the
cyclic stress amplitude to produce a specific number of cycles. The typical hold time for an
HRSG that is cycled off overnight is 14 to 16 hours. For elevated temperature cycling with hold
times an additional correction factor is needed. This creep-fatigue correction factor is not
included in either the TRD 301, Annex 1 or EN 12952-3 rules.

A-39
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Figure A-18
Fatigue Tests of 2-1/4Cr 1Mo Steel Performed at Room Temperature and at 600oC (1110oF)
with Continuous Cycling at 1 Cycle per Minute and with 30 and 300 Minute (5 hr) Tensile
Hold Times. [A-40]

Creep-Fatigue

In the previous section the detrimental effect of a stress hold period on elevated temperature
fatigue was illustrated. There are a variety of mechanisms that often occur simultaneously that
cause this loss in fatigue strength or life including:
• Introduction and relaxation of inelastic stress/strain produced during thermal transients
• Cracking of surface oxide and crack tip oxidation
• Inter- or transgranular creep cavitation
• Other, more subtle metallurgical instabilities

When there is a loss in fatigue life or strength resulting from an elevated temperature hold period
the damage mechanism is creep-fatigue. It is often difficult to distinguish elevated temperature
thermal-fatigue and creep-fatigue from metallurgical characteristics. Figure A-19 provides an
example of a creep-fatigue crack located at the toe of a T91 (modified 9% chromium) primary
reheater tube-to-header connection weld. This crack occurred in a tube to header connection in a
primary (low temperature) reheater tube bundle. Note the wide, oxide coated crack and the
extensive intergranular creep cavities around the edges of the crack. At higher temperatures
creep-fatigue cracks are often transgranular, oxide filled cracks without any visible intergranular
creep cavities.

The crack is outside surface initiated, oxide coated and exhibits and extensive network of
intergranular creep cavities along the edges of the crack. For comparison, Figure A-20 shows a

A-40
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

thermal-fatigue crack that occurred at a tube-to-header connection in an HP economizer. Note


that this crack is tight and has no visibly apparent crack surface oxide. The crack occurred at a
tube to header connection in a HP economizer.

Figure A-19
Outside Surface, Weld Toe Initiated Creep-Fatigue Crack

Figure A-20
Outside Surface Weld Toe Initiated Creep-Fatigue Crack

A-41
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Creep-fatigue cracks in the high temperature portions of primary and secondary reheaters and HP
superheaters may be straighter than the crack shown in Figure A-19 and might not exhibit any
visible creep cavities but will be wide and oxide coated. At the higher temperatures oxidation
enhanced creep-fatigue is often the dominant damage mechanism and may be intergranular,
transgranular or both.

In the absence of creep, the life of a component has been used up when the fatigue damage ratio
equals 1.0. Similarly, when there is no fatigue cycling, the life of a component is determined
when the creep damage ratio equals 1.0. However, when both creep and fatigue act on a
component, it is far more difficult to estimate the life of the component. The simplest approach
to creep fatigue assessment would be linear interpolation. Assume that the creep damage ratio is
plotted on the abscissa and the fatigue damage ratio plotted on the ordinate, as shown in Figure
A-21. A linear interpolation would utilize a straight line between (Dc = 1,Df = 0) and (Dc = 0,
Df = 1) as shown by the dashed line.

However, there is often a synergistic effect between creep and fatigue, such that this approach is
non-conservative under some conditions. Two commonly used methods for taking the creep-
fatigue interaction into account are to use a bilinear curve, as shown by the solid line in Figure
A-21, or by reducing the linear interpolation by a factor of safety, such that the new acceptance
criterion is a line parallel to the dashed line. The location of the “knee” in the bilinear criterion
depends on the material but is often around Dc = 0.1, Df = 0.1.

Figure A-21
Creep-Fatigue Interaction

A-42
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Creep-fatigue interaction is considered in ASME III NH, RCC-MR, R5, PD 6539, and BS 7910.
The ASME and RCC-MR approach is to use the bilinear interaction equation shown in Figure
A-21. The TRD 508 approach uses a linear interaction equation, which is considerably less
conservative than the bilinear equation.

The creep-fatigue rules used by ASME Section III NH are sometimes viewed as being
conservative. For low alloy steels, this may be so when the creep damage ratio exceeds 0.15,
which then sets s limit to the fatigue damage ratio of between 0 and 0.15, depending on the
actual creep damage ratio. However, when the estimated creep life fraction is 0.15 or less, the
bilinear interaction equation may be nonconservative for low alloy steel components. An
important creep fatigue interaction effect is the additional plastic strain damage from creep
relaxation of the residual stress created by yielding during the temperature rise or quench part of
each cycle. The rate of relaxation during the high temperature hold is a function of the creep
strain rate at the hold temperature. For components that operate at higher temperatures on load
the residual stress from yielding during the startup may fully relax during the 14 to 16 hours on
load that is typical of a cycled unit. Therefore, to ensure that creep fatigue damage is not
undeestimated, it has been recommended that an upper limit of 0.1 be imposed on the fatigue
usage of components operating at temperatures well into the creep regime irrespective of the
estimated creep damage fraction.[A-41]

Creep-fatigue interaction in the British standards R5, PD 6539, and BS 7910 is based on the
ductility exhaustion method, where data are available. Otherwise, the British standards use a
linear interaction equation with a factor of safety to limit the usage factors to 0.5.

Unless a detailed creep-fatigue evaluation is performed to justify a less conservative limit it is


suggested that the fatigue usage factor be limited to 0.10 (10% total fatigue damage) for pressure
parts operating above 427oC (800oF) that have been evaluated using “no-hold-time” fatigue
curves. Detailed creep fatigue assessments of HRSG pressure parts may be performed using
ASME Section III NH or using strain range partition or total strain energy models. These more
complex models are outside the scope of this report.

Fatigue Crack Growth

Most of the discussion up to this point relates to the S-N approach to fatigue. Usually, the S-N
curve represents the number of cycles to failure. However, there are some codes that use the S-N
approach to determine the number of cycles to initiate a crack, and use crack growth models to
determine the number of cycles to propagate the crack to failure. The fatigue life is then the sum
of the initiation and propagation lives.

Only the fitness for service codes and standards (ASME Section XI, API 579, R5, PD 6493, and
RCC-MR) discuss fatigue crack growth. Of these, only R5 is intended for elevated temperature
applications.

Fatigue crack growth is based on a fracture mechanics concept known as the stress intensity
factor range, which represents a measure of the stress at the tip of the crack. The stress intensity
factor range, ΔK, can be determined from the applied stress range, Δσ, by:

A-43
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Δ K = Y Δσ π a Equation A-11

where a is the crack size and Y is a factor that includes the geometry and type of loading.
Because the cyclic stress intensity factor ΔK is independent of the geometry (the geometry is
accounted for in the Y term of Eq. A-11), crack growth modeling can be applied to a number of
geometries. Both API 579 and R6 (PD 6493) provide stress intensity factor solutions for
relevant geometries (i.e., weldments and cylinders).

Crack growth per cycle of loading is typically related to the cyclic stress intensity factor ΔK by a
relationship similar to:

da / dN = C ΔK m Equation A-12

which is known as a Paris crack growth law [A-42]. There are a number of other variations on
this relationship that account for effects such as R-ratio and the maximum stress in the cycle. By
combining equations A-11 and A-12 and rearranging and integrating the crack growth terms a
relationship between fatigue life, N, and stress range can be derived:

af
da
∫ (Y
ai πa
= C Δσ m N Equation A-13

The derived equation is now similar to Eq. A-2 but includes flaw size (or weld quality) terms.
Most design codes do not consider fatigue crack growth, because it is assumed that once a crack
has initiated, there is little fatigue life remaining. The general exception is for fatigue of
weldments, as described previously. The codes that do consider fatigue crack growth are
typically fitness-for-service codes rather than design codes (fitness-for-service is the analysis of a
known or postulated defect to determine whether that defect will adversely effect the continued
operation of the component containing that defect).

Creep

In its simplest form, creep is considered to be a stress- and temperature-dependent failure


mechanism. Results of stress rupture tests are plotted as a series of iso-temperature lines, as
shown in Figure A-22. This is the basis for creep analysis in ASME Section III, Subsection NB.

The stress rupture data can normalized using the Larson-Miller parameter [A-43]

LMP = (T + 460) ( C + log t) x 10-3 Equation A-14

where LMP is the Larson-Miller parameter, which is a function of stress, T is the temperature in
degrees F, C is a material constant, and t is the life in hours. This equation can be rearranged to
solve for life as a function of temperature and LMP. Figure A-23 shows how all of the stress
rupture iso-temperature curves of Figure A-22 can be reduced to a single line.

A-44
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

More complex models exist for determining the creep life, including ductility exhaustion and
creep crack growth. Discussion of those models will be delayed until the discussion of the codes
that contain them in Appendix B.

Creep is addressed in ASME III NH, API 579, R5, PD 6593, BS 7910, RCC-MR, and TRD 508.
The ASME, RCC-MR and TRD 508 procedures are based on stress rupture curves. API 579
uses the Larson-Miller parameter, which is a condensed form of the stress rupture curves.

Creep analysis in the British standards R5, PD 6539, and BS 7910 is based on the reference
stress method, where the remaining life is calculated from a stress rupture curve using the
reference stress.

Creep crack growth is considered in the API 579 document, as well as in the British standards.
In both cases, the creep crack growth rate is calculated from the C* integral. The British
approach uses the reference stress and the strain rate to calculate C*.

Figure A-22
Typical Stress Rupture Curves (ASME Section III NH for 2¼Cr-1Mo Steel)

A-45
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

Figure A-23
Stress Rupture Curves of Figure A-22 Collapsed to a Single Line Using the Larson-Miller
Parameter

References

[A-1] “Power Boilers”, ASME Boiler and Pressure Vessel Code, Section I, 1995.
[A-2] “Rules for Construction of Nuclear Power Plant Components”, ASME Boiler and
Pressure Vessel Code, Section III, Division 1, Subsection NB, “Class 1 Components,”
1995.
[A-3] “Rules for Construction of Nuclear Power Plant Components”, ASME Boiler and
Pressure Vessel Code, Section III, Division 1, Subsection NH, “Class 1 Components in
Elevated Temperature Service,” 1995.
[A-4] “Pressure Vessels”, ASME Boiler and Pressure Vessel Code, Section ASME VIII, Div. 2,
“Alternate Rules,” 2001.
[A-5] “Rules for Inservice Inspection of Nuclear Power Plant Components”, ASME Boiler and
Pressure Vessel Code, Section XI, Appendix A, “Analysis of Flaws”, 1995.
[A-6] Fitness for Service, American Petroleum Institute, API RP 579, January 2000.
[A-7] “Specification for Design and Manufacture of Water-Tube Steam Generating Plant
(Including Superheaters, Reheaters, and Steel Tube Economizers)”, British Standards
Institute, BS 1113:1999.
[A-8] “Specification for Unfired Fusion Welded Pressure Vessel”, British Standards Institution,
BS 5500, 1985.

A-46
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

[A-9] “Specification for Unfired Fusion Welded Pressure Vessel”, British Standards Institution,
BS 5500, 1997.
[A-10] “Specification for Unfired Fusion Welded Pressure Vessel”, British Standards Institution,
BS PD 5500, 2003.
[A-11] “An Assessment Procedure for the High Temperature Response of Structures”, Berkeley
Nuclear Laboratories Report R5, 1990.
[A-12] Goodall, I.W., and Ainsworth, R.A., “R5: An Assessment Procedure for the High
Temperature Response of Structures”
[A-13] “Guide to Methods for the Assessment of the Influence of Crack Growth on the
Significance of Defects in Components Operating at High Temperatures”, British
Standards Institution, BS PD 6539, 1994.
[A-14] Harrison, R.P., Loosemore, K., Milne, I., and Dowling, A.R., “Assessment of the
Integrity of Structures Containing Defects”, Central Electricity Generating Board
R/H/R6, Rev.2, April 1980.
[A-15] “Guidance on Methods for Assessing the Acceptability of Flaws in Fusion Welded
Structures”, British Standards Institution, BS PD 6493, 1991.
[A-16] “Guide on Methods for Assessing the Acceptability of Flaws in Metallic Structures”,
British Standards Institution, BS 7910, 1999.
[A-17] ”Design and Construction Rules for Mechanical Components of FBR Nuclear Islands,
French Society for Design, Construction and Surveillance Rules for Nuclear Island
Components, RCC-MR, June 1995.
[A-18] “Calculation for Cyclic Loading Due to Pulsating Internal Pressure or Combined
Changes of Internal Pressure and Temperature”, Technical Rules for Steam Boilers,
Annex 1 (Design), TRD 301, April 1975.
[A-19] “Additional Tests on Components – Methods for the Calculation of Components Having
Time Dependent Design Strength Values”, Technical Rules for Steam Boilers, Annex 1,
TRD 508, October 1978.
[A-20] Design to Allow for Fluctuating Stress, AD-Merkblatt S2, 1988.
[A-21] Fatigue Analysis, AD-Merkblatt S2, 1995.
[A-22] “Assembly of Information on European Standards Development (CEN) Relevant to
Pressure Equipment Fatigue Design”, Centre Technique des Industries Mecanique,
Annex 6 to Pressure Components Fatigue Design in the Framework of Directive
97/23/EC on Pressure Equipment, Pressure Equipment Directive, July 2001.
[A-23]“Design and Calculation for Pressure Parts”, Water-Tube Boilers and Auxilliary
Installations, Part 3, EN 12952-3, December 2001.
[A-24] “In-service Boiler Life Expectancy Calculations”, Water-Tube Boilers and Auxilliary
Installations, Part 4,”In-service Boiler Life Expectancy Calculations”, EN 12952-4, 2002.
[A-25] Unfired Pressure Vessels, EN 13445 Part 3, draft 2001.

A-47
EPRI Proprietary Licensed Material

Review of Fatigue Assessment factors

[A-26] Miner, M.A., “Cumulative Damage in Fatigue”, Journal of Applied Mechanics, Vol 12.
No. 3, September 1945, pp. A159-A164.
[A-27] Markl, A.M.C., “Fatigue Tests of Piping Components”, Transactions of the ASME,
Volume 74, 1952, pp. 287- 303.
[A-28] ASME Section VIII, Division 2, Appendix 5, paragraph 5-112.
[A-29] Jaske, C.E., ”Fatigue Strength Reduction and Stress Concentration Factors for Welds in
Pressure Vessels and Piping”, Welding Research Council Bulletin, 432, June 1998.
[A-30] Wichman, K.R., A.G. Hopper, and J.L. Mershon, “Local Stresses in Cylindrical Shells
Due to External Loadings”, Welding Research Council Bulletin, 107, August 1965.
[A-31] Mershon, J.L., K. Mokhtarian, G.V. Ranjan, and E.C. Rodabugh, “Local Stresses in
Cylindrical Shells Due to External Loadings – Supplement to WRC Bulletin No. 107”,
Welding Research Council Bulletin, 297, August 1984..
[A-32] ASME B31.1 (2004), Appendix D, Table D-1, Note 6
[A-33] ASME Section VIII, Division 2 (2001), Article 4-6, paragraph 4-613, Note (f).
[A-34] AWS D1.1 (1996) Table C8.2 and Figure C2.20
[A-35] Rodabaugh, E.C., “Accuracy of Stress Intensification Factors for Branch Connections”,
Welding Research Council Bulletin, No. 329, December 1987.
[A-36] Rodabaugh, E.C., “Stress Indices, Pressure Design and Stress Intensification Factors for
Laterals in Piping”, Welding Research Council Bulletin, No. 360, January 1991.
[A-37] “Fatigue Crack Growth of Low-Alloy Steels in Light Water Reactor Environments”,
Welding Research Council Bulletin, 404, August 1995.
[A-38] “Evaluation of Conservatisms and Environmental Effects in ASME Code, Section III,
Class 1 Fatigue Analysis”, Sandia Laboratories Report SAND94-0187, August 1994.
[A-39] “Environmentally Assisted Fatigue Crack Initiation in Low-Alloy Steels – A Review of
the Literature and the ASME Code Design Requirements”, GE Nuclear Energy, EPRI
TR-102765, August 1993.
[A-40] H.G. Edmonds and D. J. White, J. Mech. Eng. Sci., Vol 8., (No. 3), 1966, p 310-321.
[A-41] Dooley, R.B., Shields, K.J., Paterson, S., Kuntz, T., McNaughton, W., and Pearson, M.,
“Heat Recovery Steam Generator Tube Failure Manual”, Electric Power Research
Institute, EPRI Technical Report 1004503, November 2002.
[A-42] Paris, P.C., Gomez, R.E., and Anderson, W.E., “A Rational Analytic Theory of Fatigue”,
The Trend in Engineering, University of Washington, Vol. 13, No. 1, January 1961, pp.
199-204.
[A-43] Larson, F.R., and Miller, J., “A Time-Temperature Relationship for Rupture and Creep
Stresses”, Transactions of the ASME, July 1952, pp. 765-775.

A-48
EPRI Proprietary Licensed Material

B
REVIEW OF FATIGUE DESIGN CODES AND
STANDARDS

American Codes and Standards

Within the United States, the dominant code relating to pressure vessels is the American Society
of Mechanical Engineers (ASME) Boiler and Pressure Vessel Code. The ASME Code is broken
into a number of sections that apply to different types of pressure vessels or different aspects of
pressure vessel design and construction.

ASME Section I

Section I of the ASME Code [B-1] applies to boilers, superheaters, economizers, and any other
pressure parts connected directly to the boiler. Section I is a design by rule code, and does not
provide any guidance on fatigue. As a result, other sections of the code are often used for fatigue
design. The design criteria of Section I are intended to be valid for up to 100,000 hours of creep
[B-2].

ASME Section III, Subsection NB

Section III of the ASME Code applies to nuclear power plants, and is made up of a series of
Subsections. Subsection NB [B-3] provides guidance on design of pressure vessels made of
ferritic steels operating below 371°C (698°F) or austenitic stainless steels and high-nickel alloys
operating below 427°C (800°F).

Subsection NB uses the linear cumulative damage approach to fatigue, using the S-N curve
shown in Figure B-1. The curve in Figure B-1 represents the stress amplitude, not the cyclic
stress range, Δσ. Curves are given for materials with two ultimate strength levels (interpolation
is allowed).

These curves are derived from mean fatigue data in air, reduced by a factor of 2 on stress, or a
factor of 20 on life. The kink in the curve at 1.2x104 cycles represents the transition from a
correction factor based on life to one based on stress. It should be pointed out that the intent of
these factors is not to represent a factor of safety, but to account for the differences between lab
and field data [B-4].

B-1
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

The factor of 20 consisted of the product of a factor of 2.0 to account for scatter in the data (e.g.,
minimum vs mean), a size effect factor of 2.5, and a factor of 4.0 to account for surface finish,
and the fact that the environment in the lab where the tests were performed is typically better
than in real applications. The “atmosphere” effect was not intended to consider specific
environmental conditions [B-4].

Figure B-1
ASME Section III Subsection NB Fatigue Design Curves

Subsection NB uses equivalent stresses based on the principal stresses when the direction of
stress loading does not change throughout the fatigue cycle. When the direction of loading does
change during the cycle, the Tresca criterion is used. In the ASME Code, these stresses are
referred to as stress intensities.

In the section on pressure vessels, Subsection NB allows the use of theoretical or experimental
stress concentration factors, which they also call fatigue strength reduction factors. Fatigue
strength reduction factors are limited to a value of five, except for crack-like defects, or for
piping, which has its own set of rules. For nozzles in cylindrical pressure vessels, the fatigue
strength reduction factors (which are now called stress indices) are defined as shown below for
internal pressure loading:
Longitudinal Plane Transverse Plane
Stress Inside Outside Inside Outside
σnormal 3.1 1.2 1.0 2.1
σtransverse -0.2 1.0 -0.2 2.6
σradial -t/R 0 -t/R 0
Stress intensity 3.3 1.2 1.2 2.6

B-2
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

where the normal, transverse, and radial directions are defined relative to surface of the hole
being considered (“normal” is circumferential relative to the hole) and the longitudinal and
transverse planes are defined relative to the axis of the vessel, such as a header or drum.

In the section on piping, Subsection NB specifies three separate fatigue strength reduction factors
(which are also called stress indices) that apply to pressure, bending, and thermal loadings. The
values of the three stress indices are tabulated below for several common welded connections in
piping.
Connection Pressure Moment Thermal
Girth butt welds (ground flush) 1.1 1.1 1.1
Girth butt welds (as-welded) 1.2 1.8 1.7
Girth fillet weld to socket weld 3.0 2.0 3.0
Branch connection 2.0 -- 1.7
Butt welding tees 4.0 -- 1.0
Subsection NB does not explicitly address environmental issues such as corrosion-fatigue, which
is very non-conservative for water-wetted surfaces if peak thermal mechanical stresses are high
enough to crack the protective oxide layer during the transient cyclic loading.

The Subsection NB exclusion rules indicate that fatigue is not a concern when the number of full
pressure cycles is less than number of cycles allowed from Figure B-1 for a stress level
corresponding to the ultimate strength of the material at temperature, or twice the yield strength
at temperature, whichever is less. It also requires that the temperature difference between
adjacent points to be less than Sa/2Eα (where Sa is the allowable stress amplitude for the number
of specified startup-shutdown cycles, E is the modulus of elasticity, and α is the coefficient of
thermal expansion), and the full range of mechanical loads to be less than Sa for the anticipated
number of load excursions.

ASME Code, Section III, Division 1, Subsection NH

Subsection NH [B-5] was originally written as Code Case N-47 [B-6] to address high
temperature components, and has subsequently been adopted as a separate subsection of Section
III. It applies to types 304 and 316 austenitic stainless steels up to 816°C (1500°F), 2¼Cr-1Mo
steel up to 593°C (1095°F), alloy 800H up to 760°C (1400°F) for applications other than bolting,
and alloy 718 up to 566°C (1050°F) for bolting applications.

For low temperature applications, fatigue design is covered by Subsection NB. For high
temperature applications, creep must be considered in conjunction with fatigue. Subsection NB
gives the owner authority to develop his own acceptance criteria, but provides the following
procedure in Nonmandatory Appendix T.

In the Subsection NH procedure, the maximum equivalent strain range is calculated for each
cycle type from the Von Mises stress:

B-3
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

1 3
Δε eq = (Δε x − Δε y ) 2 + (Δε y − Δε z ) 2 + (Δε z − Δε x ) 2 + (γ xy + γ yz + γ zx )
2 2 2

2 (1 + ν *) 2
Equation B-1

where ν* = 0.5 for inelastic analysis and ν* = 0.3 for elastic analysis, and the Δε terms represent
the maximum strain in each direction experienced during the cycle. A similar relationship exists
for principal, rather than component, strains, but this method is only applicable when the stress
directions do not change.

For inelastic analysis, the stress and strain concentration effects of local discontinuities are
included in this step. For elastic analysis, the stress and strain concentration effects of local
discontinuities are not included in the calculation of the strains in this step. The maximum
equivalent strain range for all cycle types is modified to account for the inelastic stress-strain
response using Neuber’s rule [B-7].

⎛ S *⎞ 2
Δε mod = ⎜ ⎟ K Δε max Equation B-2
⎝ S ⎠

where K is the equivalent elastic stress concentration factor, S* is the stress corresponding to
Δεmax in the isochronous stress-strain curve (t=0), and S is the stress corresponding to KΔεmax in
the isochronous stress-strain curve (t=0). The total strain range, εt, is then calculated from:

ε t = K v Δε mod + KΔε c Equation B-3

where Kν is a multiaxial plasticity and Poisson’s ratio adjustment factor and Δεc is the creep
strain increment during that cycle.

The Neuber rule is an important tool that can be used to estimate elastic-plastic strains under
conditions of contained plasticity. In its original form, the elastic stress concentration factor, K,
is equal to the square root of the product of the true concentration factor for stress and the true
concentration factor for strain:

K = Kσ K ε Equation B-4

which can be rearranged to the form of Eq. B-2. However, the relationship can be rearranged
further to the form:

( K σ nom ) 2
σε = = consta nt Equation B-5
E

where σ and ε are the elastic-plastic stress and strain values, respectively. The intersection of the
parabola described by this equation with the stress-strain curve produces the elastic-plastic strain.

B-4
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

For example, assume a nominal stress of 15 ksi and a stress concentration factor of 3. The
concentrated stress is 45 ksi, and the corresponding elastic strain is 0.17%, as shown in Figure
B-2. The Neuber parabola through this point intersects the actual stress-strain curve at about
0.27% strain. Thus the elastic-plastic strain at this location is 0.27%.

Figure B-2
Use of Neuber Relationship to Calculate Elastic-Plastic Strain

Once the strain range is calculated, the fatigue damage ratio is then calculated by dividing the
number of cycles for each cycle type by the allowable number of cycles for that cycle type from
the ε-N curve such as the one shown in Figure B-3, and summing over all cycles:

n
⎛n⎞
Df = ∑⎜ ⎟ Equation B-6
i =1 ⎝ N ⎠ i

Note that the S-N curve here is based on strain range, rather than stress amplitude, as in
Subsection NB. The Subsection NB curve is converted into an equivalent strain range for
purposes of comparison in this figure, using the elastic modulus. The Subsection NH curves are
more conservative than the Subsection NB curves, but only for low cycle fatigue.

The creep analysis is performed by considering the number of hours the component is exposed to
temperatures above 371°C (698°F). For each cycle type, the isochronous stress-strain curve
appropriate to the normal operating temperature is used to define the stress corresponding to the
strain calculated using the elastic approach above (Eq. B-3), and this stress is adjusted to account
for stress relaxation. As an additional conservatism, the stress is divided by a factor K’ =0.67.

B-5
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

Figure B-3
ASME Section III NH S-N Curves for 2¼Cr-1Mo Steel

The creep damage ratio is then calculated by dividing the number of hours for each cycle type by
the allowable number of hours for that cycle type determined from a stress-rupture curve such as
Figure A-23 for the maximum temperature in that cycle, and summing over all cycles:

n
⎛ Δt ⎞
Dc = ∑ ⎜⎜ ⎟⎟ Equation B-7
i =1 ⎝ Td ⎠ i

Subsection NH uses the linear damage rule to calculate the total creep-fatigue damage from the
fatigue damage ratio and the creep damage ratio. The knee in the allowable creep-fatigue
damage curve (i.e., Figure A-21) is at (0.15, 0.15) for 2¼Cr-1Mo steel and Alloy 800H, and is at
(0.3, 0.3) for 304 and 316 stainless steel.

Subsection NH points out that fatigue strength reduction factors do not exist for high temperature
applications, but nevertheless specifies a fatigue strength reduction factor of four for fillet welds
at structural attachments.

Subsection NH does not explicitly address environmental issues such as corrosion-fatigue for
water-wetted surfaces, which can reduce fatigue life by an order of magnitude, nor does it
consider crack growth by creep or fatigue. For welds, the allowable fatigue life is reduced by ½
and the creep time to failure is adjusted by a weld strength reduction factor R.

B-6
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

ASME Code, Section VIII

Section VIII of the ASME Code applies to stationary pressure vessels operating at pressures
above 15 psig, and includes both fired and unfired pressure vessels not specifically covered in
Sections I or III.

Section VIII is divided into two divisions: Division 1 utilizes design by rule, whereas Division 2
utilizes a design by analysis approach. Division 1 considers non-pressure part attachments, but
Division 2 considers only the weld attaching the part to the vessel. Division 1 uses maximum
principal stresses, whereas Division 2 uses the maximum shear stress (Tresca) theory.

In general, the Division 2 rules are more restrictive in the choice of materials that may be used,
but allows the use of higher stresses than Division 1.

Unfired boilers are required to be designed to Section I or Section VIII, Division 1, whereas
evaporators can be designed to Division 1 or Division 2 of Section VIII. This review will focus
only on Division 2 [B-8] as there are no fatigue rules in Division 1.

The fatigue design rules of Section VIII, Division 2 are the same as Section III, Subsection NB,
except that piping is not considered in Section VIII. As a result, the piping stress indices used as
fatigue strength reduction factors are not used in Section VIII.

Section VIII does not explicitly address environmental issues such as corrosion-fatigue, nor does
it consider fatigue crack growth.

Section VIII provides two sets of exclusion rules for fatigue. The first criterion specifies that a
fatigue analysis is not required if the total number of cycles

Np + Nc + NΔT < 1000 Equation B-8

where Np is the number of full-range pressure cycles, Nc is the number of operating cycles where
the pressure range exceeds 20% of the full pressure, and N T is the number of cycles where the
metal temperature difference between two points exceeds a given amount. The Code specifies
that thermal cycles with a ΔT less than 50°F do not count, but thermal cycles with greater
temperature differences should be counted at a rate that increases with the magnitude of the
temperature difference, as shown in Figure B-4. When ΔT > 450°F (232°C), each cycle counts
as 20 cycles.

The second exclusion rule is basically the same as Section III, Subsection NB.

B-7
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

Figure B-4
Multiplier on Cycles for Metal Temperature Differentials in ASME Section III Subsection NB
Exclusion Rules

ASME Section XI

Section XI of the ASME Code [B-9] is a fitness-for-service code that provides for evaluation of
crack-like flaws found during in-service inspection of nuclear components. Fatigue evaluation is
based on fracture mechanics, using the Paris crack growth rule described in Eq. A-12. Crack
growth constants are defined for air and water environments, and include R-ratio effects. The
crack growth curves for water utilize a bilinear Paris law relationship. Figure B-5 shows the
crack growth curves from Section XI. Code Case N643 [B-10] provides additional crack growth
curves for material susceptible to environmentally assisted crack growth. The N643 curves are a
function of the load rise time as well as the R-ratio.

ASME Section XI is not intended for high temperature applications.

API RP 579

The American Petroleum Institute also provides design criteria for pressure vessels, although it is
intended for use in refineries and petrochemical plants. However, it has also introduced a
fitness-for-service code, RP 579 [B-11] that is often used in power plant applications to address
damage mechanisms similar to those found in petrochemical service. API RP 579 is not a code,
it is a “Recommended Practice”, which means that it is not mandatory.

B-8
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

Figure B-5
ASME Section XI Fatigue Crack Growth Curves

The main body of API 579 does not address fatigue, but Appendix F5 gives an overview of
various options for fracture mechanics based fatigue crack growth, including corrosion-fatigue.
It also provides sources of data such as the ASME Section XI crack growth curves.

API 579, Appendix F6 also provides S-N curves for use in fatigue initiation analyses. For
unnotched specimens, it references the ASME Section VIII S-N curve. For welded components,
it provides S-N curves for a variety of joint classes, as shown in Figure B-6 The source of these
curves is not given, but appears to be European, as the classes are designated by the stress in
MPa associated with a life of 2x106 cycles. The figures that explain the weld details are identical
to those in BS PD 5500 (discussed later).

The fatigue curves were derived from fatigue test data obtained from welded steel specimens,
fabricated to normal workmanship standards, and tested in air. The curves are drawn two
standard deviations of log(N) below the mean. This represents a 98% lower bound. A 99.9%
lower bound can be obtained by using the next lower curve.

For low cycle fatigue, a pseudo-elastic stress range can be used with these curves, based on the
strain range times the modulus of elasticity.

Although not explicitly stated, it appears that the correct stress value to be used with the S-N
curves is the principal stress at the nominal stress location. Class 80 may be used for welds
designated as Class 63 or Class 50 if the hot spot stress range is used in place of the nominal
stress range.

B-9
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

Figure B-6
API 579 S-N Curves

The S-N curves can be used for any welded steel, as the fatigue strength of weldments is
independent of tensile strength. An adjustment factor is given for applying these curves with
other materials, or at other temperatures (below the creep range). An adjustment factor is
provided to account for thickness effects. Although environmental effects are discussed, no
direct guidance is provided for adjusting these S-N curves to account for the environment.

There is a chapter in API 579 on creep, but it is “currently being developed”. However,
Appendix F7 provides several methods for assessing creep life. The first is the Larson-Miller
parameter, described in Appendix A of this report. Appendix F7 of API 579 also discusses the
Materials Property Council (MPC) Omega program, which was developed for the refining and
petrochemical industry to assess creep. The remaining life of a component is given by

1
t= Equation B-9
Ω mε&co

where Ωm is a multiaxial damage parameter and εco is the initial creep strain rate. Coefficients
are provided to calculate Ωm and εco for a variety of materials, including carbon steel, 2¼Cr-1Mo
(in three different heat treatments) and 9Cr-1Mo steel.

Creep crack initiation is not explicitly addressed, although two references are cited for fatigue
initiation for components operating in the creep regime. Two alternative approaches to creep
crack growth are given.

B-10
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

Other American Codes

There are other American Codes that address fatigue in contexts other than pressure vessels.
These include the American Welding Society Structural Welding Code [B-12] and the American
Association of State Highway Transportation Officials fatigue design guides [B-13, B-14]. In
general, these codes provide guidance on welds in beams and columns, but can provide useful
guidance on fillet weld attachments.

British Codes and Standards

Two types of British standards will be addressed here. The first type of documents are true
standards, published by the British Standards Institute (BSI). The second type of documents are
guidance documents, that are not mandatory and are intended to provide technical guidance on
specific issues. Some of these are published by BSI, and are preceded by a PD designation (for
Published Document) to distinguish them from true standards (this is comparable to the RP
designation used by API for recommended practices). The advantage of these guidance
documents is that they can be published more quickly than normal standards, with their long
development time.

However, there are additional sources for guidance documents. These include agencies such as
the Health and Safety Executive (HSE) and the Central Electricity Generating Board (CEGB).
The CEGB has been restructured and privatized, so the documents that are described below as
being generated by the CEGB are available through Nuclear Electric.

BS 1113

The British design standard for water-tube steam generating equipment is BS 1113 [B-15]. That
standard discussed the need for fatigue analysis, but did not specifically require one. BS 5500
was typically used when fatigue analysis was required. BS 1113 has been partially superceded
by EN 12952.

BS PD 5500

The British design document for unfired pressure vessels is BS PD 5500. It was originally a
design standard (BS 5500) but is now a “Published Document”.

The original BS 5500 fatigue design rules [B-16] were essentially the same as the ASME
Section III and Section VIII rules. The primary difference between the ASME and BS 5500
approaches was the basis for the S-N curve. The S-N curve from BS 5500 is shown in Figure B-
7, along with the ASME S-N curve (for low strength steels) for comparison.

The BS 5500 S-N curve was based on strain controlled fatigue tests of smooth ground butt
welds. The design curve was drawn four standard deviations below the mean, which represents a

B-11
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

99.997% lower bound. The design curve was a factor of 15 below the mean curve on life, and
2.2 below the mean on stress [B-17].

Figure B-7
BS 5500 S-N Curve, Compared to ASME III NB

BS 5500 used fatigue strength reduction factors (FSRFs) to account for the stress concentration
at the toe of welds. The value of the fatigue strength reduction factor for fillet welds is 2.5,
compared to 4.0 for the ASME Code, but this difference is minimal, when the extra conservatism
of the BS 5500 curve is considered.

During the 1980s, there was a great deal of work being done in other industries (primarily
offshore oil production) that produced S-N data for as-welded weldments, as opposed to the
weldments ground flush that were the basis for BS 5500. Enquiry Case BS 5500/79 was issued
to resolve issues raised by that work, and BS 5500 was revised in 1988 [B-18], incorporating
these data into new fatigue design curves. Due to the European standardization process, this
standard has been withdrawn but still exists in the form of a published document, BS PD 5500
[B-19].

These new fatigue curves were drawn based on a lower bound 95% confidence limit,
corresponding to a 97.7% probability of survival [B-20]. For the simplified method of analysis,
the probability of survival was increased to 99.9% by limiting the fatigue usage factor to 0.6.

The BS PD 5500 fatigue design curves are based on six weld detail classes, plus a class for
unwelded material (Class C) as shown in Figure B-8. Class D applies to unwelded parts of the
vessel where there is the possibility of weld repair. The weld classes were selected to be relevant
to pressure vessel design details, and depend on the direction of loading. In general, the fatigue

B-12
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

curves of BS PD 5500 are less conservative than the original BS 5500 curves for low cycle
fatigue, and more conservative for high cycle fatigue.

Figure B-8
BS PD 5500 S-N Curves

The BS PD 5500 weld classes are based on the assumption that the welds are fully inspected to
eliminate weld defects. When the weld is not inspected, a lower weld class can be used to
account for possible weld defects. The rules allow for fatigue improvement methods, such as
weld toe grinding, by allowing the next higher weld class to be used. However, the user is
cautioned that these fatigue life improvement techniques may be ineffective in the presence of a
corrosive environment, where fatigue cracks may initiate at corrosion pits.

The BS PD 5500 rules include consideration of the thickness effect. The allowable stress is
reduced for thickness greater than 22 mm by the following relationship:

0.25
⎛ 22 ⎞
f th = ⎜ ⎟ Equation B-10
⎝ t ⎠

where t is the thickness of the part in mm.

The S-N curves in BS PD 5500 are to be used with the nominal stress. When the stress analysis
considers the local stress concentration effects, the design curve for butt welds may be used, with
the concentrated stress. This is similar to the “hot spot” stress approach used in welded tubular
joints for offshore structures.

BS PD 5500 recommends using the maximum principal stress range with these S-N curves when
the direction of loading does not change throughout the cycle. When the direction of principal

B-13
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

stress changes during the cycle, the procedure used is similar to the ASME Code procedure,
except that the algebraic difference of each component stress is calculated for the load cycle, and
this is used to calculate the principal stress range. The largest principal stress range is then used
with the S-N curves. It is also noted that it is conservative to use the difference between the
largest and smallest principal stresses occurring during the cycle when the cyclic loading is
complex.

BS PD 5500 does not consider environmental effects directly, and does not address high-
temperature applications.

The BS PD 5500 exclusion rules are similar to the ASME Section VIII Division 1 rules.

R5 and R6

The CEGB originally developed the R6 procedure [B-21] as a fitness-for-service procedure for
determining when a defect will fail by elastic-plastic fracture mechanics (EPFM). The original
procedure was an elastic-plastic modification to the two-criteria approach to fracture, where
defects were assessed against brittle fracture using linear elastic fracture mechanics (LEFM) and
against fully plastic, or limit load, failure. The R6 procedure was adopted by BSI and
incorporated into PD 6493, which included guidance on fatigue. The R5 procedure [B-22] was a
high-temperature version of R6, and was incorporated into PD 6539. Although R5 considers
high temperature applications, it was intended to cover conditions relevant to plant components
(i.e, a few hundred cycles of loading) and does not consider high cycle fatigue.

Because they are both fitness-for-service codes, the primary focus of R5 and R6 is on crack
propagation rather than crack initiation. The R5 fatigue design curve is based on fatigue failure
data, which includes both an initiation life Ni and a crack growth life Ng. The initiation life can
be derived from the total life N from the following empirical equation:

ln( N i ) = ln( N ) − 8.06 N −0.28 Equation B-11

Weldments are assumed to contain micro-cracks, and it is often suggested that it is more
appropriate to calculate the life of a weldment based on crack growth from an initial size of 0.2
mm deep to a failure depth (e.g., 1 mm).

Instead, the R5 approach uses fatigue strength reduction factors (FSRFs) for each class of weld
to reduce the fatigue life from that shown in the fatigue design curve. The R5 approach has been
shown [B-23] to provide equivalent results as the initiation-growth model described above when
an FSRF of 1.5 is used (Class 1 welds).

The R5 procedure allows the use of the ductility exhaustion method for creep-fatigue interaction,
if the appropriate material data are available. This method calculates the fatigue damage fraction
as in other codes, but uses the following relationship to calculate the creep damage fraction:

B-14
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

N
⎛ε ⎞
D = ∑ ni ⎜⎜ i ⎟⎟ Equation B-12
i =1 ⎝εr ⎠

where εi is the creep strain in the ith cycle, εr is the creep ductility, and ni is the number of cycles
of type i. The fatigue damage fraction and the creep damage fraction are then summed. If the
creep ductility data are not available, then the linear damage summation approach of ASME
Section III NH is used.

Because R5 is a fitness for service procedure, it uses the approach described above to determine
the initiation of a macroscopic crack. The remaining life is calculated using fracture mechanics,
based on the sum of the fatigue crack growth and the creep crack growth for each cycle.

For creep crack growth, R5 relies on a simplified approach using the reference stress method [B-
24]. The remaining life is calculated from a stress rupture curve using the reference stress, which
is defined as:

σ ref = P σ y / PL Equation B-13

where P is the applied load, σy is the yield stress, and PL is the limit load. The limit load depends
on the crack size, although a limit load is defined for the uncracked geometry, so this allows the
reference stress to be calculated with or without a crack. As a crack grows, the limit load
decreases and the reference stress therefore increases.

Creep crack growth is evaluated using the C* integral:

= A (C *)
da q
Equation B-14
dt

where A and q are material constants. The value of C* can be estimated from:

C* = σ ref ε&ref R' Equation B-15

where ε&ref is the strain rate at the reference stress and R' is a length scale related to the applied
stress intensity factor K:

K2
R' = Equation B-16
σ ref 2

BS PD 6493

As noted above, the R6 method was incorporated into BS PD 6493 [B-25]. Because BS
PD 6493 is a fitness-for-service standard, its focus is on fatigue crack growth rather than
initiation. It provides recommended crack growth rate curves for two R-ratios. The low R-ratio

B-15
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

(R<0.5) curve is identical to the ASME Section XI air curve in Figure B-5. The high R-ratio
curve (recommended for weldments) is a factor of two above this curve.

BS PD 6493 provides additional crack growth rate curves for water environment, but these were
created for a marine environment (saltwater) for offshore structures, and are not appropriate for
pressure vessels. BS PD 6493 also provides a high-temperature correction for the crack growth
rate, for use up to 600°C (1112°F):

3
FHT = ⎛⎜ 100°C ⎞
E
⎟ Equation B-17
⎝ E HT ⎠

where E100°C and EHT are the elastic moduli at 100°C (212°F) and the operating temperature,
respectively. The crack growth rate is multiplied by this factor FHT.

BS PD 6493 also provides a series of 10 S-N curves for crack initiation. The first six of these are
identical to the BS PD 5500 curves for classes D, E, F, F2, G, and W.

BS PD 6493 has been superceded by BS 7910.

French Codes and Standards

RCC-MR

RCC-MR [B-26] was developed in France as a high temperature extension of RCC-M, the code
for design of French pressurized water reactors, to be applicable to breeder reactors. The basic
rules are similar to ASME Section III Subsection NH rules, but provide more detailed
instructions on how to carry out the analysis.

RCC-MR considers both fatigue crack initiation and crack growth. Fatigue crack growth is
based on an effective ΔK, based on the J estimation scheme, and is added to creep crack growth,
which is calculated using the C* integral.

RCC-MR uses a creep-fatigue interaction rule similar to the ASME linear damage rule, except
that a Tresca criterion is used in place of the Von Mises criterion of Eq. B-1. The strain range is
amplified to account for plasticity and creep effects in a slightly different way than the ASME
procedure, and the effective stress is divided by 0.9 instead of 0.67 as in the ASME procedure
before computing the creep damage. RCC-MR is the only code with creep data for P91 material.

B-16
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

German Codes and Standards

TRD 301

The TRD standards are the German technical rules for steam boilers. TRD 301 provides for the
design of cylindrical shells under internal pressure. Annex 1 of TRD 301 [B-27] provides the
rules for fatigue design.

The S-N curves of TRD 301 are based on incipient crack formation on unnotched bars. Curves
are provided at various temperatures up to 600°C (1112°F), as shown in Figure B-9. It is
believed that the fatigue design curves employ a factor of either 1.5 on stress or 10 on life
(whichever produces the lowest life [B-2].

These S-N curves are corrected by a surface finish correction factor that depends on the strength
of the material. A mean stress correction based on the Gerber rule is applied to the stress values
used to compare against these curves.

TRD 301 provides equations for calculating the stresses at branch connection bore holes for
pressure and thermal loading conditions. It also specifies the following fatigue strength
reduction factors:
2.6 for set-through and full penetration welded branches
2.9 for welded-on branches
5.0 for expanded or expanded and welded tube joints
2.0 for thermal stresses
and includes a surface finish correction factor ranging from 1.0 to 1.4, depending on the material
strength. TRD 301 also provides equations for calculating stresses as a function of the rate of
change of temperature. These are used to define allowable temperature change rates (dT/dt) as
well as allowable local thermal gradients (dT/dx). It is important to appreciate that the
temperature change rates calculated by TRD 301 produce quasi steady state temperature
gradients in headers. Thus for cold and cool warm starts when condensation heating causes rapid
step increases to saturation temperature of internal surfaces of steam filled pressure parts there
must be a pressure hold for as long as is necessary to establish the quasi steady state temperature
gradient in header or drum wall before ramping pressure and temperature at the ramp rate
calculated by TRD 301.

TRD 301 restricts the number of cold starts such that the fatigue usage from cold starts alone
must be less than 0.2. For variable amplitude loading, TRD 301 relies on linear cumulative
damage, with an allowable fatigue usage factor of 0.5.

For components in contact with water, TRD 301 provides protection of the magnetite layer by
placing minimum and maximum limits on the hoop stress.

The exclusion rules for TRD 301 exempt vessels from more detailed fatigue assessment for
internal pressure and temperature changes for up to 10,000 cycles from ambient condition for

B-17
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

carbon or low alloy steel if the maximum allowable working pressure does not exceed 3.2 MPa
(460 psig) or the membrane stress due to design pressure does not exceed 150 MPa (22 ksi).

Figure B-9
TRD 301 Fatigue Design Curves

TRD 508

Annex 1 of TRD 508 [B-28] provides guidance on the calculation of the service life under creep
and fatigue loadings, and is based on a life fraction concept (they use the term exhaustion, but it
is not used here to eliminate confusion with the ductility exhaustion method). The total life
fraction is the sum of the fraction (given in percent) due to fatigue and the fraction due to creep.
The fatigue life fraction is calculated from TRD 301, and the creep life fraction is based on linear
summation of damage from individual hold times compared with stress rupture data.

In effect, this procedure is equivalent to the dashed line in Figure A-21, which has been shown to
be nonconservative under certain conditions.

Other German Codes and Standards

The 1988 edition of AD-Merkblatt S2 [B-29] was similar to TRD 301, in that it used the same
approach, had the same factors of safety and temperature correction. The design curves are
similar to the unwelded material curves of EN 12952 (discussed below, and shown in Figure B-
10), except that the AD-Merkblatt S-N curves had a correction factor of 1.5 on stress and 10 on
life (the EN 12952 curves are mean failure curves). The AD-Merkblatt S2 procedure used an

B-18
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

approach of defining three levels of stress concentrations similar to that used in EN 12952
(discussed below).

The 1995 edition of AD-Merkblatt S2 [B-30] took on the welded joint approach like BS PD
5500, which is based on the structural stress rather than the nominal stress. There are only four
weld details in the AD-Merkblatt S2 code, and these are defined in a way that systematically puts
a detail into a lower class than in the BS PD 5500 procedure.[B-31]

European Community Codes and Standards

EN 12952

The European standard for the design of water tube boilers is EN 12952. This standard appears
to have been based on the German TRD 301 and 1988 AD-Merkblatt standards described above.
Fatigue is addressed in Part 3 of EN 12952, [B-32] and is applicable up to 600°C (1112°F), but is
non-conservative, possibly by an order of magnitude in respect to creep-fatigue interaction.
Creep is addressed in Part 4 of EN 12952. [B-33].

The EN 12952 procedure is based on the maximum principal stress range when the direction of
stresses remains constant. Corrections are made to the stress range when part of the cycle is in
the plastic range.

EN 12952 presents a series of room-temperature fatigue curves for unnotched specimens of


different material strengths, as shown in Figure B-10. The data in this figure represent mean
failure behavior, so a correction factor of 1.5 on stress and 10 on cycles is required to be applied
to this curve.

Three fatigue strength reduction factors, K1, K2, and K3, are defined (for slight, moderate, and
pronounced notched effects, respectively) to account for real notch geometries. An additional
factor considers the surface notch effect due to mill scale. Both the fatigue strength reduction
and mill scale factors are a function of the number of cycles to failure as well as the material
strength.

B-19
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

Figure B-10
Fatigue Design Curves of EN 12952-3

In addition to the notch FSRFs, EN 12952 presents FSRFs for nozzles under pressure and
thermal loading. Figure B-11 shows the FSRF for a cylindrical vessel under pressure loading.
The parameter zeta, ζ is a non-dimensional ratio defined as:

db ds
ζ = Equation B-18
ds 2t s

where d is the mean diameter, t is the thickness, and the subscripts b and s refer to the branch and
shell, respectively. It should be noted that Figure B-11 does not match the corresponding figure
in EN 12952 because of a mistake in the standard. The mistake is minor, except for tb/ts = 0. It
is not clear whether the error is in the figure or the equation.

B-20
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

Figure B-11
EN 12952 SCF for Pressurized Nozzle

EN 12952 also provides a fatigue strength reduction factor for thermal stresses, as shown in
Figure B-12. The ferritic curve appears to be the same as the fatigue strength temperature
correction curve listed in International Institute of Welding document IIW XIII-1539-96/XV-
845-96. The thermal stress is based on a through-wall temperature gradient.

No explicit guidance is provided for stress concentrations due to external forces and moments on
branches. Since the temperature correction curves have no loading rate or hold time factor it is
assumed that these do not include a creep-fatigue interaction correction factor. Until further
research provides clarification of this it is recommended that an additional reduction factor of 10
on cycles to cracking be applied for components operating above 427oC (800oF) to account for
creep-fatigue. This extra correction factor is perceived to be especially important for Grade 91
(modified 9% chromium) components since metallurgical evaluations of the failures of HRSG
pressure parts fabricated from this material have almost always revealed a creep-fatigue rather
than thermal-fatigue damage mechanism.

B-21
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

Figure B-12
Stress Concentration Factor for Thermal Stresses

An additional correction factor, shown in Figure A-17, accounts for the temperature effect, up to
600°C (1112°F). The temperature used in this curve is a weighted mean reference temperature,
equal to 0.75 times the maximum temperature in the cycle plus 0.25 times the minimum
temperature in the cycle. This factor represents the reciprocal of an FSRF because the
temperature correction factor is a multiplier for the allowable strength rather than the applied
stress.

As discussed in Appendix A this correction factor only accounts for elevated temperature fatigue
without hold times. With hold times an additional correction factor of 10 on cycles to cracking
should be used or a comprehensive creep-fatigue analysis should be performed to justify a less
conservative creep-fatigue correction factor.

EN 12952 uses linear cumulative damage to account for variable amplitude loading, where the
usage factor is limited to 1.0. If the numbers and types of transient events are

not known, 2000 cold starts are assumed. When the fatigue loads have not been specified and
2000 cold starts have been assumed, the usage factor is limited to 0.4.

The exclusion rules of EN 12952 exclude welded components joining materials with similar
coefficients of thermal expansion if previous experience has shown that the loading conditions

B-22
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

will cause only insignificant fatigue damage. When external loads are significant, parts that
satisfy all of the following criteria are excluded from the requirement of a fatigue analysis:
• the part is designed for sustained pressure in accordance with EN 12952-3
• the number of transients with >50% MAWP pressure change is less than 3000
• the number of transients with <50% MAWP pressure change is less than 10,000
• the mechanical loads on branches are limited
• the local temperature differential for satisfy an interaction equation

Unfortunately, neither of the last two factors can be quantified due to errors in the standard. The
bending moment appears to be limited to a value where the bending stress equals the allowable
stress for the material. The interaction equation between ΔTcold start and ΔThot start can not be
evaluated due to a lack of defined variables.

For components in contact with water, EN 12952 provides protection of the magnetite layer by
placing minimum and maximum limits on the hoop stress. The minimum limits are imposed
because it is assumed that the magnetite layer forms under operational conditions, and the oxide
layer will therefore be in compression during a shutdown.

EN 12952 also provides a procedure for calculating the allowable through-wall temperature
difference and the temperature ramp rate for startup and shutdown similar to that in TRD 301. It
is important to appreciate that the temperature change rates calculated by EN 12952 produce
quasi steady state temperature gradients in headers and do not account for very rapid large
changes in heating and cooling rates caused by condensation heating, condensate quenching or
severe steam temperature ramps up and down that often occur at attemperator outlets during
loading.at startups Thus for cold and cool warm starts when condensation heating causes rapid
step increases to saturation temperature of internal surfaces of steam filled pressure parts there
must be a pressure hold for as long as is necessary to establish the quasi steady state temperature
gradient in header or drum wall before ramping pressure and temperature at the ramp rate
calculated by EN 12952.

The creep rules of EN 12952 are essentially the same as TRD 508.

EN 13445

The European standard for unfired pressure vessels in EN 13445. [B-34] This standard appears
to have been built based on the 1995 AD-Merkblatt standard described above and the welded
joint approach used in the BS PD 5500 and Eurocode procedures. The standard does not apply
to components acting in the creep regime (< 380°C (716°F) for ferritic steels and < 500°C
(932°F) for austenitic stainless steels).

EN 13445 is unique in that it provides design criteria based on equivalent stresses and a second
set of criteria based on principal stresses. Nominal stresses are used for weld metal in directly
loaded fillet welds at partial penetration welds that are not seam welds. For all other welded

B-23
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

components, the stresses should be based on the structural stress extrapolated to the crack
initiation site (e.g., the toe of the weld).

The EN 13445 fatigue design curves (shown in Figure B-13) are based on welded laboratory
specimens tested under load control, and are approximately three standard deviations below the
mean curve. This represents a 99.9% lower bound. Each of the weld classes represents a
specific weld detail. The class name corresponds with the allowable stress (in MPa) at 2x106
cycles.

Design details that are not shown in one of the weld classes should be evaluated as Class 32.
However, if fatigue data are available for an unclassified design detail, guidance is provided on
the factors of safety that should be applied (as a function of the number of test specimens) to
develop a design curve for that detail.

Figure B-13
Fatigue Design Curves of EN 13445

Note that EN 13445 allows consideration of the beneficial effect of fatigue improvement
techniques, such as toe dressing. The standard cautions that fatigue improvement techniques can
not affect the fatigue performance when it is governed by other features, such as weld roots, and
also is not effective when the weld is exposed to a corrosive environment that can cause pitting
(which act as fatigue initiation sites).

The thickness correction factor is the same as in BS PD 5500 (Eq. B-10) except that the
numerator is 25 mm instead of 22 mm. Mean stress effects are not considered, except for
unwelded components. Corrections are made to the stress range when part of the cycle is in the
plastic range. EN 13445 uses the same temperature correction as in EN 12952 (shown in Figure
A-17).

B-24
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

The endurance limits, shown by the dashed lines in Figure B-13, are applicable only to constant
amplitude loading. For variable amplitude loading, linear cumulative damage is employed, and
the fatigue usage factor is limited to 1.0 (i.e., no correction factor, as it has already been applied
to the design curves).

EN 13445 recommends an adjustment to the fatigue strength to account for environmental


conditions, but does not provide any guidance on the values to use. They do recommend a
higher inspection frequency if there is a lack of confidence in the environmental factor used.
The criteria for protecting the magnetite protection layer is the same as EN 12952 and TRD 301.

The EN 13445 exclusion rules are limited, but allow up to 500 full pressure cycles. Other small
pressure fluctuations can be ignored.

EN 13445 provides a separate design procedure for unwelded components, using fatigue design
curves for four strength levels, as shown in Figure B-14. These curves were generated from
unnotched specimens tested under alternating load (R=0), and represent factors of safety of 1.5
on stress and 10 on life from the mean failure data. These curves are adjusted for mean stress,
thickness (using a different thickness correction), temperature, and surface finish.

Figure B-14
EN 13445 Design Curve for Unnotched Material

B-25
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

Other European Codes and Standards

The Eurocode 3 is a design code for steel structures, which provides a series of fatigue curves for
weldments similar to BS PD 5500. The Eurocode weld classes range from 36 to 160, which in
this case represents the allowable stress (in MPa) at 107 cycles, not 2x106 cycles, as in other
codes. [B-35]

Comparison Of Fatigue Code Attributes

This review of international codes and standards relevant to fatigue issues in HRSG components
is by no means comprehensive, but it includes a sufficient number of different types of codes and
standards to examine different methods of solving the same problem. Deficiencies in one
standard may be resolved by using a correction factor derived from another standard. The table
below provides a brief summary of the features of each code or standard. The key features are
discussed and compared in Table B-1.

Exclusion Rules

Exclusion rules are criteria that in effect state that fatigue is not considered to be a concern for a
given component design and anticipated operating transient forecast.

The first and foremost exclusion rule is experience. If a specific design configuration has
significant operational experience without experiencing a fatigue failure, then there is no real
reason to perform a fatigue analysis, unless that configuration is used outside the experience
base. That is the underlying assumption of the design by rule standards, such as ASME Sections
I and VIII (Div. 1).

However, in practice there is little or no relevant experience with large numbers of start-stop
cycled HRSGs which resemble the current design offerings, so no manufacturer can demonstrate
adequate experience with cycling. Furthermore, even if such experience had been acquired on a
10+ year old HRSG design, even seemingly small design changes can substantially increase the
fatigue risk. Examples of adverse design changes include:

Interconnecting drains from different HPSH sections, which is certain to cause condensate
migration problems

Modifying the position of attemperators in HPSH sections closer to the HP drum, which will
cause even greater overspraying problems at startup to those widely experienced with current
designs

B-26
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

Table B-1
Comparison of Fatigue Codes

S2
1

2
Di v

Di v

t
H

la t
B

00

93

39
9
N

57
I

5
r kb
55

64

65
VI I

VI I

8
XI

MR
II I

II I

13

00

10

95

44
30

50
I

e
PD

PD

PD
ME

ME

ME

ME

ME

ME

12

13
11

55

79
IR

-M
C-

D-

D-
RC

EN
AD

EN
AS

AS

AS

AS

AS

AS

AP

BS

BS

BS

BS

BS

BS

TR

TR
R5

R6
A m e ric a n C o d e s / S t a n d a rd s B rit is h C o d e s / S t a n d a rd s F re n c h G e rm a n E u ro p e a n
D e s ig n b y R u le X X X
D e s ig n b y A n a ly s is X X X X X X
F it n e s s -fo r-S e rvic e X X X X X X X X
F a t ig u e X X X X X X X X X X X X X X X
E x c lu s io n R u le s A ll X X A ll X X X X X X X
E x p e rie n c e X X X
N u m b e r o f s t a rt s X X
To t a l n u m b e r o f c y c le s X X X X X
R a m p ra t e s X
Th ro u g h -w a ll T X X X X
M u lt ia x ia l S t re s s e s
M a x P rin c ip a l X X X X X X
Tre s c a X* X X X
V o n M is e s X*
S -N C u rve s X X X X X X X X X X
B a s is * * U U U 2 UW W U W U 2
C o rre c t io n fa c t o r o n s t re s s 2 2 2 2.2 1.5 1.5
C o rre c t io n fa c t o r o n life 20 20 20 15 10 10
S t d . D e v. b e lo w m e a n 2 4 3
R e le va n t F S R F s X X X X X
C u rve s fo r W e ld m e n t s X X X X X
H ig h Te m p e ra t u re X X X X X
F a t ig u e C ra c k G ro w t h X X X X X
C u rve s fo r W e ld m e n t s X
H ig h Te m p e ra t u re X
E n viro n m e n t a l E ffe c t s
P ro t e c t io n o f M a g n e t it e L a y e r X X X X
M u lt ip lie r o n L ife
W a t e r C u rve s X X
C re e p X X X X X X X
S t re s s -R u p t u re C u rve s X X X
L a rs o n -M ille r P a ra m e t e r X
D u c t ilit y E x h a u s t io n X X X
O m ega X
C re e p C ra c k G ro w t h X X X X X
C re e p -F a t ig u e In t e ra c t io n X X X X X
K n e e in D a m a g e D ia g ra m X X
L in e a r w / F S 0.5 1
* In d ic a t e s e q u iva le n t s t re s s e s s h o u ld b e c a lc u la t e d u s in g t h is m e t h o d w h e n t h e d ire c t io n o f lo a d in g c h a n g e s
* * B a s is fo r S -N c u rve s : U = u n n o t c h e d b a s e m e t a l s p e c im e n , U W = u n n o t c h e d w e ld m e t a l, W = w e ld m e n t s , 2 = 2 c u rve s (1 u n n o t c h e d & 1 w e ld e d )

B-27
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

There are two basic types of exclusion rules in the various Codes and Standards: those that limit
the number of startup cycles and load changes, and those that limit the stresses. ASME Section
III NB limits the total number of full pressure cycles to the number of cycles allowed for a stress
level corresponding to the ultimate strength of the material at temperature, or twice the yield
strength at temperature, whichever is less. For a 2¼Cr-1Mo material, this corresponds to about
4000 full pressure cycles.

Since the dominant cause of fatigue in HRSG pressure parts is not pressure cycles, but rather
transient thermal mechanical stresses during major dynamic upset operating conditions this
exclusion rule is not applicable for HRSGs. The example in Appendix A (Figure A-13) would
not require a fatigue analysis under this criterion despite the fact that cyclic forces and bending
moments caused by tube-to-tube temperature differencescould be large enough to cause fatigue
failure in a low number of cycles.

The Section VIII exclusion rules specify that a fatigue analysis is not required if the total number
of cycles (including full-range pressure cycles, cycles where the pressure range exceeds 20% of
the full pressure, and where the metal temperature difference between two points exceeds a given
amount) is less than 1000. A fatigue analysis would be required for our Appendix A example
under this criterion because of the tube-to-tube temperature differences.

The BS PD 5500 exclusion rules are similar to the ASME Section III NB rules.

TRD 301 restricts the number of cold starts such that the fatigue usage from cold starts alone
must be less than 0.2. Using the assumed 2000 cold starts, this means the peak stress must
correspond to a life of 10,000 cycles, or about 55 ksi (379 MPa). For an FSRF of three, this
results in an allowable nominal stress of 18 ksi (124 MPa). A fatigue analysis would not be
required for our example under this criterion based on steady state or quasi-steady state
conditions.

In HRSGs the dominant cause of fatigue damage is usually not steady state temperature
differences or quasi-steady state conditions produced by linear steam temperature ramps but
transient thermal mechanical stresses developed by severe temperature gradients during major
upset operating conditions, including startups, shutdowns, when attemperation commences, etc.
If exclusion rules are used then they need to include a detailed consideration of both the quasi-
steady state and the often very short lived peak transient thermal mechanical stresses. Unless the
owner/operator or HRSG manufacturer has previous experience with a very similar design, it is
unlikely that the magnitude and frequency of the peak transient thermal mechanical stresses will
be known during the design phase. To overcome this deficiency requires good engineering
judgment regarding the potential for off-design thermal mechanical transients and subsequent
verification of these judgments by diagnostic/troubleshooting monitoring during commissioning
or during the warranty period.

The exclusion rules for TRD 301 exempt vessels from a more detailed fatigue assessment for
internal pressure and temperature changes for up to 10,000 cycles from ambient condition for
carbon or low alloy steel if the maximum allowable working pressure does not exceed 460 psig
(3.2 MPa). This exclusion rule seems overly generous because the fatigue life depends on the

B-28
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

stress, not the pressure (a thin low-pressure vessel can have higher stresses than a thicker high-
pressure vessel). TRD 301 also excludes components where the membrane stress due to design
pressure does not exceed 22 ksi (150 MPa). Our example would be inappropriately excluded
from the requirement of a fatigue analysis by this criterion.

The other type of exclusion criteria is one based on stress limits. ASME III NB excludes
components where the temperature difference between adjacent points is less than Sa/2Eα (where
Sa is the allowable stress amplitude for the number of specified startup-shutdown cycles) and
where the full range of mechanical loads is less than Sa for the anticipated number of load
excursions.

For the 8000 hot and cold starts specified in our example, the temperature criterion corresponds
to a ΔT of 99°F (37°C). This ΔT means that if the metal temperature gradient through the
thickness at any point is greater than 99°F, a detailed fatigue analysis is required. Similarly, if
the temperature difference of two points on the same surface, but separated by 2 Rt or less is
greater than 99°F, a detailed fatigue analysis is required.

The mechanical loads created by the assumed differential thermal expansion creates a bending
stress of only a few hundred psi. This is far below the allowable stress for 8000 cycles, and the
component would be excluded from consideration of fatigue under the ASME Section III NB
criterion provided the temperature criterion above is met.

The exclusion rules of EN 12952 are a combination of cycle limits and stress limits. The EN
12952 criteria require that both sets of limits be met:
• the number of cold starts (>50% pressure change) is less than 3000, and
• the number of hot starts and load changes (<50% pressure change) is less than 10,000, and
• the mechanical loads on branches are limited , and
• the local temperature differential for cold starts and hot starts satisfy an interaction equation

Because of the errors in the EN 12952 standard, the last two criteria cannot be evaluated. A
fatigue analysis would not be required for our example under the cycle criteria but probably
would be required under the last two criteria because the loading mechanism is primarily the
forces and bending moments caused by tube-to-tube temperature differences.

TRD 301 and EN 12952 also provide rules that, although they are not exclusion rules, provide
guidance on operational limits, specifically the maximum allowable through-wall temperature
difference (defined as the difference of the inside surface temperature and the mean wall
temperature) and the maximum ramp rate (the time rate of change of the metal temperature). For
the geometry assumed, a maximum allowable ΔT of 26°F (14°C) is calculated. The maximum
allowable ramp rate is calculated to be 350°F (194°C).

B-29
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

Conclusions

This appendix has reviewed international codes and standards relating to fatigue and creep of
pressure boundary components typical of heat recovery steam generators. Each of the codes and
standards has its own strengths and weaknesses, and no one single standard provides everything
that a HRSG designer needs to perform an adequate fatigue design.

The ASME Codes are inadequate in their treatment of welds. The weld-based fatigue design
standards are inadequate in the area of structural stress concentrations. Most of the creep codes
do not adequately address newer materials such as P91, and the treatment of weld metal and heat
affected zone is inadequate. Many of the codes have exclusion rules that nonconservatively
conclude that fatigue is not a consideration. However, between all of the various codes and
standards, all of the elements necessary to perform a proper fatigue evaluation exist.

The problem is not in defining the acceptance criteria, it is in implementing the fatigue analysis
procedure. None of the codes and standards provides adequate guidance on the details of fatigue
design. Specifically, the area of how to address structural discontinuities is lacking. Guidance
on specific structural discontinuities relevant to HRSGs is sorely needed.

Many of the fatigue problems facing HRSGs are not the result of inadequate design codes. They
are the result of HRSG designs that are intolerant of transient temperature differences in
conjunction with more severe thermal conditions during major upsets at startups and shutdowns
than those anticipated in the design process. These include unanticipated thermal transients and
temperature differences that result in bowing of the header and interconnecting pipes applying
moments at the tube to header connections of offset tubes. Additional guidance is needed in
specifying the specific transients to be evaluated in the fatigue design, or there must be
assurances (backed up by testing) that these unanticipated thermal events do not occur.

The technology for ensuring fatigue-free performance of HRSG components exists. Some
manufacturers are better at using this technology than others. Development of design guidelines
specific to HRSG components will result in more operationally flexible HRSG designs capable
of extensive cycling operation with increased performance across the industry.

References

[B-1] “Power Boilers”, ASME Boiler and Pressure Vessel Code, Section I, 1995.

[B-2] “Comparison of Fatigue Assessment Techniques for Heat Recovery Steam Generators”,
American Boiler Manufacturers Association Task Group on Cyclic Service.

[B-3] “Rules for Construction of Nuclear Power Plant Components”, ASME Boiler and
Pressure Vessel Code, Section III, Division 1, Subsection NB, “Class 1 Components,”
1995.

B-30
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

[B-4] Cooper, W.E., “The Initial Scope and Intent of the Section III Fatigue Design
Procedures”, Pressure Vessel Research Council Workshop on Environmental Effects on
Fatigue Performance, 1992.

[B-5] “Rules for Construction of Nuclear Power Plant Components”, ASME Boiler and
Pressure Vessel Code, Section III, Division 1, Subsection NH, “Class 1 Components in
Elevated Temperature Service,” 1995.

[B-6] “Class 1 Components in Elevated Temperature Service”, ASME Boiler and Pressure
Vessel Code, Code Case N-47-29, December 3, 1990.

[B-7] Neuber, H., “Theory of Stress Concentrations for Shear-Strained Prismatical Bodies with
Arbitrary Nonlinear Stress-Strain Law”, Journal of Applied Mechanics, December 1961,
pp. 544-550.

[B-8] “Pressure Vessels”, ASME Boiler and Pressure Vessel Code, Section ASME VIII, Div. 2,
“Alternate Rules,” 2001.

[B-9] “Rules for Inservice Inspection of Nuclear Power Plant Components”, ASME Boiler and
Pressure Vessel Code, Section XI, Appendix A, “Analysis of Flaws”, 1995.

[B-10] “Fatigue Crack Growth Rate Curves for Ferritic Steels in PWR Water Environment,
Section XI, Division 1”, ASME Boiler and Pressure Vessel Code, Code Case N-643, May
24, 2000.

[B-11] Fitness for Service, American Petroleum Institute, API RP 579, January 2000.

[B-12] Structural Welding Code, American Welding Society, AWS D1.1, 1996

[B-13] Guide Specification for Fatigue Design of Steel Bridges, American Association of State
Highway and Transportation Officials, AASHTO, 1989.

[B-14] AASHTO LRFD Guide Specification for Fatigue Design of Steel Bridges, American
Association of State Highway and Transportation Officials, AASHTO, 1994.

[B-15] “Specification for Design and Manufacture of Water-Tube Steam Generating Plant
(Including Superheaters, Reheaters, and Steel Tube Economizers)”, British Standards
Institute, BS 1113:1999.

[B-16] “Specification for Unfired Fusion Welded Pressure Vessel”, British Standards Institution,
BS 5500, 1985.

[B-17] Barsom, J.M., and Vecchio, R.S., “Fatigue of Welded Components”, Welding Research
Council Bulletin 422, June 1997.

B-31
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

[B-18] “Specification for Unfired Fusion Welded Pressure Vessel”, British Standards
Institution, BS 5500, 1997.

[B-19] “Specification for Unfired Fusion Welded Pressure Vessel”, British Standards Institution,
BS PD 5500, 2003.

[B-20] Maddox, S.J., “Fatigue Analysis of Pressure Vessels in BS 5500”, Welding Research
Council Bulletin 374, August 1992.

[B-21] Harrison, R.P., Loosemore, K., Milne, I., and Dowling, A.R., “Assessment of the
Integrity of Structures Containing Defects”, Central Electricity Generating Board
R/H/R6, Rev.2, April 1980.

[B-22] “An Assessment Procedure for the High Temperature Response of Structures”, Berkeley
Nuclear Laboratories Report R5, 1990.

[B-23] Connors, D.C., ed., “R5/R6 Newsletter”, Nuclear Electric, No. 28, December 2003.

[B-24] Goodall, I.W., and Ainsworth, R.A., “R5: An Assessment Procedure for the High
Temperature Response of Structures”

[B-25] “Guidance on Methods for Assessing the Acceptability of Flaws in Fusion Welded
Structures”, British Standards Institution, BS PD 6493, 1991.

[B-26] Design and Construction Rules for Mechanical Components of FBR Nuclear Islands,
French Society for Design, Construction and Surveillance Rules for Nuclear Island
Components, RCC-MR, June 1995.

[B-27] “Calculation for Cyclic Loading Due to Pulsating Internal Pressure or Combined
Changes of Internal Pressure and Temperature”, Technical Rules for Steam Boilers,
Annex 1 (Design), TRD 301, April 1975.

[B-28] “Additional Tests on Components – Methods for the Calculation of Components Having
Time Dependent Design Strength Values”, Technical Rules for Steam Boilers, Annex 1,
TRD 508, October 1978.

[B-29] Design to Allow for Fluctuating Stress, AD-Merkblatt S2, 1988.

[B-30] Fatigue Analysis, AD-Merkblatt S2, 1995.

[B-31] “Assembly of Information on European Standards Development (CEN) Relevant to


Pressure Equipment Fatigue Design”, Centre Technique des Industries Mecanique,
Annex 6 to Pressure Components Fatigue Design in the Framework of Directive
97/23/EC on Pressure Equipment, Pressure Equipment Directive, July 2001.

B-32
EPRI Proprietary Licensed Material

Review of Fatigue Design Codes and Standards

[B-32] “Design and Calculation for Pressure Parts”, Water-Tube Boilers and Auxilliary
Installations, Part 3, EN 12952-3, December 2001.

[B-33] “In-service Boiler Life Expectancy Calculations”, Water-Tube Boilers and Auxilliary
Installations, Part 4,”In-service Boiler Life Expectancy Calculations”, EN 12952-4, 2002.

[B-34] Unfired Pressure Vessels, EN 13445 Part 3, draft 2001.

[B-35] Maddox, S.J., “Fatigue Design Review Task 5 – Assembly of Available Fatigue Data
Relevant to Pressure Equipment Design”, Annex 5 to Pressure Components Fatigue
Design in the Framework of Directive 97/23/EC on Pressure Equipment, Pressure
Equipment Directive, September 2001.

B-33
EPRI Proprietary Licensed Material

C
ESTIMATING THE FLEXIBILITY/TUBE-TO-TUBE
TEMPERATURE DIFFERENCE INFLUENCED FORCES
AND MOMENTS AT TUBE-TO-HEADER
CONNECTIONS

There are numerous tube bundle configurations used in HRSGs. Hand calculations can be used
for many of the common bundle configurations to determine the forces and moments acting at
the tube-to-header connections due to a combination of tube row-to-row or element-to-element
temperature differences. Four configurations with arbitrary dimensions and dogleg bend angles,
and any desired number of tube elements that have been modeled in this way are shown in
Figure C-1. Other more complex bundle configurations could be modeled using this approach.

The determination of the displacements, forces and moments can also be performed using
commercially available piping flexibility codes. This is often performed by designers in two
steps. The first step being to estimate the forces, moments, displacements and code stress values
in the header. The header models are often simplified by assuming a single tube row that has the
equivalent stiffness of the combined rows of tubes in the harp. This type of model provides
adequate code checks but doesn’t provide meaningful data for a fatigue assessment, especially if
temperature differences are anticipated or have been measured between tubes within individual
rows and between rows.

A second code flexibility analysis is then performed to assess the tubing displacements, forces,
moments and code stresses. The analysis often assumes that the headers are infinitely stiff and
only includes the tubes at the ends of the headers and on each side of any header partition plates.
This is also an acceptable approach for design checks but lacks the detail needed for a fatigue
assessment of the tube to header connections.

It should also be recognized that most commercially available piping flexibility analysis codes
only provide estimates of code stresses, which might not have appropriate estimations of the
connection stress intensification factors or local micro-notch weld discontinuity stress
concentration factors. To overcome these deficiencies the analysis should be performed using a
model that includes the actual tube bundle configuration and all tube rows. The forces and
moments output from the model should be used to independently estimate the tube to header
nominal, intensified, and weld discontinuity concentrated stress values.

C-1
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

Figure C-1
Geometry Characteristics of Selected Types of Tube Bundles

C-2
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections
The forces and moments resulting from the constrained differential thermal expansion of the
tubes within the bundle should be combined with the pressure load and local thermal expansion
discontinuity loads that result from more rapid heating/cooling of the tube compared with the
header. These combined stresses must then be converted to local tube to header stress
magnitudes and ranges using fatigue strength reduction/stress intensification /stress
concentration factors such as those in EN 13480-3.

Fatigue life consumption estimations can be estimated with these stress range values using the
fatigue life rules in EN 12952-3. For tubes operating at temperatures above 427oC (800oF) the
estimated cycles to cracking should be reduced by a factor of 10 to account for creep-fatigue
damage. For water-touched tubes the oxide cracking checks in EN 12952-3 should be made and,
if exceeded, the estimated cycles to cracking should be reduced by a factor of 10 to account for
corrosion fatigue damage.

The following is a derivation of closed-form solution for stress analysis of tube-header


connection due to differential tube row temperatures. The model includes two headers, a straight
tube connecting them, and a second tube that comes out at an angle and bends to run parallel to
the first tube, but offset from the centerline of the headers and the other tube. The upper and
lower bends for the tube on the left side are called dogleg bends. This represents tube bundle
configuration #1 in Figure C-1, with a single pass, two parallel row “pant leg” configuration with
“dogleg” bends on the lower and upper sections of one of the two tube rows.

The model was based on matching the displacements from the thermal expansion of one tube
with the thermal expansion of the second tube. This requires determination of the stiffness of the
dogleg tube. This model was confirmed with finite element analysis.

Temperatures can be defined for both headers and at the top and bottom of each of the two tubes.
Where temperatures are defined at both ends of a tube, the thermal expansion is based on the
average of those two temperatures.

The results are equations for determining the axial stress and bending stress in the dogleg tube at
the tube to header connection. No consideration is given here for stress concentrations, which
must be added to the solution.

C-3
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

C-4
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

C-5
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

C-6
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

C-7
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

C-8
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

C-9
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

C-10
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

C-11
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

C-12
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

C-13
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

C-14
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

Similar derivations have been made for the tube bundle geometries with arbitrary numbers of
tube elements and any desired combination of top and bottom tube temperatures input for every
tube row and element. An example is given in Figures C-2 through C-4

Figure C-2
Geometry of a single pass, two parallel row tube bundle with dogleg bends at the bottom
of the leading row and top of the training row.

C-15
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

Figure C-3
Example top and bottom tube metal temperatures for a 32 element tube bundle with the
configuration and geometry shown in Figure C-2.

C-16
EPRI Proprietary Licensed Material

Estimating the Flexibility/Tube-to-Tube Temperature Difference Influenced Forces and Moments at Tube-to-Header
Connections

20000

15000

10000
Nominal Stress, psi

5000

-5000
Row 1 - Axial stress
-10000 Row 1 - Bending stress

Row 2 - Axial stress


-15000
Row 2 - Bending stress
-20000
0 5 10 15 20 25 30 35
Tube Element Number

Figure C-4
Estimated nominal axial and bending stress at the tube to header connections for the
geometry shown in Figure C-2 and the tube metal temperatures shown in Figure C-3.
To estimate the fatigue life these nominal stresses would be magnified by a stress
concentration factor and the tube temperatures for the remainder of the operating cycle
would need to be measured to determine if the stresses will just return to near zero or will
reverse and produce a stress range that exceeds these values.

C-17
EPRI Proprietary Licensed Material

D
EXAMPLE FATIGUE ANALYSIS OF AN LP
ECONOMIZER (PREHEATER) WITH TUBE-TO-TUBE
AND ROW-TO-ROW TEMPERATURE DIFFERENCES

This appendix provides an example fatigue life estimation performed for a preheater tube bundle
that experiences thermal quenching of a few tubes. The severity of the quenching evaluated is
outside the design-basis tube temperature distribution. The design-basis cyclic life is compared
to the life under the upset conditions. The analysis performed is based on estimated inelastic
stress and strain values rather than the commonly used linear elastic and virtual elastic stress
based methods that are used in most design codes. As will be shown, the inelastic stress/strain
based approach provides significantly more insight into the physical damage and local stress-
strain history experienced at the toe of the tube to header welds. In addition to estimating the
minimum and median fatigue life, this approach also provides estimations of the residual stress
and strain magnitudes at the weld toe on both the compression and tension side of the tube to
header connection.

Background Information

The three pass, three row preheater bundle shown in Figure D-1 has a worst case design basis
temperature profile as shown in Figure D-2. The tubes were fabricated from SA 192 tubing
(carbon steel). The reader should be aware that the example below used to illustrate the analysis
steps is not a worst case example. Far more severe transients have been measured in LP
economizers.

In addition to the nonuniform, steady state, side to side temperature distribution associated with
the three pass design and header partition plates, the analysis assumed that a few tubes in Row 3
located near the inlet pipe nozzle and lower header partition plate were selectively quenched with
cold feedwater during a cold start to a value of approximately 106oF (41oC). The few tubes that
are selectively quenched have a temperature that is approximately 70oF (39oC) cooler than their
neighbors in the first pass and approximately 100oF (56oC) cooler than the nearby tubes in the
second pass on the opposite side of the header partition plate. Using a flexibility model, the
estimated maximum non-intensified axial and bending stresses at the tube to header connection
(which occur in the Row 3 tubes that are selectively quenched) are:

Saxial = 3,085 psi (21 MPa)

Sbending = 22,115 psi (153 MPa)

D-1
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences

Figure D-1
Feedwater heater tube bundle attributes

Figure D-2
Design-basis tube and header temperatures. In addition to the nonuniform, steady state,
side to side temperature distribution associated with the three pass design and header
partition plates, the analysis assumed that a few tubes in Row 3 located near the inlet pipe
nozzle and lower header partition plate were selectively quenched with cold feedwater
during a cold start to a value of approximately 106oF (41oC).

D-2
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences
The tubes with the highest magnitude stresses were found to be the tubes that were selectively
quenched and located in the vicinity of the lower header inlet pipe nozzle and lower header
partition plate.

Material Properties

The estimated effective temperature of the thermal cycle was estimated using the relationship in
EN12952-3:

Teff = 0.75(Tmax) +0.25 (Tmin) = 0.75(176oF)+ 0.25(70oF) = 150oF (65oC) Equation D-1

The room temperature material property information in Appendix F for SAE 1005 material
(which is similar to SA 192 steel) was used in the assessment:

E = 29.1E6 psi (200,690 MPa)


H’ = 77,000 psi (531 MPa)
n’ = 0.16

Using these properties provided an estimate of the cyclic stress-strain curve response of the
material (Figure D-3):

εta = σa/E + (σa/H’)1/n’= σa/29.1 x 106 psi + (σa/77,000 psi)1/0.16 Equation D-2

where:
εta = total strain amplitude
σa = stress amplitude (psi)

The SAE 1005 steel fatigue curve values in Appendix F were used for the strain amplitude
versus reversals to failure curve:

σf' = 93,000 psi (641 MPa)


b = -0.109
E = 29.1E6 psi (200,690 MPa)
εf' = 0.100
c = -0.390

Δε/2 = σf'/E (2Nf)b + εf' (2Nf)c Equation D-3

Δε/2 = 93,000 psi/29.1E6 psi (2Nf)-0.109 + 0.100(2Nf)-0.390 Equation D-4

where:
Δε/2 = total (elastic + plastic) strain amplitude
Nf = cycles to failure, (2Nf = reversals to failure)

The resulting strain amplitude versus reversals to failure curve is shown in Figure D-4.

D-3
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences

Figure D-3
Ramberg-Osgood cyclic stress strain curve

Figure D-4
Coffin-Manson strain amplitude versus median reversals to failure curve for SAE 1005
carbon steel at room temperature. SA 192 tubing is anticipated to have similar fatigue
behavior.

D-4
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences

Nominal Elastic Stresses

The fluctuating stress values that must be considered in this tube to header fatigue analysis are:
1. Internal pressure stress
2. The axial and bending stress associated with the external forces and moments resulting
from tube to tube and row to row temperature differences
3. External forces and moments associated with water hammer and piping loads were
believed to be insignificant and were ignored in the analysis

The maximum stress on the outside surface of the tube resulting from internal pressure is the
axial stress which is estimated from the product of the gauge pressure and the cross sectional
area of the tube bore divided by the cross sectional area of the tube. An axial pressure stress
value of 164 psi (1.1 MPa) was estimated for the 87 psig (0.6 MPag) operating pressure. The
tube hoop stress is 410 psi (2.8 MPa).

In most design codes the pressure stress values used for fatigue assessment of tube to header
regions is the value of the borehole stresses at the header inside surface and borehole intersection
at the longitudinal plane of the header (this location is sometimes referred to as the “crotch
corner”). Since the crotch corner stresses are very localized and don’t extend to the outside
surface at the tube to header connection these much higher crotch corner pressure stress values
were not included in the evaluation of the fatigue life of the tube side tube to header fillet weld
toe. The minimum internal pressure stress is zero when the unit is offline and the internal
pressure has decayed to ambient. Throughout the overall loading cycle the minimum principal
pressure stress equal to the radial pressure stress which is zero at the outside surface of the tube.

The stresses resulting from external forces and moments associated with tube to tube and tube
row to row temperature differences were determined with a flexibility analysis. Maximum
nominal stress values of Saxial = 3,085 psi (21 MPa) and Sbending = 22,115 psi (153 MPa) were
determined for the simulated tube temperature differences associated with a cold feedwater
quenching event. The minimum external force and moment stress is zero and occurs when the
unit is off line and the tube temperatures are uniform.

Structural Elastic Stresses

The nominal stress values are significantly intensified by the gross structural discontinuity
associated with the tube to header connection to obtain an estimate of the structural stress. The
gross structural stress concentration factor for the tube to header connection was estimated using
Welding Research Council Bulletin WRC 297. This gross structural stress concentration factor
was applied to the internal pressure axial stress, the external force axial stress and external force
bending stress. A gross structural stress concentration (stress intensification factor) of 2.97 was
obtained with the WRC 297 relationships and the tube and header configuration shown in Figure
D-1. This value is similar to the value obtained from finite element modeling of the tube to
header connection.

D-5
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences
Gross structural stress concentration factors for branch connections are included in EN 13480-3,
ASME B31.1, ASME Section III, and PD 5500, Annex G but these do not consider the stresses
at the fillet weld toe located on the branch side of the connection welds where fatigue failures are
most likely to occur. The gross structural stress concentration factors in these codes only
consider the header side of the connections. The WRC 297 branch side of a cylinder to branch
connection gross structural stress concentration factor solution is also somewhat limited. Many
common tube to header configurations are outside the dimensional limits of the WRC 297
solution. When these cases are encountered more detailed analysis or fatigue testing of the
connections will be required to develop an appropriate gross structural stress concentration
factor. For the case illustrated in this appendix the WRC 297 was valid.

At the tube side weld toe on the tension side of the tube to header connection (Position A in
Figure D-1) during the quenching event the maximum combined axial plus bending principal
stress was:

SA, max principal = 2.97 (164 psi + 3,085 psi + 22,115 psi) = 75,330 psi (520 MPa)

At the tube side weld toe on the compression side of the tube to header connection (Position B in
Figure D-1) during the quenching event the minimum combined axial plus bending principal
stress was:

SB, min principal = 2.97 (164 psi + 3,085 psi - 22,115 psi) = -56,033 psi (-386 MPa)

Peak Maximum Principal Elastic Stresses

The structural stress must be further amplified to account for local micro-notch (local structural)
stress concentration factors associated with local variations in weld discontinuity characteristic.
There are a variety of different approaches used to address the stress concentration factor
associated with local weld discontinuities. In some design codes a series of design curves that are
assigned classifications according to the weld detail characteristics are used. In other cases some
bounding stress concentration factors for specified weld detail characteristics are used. For this
illustration the approach included in EN 12952-3 was used.

In this case there are three sets of tables that are provide illustrations of weld details with slight
(Group K1), moderate (Group K2) and severe (Group K3) micro-notch configurations. Set-on or
set-in nozzles with complete joint penetration welds (as was the case being analyzed) are
assigned to the slight (Group K1) group. For each micro-notch severity group there is a set of
curves that provide the micro-notch stress concentration factor. These local structural
discontinuity stress concentration factors are referred to in En 12952-3 as correction factors Ck1,
Ck2 and Ck3 for slight, moderate and severe micro-notch severities respectively.

For each of these groupings the correction factor is a function of the cycles to cracking and the
base metal room temperature tensile strength. For this analysis the correction factor Ck1 with a
room temperature tensile strength of 58 ksi (400 MPa) and up to 35,000 cycles to cracking was
used. The value of Ck1 for these conditions was 1.5.

D-6
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences
The resulting concentrated, peak, maximum and minimum linear elastic stresses at locations A
and B are:

σelastic,peak,A = 1.5 (75,330 psi) = 112,995 psi (779 MPa)

σelastic,peak,B = 1.5 (-56,033 psi ) = -84,050 psi (580 MPa)

Other Stress Concentration Factors

There is no need to further increase the peak elastic stress values for factors such as cracked mill
scale, surface roughness, weld toe angle or weld joint or weld quality attributes since the micro-
notch stress concentration factor given above already includes these factors. If this analysis was
being performed for the header borehole then a surface roughness or mill scale micro-notch
stress concentration factor would be used instead of the weld grouping based correction factors.

Equivalent, Multiaxial Elastic Stress (Stress Intensity) Values

During the complete cold start, full load and shutdown operating cycle the outside surface of the
tube to header connection is cycled between a nominal stress value of 0 to the nominal tensile
and compressive stress values documented above. The peak stress intensity on the tension side of
the connection when during the quenching event is:

SA = peak maximum principal stress – peak minimum principal stress


= 112,995 psi - 0 psi = 112,995 psi (779 MPa)

Note that in this case the minimum principal stress was the radial pressure + thermal stress which
is zero throughout the loading event.

On the compression side of the connection the maximum principal stress during the quench event
is the internal pressure hoop stress and the minimum principal stress is the combined axial and
bending stresses. The stress intensity during the quenching event is:

SB = 1216 psi - -84,050 psi = 85,875 psi (592 MPa)

The first term in this equation is the internal pressure hoop stress value increased by the gross
and local structural stress concentration factors. It could perhaps be justified to not increase the
nominal mean diameter hoop stress value by these gross and local structural stress concentration
factors, but in this case this change would have an insignificant effect on the estimate fatigue life.

Estimating the Fatigue Life Using the Markl Equation

It is possible to estimate the fatigue life directly using the structural stress values already
calculated. Recall from Appendix A that the Markl fatigue curve was benchmarked with tests of
pipe to pipe connections with butt welds. For a Group K1 weld detail no additional correction is

D-7
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences
needed to account for micro-notch (local structural) stress concentration. Using the structural
stress range of 75,330 psi and the Markl relationship:

Nf = (490,000 psi /elastic structural stress range)5

Produces an estimated average life Nf of:

Nf = (490,000 psi /75,330 psi)5 = 11,645 cycles

The minimum cycles to failure can be estimated by increasing the stress range by a factor of 2
which produces an estimated minimum life of 364 cycles. Using the Markl equation the
compression side of the connection has an average and minimum life of 51,139 cycles and 1,598
cycles. Somewhat fewer cycles to failure would be estimated if a mean stress corrected structural
stress range were used with the Markl equation. These life estimates are well within an order of
magnitude of the estimated cycles to failure using the inelastic, strain based life estimations
shown below.

Estimating the Elastic Plastic Stress and Total Strain Values During the
First Loading Cycle

It is useful and informative to estimate the stress strain history at the toe of the tube to header
weld at locations A and B during the loading (thermal quenching) and unloading (returning back
to off-line conditions). This inelastic behavior can be estimated using detailed finite element
modeling with incremental solutions to keep the estimated stresses and strains on the cyclic
stress strain curve of the material or can be estimated using a simple stress-strain conservation
rule (i.e., the Neuber rule). The peak linear elastic stress times the peak linear elastic strain
equals the elastic plastic stress times the total elastic plastic strain:

[S(SCF)]2/E = (σelastic-plastic)(εta) Equation D-5

Equation D-5 can be solved for the elastic-plastic and total strain values by either substituting
Equation D-2 into the equation for the total strain amplitude term and iteratively searching for
the elastic plastic stress value and then using equation D-2 to solve for the total strain amplitude.
Alternatively the strain amplitude can be estimated directly by assuming elastic perfectly plastic
conditions and setting the elastic plastic stress value to the cyclic yield stress, YS’. In this case
the total strain amplitude can be determined as follows:

εta= [S(SCF)]2/E/(YS’) Equation D-6

For locations A and B the elastic strain energy values are:

[S(SCF)]2/E = (σelastic,peak,A)2/E Equation D-7

At location A: (112,995 psi)2/29.1E6 psi = 439 in-lb/in3 (439 psi, 3.03 MPa)
At location B: (79,926 psi)2/29.1E6 psi = 253 in-lb/in3 (253 psi, 1.75 MPa)

D-8
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences
Using an estimated cyclic yield stress of 35,000 psi leads to the following elastic perfectly plastic
estimates of total strain amplitude:

For location A: εta= [S(SCF)]2/E/(YS’) = 439 in-lb/in3/55,000 psi = 0.0125 (1.25%)


For location B: εta= 220 in-lb/in3/55,000 psi = -0.0072 (-0.72%)

Iteratively solving the Ramberg-Osgood relationship for the cyclic stress strain curve produces
the following estimated elastic plastic stress and total strain amplitude values:

[S(SCF)]2/E = (εta)( σa) = σa2/E + σa (σa/H’)1/n’ Equation D-8

εta = σa/E + (σa/H’)1/n’ Equation D-9

For location A: σelastic-plastic = 37,019 psi, εta= 0.0118


For location B: σelastic-plastic = -34,176 psi, εta= -0.0074

Estimating the Elastic Plastic Stress and Total Strain Values after
Unloading from the Peak Stress Loading Cycle

After removal of the quench load and return to the off line conditions there may or may not be
local yielding at locations A and B.

The change in stress, Δσ and total strain, εr and final residual stress, σr and strain, εr values can
be estimated using the Neuber rule:

[ΔS(SCF)]2/E = (Δσ)( Δε) Equation D-10

or

Δε = [ΔS(SCF)]2/ [(Δσ)(E)] Equation D-11

For elastic perfectly plastic conditions:

Δσ = ΔS(SCF) when YS’ ≤ ΔS(SCF) ≤ 2YS’ Equation D-12

or

Δσ = 2 YS’ when ΔS(SCF) ≥ 2YS’ Equation D-13

σr = σa - Δσ Equation D-14

εr = εat - Δε Equation D-15

For locations A and B the stress and strain changes and residual stress and strain values
associated with the unloading event based on elastic perfectly plastic conditions are:
D-9
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences
For Location A:
[S(SCF)]2/E = 439 in-lb/in3 (3.03 MPa)
εta= 0.0125 (1.25%)
σelastic,peak,A = 112,995 psi (779 MPa)

Since the stress change during the unloading event, σelastic,peak,A > 2 YS’:
Δσ = σelastic,peak,A = 2 x YS’ = 70,000 psi (483 MPa)
Δε = [ΔS(SCF)]2/ [(Δσ)(E)] = 385 in-lb/in3 /(70,000 psi) = 0.0063 (0.63%)
εr = εat - Δε = 0.0125 - 0.0063 = 0.0063 (0.63%)
σr = σa - vσ = 35,000 psi – 70,000 psi = -35,000 psi (-241 MPa)

For location B:
[S(SCF)]2/E = 253 in-lb/in3 (1.75 MPa)
εta= -0.0072 (-0.72%)
σelastic,peak,B = 85,875 psi (592 MPa)

Since the stress change during the unloading event, σelastic,peak,B > 2 YS’:
Δσ = 70,000 psi (483 MPa)
Δε = [ΔS(SCF)]2/ [(Δσ)(E)] = 253 in-lb/in3/ 70,000 psi = 0.0036 (0.367%)

Noting that Location B was initially loaded in compression:


εr = εat - Δε = -0.0072 + 0.0036 = -0.0036 (-0.36%)
σr = σa - Δσ = -35,000 psi + 70,000 psi = +35,000 psi (+241 MPa)

The Neuber rule can also be used in combination with the Ramberg-Osgood cyclic stress strain
curve relationship to estimate the stress and strain changes and residual stress and strain values
associated with the unloading event based on more precise cyclic strain hardening behavior. In
this case the combined Neuber, Ramberg-Osgood equation is recast in terms of strain amplitude
and stress amplitude changes:

Strain energy balance: [ΔS(SCF)]2/E = (Δε)(Δσ) Equation D-16

Cyclic stress-strain curve relationship: Δε/2 = (Δσ/2)/E + [(Δσ/2)/H’]1/n’Equation D-17


or

Δε = (Δσ)/E + 2[(Δσ/2)/H’]1/n’ Equation D-18


2 2 1/n’
[ΔS(SCF)] /E = (Δε)(Δσ) = (Δσ) /E + 2(Δσ)[(Δσ/2)/H’] Equation D-19

For location A:
E = 29.1E6 psi (200,690 MPa)
H’ = 77,000 psi (754 MPa)
n’ = 0.16
[S(SCF)]2/E = 439 in-lb/in3 (3.03 MPa)
σelastic-plastic = 37,160 psi, εta= 0.0118
(Δσ)2/29.1E6 psi + 2(Δσ)[(Δσ/2)/ 77,000 psi]1/0.16 = 439 in-lb/in3

D-10
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences
Solving Equation G-19 using an iterative process (e.g., using Solver in an Excel spreadsheet)
results in an estimated stress range value of 59,618 psi (411 MPa).
Δε = 59,618 psi/29.1E6 psi + 2(59,618 psi/2/70,000 psi)1/0.16 = 0.0074 (0.74%)
εr = εat - Δε = 0.0118 - 0.0074 = 0.0044 (0.44%)
σr = σa - Δσ = 37,160 psi –59,618 psi = -22,458 psi (-155 MPa)
Mean stress, σm = (σelastic-plastic + σr)/2 = (37,160 psi - 22,458 psi)/2 = 7,351 psi (51 MPa)

For location B:
σelastic-plastic = -34,176 psi, εta= -0.0074
[S(SCF)]2/E = 253 in-lb/in3 (1.75 MPa)
Stress range (using iterative solution) = 53,948 psi (372 MPa)
Δε = 53,948 psi/29.1E6 psi + 2(53,948 psi /2/70,000 psi)1/0.16 = 0.0047 (0.47%)
εr = εat - Δε = -0.0074 + 0.0047 = -0.0027 (-0.27%)
σr = σa - Δσ = -34,176 psi + 53,948 psi = 19,772 psi (136 MPa)
Mean stress, σm = (σelastic-plastic + σr)/2
= (-34,176 psi + 19,772 psi)/2 = -7,202 psi (-50 MPa)

The overall predicted stress strain histories during loading and unloading of Locations A and B
are shown in Figure D-5. The side of the tube to header connection (Location A) that is initially
loaded in tension has a compression residual stress and small tensile residual strain after the
thermal loads are removed. The opposite side of the tube to header connection (Location B) that
is initially loaded in compression has a tensile residual stress and small compressive residual
strain after the thermal loads are removed.

The reader should consider the consequences of this type of stress history if it were to occur on a
HP superheater or reheater tube operating within the creep regime of the material. In this case,
the residual stresses would continue to promote addition permanent strains and creep fatigue
damage by creep relaxation or creep cavitation during steady state operation after the thermal
quench event is no longer active.

Estimating the Cyclic Life with Mean Stress Corrections

The cyclic life can now be estimated using the Coffin-Manson relationship (Equation D-3):

Δε/2 = σf'/E (2Nf)b + εf' (2Nf)c

To account for mean stress the Coffin-Manson equation is modified as follows:

Δε/2 = σf' (1-σm/σf')/E (2Nf)b + εf' (1-σm/σf')c/b(2Nf)c Equation D-20

For the material and temperature conditions associated with the load cycle being assessed the
“Morrow modified” Coffin-Manson equation is:

Δε/2 = 93,000 psi (1-σm/93,000 psi )/29.1E6 psi (2Nf) )-0.109


+ 0.10(1-σm/93,000 psi ))-0.39/-0.109 (2Nf) )-0.39

D-11
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences
The cycles to failure, Nf can be readily determined using a spreadsheet and an iterative goal seek
function (e.g., EXCEL Solver).

For Location A:
Strain range, Δε = 0.0074 (0.74%)
Mean stress, σm = 7,351 psi (51 MPa)
Median cycle to failures, Nf = 2,905 cycles
Minimum cycles to failure, Nf/10 = 291 cycles

For location B:
Strain range, Δε = 0.0047 (0.47%)
Mean stress, σm = -7,202 psi (-50 MPa)
Median cycle to failures, Nf = 58,027 cycles
Minimum cycles to failure, Nf/10 = 5,803 cycles

These calculations are summarized in Tables D-1 and D-2 for metric and English units
respectively. This example illustrates the steps used for tube to header fatigue assessments made
using inelastic strain range and mean stress values.

This example also illustrates the need to consider “upsets” in the design-basis assumption. The
same analysis performed using the design-basis temperatures without the selectively quenched
tubes produced 1,330 psi (9.2 MPa) axial and 9,537 psi (65.8 MPa) bending nominal stress
values and an estimated average and minimum cycles to failure for the tension side of the
connection of 480,000 cycles and 48,000 cycles respectively.

The Row 3 tube quenching event is anticipated to reduce the life to of a few hundred thermal
quench cycles. Since these quench events often occur many times during each startup it is
anticipated that cracking of these tubes may occur in even fewer shutdown/startup cycles. If not
corrected, these quench events will undoubtedly produce cracking on the tensile side of the tube
to header connections in far less than the anticipated lifetime of the HRSG.

D-12
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences

Table D-1
Cyclic Stress-Strain and Fatigue Evaluation

Input
o
1 Maximum cycle temperature: T^ 80.0 C
o
2 Minimum cycle temperature: T_ 21.1 C
o
3 Effective cycle temperature: T* 65 C T* = 0.75(T^) +0.25 (T_)
4 Operating pressure P 0.60 MPa
5 Nominal tube to header axial stress: Saxial 21 MPa
6 Nominal tube to header bending Sbending 153 MPa
stress:
7 Axial pressure stress: Spressure, 1.1 MPa
axial
8 Hoop pressure stress: Spressure, 2.8 MPa
hoop
9 Tube OD: do 38.100 mm
10 Tube nom. Thickness: tn 3.658 mm
11 Tube mean diameter: dm 34.442 mm
12 Header OD: Do 219.1 mm
13 Header nom. Thickness: Tn 15.09 mm
14 Header mean diameter: Dm 204.0 mm
Material Properties at T*
15 Tube material: SA192
16 Room temperature tensile strength: UTS 345 MPa
17 Modulus: E 200,690 MPa
18 Cyclic yield stress: YS' 241 MPa
19 Ramberg-Osgood strength coefficient: H' 531 MPa
20 Ramberg-Osgood strain hardening n' 0.16
exponent:
21 Fatigue strength coefficient: σf' 641 MPa
22 Fatigue strength exponent: b -0.109
23 Fatigue ductility coefficient: εf' 0.1
24 Fatigue ductility exponent: c -0.39
1st load reversal
Tension Compression
side side
25 Max. combined nominal stress: 175 2.8 MPa
26 Min. combined nominal stress: 0 -130 MPa
27 SIF: 2.97 2.97 Gross structural SIF from WRC
Bulletin 297
28 SCF: 1.50 1.50 Local structural notch SCF
from EN12952-3 B-4, Group
K1, Rm = 400 MPa
29 σelastic, max principal: 520 8.4 MPa Elastic structural stress = Nom.
stress x SCF(gross structural)
30 σelastic, min. principal: - (386)
31 σelastic, peak, max. principal: 779 12.6 MPa Peak local stress = structural
stress x SCF (micro-notch)
32 σelastic, peak, min. principal: - (580)
33 Smax: 779 592 Max peak Tresca stress
(stress intensity) during entire
cycle

D-13
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences
34 Smin: - - Minimum peak stress intensity
during entire cycle
35 ΔS: 779 592
36 Peak elastic stress intensity x strain: 3.03 1.75 N- [S(SCF)]2/E = (σelastic,peak,A)2/E
mm/
mm3
37 Elastic perfectly plastic stress 241 -241 MPa
amplitude at end of 1st load:
38 Elastic perfectly plastic strain amplitude 0.0125 (0.0072) εta= [S(SCF)]2/E/(YS’)
at end of 1st load:
39 Elastic plastic stress x strain: 3.03 1.75 N- [S(SCF)]2/E = (εta)(σa) = σa2/E
mm/ + σa (σa/H’)1/n’
mm3
40 Ramberg Osgood stress amplitude at 256 236 MPa Compression side stress in
end of 1st load: negative
41 Ramberg Osgood strain amplitude at 0.0118 0.0074 εta = σa/E + (σa/H’)1/n’
end of 1st load:
After unloading (2nd reversal)
42 2 x YS': 483 483 MPa
43 Peak elastic stress intensity change 779 592 MPa
during unloading:
44 Peak elastic stress intensity change x 3.03 1.75 N- [dS(SCF)]2/E
strain change during unloading: mm/
mm3
45 Δσelastic-perfectly plastic: 482.8 482.8 MPa Δσ = peak elastic stress
change during unloading if < 2
YS’ or = 2YS'
46 Δεelastic-perfectly plastic: 0.0063 0.0036 Δε = [ΔS(SCF)]2/[(2)(Y')(E)]
47 σr, elastic-perfectly plastic: -241 241 MPa σr = σa - Δσ
48 yr, elastic-perfectly plastic: 0.0063 (0.0036) εr = εat - Δε
49 Elastic plastic stress x strain change: 3.03 1.75 N- Elastic plastic stress x strain
mm/ change = (Δσ)2/E +
mm3 2(Δσ)[(Δσ/2)/ H']1/n'
50 ΔσRamberg-Osgood: 411 372 MPa
51 ΔεRamberg-Osgood: 0.0074 0.0047 Δε = (Δσ)/E + 2[(Δσ/2)/H’]1/n’
52 σr, Ramberg-Osgood: (155) 136 MPa σr = σa - Δσ
53 εr, Ramberg-Osgood: 0.0045 -0.0027 εr = εat - Δε
54 σm, Ramberg-Osgood: 51 (50) Mean stress, σm = (σelastic-plastic
+ σr)/2
Morrow modified Coffin Manson
Fatigue Life
55 10,000 x ΔεRamberg-Osgood: 73.68 47.02 Increase strain range by
10,000X for goal seeker
56 10,000 x ΔεRamberg-Osgood: 73.68 47.02 ε x 10,000 =(10,000)(2){σf' (1-
σm/σf')/E (2Nf)b + εf' (1-
σm/σf')c/b(2Nf)c}
57 Nf: 2,893 57,797 cycles Median cycles to failure
58 Nf/10: 289 5,780 cycles Minimum cycles to failure

D-14
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences

Table D-2
Cyclic Stress-Strain and Fatigue Evaluation

Input
o
1 Maximum cycle temperature: T^ 176 F
o
2 Minimum cycle temperature: T_ 70 F
o
3 Effective cycle temperature: T* 150 F T*= 0.75(T^) +0.25 (T_)
4 Operating pressure P 87.00 psi
Nominal tube to header axial
5 Saxial 3,085 psi
stress:
Nominal tube to header bending
6 Sbending 22,115 psi
stress:
Spressure,
7 Axial pressure stress: 164 psi
axial
Spressure,
8 Axial pressure stress: 410 psi
hoop
9 Tube OD: do 1.500 inch
10 Tube nom. Thickness: tn 0.144 inch
11 Tube mean diameter: dm 1.356 inch
12 Header OD: Do 8.625 inch
13 Header nom. Thickness: Tn 0.594 inch
14 Header mean diameter: Dm 8.031 inch
Material Properties at T*
15 Tube material: SA192
Room temperature tensile
16 Rm 50000 psi
strength:
29,100,0
17 Modulus: E psi
00
18 Cyclic yield stress: YS' 35,000 psi
Ramberg-Osgood strength
19 H' 77,000 psi
coefficient:
Ramberg-Osgood strain
20 n' 0.16
hardening exponent:
21 Fatigue strength coefficient: σf' 93,000 psi
22 Fatigue strength exponent: b -0.109
23 Fatigue ductility coefficient: εf' 0.1
24 Fatigue ductility exponent: c -0.39
1st load reversal
Tension Compression
side side
25 Max. combined nominal stress: 25,364 410 psi
26 Min. combined nominal stress: 0 -18,866 psi
Gross structural SIF from WRC Bulletin
27 SIF: 2.97 2.97
297
Local structural notch SCF from
28 SCF: 1.50 1.50
EN12952-3 B-4, Group K1, Rm = 58 ksi
Elastic structural stress = Nom. stress x
29 σelastic, max principal: 75,330 1,217 psi
SCF(gross structural)
30 σelastic, min. principal: - (56,033) psi
Peak local stress = structural stress x
31 σelastic, peak, max. principal: 112,995 1,825 psi
SCF (micro-notch)

D-15
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences
32 σelastic, peak, min. principal: - (84,050) psi
Max peak Tresca stress (stress
33 Smax: 112,995 85,875 psi
intensity) during entire cycle
Minimum peak stress intensity during
34 Smin: - - psi
entire cycle
35 ΔS: 112,995 85,875 psi
Peak elastic stress intensity x in-
36 438.76 253.42 [S(SCF)]2/E = (σelastic,peak,A)2/E
strain: lb/in3
Elastic perfectly plastic stress
37 35,000 35,000 psi
amplitude at end of 1st load:
Elastic perfectly plastic strain
38 0.0125 (0.0072) εta= [S(SCF)]2/E/(YS’)
amplitude at end of 1st load:
in- [S(SCF)]2/E = (εta)( σa) = σa2/E + σa
39 Elastic plastic stress x strain: 438.76 253.42
lb/in3 (σa/H’)1/n’
Ramberg Osgood stress
40 37,160 34,176 psi Compression side stress in negative
amplitude at end of 1st load:
Ramberg Osgood strain
41 0.0118 0.0074 εta = σa/E + (σa/H’)1/n’
amplitude at end of 1st load:
After unloading (2nd reversal)
42 2 x YS': 70,000 70,000 psi
Peak elastic stress intensity
43 112,995 85,875 psi
change during unloading:
Peak elastic stress intensity
in-
44 change x strain change during 438.76 253.42 [dS(SCF)]2/E
lb/in3
unloading:
Δσ = peak elastic stress change during
45 Δσelastic-perfectly plastic: 70000 70000 psi
unloading if < 2 YS’ or = 2YS'
46 Δεelastic-perfectly plastic: 0.0063 0.0036 Δε = [ΔS(SCF)]2/ [(2)(Y')(E)]
47 σr, elastic-perfectly plastic: -35,000 105,000 psi σr = σa - Δσ
48 εr, elastic-perfectly plastic: 0.0063 (0.0036) εr = εat - Δε
Elastic plastic stress x strain in- Elastic plastic stress x strain change =
49 438.76 253.42
change: lb/in3 (Δσ)2/E + 2(Δσ)[(Δσ/2)/ H']1/n'
50 ΔσRamberg-Osgood: 59,618 53,948 psi
51 ΔεRamberg-Osgood: 0.0074 0.0047 Δε = (Δσ)/E + 2[(ϖσ/2)/H’]1/n’
52 σr, Ramberg-Osgood: (22,458) 19,772 psi σr = σa - Δσ
53 εr, Ramberg-Osgood: 0.0044 -0.0027 εr = εat - Δε
54 σm, Ramberg-Osgood: 7,351 (7,202) Mean stress, σm = (σelastic-plastic + yr)/2
Morrow modified Coffin
Manson Fatigue Life
Increase strain range by 10,000X for
55 10,000 x ΔεRamberg-Osgood: 73.59 46.97
goal seeker
Δε x 10,000 =(10,000)(2){σf' (1-σm/σf')/E
56 10,000 x ΔεRamberg-Osgood: 73.59 46.97
(2Nf)b + εf' (1-σm/σf')c/b(2Nf)c}
57 Nf: 2,905 58,027 cycles Median cycles to failure
58 Nf/10: 291 5,803 cycles Minimum cycles to failure

D-16
EPRI Proprietary Licensed Material

Example Fatigue Analysis of aN LP Economizer (Preheater) with Tube-to-Tube and Row-to-Row Temperature
Differences

Figure D-5
Schematic of the stress-strain history for Locations A and B during thermal quench
loading and subsequent shutdown unloading events.

D-17
EPRI Proprietary Licensed Material

E
ESTIMATING THE BOREHOLE THERMAL SHOCK
(LOADING MODE 1) LIMITS FOR RISER/DOWNCOMER
PIPE NOZZLE TO DRUM, NOZZLE PIPE TO HEADER
OR TUBE TO HEADER BOREHOLES

One of the primary startup and shutdown rate limiting conditions is the local thermal stress that
occur when the temperature of the inside surface of a header or drum boreholes are transiently
different than the header or drum midwall temperature. During design basis startups the borehole
surface temperatures are considered to be heated above the midwall temperatures, leading to
compressive surface stresses. During shutdowns the borehole surface is cooled more rapidly than
the midwall temperature resulting in borehole surface tensile stress.

Undesirable, off design basis startup conditions such as forward flow of condensate or
introduction of unflashed attemperator water in superheaters and reheater or introduction of cold
feedwater into economizers will quench cool the borehole surfaces during startups and produce
transient tensile rather than compressive surfaces stresses. After this transient quench cooling
condition subsides the borehole heating will continue shifting the surface stresses back to
compressive values.

During shutdowns or pre-start purge events prior to hot restarts the gas temperatures will drop
sufficiently to produce condensation of steam in the superheaters and reheaters. This condensate
will quench cool the boreholes, especially on lower headers and manifolds. This transient
condensate quench cooling condition will exacerbate the magnitude of borehole tensile stresses.

There is also a locally increased pressure stress associated with boreholes which will be additive
to tensile (cooling, quench cooling) thermal stresses or in opposite sign to borehole compressive
(heating) thermal stresses. The peak combined thermal plus pressure borehole tensile surface
stress value will occur when the pressure is high and the borehole is being cooled or transiently
quench cooled. This will be associated with the beginning of the cooldown for design basis
condition or during off-design quench cooling events that may occur either during startups or
shutdowns. The largest compressive borehole surface stress value will occur when the pressure
is low and the borehole is being rapidly heated (e.g., at the beginning of a cold startup).

The fatigue damage and potential for borehole cracking will be dependent on the cyclic stress
range and the mean stress. The strain rate will influence the rate of damage in water touched
boreholes if corrosion fatigue is active. For creep fatigue damage the magnitude of stress and
temperature during steady state operating periods will have a significant impact on the damage
accumulation rate. If the borehole surface has experience compressive yielding during rapid

E-1
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum,
Nozzle Pipe to Header or Tube to Header Boreholes
heating of the inside surface then the tensile pressure stress and tensile residual thermal stress
will both be active during steady load periods of operation and accelerated creep damage will
occur as the residual thermal stress attempts to relax away.

When severe quenching of the borehole occurs the borehole surface will locally yield in tension
and after the quenching event is no longer active a residual compressive surface stress will be
present. During steady state elevated temperature operation an oxide scale will grow on the
surface which has lower stress than the normal pressure only steady state stress. During
subsequent shutdowns and cooling of the borehole the tensile surface strains produced will crack
the surface oxide and accelerate crack initiation and growth of shallow cracks [E-1, E-2]. This is
one of the creep fatigue damage mechanisms that is also referred to as oxidation enhanced
fatigue or environmentally assisted fatigue. Oxide cracking on borehole surfaces will reduce the
fatigue life well below the life calculated with the temperature corrected fatigue curves included
in most commonly used HRSG pressure part design codes.

The most commonly used design codes for assessing the permissible rate of pressure/fluid
temperature change or the permissible borehole surface to header/manifold/drum midwall
temperature are TRD 301, Annex 1 or EN12952-3, Section 5.5, Chapter 13, and Annex B and C.
PD 5500, Annex G is also useful for these assessments and has a somewhat more complex but
less conservative thermal stress borehole gross structural stress concentration solution than
equations included in TRD 301 Annex 1 or EN 1252-3. The PD 5500 borehole thermal stress
solution also allows the heat transfer coefficients in the borehole to be higher than on the header
inside surface which is more realistic for the conditions that are associated with severe borehole
thermal stresses. These three codes have very similar approaches and design rules with minor
differences in the stress solutions and fatigue curves included.

Two borehole fatigue assessment case studies are included in this appendix. Table E-1 provides
an example of a borehole fatigue analysis for a carbon steel riser pipe/downcomer pipe nozzle to
drum borehole performed using the TRD 301, Annex 1 rules with minor modifications. Table E-
2 provides and example of a fatigue assessment of the tube boreholes in a 9Cr-1Mo-V-Cb (Grade
91) HP superheater outlet header performed in accordance with the EN 12952-3 rules.

In this case the header design met the EN12952-3 requirements and produced an estimated
design basis fatigue life usage factor of 0.33 (33%) but failed to meet the EPRI recommendation
that the fatigue usage factor for components that operate at temperatures above 427oC (800oF) be
limited to 0.10 (10%) to account for creep fatigue damage (which is not included in the current
design rules).

Case Study 1 – HP Steam Drum Borehole Fatigue Assessment

Table E-1 provides an example of a HP drum borehole fatigue assessment performed with a
slightly modified version of TRD 301 Annex 1 rules. The shaded regions show the required input
values. In rows 1 through 10 of Table E-1 the key dimensions, design details and fabrication
limits of the nozzle/drum connection are shown. The assumed numbers of operating cycles per
year are shown in row 13. The minimum and maximum values of drum pressure for each type of
operating cycle are documented in rows 16 through 18. The minimum and maximum values of

E-2
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum,
Nozzle Pipe to Header or Tube to Header Boreholes
drum water temperature for each type of operating cycle are shown in rows 19 and 20. The
startup and shutdown drum water temperature ramp rates are shown in rows 23 and 25
respectively. The remaining rows in the table were derived from these input values.

Rows 26 through 35 of Table E-1 list the room temperature or elevated temperature material
properties. The elevated temperature properties are based on an effective temperature for the
cyclic loading event given by:

T* = 0.75 T^ + 0.25 T_ Equation E-1

where:
T* = Effective temperature
T^ = Maximum temperature during operating cycle
T_ = Minimum temperature during operating cycle

Rows 36 through 38 in Table E-1 are estimated gross and local structural stress concentration
factors for header and borehole dimensions, surface finish, the connection and weld design and
the characteristic of the weld root. Rows 39 through 62 are the estimated peak, thermal and
pressure stresses, stress ranges and mean stress or elastic/plastic stress corrected stress ranges at
the beginning and end of each thermal/pressure operating cycle. The corrected stress ranges are
used to estimate the cycles to cracking for each operating transient (Row 63). The lifetime
fatigue life consumption for each type of operating cycle is presented in Row 65 and summed up
for all types of operating transients in Row 65. If this latter value is less than 0.5 then the fatigue
life is determined to be acceptable.

Rows 67 through 70 of Table E-1 are estimates of the stresses in the waterside oxide. If the
maximum estimated compressive stress in the oxide (based on the assumption that the oxide
grows stress free at the stress level associated with normal operating pressure with no thermal or
residual stress) has a magnitude less than 600 MPa and the tensile stress has a magnitude less
than 200 MPa the oxide is not predicted to crack and the decreased fatigue life resulting from
oxide cracking and corrosion-fatigue damage will be avoided. If the oxide scale stress was
outside these limits then the design or operating limits would need to be altered.

Case Study 2 – HP Superheater Outlet Header Tube Borehole Fatigue


Assessment

Table E-2 provides an example of a HP superheater outlet header tube borehole fatigue
assessment performed in accordance with the EN 12952-3 rules. In rows 1 through 10 of Table
E-2 the key dimensions, design details and fabrication limits of the header and stub tubing are
input.

The assumed numbers of operating cycles per year are shown in row 11. In this case it was
assumed that either a shutdown and cold start or a shutdown and warm start would occur once a
week throughout the year and 180 nightly shutdowns followed by hot starts would occur each
year and approximately two load change events would occur each day throughout the year.

E-3
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum,
Nozzle Pipe to Header or Tube to Header Boreholes
The minimum and maximum values of steam pressure for each type of operating cycle are
documented in rows 14 through 17. The minimum and maximum values of drum water
temperature for each type of operating cycle are shown in rows 18 through 20. The startup and
shutdown drum water temperature ramp rates are shown in rows 23 and 25 respectively. The
remaining rows in the table were derived from these input values.

Rows 28 through 38 of Table E-2 list the room temperature or elevated temperature material
properties. The elevated temperature properties are based on an effective temperature for the
cyclic loading event given by equation E-1 above.

Rows 39 through 48 in Table E-2 document the estimated gross structural stress concentration
factors and the temperature gradient and thermal stress factors. The normal, minimum and
maximum pressure stresses including the gross structural borehole stress concentration factor are
presented in Rows 49 through 51. The inside surface to midwall temperature gradients and
associated borehole thermal stress values adjusted with the gross structural stress concentration
factor for borehole thermal shock stress are documented in Rows 52 through 55.

The minimum and maximum combined thermal and pressure stress values are documented in
Rows 56 and 59 respectively. The thermal and pressure stresses given at this juncture are Tresca
multiaxial stress values (i.e., stress intensities). For example, the pressure stress calculated here
was the mean diameter hoop stress of the header amplified by a gross structural stress
concentration factor of 3.15 minus the radial pressure stress (which has a value of minus the
gauge pressure).

Rows 57 and 60 list the normal operating pressure stress plus 200 MPa and minus 600 MPa
respectively. These oxide cracking stress limits are compared with the minimum or maximum
combined thermal and pressure stress values. This oxide cracking check is only required to be
determined by EN 12952-3 for water touched components, however for elevated temperature
components such as this HP superheater it is useful to determine if oxide cracking is likely to be
active in the boreholes. Note that the structural stress values (adjusted with the gross structural
stress concentration factors) rather than the peak stress values (adjusted with both the gross and
local structural stress concentration factors) are used to make the oxide cracking check.

The mean value and range of the combined thermal and pressure stress intensity values are
documented in Rows 62 and 63 respectively. These mean stress intensity and stress intensity
range values are next corrected with local structural (micro-notch) stress concentration factors to
account for borehole surface roughness or mill scale. The micro-notch corrected (i.e., peak)
mean stress intensity and stress intensity range values are presented in Rows 69 and 70
respectively.

The peak stress range is next corrected for mean stress and/or plasticity factors in Rows 72 to 75.
The specific correction used is a function of the peak stress intensity range divided by the yield
strength at the effective cycle temperature. When this ratio is less than one the only correction
made is a Gerber mean stress correction. When the ratio is between one and two the mean stress
is assumed to be the cycle temperature yield strength minus half the peak stress intensity range
and this is used in the Gerber mean stress correction relationship. When the ratio is greater than

E-4
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum,
Nozzle Pipe to Header or Tube to Header Boreholes
two an elastic plastic correction is made using the Neuber rule and elastic perfectly plastic
behavior with the cycle temperature yield strength assumed to be the cyclic yield stress. This
elastic plastic strain range estimate is then converted back to a virtual elastic stress value by
multiplying the estimated total strain range by the cycle temperature modulus of elasticity value.
The conversions to strain range and back to virtual stress are combined to give the equation
shown in Row 74. For the case study shown the peak stress intensity range to cycle temperature
yield strength ratio was between 1 and 1.5 (see Row 71) so the mean stress modified Gerber
relationship was used to determine the final corrected stress intensity range values shown in Row
75.

The plasticity/mean stress corrected stress intensity range value is next corrected for service
temperature using the temperature correction factors in Row 76. The fatigue life is then
estimated using this temperature corrected stress intensity range two ways. First the life is
estimated with the temperature corrected stress intensity range increased by 1.5. The resulting
life is given in Row 80. Next the average life is determined without the stress adjustment factor
and then reduced by a factor of 10 on cycles. The mean and minimum life estimates for this
approach are given in Rows 81 and 83. The estimated cycles to cracking for each type of
shutdown/startup transient is the lower of the lives estimated with a factor of 1.5 on stress or 10
on cycles.

The usage factor (i.e., fatigue life fraction) for each type of operating transient is determined by
the ratio of the total anticipated number of transients divided by the estimated fatigue life. The
individual usage factors for each type of transient are shown in Row 84. The total combined
usage factor for all the transients is documented in Row 86. Since the design basis operating
history was specified EN 12952-3 permits the life usage factor to be up to 1.00 (100%). The
estimated life usage factor was 0.33 (33%) which is satisfactory to the code requirements.

On the other hand, since this fatigue life estimation does not include any consideration of creep
fatigue damage and the header is anticipated to operate at temperatures near 567oC (1053oF) it is
recommended the life usage factor be limited to 0.10 (10%). To meet this requirement would
require modification of the design (using a smaller, thinner header) or the operating limits or
would require a more detailed assessment of the pressure and thermal stress magnitudes using
other analytical methods (e.g., PD 5500, Annex G) or finite element modeling.

Consideration of Operating Limits for Borehole Cracking Prevention

During the review of these TRD 301, Annex 1/ EN 12952-3, Section 13 calculations, careful
consideration should be given to the fluid temperature ramp rate values assumed for both the
startup and shutdown portions of each cycle. For example, for the analysis shown in Table E-1
the assumed shutdown ramp rate of -0.50oC/min (-0.9oF/min) might be questioned. A review of
similar units indicates that HP drum water temperature decay rates of -0.19oC/min (-0.33oF/min)
or less can be achieved without using controlled, fired combustion turbine exhaust controlled
shutdowns for relatively new, large, triple pressure units. The design-basis shutdown ramp rate
of -0.50oC/min (-0.9oF/min) is probably reasonable and will allow for some loss in pressure
tightness of the HP system as the unit ages. The startup temperature ramp rate of 3oC/min
(5.4oF/min) will allow the HP drum temperature to be raised from 5oC to 333oC in approximately

E-5
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum,
Nozzle Pipe to Header or Tube to Header Boreholes
110 minutes (~ 2 hrs) which is within the startup times being achieved for large, triple pressure
HRSGs.

For most HRSGs there are critical HP saturation temperature milestones that must be achieved
and controlled during startups. For instance, in order to start heating the cold steam turbine
which is often is the item that controls the overall duration of the startup during cold starts once
acceptable steam chemistry is achieved. There is thus some urgency to raise the pressure to the
minimum acceptable HP pressure to roll off the steam turbine (often 3.4 MPa to 4.1 MPa (500
psig to 600 psig) but sometimes specified as higher by the steam turbine designer) in less than an
hour in order to start heating the steam turbine during the initial hold. In this case the saturation
temperature would be increased to 243oC to 254oC (470oF to 490oF) in less than an hour and
saturation temperature ramp rates closer to 4.4oC (8oF/min) would be expected.

Another critical pressure milestone that often must be met during startups within a specified time
interval is about 6.2 MPa to 6.9 MPa (900 psig to 1000 psig) (Tsat = 279oC to 286oC (534oF to
547oF)) because the SCR catalyst cannot be put into service to meet emissions limits until the gas
temperature downstream of the HP evaporator exceeds about 288oC (550oF). If a somewhat
severe drum water temperature ramp rates in the range of 5oC/min to 6oC/min (9oF.min to
11oF/min) was anticipated then the design would need to be tolerant of these more severe
operating conditions.

Cold starts from ambient temperatures are potentially particularly damaging because
condensation heating develops a nonlinear, high temperature gradient at the inner wall of the HP
drum and thicker HP superheater and reheater headers and will produce top to bottom
temperature differences and humping of the HP drum due to the differences in heat transfer
between water and saturated steam with condensation heating.

The two primary ways that an operator or control system has available to control the initial
pressure increase during startups are: (1) minimizing the combustion turbine heat input by
holding the combustion turbine at minimum load (say 5% of baseload generator output, which
will typically produce a combustion turbine exhaust heat discharge in the range of approximately
35% to 45% of the baseload value) and/or adjusting the diverter damper opening or inlet gas
vane settings of the combustion turbine and (2) controlling the flow through the HP startup vent
and HP bypass inlet pressure control valve. Most units have the HP bypass inlet pressure control
valves sized to pass the maximum steam mass flow corresponding to baseload steam flow rates
when the HP pressure is at the maximum design pressure. During startups, at much lower
pressures, significantly larger bypass control valve areas are required to provide a means of
controlling the rate of pressure rising.

It is interesting to consider whether the use of slow, controlled shutdowns could be used to
justify an increase in the subsequent startup rates. This would certainly be possible if spin-cooled
shutdowns are being used when they are not justified, but for the most part the shutdowns should
achieve two objectives: (1) raising the pressure during the HP pressure during the first part of the
deloading so that the HP pressure after combustion turbine firing ceases is as high as possible
and (2) steam cooling the HPSH headers at a predetermined, controlled rate by controlling the
rate of combustion turbine exhaust gas temperature decay. After combustion turbine shutdown

E-6
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum,
Nozzle Pipe to Header or Tube to Header Boreholes
the cooling will occur by natural convection and leakage at isolation points and will not be
readily controlled beyond the natural cooling and pressure decay that is inherent in the design.

A similar philosophy should be considered when evaluating the shutdown and startups rates and
estimated fatigue damage for the different types of operating cycles. For example, if the startup
and shutdown procedures for hot and warm starts were designed to be more benign than the
conditions initially assumed in the design basis then this would create an extra fatigue margin
that could be used to justify faster ramp rates during cold starts. This “balancing of the fatigue
damage” among the different operating cycles and between the shutdown and startup portions of
each operating cycle is a key and useful feature of the fatigue design rules discussed above.

Limitations of Current Design Code for Fatigue Assessment of Drum or


Header Boreholes

The primary limitations of these simplified borehole cyclic pressure/thermal shock fatigue design
rules are:
1. There are numerous typographical errors in EN 12952-3:2001(E). When initially performing
calculations in accordance to this code it is important that each step in the process is carefully
checked. The primary errors can be readily found by duplicating the example calculations in
Annex C (which also has a few typographical errors).
2. The thermal stress relationships included do not allow the beneficial effect of hold times to
be readily determined
3. The thermal stress relationships do not account for the detrimental effect of top-to-bottom
header/drum/manifold temperature differences that occur at the same time as the local
borehole thermal stresses
4. The borehole thermal stress relationships are based on simplified and conservative
assumptions with regard to the temperature gradient across the wall thickness. This
conservatism can be overcome to some extent by using the temperature gradient and stress
solutions in PD 5500, Annex G or by estimating the temperature gradients and stresses with a
finite element model. When this approach is taken then it is important to carefully review
each of the micro-notch (local structural) stress correction factors in EN12952-3 (e.g., weld
root finish, surface finish, weld root gap, out-of roundness, ) and apply them to the finite
element calculated minimum and maximum stress values.
5. The fatigue curves used in EN12952-3 were based on smooth bar, fully reverse loaded (zero
mean stress) laboratory fatigue test data performed in air with no hold times. The median or
average fatigue life values are required to be adjusted with a factor of 1.5 on the stress range
or a factor of 10 on cycles to cracking (i.e., whichever factor produces the lowest life).
6. The EN12952-3, PD 5500 and TRD 301, Annex 1 fatigue curve temperature adjustment
curves were based on rapidly loaded laboratory fatigue test data with no hold times. These
tests and temperature correction factors do not adequately address creep-fatigue performance.
For evaluations of HP superheater or reheater headers/manifold boreholes long hold times at
elevated temperature will increase the amount of damage per operating cycle to levels that
are significantly greater than predicted by the EN12952-3 temperature adjusted fatigue

E-7
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum,
Nozzle Pipe to Header or Tube to Header Boreholes
curves. It is recommended that the fatigue usage factor for components that operate above
425oC (800oF) not exceed values of 0.10.
7. For water touched components such as LP economizer (preheater) inlet header it is critical
that oxide cracking strain limits in EN12952-3 be met. If the oxide strain limits are exceeded
then corrosion fatigue cracking is likely and will occur far more rapidly than will be
predicted with the lower bound EN12952-3, PD 5500 and TRD 301, Annex 1 fatigue curves.
If the oxide strain limits are predicted to be exceeded then it is recommended that the fatigue
usage factor not exceed values of 0.10.

For certain designs these limitations will not be a concern and the simple design rules will be
sufficient to demonstrate that the design and anticipated operating conditions are acceptable. In
other cases more detailed, complex analysis will be required to demonstrate that the design and
operating conditions are acceptable. To optimize startup and shutdown procedures, it is likely
that analytical methods that allow the benefit of hold time to be included in the assessment will
be needed.

References
[E-1] Challenger, K.D., A.K. Miller, and C.R. Brinkman, “An Explanation for the Effects of
Hold Periods on the Elevated Temperature Fatigue Behavior of 2-1/4Cr-1Mo Steel,”
Journal of Engineering Materials and Technology, Jan. 1981, Vol. 103, pp. 7-14.
[E-2] Hecht, R.L., and J.R. Weertman, “Periodic Oxide Cracking on Fe2.25Cr1Mo Produced
by High-Temperature Fatigue Tests with a Compression Hold,” Metallurgical
Transactions A. Vol. 24A, Feb. 1993, pp. 327 – 333.

E-8
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum, Nozzle Pipe to Header or Tube to Header
Boreholes

Table E-1
Fatigue analysis of a borehole region of a 1890-mm OD x 106-mm wall (74.4-inch OD x 4.2-inch wall) SA209 drum with set-in
nozzles. The shaded regions show the required input values
Calculations for cyclic loading due to pulsating internal pressure or combined changes of internal
pressure and temperature
A B C D E F G
Symbol Value Units
Design Data & Analysis
Assumptions
1 Type and (nominal) dimensions of 1890 mm OD x 106 mm wall SA 299 HP Drum with set in nozzles with full
detail: penetration welds and machined root
2 Material (CS,11,22,91): CS
3 Seamless(1) or longitudinally 0
welded (0):
4 Water-touched (Yes=1, No=0): 1
5 Connection configuration: 2
1 = set through with full penetration weld
2 = set-on or set in with full penetration weld
3 = flanged:
6 Weld root finish: 1
1 = machined CJP weld
2 = not machined or PJP weld
3 = Y-branch, not machined:
7 Design wall thickness of header sb 106 mm
= specified thickness (if
longitudinally welded)
= 1.15 x specified thickness (if
seamless), or
= measured thickness:
8 Outside diameter of header: do 1890 mm
9 Inside diameter of header di 1678 mm
= do - 2(sb):
10 Inside diameter of branch opening: dAi 382 mm
11 Out of roundness: U 0.5 %

E-9
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum, Nozzle Pipe to Header or Tube to Header
Boreholes
Calculations for cyclic loading due to pulsating internal pressure or combined changes of internal
pressure and temperature
A B C D E F G
Symbol Value Units
o
12 Opening angle for Y-shaped phiA Not
branches: used
Shutdown plus:
Transient Cold Warm Hot start Load
start start change
13 #/yr 48 192 24 720
14 Years 25 25.0 25 25
15 Total #, n: 1200 4800 600 18000
16 Normal operating pressure: p4 MPa 15.30 15.30 15.30 15.30
17 Min. operating pressure during p_ MPa 0.00 0.30 3.80 8.90
cycle:
18 Max. operating pressure during p^ MPa 13.25 13.25 13.25 13.25
cycle:
o
19 Min. operating temperature during T_ C 20 144 249 303
cycle:
o
20 Max. operating temperature during T^ C 333 333 333 333
cycle:
o
21 Maximum temperature change = T^ deltaT^ C 313 189 84 29
- T_:
o
22 Governing cycle temperature, T* C 255 286 312 325
T*=0.75T^+0.25T_:
o
23 Startup temperature ramp rate at vT1 C/mi 3.0 3.0 2.5 1.0
p_ (positive ramp!): n
24 Ratio of shutdown to startup ramp Y 0.17 0.17 0.20 0.50
rate (-vT2 / vT1):
o
25 Shutdown temperature ramp rate at vT3 C/mi -0.5 -0.5 -0.5 -0.5
p^ (negative ramp!): n
26 Modulus of elasticity at T* ET* MPa 189494 186063 182788 180958
E(C-Steel)=213160- 69.1 x T*
-0.01824 x T*2
27 Minimum yield strength at T* YST* MPa 222.8 216.6 211.4 208.8
28 Differential thermal expansion BLT* 1/oC 1.31E- 1.33E- 1.34E- 1.35E-
coefficient at T* 05 05 05 05
29 Linear thermal expansion B'LT* 1/oC 1.29E- 1.31E- 1.32E- 1.33E-
coefficient (used to estimate 05 05 05 05
density)
30 Density at T* Den kg/m3 7779 7769 7760 7755
=7850 kg/m3 / [1+B'LT* x (T*-20)]^3,

E-10
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum, Nozzle Pipe to Header or Tube to Header
Boreholes
Calculations for cyclic loading due to pulsating internal pressure or combined changes of internal
pressure and temperature
A B C D E F G
Symbol Value Units
(used to estimate thermal
diffusivity):
31 Thermal conductivity (used to Alpha J/sec- 44.82 44.17 43.60 43.29
estimated thermal diffusivity): m-oC
32 Heat Capacity (used to estimate Cp J/kgo 550.56 561.42 571.27 576.67
thermal diffusivity): C
33 Thermal diffusivity at T* aT* mm2/ 627.62 603.15 582.11 571.10
= (60 sec/min) x (10002 mm2/m2) x min
Alpha / (Den x Cp)
34 Minimum tensile strength at room UTS MPa 517.2 517.2 517.2 517.2
temperature
35 Minimum yield strength at room YS MPa 275.9 275.9 275.9 275.9
temperature

36 Surface finish factor (YS) f3 1


= 1.0 for YS <= 360 MPa
= 1.2 for 360 < YS <=600
=1.4 for YS>600:
37 Theoretical stress concentration for amo 2.9
membrane stress
=2.6 (set through full penetration
welded branches and forged
nipples)
=2.9 (set on or set in with full
penetration welded branches)
=3.2 (flanged):
38 Weld root gap factor f4 f4 1.000
=1 for machined root:

E-11
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum, Nozzle Pipe to Header or Tube to Header
Boreholes
Calculations for cyclic loading due to pulsating internal pressure or combined changes of internal
pressure and temperature
A B C D E F G
Symbol Value Units
Fatigue Life Calculations and Corrosion-Fatigue
Oxide Cracking Checks

39 Mean diameter, dm = di + sb: dm mm 1784 1784 1784 1784


40 uo = 1 + 2 x sb/di: uo 1.126 1.126 1.126 1.126
41 W = 0.35 / (BLT* x ET*): W mm2- 0.143 0.144 0.145 0.145
o
C/N
2 2
42 phif = (1/8) x ((uo -1)(3uo -1)- phif -0.354 -0.354 -0.354 -0.354
4uo4ln(uo))/[(uo2-1)(uo-1)2]:
43 Temperature gradient factor V 1/min -0.158 -0.152 -0.146 -0.144
V = aT* / (phif x sb2):
44 Out of roundness factor at MAWP, fu(p4) 0.107 0.107 0.107 0.107
fu(p4) = 1.5 x (dm/sb) x (U/100) / [(1+
0.455 x (p4/ET*) x (dm/sb)3]
45 Effective stress concentration am(p4) 3.11 3.11 3.11 3.11
factor, am(p4) = amo x f4 + 2 x fu(p4)
46 Peak pressure stress at MAWP, Sip4 MPa 401.0 401.0 400.9 400.8
Sip4 = am(p4) x p4*dm/(2sb), :
47 Out of roundness factor at highest fu(p^) 0.110 0.109 0.109 0.109
pressure, fu(p^)
= 1.5 x (dm/sb) x (U/100) / [(1+
0.455 x (p^ / ET*) x (dm/sb)3]
48 Effective stress concentration am(p^) 3.12 3.12 3.12 3.12
factor,
am(p^) = amo x f4 + 2 x fu(p^)
49 Maximum peak pressure stress at Sip^ MPa 347.8 347.7 347.7 347.6
the beginning of the shutdown,
Sip^ = am(p^) x p^*dm/(2Sb)
50 Out of roundness factor at lowest fu(p_) 0.126 0.126 0.121 0.114
pressure,
fu(p_) = 1.5 x (dm/sb) x (U/100) / [(1+
0.455 x (p_/ET*) x (dm/sb)3]
51 Effective stress concentration am(p_) 3.15 3.15 3.14 3.13
factor,
am(p_) = amo x f4 + 2 x fu(p_)

E-12
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum, Nozzle Pipe to Header or Tube to Header
Boreholes
Calculations for cyclic loading due to pulsating internal pressure or combined changes of internal
pressure and temperature
A B C D E F G
Symbol Value Units
52 Minimum peak pressure stress at Sip_ MPa 0.0 8.0 100.5 234.3
beginning of startup transient,
Sip_ = am(p_) x p_*dm/(2Sb)
o
53 Temperature gradient between the dTt1 C -19.01 -19.78 -17.08 -6.96
mid and inner wall thickness at the
beginning of the startup when p =
p_ (should be a negative
temperature gradient) = vt1/V:
54 Thermal stress during startup x 2, Sit1 MPa -132.8 -137.5 -118.0 -47.9
Sit1 = vt1/ W V (or = dTt1 / W)
(should be negative = compression
at ID surface):
o
55 Temperature gradient across the dTt2 C 3.17 3.30 3.42 3.48
mid and inner wall thickness at the
beginning of the shutdown when p
= p^ ( should be a positive
temperature gradient) = vt1/V
56 Thermal stress during shutdown x Sit2 MPa 22.1 22.9 23.6 23.9
2, Sit2 = vt2/ W x V (or = dTt2 / W)
(should be positive = tension at ID
surface)
57 Minimum peak (pressure + Si_ MPa -132.8 -129.6 -17.5 186.4
thermal) stress during startup,
Si_ = Sip_ + Sit1
58 Maximum peak (pressure + Si^ MPa 369.9 370.7 371.3 371.6
thermal) stress during shutdown,
Si^ = Sip^ + Sit2
59 Stress range for complete dSi MPa 502.7 500.2 388.8 185.2
start/stop cycle, dSi = Si^ - Si_
60 Total stress range for elastic-plastic 2Sa MPa 567.2 577.6 357.4 82.1
conditions (with surface finish
correction and elastic/plastic stress
adjustment),
2Sa = dSi x f3 x dSi/(2 x YST*)

E-13
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum, Nozzle Pipe to Header or Tube to Header
Boreholes
Calculations for cyclic loading due to pulsating internal pressure or combined changes of internal
pressure and temperature
A B C D E F G
Symbol Value Units
61 Total stress range for elastic 2Sa MPa 504.3 502.4 389.2 195.0
conditions (with surface finish and
Gerber mean stress correction),
2Sa = dSi x f3 x (2 x UTS)2 / [(2 x
UTS)2 - (2 x YST* - dSi)2]
62 Adjusted stress range, relevant to 2Sa' MPa 567.2 577.6 389.2 195.0
smooth base metal specimens with
fully reversed loading
63 Cycles to crack initiation at 2Sa' ni^ stop/s 24409 22443 93914 165192
tarts 17
64 Safety factor of cycles to cracking SF(ni^) 1 1 1 1
65 Fatigue damage fraction, n/ni^ 0.0492 0.2139 0.0064 0.0011
66 Total cyclic damage fraction check, Is the sum of the ni/ni^ less 0.2705 Fatigue Life OK
than 0.5?
67 Peak pressure stress @ MAWP - S1 MPa -199.0 -199.0 -199.1 -199.2
magnetite compressive strength
S1 = Sip4 - 600
68 Peak pressure stress @ MAWP + S2 MPa 601.0 601.0 600.9 600.8
magnetite tensile strength,
S2 = Sip4 + 200
69 If Si_ is less than S1 then Check1 OK OK OK OK
compressive cracking of the Fe3O4
may occur
70 If Si^ is greater than S1 then tensile Check2 OK OK OK OK
cracking of the Fe3O4 may occur

E-14
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum, Nozzle Pipe to Header or Tube to Header
Boreholes

Table E-2
Example EN12952-3 Section 13 & Appendix B & C 2001 Edition Calculations for a 323.8 mm OD x 28.6 mm Wall P91 HP
Superheater Outletlet Header Borehole Subjected to Pressure Changes and Thermal Shock Loadings

Design Data & Analysis Assumptions

Symbol Value Units

Type and (nominal) dimensions of detail: 324 mm OD x 29 mm wall P91 HP superheater outlet header with set in
nozzles with full srength, partial joint penetration welds
1 Material (CS,11,22,91): 91
2 Seamless(1) or longitudinally welded (0): 1
3 Water-touched (Yes=1, No=0): 0
4 Design wall thickness of header, ems 28.58 Mm
= specified thickness (if longitudinally
welded),
= 1.15 x specified thickness (if seamless),
or = measured thickness:
5 Outside diameter of drum/header: do 323.8 Mm
6 Inside diameter of drum/header, = do - 2(sb): 266.64 Mm
7 Mean diameter of drum/header: dms 295.22 Mm
8 Outside diameter of branch: dob 129.8 Mm
9 Branch mean wall thickness: emb 11.46 Mm
10 Mean diameter of branch: dmb 118.3 Mm
Normal shutdown plus:
Transient: Cold Warm Hot start Load
start start change
11 #/yr 40 12 180 600
12 Operating conditions, pressure and thermal loadings: Years 30 30 30 30
13 Number of loadings n Total #, n: 1200 360 5400 18000
2
14 Calculation pressure: pc N/mm 14.20 14.20 14.20 14.20
2
15 Normal operating pressure: po N/mm 12.59 12.59 12.59 12.59
2
16 Min. operating pressure during cycle: pmin N/mm 0.00 0.30 3.80 8.41
2
17 Max. operating pressure during cycle: pmax N/mm 12.59 12.59 12.59 12.59
o
18 Calculation temperature: tc C 300 258 258 258
o
19 Min. operating temperature during cycle: tmin C 20 152 250 567
o
20 Max. operating temperature during cycle: tmax C 567 567 567 567
o
21 Maximum temperature change = T^ - T_: C 547 415 317 0
o
22 Governing cycle temperature, t* C 430.3 463.3 487.8 567.0
T*=0.75T^+0.25T_:
o
23 Estimated startup temperature ramp rate vt1 C/min 25.0 26.4 35.0 10.0
at p_ (positive ramp!) = V x dt1:
o
24 vt1 C/sec 0.42 0.44 0.58 0.17
o
25 Shutdown temperature ramp rate at p^ vt2 C/min -5.0 -5.0 -5.0 -5.0
(negative ramp!):

E-15
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum, Nozzle Pipe to Header or Tube to Header
Boreholes

Design Data & Analysis Assumptions

Symbol Value Units


o
26 vt2 C/sec -0.083 -0.083 -0.083 -0.083
27 Ratio of shutdown to startup ramp rate (-vT2 / vT1): 0.20 0.19 0.14 0.50
Material properties:
2
28 Modulus of elasticity at T*, E E N/mm 184617 181315 178754 169828
2
29 Minimum yield strength at T* Re(t*) N/mm 315.5 292.0 270.5 172.2
o o o
30 Coefficient of expansion going from 20 C to T* + 10 C 1/ C 1.21E-05 1.22E-05 1.23E-05 1.25E-05
o o o
31 Coefficient of expansion going from 20 C to T* - 10 C 1/ C 1.20E-05 1.22E-05 1.22E-05 1.24E-05
o o
32 Coefficient of expansion going from 20 C to T* 1/ C 1.21E-05 1.22E-05 1.23E-05 1.25E-05
o
33 Differential thermal expansion coefficient betaL 1/ C 1.33E-05 1.35E-05 1.36E-05 1.38E-05
at T*
o
34 Thermal conductivity (used to estimated thermal diffusivity): J/sec-m- C 28.84 29.05 29.18 29.42
2
35 Thermal diffusivity at T* = (60 sec/min) x Dth mm /sec 5.60 5.36 5.18 4.49
2 2 2
(1000 mm /m ) x Alpha / (Den x Cp)
2
36 Minimum tensile strength at room Rm N/mm 586.2 586.2 586.2 586.2
temperature
2
37 Minimum yield strength at room temperature N/mm 413.8 413.8 413.8 413.8
38 Poisson's ratio ny 0.3 0.3 0.3 0.3
Gross structural stress concentration and other factors:
39 emb/ems 0.4009 0.4009 0.4009 0.4009
40 zeta = dmb/dms x SQRT(dms/2/ems) zeta 0.9110 0.9110 0.9110 0.9110
41 Pressure stress concentration factor, alpham 3.15 3.15 3.15 3.15
m (13.4.5, fig13.4-5)
42 z = dmb/dms z 0.4009 0.4009 0.4009 0.4009
43 Heat transfer coefficient, h h 1000 1000 1000 1000
44 Temperature stress concentration factor, alphat 1.160 1.160 1.160 1.160
m (13.4.7, fig.13.4-8)
45 uO = do/(dmb/dms), uo = 1 + 2 x sb/di: uO 1.2144 1.2144 1.2144 1.2144
46 Shape factor for thermal gradient,gamma gamma -0.3676 -0.3676 -0.3676 -0.3676
(13.4.10 and fig. 13.4-6) gamma
= (1/8) x ((uo2-1)(3uo2-1)-
4uo4ln(uo))/[(uo2-1)(uo-1)2]:)
2 o
47 W=alphat x betaL x E/(1-ny): W N/mm / C 4.0831 4.0529 4.02457 3.89837
48 Temperature gradient factor, V 1/s -0.0186 -0.0178 -0.0172 -0.0149
V = Dth/gamma/ems^2:
Temperature gradients, stresses and oxide cracking checks:
2
49 Normal operating pressure stress, Sp,o N/mm 217.42 217.42 217.42 217.42
Sp,o = (alpham x dms/2/ems+1) x po
2
50 Min. pressure stress, Sp,min N/mm 0.00 5.18 65.62 145.23
Sp,min = alpham x dms/2/ems+1) x pmin

E-16
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum, Nozzle Pipe to Header or Tube to Header
Boreholes

Design Data & Analysis Assumptions

Symbol Value Units


2
51 Max. pressure stress, Sp,max N/mm 217.42 217.42 217.42 217.42
Sp,max = alpham x dms/2/ems+1) x pmax
o
52 Temperature gradient at the beginning of dt1 C -22.35 -24.63 -33.84 -11.15
the startup when p = p_ (should be a
negative temperature gradient) = vt1/V:
o
53 Temperature gradient at the beginning of dt2 C 4.47 4.67 4.83 5.57
the shutdown when p = p^ (should be a
positive temperature gradient) = vt2/V:
2
54 Min thermal stress (beginning of startup), St,min N/mm -91.27 -99.83 -136.18 -43.46
St,min = W x dt1:
2
55 Max thermal stress (begining of St, max N/mm 18.25 18.91 19.45 21.73
shutdown), St,max = W x dt2:
2
56 Min. combined pressure + thermal stress f1 N/mm -91.27 -94.65 -70.55 101.77
(beginning of startup), f1
2
57 Min. combined stress to prevent oxide cracking: N/mm -395.17 -395.17 -395.17 -395.17
58 Oxide cracking check #1 (min oxide stress OK OK OK OK
@ beginning of startup):
2
59 Max combined pressure + thermal stress f2 N/mm 235.67 236.33 236.87 239.15
(beginning of shutdown), f2
2
60 Max. combined stress to prevent oxide cracking: N/mm 404.83 404.83 404.83 404.83
61 Oxide cracking check #2 (max oxide stress OK OK OK OK
@ beginning of shutdown):
2
62 Mean stress, fv = (f1+f2)/2 fv N/mm 72.2 70.8 83.2 170.5
2
63 Stress range, deltafv = f2-f1 deltafv N/mm 326.9 331.0 307.4 137.4
Micro-notch stress concentration factors
(as appropriate):
64 Mill scale factor (B.5.2, Fig. B-11), Cko: Cko 1.30 1.30 1.30 1.30
65 Weld configuration group K1 (slight notch Ck1 1.00 1.00 1.00 1.00
effect) factor (B5.5, Tables B-1 and B-4),
Ck1:
66 Weld configuration group K2 (moderate Ck2 1.00 1.00 1.00 1.00
notch effect) factor (B5.5, Tables B-2 and
B-4), Ck2:
67 Weld configuration group K3 (pronounced Ck3 1.00 1.00 1.00 1.00
notch effect) factor (B5.5, Tables B-3 and
B-4), Ck3:
68 Ck = Cko x Ck1 x Ck2 x Ck3: Ck 1.30 1.30 1.30 1.30
Stress range corrected for notch and
mean stress/plasticity factors:
2
69 Notch factor corrected mean stress, fv* = fv* N/mm 93.86 92.09 108.11 221.60
Ck x fv

E-17
EPRI Proprietary Licensed Material

Estimating The Borehole Thermal Shock (Loading Mode 1) Limits for Riser/Downcomer Pipe Nozzle to Drum, Nozzle Pipe to Header or Tube to Header
Boreholes

Design Data & Analysis Assumptions

Symbol Value Units


2
70 Notch factor corrected elastic stress range 2fva* N/mm 425.03 430.27 399.65 178.59
(2 x stress amplitude), 2fva* = Ck x fv
71 2fva*/Re(t*): 1.35 1.47 1.48 1.04
2
72 Mean stress/plasticity corrected stress 2fa* N/mm 436.21 441.16 413.72 208.37
range (for 2fva*/Re(t*)<1), 2fa* = 2fva*/(1-
(fv*/Rm)^2):
2
73 Mean stress/plasticity corrected stress 2fa* N/mm 438.58 437.81 405.54 182.23
range (for 1<2fva*/Re(t*)<2), 2fa* =
2fva*/{1-[(Re(t*)-2fva*/2)/Rm]^2}:
2
74 Mean stress/plasticity corrected stress 2fa* N/mm 286.26 316.96 295.26 92.63
range (for 2fva*/Re(t*)>2), 2fa* =
(2fva*^2)/[2*Re(t*)]:
2
75 Final corrected stress range, 2fa*: 2fa* N/mm 438.58 437.81 405.54 182.23
Fatigue life calculations:

76 Fatigue curve stress range temperature Ct* 0.69 0.64 0.60 0.46
correction, Ct*
77 Temperature corrected stress range, 2fa*(t*) 637.66 685.56 675.92 393.83
2fa*(t*) = 2fa*/Ct*:
78 Fatigue curve statistical scatter stress Ss 1.50 1.50 1.50 1.50
correction factor, Ss:
79 Final notch, mean stress/plasticity, 2fas 956.5 1028.3 1013.9 590.7
temperature and statistical corrected
stress range, 2fas = 2fa*(t*)*Ss:
80 Minimum cycles to cracking at 2Sfas, NAs stop/starts 45751 35585 37329 577700
NAs:
81 Mean cycle to cracking at 2faL, NAL NAL 318389 201629 219122 1000000
0
82 Statistical factor to adjust from mean to SL 10 10 10 10
minimum cycles to cracking, SL:
83 Minimum cycles to cracking with factor of NAL/SL 31839 20163 21912 1000000
SL on cycles to cracking: NAL/SL:
84 Maximum fatigue damage fraction, n/ni^ = n/N 0.0377 0.0179 0.2464 0.0312
n/min(NAs, NAL/SL):
85 Allowable min fatigue life usage factor: (n/N)allowable 0.100
86 Total cyclic damage fraction check, Is the sum of the ni/ni^ less than the allowable 0.333 Min. fatigue life consumption is
usage factor (=1.00, 0.40 or 0.10)? too high

E-18
EPRI Proprietary Licensed Material

F
MONOTONIC AND CYCLIC PROPERTIES OF HRSG
PRESSURE PART MATERIALS

This appendix provides a compilation of data for the three following commonly used HRSG
pressure part materials:
Carbon steel
2.25%Cr – 1 Mo steel (Grade 22)
9Cr-1Mo-V-Cb (Grade 91)

The chemical composition of the tubing versions of these materials is provided in the table
below.
Composition of Tubing

Material Grade Carbon Mn P S max Si Cr Mb V Nb N Ni Al

max.

Carbon steel SA 192 0.06- 0.27- 0.035 0.035 0.25 max ___ ___ ___ ___ ___ ___ ___
0.18 0.63

Carbon steel SA178A 0.06- 0.27- 0.035 0.035 ___ ___ ___ ___ ___ ___ ___ ___
0.18 0.63

Carbon- SA178D 0.27 (max.) 1.00- 0.030 0.015 0.10 (min.)


manganese 1.50
steel

2.25Cr-1Mo SA213 T22 0.05- 0.30- 0.025 0.025 0.50 (max.) 1.90- 0.87- ___ ___ ___ ___ ___
0.15 0.60 2.60 1.13

9Cr-1Mo-V-Cb SA213 T91 0.08- 0.30- 0.020 0.010 0.20- 8.00- 0.85- 0.18- 0.06- 0.03- 0.40 0.04
0.12 0.60 0.50 9.50 1.05 0.25 0.10 0.070 (max.) (max.)

F-1
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

The limits of the required mechanical properties of the tubing versions of these materials are
provided in the following table.

Material Grade Room Temperature Room Temperature Elongation in Hardness


Tensile Strength Yield Strength 2” (50 mm)

Carbon steel SA 192 47 ksi (325 MPa) 26 ksi (180 MPa) 35% 77 HRB max

Carbon steel SA178A 47 ksi (325 MPa) 26 ksi (180 MPa) 35% ~

Carbon-manganese steel SA178D 60 ksi (415 MPa) 37 ksi (255 MPa) 30% ~

2.25Cr-1Mo SA213 T22 60 ksi (415 MPa) 30 ksi (255 MPa) 30% 85 HRB max

9Cr-1Mo-V-Cb SA213 T91 85 ksi (585 MPa) 60 ksi (415 MPa) 20% 25 HRC max

Generic materials properties were obtained from published literature.

The success of estimating the cyclic performance/life of these materials depends to a large extent
on the availability and proper use of reliable material property data over a range of temperatures.
Tables F-1 through F-3 summarize the following code based material properties for carbon, 2-
1/4Cr – 1 Mo steel and 9Cr-1Mo-V-Cb steels. The carbon steel properties are for moderately
high strength carbon steel and might not be appropriate for modeling of SA192 and SA 178
Grade A and B tubing. The following properties are included in these tables:

Thermal expansion coefficient, TE


Thermal conductivity, TC
Thermal diffusivity, TD
Modulus of elasticity, E
Yield stress, YS
Ultimate tensile strength, UTS

Monotonic (single load tensile test) and cyclic properties of carbon, 2-1/4Cr – 1 Mo steel and
9Cr-1Mo-V-Cb steels are summarized in Table F-4 and provided for individual materials and
test conditions in Tables F-5 through F-19. The following material properties are included in
these tables.

Cyclic (Ramberg-Osgood equation) strength coefficient, H’


Cyclic (Ramberg-Osgood equation) strain hardening exponent, n’
Coffin-Manson fatigue ductility coefficient, εf’
Coffin-Manson fatigue ductility exponent, c
Coffin-Manson fatigue strength coefficient, σf’
Coffin-Manson fatigue strength exponent, b

F-2
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

The key equations that are associated with these material properties are:

Cyclic Stress Strain Curve

Ramberg-Osgood equation representing the plastic strain portion of a cyclic stress-strain curve
is:

σ = H’ εpan’ (1)

Log(σ) = Log(H’) + n’ Log(εpa) (1a)


where:
σ = elastic-plastic stress amplitude in a fatigue test after initial strain hardening or
softening has occurred (i.e., measured when half the fatigue life has been consumed)
εpa = plastic strain amplitude in a fatigue test measured when half the fatigue life has been
consumed
εpa = εt – εe = εt – σ/E (2)

εt, εe = total and elastic strains in a tensile test

Rearranging equations (1) and (2) provides the overall cyclic stress strain curve equation:

εt = σ/E + (σ/Η’)1/n’ (3)

The relationship between cyclic stress range and cyclic strain range is given as:

εt/2 = σ/2E + (σ/2Η’)1/n’ (4a)

Δε = Δσ/E + 2(Δσ/2Η’)1/n’ (4b)

Strain Life Curve

Thermal fatigue tests with fully reversed loading (zero mean stress) are modeled using the
Coffin-Manson fatigue life equation:

εa = σf’/E (2Nf)b + εf’ (2Nf)c (5)

where:
εa = total elastic + plastic strain amplitude
2Nf = reversals to failure = 2 x cycles to failure, Nf

The empirical coefficients and exponents are derived from the elastic stress and plastic strain
portions of experimental fatigue tests using the following relationships:

F-3
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

εa = εea + εpa = σa/E + εpa (6)

εea = σa/E = σf’/E (2Nf)b (7a)

Rearranging equation (7a) so that the reversal to failure is the independent variable and taking
the logarithmic values of the variables provides the regression equation for the elastic strain
portion of the Manson-Coffin equation:

Log (2Nf) = 1/b Log (εea E) – 1/b Log (σf’) (7b)

Inspection of equation (7b) reveals:

b = 1/slope of regression equation (8)

σf’ = 10-[intercept of regression equation/slope of regression equation] (9)

The plastic strain portion of the Coffin-Manson equation is:

εpa = εf’ (2Nf)c (10a)

Log (2Nf) = 1/c Log (εpa) – 1/c Log (εf’) (10b)

Regression of equation (10b) provides the empirical constants:

c = 1/slope of regression equation (11)

εf’ = 10-[intercept of regression equation/slope of regression equation] (12)

F-4
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-1
Carbon Steel (e.g.,SA299, >1" thick) Material Properties Ref: ASME Section II, Part D (2001)

Temp TC TD TEmean E YS UTS Temp TC TD TEmean E YS UTS


o
F Btu/hr-ft-oF ft2/hr in/in/oF psi ksi ksi o
C W/m-oC mm2/min mm/mm/oC N/mm2 N/mm2 N/mm2
70 27.5 0.529 6.40E-06 2.95E+07 40.0 75.0 21.1 47.56 819 1.15E-05 2.03E+05 275.9 517.2
100 27.6 0.512 6.50E-06 40.0 75.0 37.8 47.73 793 1.17E-05 275.9 517.2
150 27.6 0.496 6.60E-06 37.6 65.6 47.73 768 1.19E-05 259.3
200 27.6 0.486 6.70E-06 2.88E+07 36.6 75.0 93.3 47.73 753 1.21E-05 1.99E+05 252.4 517.2
250 27.4 0.467 6.80E-06 36.0 121.1 47.39 723 1.22E-05 248.3
300 27.2 0.453 6.90E-06 2.83E+07 35.4 75.0 148.9 47.04 701 1.24E-05 1.95E+05 244.1 517.2
350 27 0.44 7.00E-06 176.7 46.70 681 1.26E-05
400 26.7 0.428 7.10E-06 2.77E+07 34.2 75.0 204.4 46.18 663 1.28E-05 1.91E+05 235.9 517.2
450 26.3 0.413 7.20E-06 232.2 45.49 639 1.30E-05
500 25.9 0.398 7.30E-06 2.73E+07 32.6 75.0 260.0 44.79 616 1.31E-05 1.88E+05 224.8 517.2
550 25.5 0.387 7.30E-06 287.8 44.10 599 1.31E-05
600 25 0.374 7.40E-06 2.66E+07 30.7 75.0 315.6 43.24 579 1.33E-05 1.83E+05 211.7 517.2
650 24.5 0.36 7.50E-06 29.6 75.0 343.3 42.37 557 1.35E-05 204.1 517.2
700 24 0.346 7.60E-06 2.55E+07 28.6 75.0 371.1 41.51 536 1.37E-05 1.76E+05 197.2 517.2
750 23.5 0.332 7.70E-06 27.7 74.1 398.9 40.64 514 1.39E-05 191.0 511.0
800 23 0.318 7.80E-06 2.42E+07 26.8 68.9 426.7 39.78 492 1.40E-05 1.67E+05 184.8 475.2
850 22.6 0.305 7.90E-06 26.0 62.8 454.4 39.09 472 1.42E-05 179.3 433.1
900 22.1 0.291 7.90E-06 2.24E+07 25.3 56.0 482.2 38.22 451 1.42E-05 1.54E+05 174.5 386.2
950 21.5 0.277 8.00E-06 24.6 49.2 510.0 37.18 429 1.44E-05 169.7 339.3
1000 21 0.263 8.10E-06 2.04E+07 23.8 43.2 537.8 36.32 407 1.46E-05 1.41E+05 164.1 297.9
1050 20.5 0.249 8.10E-06 565.6 35.45 386 1.46E-05
1100 19.9 0.237 8.20E-06 1.80E+07 593.3 34.42 367 1.48E-05 1.24E+05
TC = thermal conductivity (Btu/hr-ft-oF x 1.7295 = W/m-oC)
TD = thermal diffusivity = thermal conductivity/(density x specific heat), (ft2/hr x 1548.38 = mm2/min)
TEmean = mean thermal expansion coefficient in going from 70oF to the indicated temperature (in/in-oF x 1.8 = mm/mm-oC)
E = modulus of elasticity (psi x 6.896E-03 = N/mm2 = MPa)

F-5
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-2
Grade 22 (2-1/4 Cr – 1 Mo steel) Material Properties Ref: ASME Section II, Part D (2001
Temp TC TD TEmean E YS UTS Temp TC TD TEmean E YS UTS
o
F Btu/hr-ft-oF ft2/hr in/in/oF psi ksi ksi o
C W/m-oC mm2/min mm/mm/oC N/mm2 N/mm2 N/mm2
70 20.9 0.408 6.40E-06 3.06E+07 30.0 60.0 21.1 36.15 632 1.15E-05 2.11E+05 206.9 413.8
100 21 0.397 6.50E-06 30.0 60.0 37.8 36.32 615 1.17E-05 206.9 413.8
150 21.2 0.391 6.60E-06 28.6 60.0 65.6 36.67 605 1.19E-05 197.2 413.8
200 21.3 0.385 6.70E-06 2.98E+07 28.0 59.9 93.3 36.84 596 1.21E-05 2.06E+05 193.1 413.1
250 21.4 0.378 6.80E-06 27.6 121.1 37.01 585 1.22E-05 190.3
300 21.5 0.371 6.90E-06 2.94E+07 27.2 58.2 148.9 37.18 574 1.24E-05 2.03E+05 187.6 401.4
350 21.5 0.364 7.00E-06 176.7 37.18 564 1.26E-05
400 21.5 0.357 7.10E-06 2.88E+07 26.9 58.2 204.4 37.18 553 1.28E-05 1.99E+05 185.5 401.4
450 21.5 0.349 7.20E-06 232.2 37.18 540 1.30E-05
500 21.4 0.341 7.30E-06 2.83E+07 26.9 58.2 260.0 37.01 528 1.31E-05 1.95E+05 185.5 401.4
550 21.3 0.332 7.30E-06 287.8 36.84 514 1.31E-05
600 21.1 0.323 7.40E-06 2.77E+07 26.9 58.2 315.6 36.49 500 1.33E-05 1.91E+05 185.5 401.4
650 20.9 0.314 7.50E-06 26.9 58.2 343.3 36.15 486 1.35E-05 185.5 401.4
700 20.7 0.305 7.60E-06 2.71E+07 26.9 58.2 371.1 35.80 472 1.37E-05 1.87E+05 185.5 401.4
750 20.5 0.295 7.70E-06 26.8 58.2 398.9 35.45 457 1.39E-05 184.8 401.4
800 20.2 0.285 7.80E-06 2.63E+07 26.6 58.2 426.7 34.94 441 1.40E-05 1.81E+05 183.4 401.4
850 20 0.274 7.90E-06 26.2 58.2 454.4 34.59 424 1.42E-05 180.7 401.4
900 19.7 0.264 7.90E-06 2.56E+07 25.6 58.2 482.2 34.07 409 1.42E-05 1.77E+05 176.6 401.4
950 19.4 0.252 8.00E-06 24.8 57.5 510.0 33.55 390 1.44E-05 171.0 396.6
1000 19.1 0.241 8.10E-06 2.46E+07 23.7 53.9 537.8 33.03 373 1.46E-05 1.70E+05 163.4 371.7
1050 18.8 0.229 8.10E-06 565.6 32.51 355 1.46E-05
1100 18.5 0.217 8.20E-06 2.37E+07 593.3 32.00 336 1.48E-05 1.63E+05
1150 18.3 0.205 8.30E-06 621.1 31.65 317 1.49E-05
1200 18 0.192 8.30E-06 2.25E+07 648.9 31.13 297 1.49E-05 1.55E+05
1250 17.7 0.179 8.40E-06 676.7 30.61 277 1.51E-05
1300 17.2 0.163 8.40E-06 2.11E+07 704.4 29.75 252 1.51E-05 1.46E+05
TC = thermal conductivity (Btu/hr-ft-oF x 1.7295 = W/m-oC)
TD = thermal diffusivity = thermal conductivity/(density x specific heat), (ft2/hr x 1548.38 = mm2/min)
TEmean = mean thermal expansion coefficient in going from 70oF to the indicated temperature (in/in-oF x 1.8 = mm/mm-oC)
E = modulus of elasticity (psi x 6.896E-03 = N/mm2 = MPa)

F-6
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-3
Grade 91 (9Cr-1Mo-V-Cb steel) Material Properties Ref: ASME Section II, Part D (2001)

Temp TC TD TEmean E YS UTS Temp TC TD TEmean E YS UTS


o o 2 o o o 2 o
F Btu/hr-ft- F ft /hr in/in/ F psi ksi ksi C W/m- C mm /min mm/mm/ C N/mm2 N/mm2 N/mm2
70 12.8 0.256 5.80E-06 3.10E+07 60.0 85.0 21.1 22.14 396 1.04E-05 2.14E+05 413.8 586.2
100 13.1 0.258 5.90E-06 60.0 85.0 37.8 22.66 399 1.06E-05 413.8 586.2
150 13.6 0.26 5.90E-06 57.1 85.0 65.6 23.52 403 1.06E-05 393.8 586.2
200 14 0.261 6.00E-06 3.02E+07 55.9 85.0 93.3 24.21 404 1.08E-05 2.08E+05 385.5 586.2
250 14.4 0.263 6.10E-06 55.2 121.1 24.90 407 1.10E-05 380.7
300 14.7 0.262 6.20E-06 2.98E+07 54.8 85.0 148.9 25.42 406 1.12E-05 2.06E+05 377.9 586.2
350 15 0.261 6.20E-06 176.7 25.94 404 1.12E-05
400 15.2 0.259 6.30E-06 2.91E+07 54.7 84.7 204.4 26.29 401 1.13E-05 2.01E+05 377.2 584.1
450 15.4 0.256 6.30E-06 232.2 26.63 396 1.13E-05
500 15.6 0.253 6.40E-06 2.87E+07 54.7 84.4 260.0 26.98 392 1.15E-05 1.98E+05 377.2 582.1
550 15.8 0.249 6.50E-06 287.8 27.33 386 1.17E-05
600 15.9 0.245 6.50E-06 2.81E+07 54.5 83.1 315.6 27.50 379 1.17E-05 1.94E+05 375.9 573.1
650 16 0.24 6.60E-06 54.0 81.8 343.3 27.67 372 1.19E-05 372.4 564.1
700 16 0.235 6.60E-06 2.74E+07 53.2 80.0 371.1 27.67 364 1.19E-05 1.89E+05 366.9 551.7
750 16.1 0.229 6.70E-06 52.0 77.6 398.9 27.84 355 1.21E-05 358.6 535.2
800 16.1 0.22 6.70E-06 2.68E+07 50.4 74.7 426.7 27.84 341 1.21E-05 1.85E+05 347.6 515.2
850 16.2 0.214 6.80E-06 48.5 71.1 454.4 28.02 331 1.22E-05 334.5 490.3
900 16.1 0.207 6.80E-06 2.62E+07 46.1 66.9 482.2 27.84 321 1.22E-05 1.81E+05 317.9 461.4
950 16.1 0.199 6.90E-06 43.4 62.2 510.0 27.84 308 1.24E-05 299.3 429.0
1000 16.1 0.192 6.90E-06 2.54E+07 40.2 57.0 537.8 27.84 297 1.24E-05 1.75E+05 277.2 393.1
1050 16 0.182 7.00E-06 565.6 27.67 282 1.26E-05
1100 16 0.174 7.00E-06 2.45E+07 593.3 27.67 269 1.26E-05 1.69E+05
1150 15.9 0.165 7.10E-06 621.1 27.50 255 1.28E-05
1200 15.8 0.155 7.10E-06 2.33E+07 648.9 27.33 240 1.28E-05 1.61E+05
1250 15.7 0.143 7.10E-06 676.7 27.15 221 1.28E-05
1300 15.6 0.131 7.20E-06 2.20E+07 704.4 26.98 203 1.30E-05 1.52E+05
TC = thermal conductivity (Btu/hr-ft-oF x 1.7295 = W/m-oC)
TD = thermal diffusivity = thermal conductivity/(density x specific heat), (ft2/hr x 1548.38 = mm2/min)
TEmean = mean thermal expansion coefficient in going from 70oF to the indicated temperature (in/in-oF x 1.8 = mm/mm-oC)
E = modulus of elasticity (psi x 6.896E-03 = N/mm2 = MPa)

F-7
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-4
Summary of Monotonic and Cyclic Properties
Material Table Mtrl. Test condition H/H’ n/n’ εf/εf’ σf/σf’ b c
condition MPa MPa
SAE 1005 steel F-5 Hot rolled RT, 4 x 10- --/462 0.16/0.12 1.6/0.10 841/641 -0.109 -0.390
3
/sec

C-Steel Plate (20mm) F-6 N&T RT, 10-3/sec - - /757 - - /0.128 1.02/0.226 1017/1030.2 -0.1045 -0.4699
C-Steel Plate (20mm) F-7 N&T 300oC, 10-3/sec - - /741 - - /0.093 1.05/4.16 931/1025.8 -0.0965 -0.8341
C-Steel Plate (50mm) F-8 N&T 300oC, 10-3/sec --/741 --/0.082 1.08/7.650 926/727.13 -0.0468 -0.9681
C-Steel Plate (100mm) F-9 N&T 300oC, 10-3/sec --/748 --/0.100 0.799/0.9114 870/937.29 -0.0873 -0.6670
C-Steel Plate (20mm) F-10 N&T 400oC, 10-3/sec --/666 --/0.109 1.661/2.986 984/959 -0.1118 -.07927

C-Steel WM (20mm) F-11 RT, 10-3/sec - - /695 - - /0.110 1.309/1.121 --/1034.0 -0.1147 -0.6561
C-Steel WM (20mm) F-12 300oC, 10-3/sec - - /695 - - /0.066 0.8675/0.345 --/845.88 -0.0727 -0.5904
C-Steel WM (20mm) F-13 400oC, 10-3/sec --/549 --/0.067 1.204/4.826 --/1142.11 -0.1335 -0.9255

P22 Plate F-14 N&T RT, 10-3/sec 875/796 0.079/0.100 1.61/0.894 1648/871.01 -0.0639 -0.7061
P22 Plate F-15 N&T 500oC, 10-3/sec 819/652 0.120/0.105 1.77/1.13 1276/611 -0.0554 -0.7799
P22 Plate F-16 N&T 600oC, 10-3/sec 817/428 0.082/0.082 2.12/3.14 1062/456 -0.0566 -0.9629

P91 Plate F-17 N&T RT, 10-3/sec 710/975 0.047/0.117 1.43/0.577 1328/1052 -0.0766 -0.6098
P91 Plate F-18 N&T 550oC, 10-3/sec 482/609 0.054/0.142 2.04/0.179 1010/647.8 -0.0979 -0.5140
P91 Plate F-19 N&T 600oC, 10-3/sec 330/443 0.042/0.121 2.41/0.531 743/374 -0.0589 -0.6370

F-8
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-5
Materials Characterization Sheet
-3
Material: SAE 1005 to 1009 Carbon Steel Data sheet number: SAE 1005/9 (RT/4 x 10 /sec)

Condition: Test condition:

Heat treatment: Hot rolled Temperature:_RT


-3 -1
Hardness: 90 BHN Rising load strain rate: 4 x 10 /sec to 1 x 10 /sec
Product form: 3.3 mm to 4.0 mm (0.13” to 0.155” ) thick sheet Tensile hold time: None
-3 -1
Falling load strain rate: 4 x 10 /sec to 1 x 10 /sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 200,690 MPa (29.1 x 10 psi) Strength coefficient, H’: 462 MPa (67,000 psi)
Yield strength, YS: 262 MPa (38,000 psi) Strain hardening exponent, n’: 0.12
Ultimate strength, UTS: 345 MPa (50,000 psi)
Reduction in area, %RA: 80% Fatigue strength coefficient, σf’ : 641 MPa (93,000 psi)
True fracture strength, σf: 848 MPa (123,000 psi) Fatigue strength exponent, b : -0.109
True fracture ductility, εf: 1.6 Fatigue ductility coefficient, εf’: 0.100
Fatigue ductility exponent, c : -0.390

Comments: Microstructure :

Chemistry: C Mn P S Ferrite grain size : ASTM = 9 to 10


0.05 to 0.09 0.25 to 0.35 0.04 max 0.04 max
Inclusions : Not listed

Processing: Hot rolled sheet Microstructure : Ferrite

Stress-strain curve: Strain amplitude vs. reversals to complete separation into two
parts :

Reference: Landgraf & LaPointe, 1972

F-9
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-6
Materials Characterization Sheet
-
Material: SB450 Carbon Steel Data sheet number: SB450 NRIM DATA SHEET NO. 67 (RT/10
3
/sec)
Condition: Test condition:

Heat treatment: Hot rolled, and tempered Temperature:_RT


-3
Hardness:____________________________________ Rising load strain rate: 10 /sec
Product form: 1500 mm x 3000 mm x 20 mm plate Tensile hold time: None
Falling load strain rate: 10-3/sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 203,000 MPa (29.5 x 10 psi) Strength coefficient, H’: 757 MPa (109,800 psi)
Yield strength, YS: 328 MPa (47,600 psi) Strain hardening exponent, n’: 0.128
Ultimate strength, UTS: 503 MPa (73,000 psi)
Reduction in area, %RA: 64% Fatigue strength coefficient, σf’ : 1030 MPa (149,400 psi)
True fracture strength, σf: 1648 MPa (239,000 psi) Fatigue strength exponent, b : -0.1045
True fracture ductility, εf: 1.022 Fatigue ductility coefficient, εf’: 0.2261
Fatigue ductility exponent, c : -0.4699

Comments: Microstructure :

Chemistry: C Mn P S Si Cr Mo N Ferrite grain size : JIS G 0552 (1977) = 9


0.20 0.83 0.014 0.004 0.25 0.02 0.002 0.009
Ni Al Cu Inclusions : 0.03 area % inclusions deformed by plastic work
0.02 0.026 0.01 (sulphide, silicate, etc.) + no area % discontinuous arrays (alumina)
inclusions + no area % isolated (granular oxide) inclusions

Processing: Vacuum degassing, LD converter, Hot rolled Microstructure : Not documented


o
650 C/48min, furnace cooled

Stress-strain curve: Strain amplitude vs. reversals to complete separation into two
parts :

Reference: NRIM FDS No. 67, 1991

F-10
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-7
Materials Characterization Sheet
Material: SB450 Carbon Steel Data sheet number: SB450 NRIM DATA SHEET NO. 67
-3
(300C/10 /sec)
Condition: Test condition:

Heat treatment: Hot rolled, and tempered Temperature:_300C


-3
Hardness:____________________________________ Rising load strain rate: 10 /sec
Product form: 1500 mm x 3000 mm x 20 mm plate Tensile hold time: None
Falling load strain rate: 10-3/sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 162,000 MPa (23.5 x 10 psi) Strength coefficient, H’: 741 MPa (107,500 psi)
Yield strength, YS: 234 MPa (33,900 psi) Strain hardening exponent, n’: 0.093
Ultimate strength, UTS: 467 MPa (67,700 psi)
Reduction in area, %RA: 65% Fatigue strength coefficient, σf’ : 1026 MPa (148,800 psi)
True fracture strength, σf: 931 MPa (135,000 psi) Fatigue strength exponent, b : -0.0965
True fracture ductility, εf: 1.050 Fatigue ductility coefficient, εf’: 4.159
Fatigue ductility exponent, c : -0.8341

Comments: Microstructure :

Chemistry: C Mn P S Si Cr Mo N Ferrite grain size : JIS G 0552 (1977) = 9


0.20 0.83 0.014 0.004 0.25 0.02 0.002 0.009
Ni Al Cu Inclusions : 0.03 area % inclusions deformed by plastic work
0.02 0.026 0.01 (sulphide, silicate, etc.) + no area % discontinuous arrays
(alumina) inclusions + no area % isolated (granular oxide)
inclusions
Processing: Vacuum degassing, LD converter, Hot rolled
o
650 C/48min, furnace cooled Microstructure : Not documented

Stress-strain curve : Strain amplitude vs. reversals to complete separation into


two parts :

Reference: NRIM FDS No. 67, 1991

F-11
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-8
Materials Characterization Sheet
-
Material: SB450 Carbon Steel Data sheet number: SB450 NRIM DATA SHEET NO. 67 (400C/10
3
/sec)
Condition: Test condition:

Heat treatment: Hot rolled, and tempered Temperature:_400C


-3
Hardness:____________________________________ Rising load strain rate: 10 /sec
Product form: 1500 mm x 3000 mm x 20 mm plate Tensile hold time: None
Falling load strain rate: 10-3/sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 184,000 MPa (26.7 x 10 psi) Strength coefficient, H’: 666 MPa (96,5900 psi)
Yield strength, YS: 215 MPa (31,200 psi) Strain hardening exponent, n’: 0.109
Ultimate strength, UTS: 413 MPa (59,900 psi)
Reduction in area, %RA: 81% Fatigue strength coefficient, σf’ : 959.4 MPa (139,100 psi)
True fracture strength, σf: 984 MPa (142,700 psi) Fatigue strength exponent, b : -0.1118
True fracture ductility, εf: 1.661 Fatigue ductility coefficient, εf’: 2.9860
Fatigue ductility exponent, c : -0.7927

Comments: Microstructure :

Chemistry: C Mn P S Si Cr Mo N Ferrite grain size : JIS G 0552 (1977) = 9


0.20 0.83 0.014 0.004 0.25 0.02 0.002 0.009
Ni Al Cu Inclusions : 0.03 area % inclusions deformed by plastic work
0.02 0.026 0.01 (sulphide, silicate, etc.) + no area % discontinuous arrays (alumina)
inclusions + no area % isolated (granular oxide) inclusions

Processing: Vacuum degassing, LD converter, Hot rolled Microstructure : Not documented


o
650 C/48min, furnace cooled

Stress-strain curve : Strain amplitude vs. reversals to complete separation into two
parts :

Reference: NRIM FDS No. 67, 1991

F-12
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-9
Materials Characterization Sheet
-
Material: SB450 Carbon Steel Data sheet number: SB450 NRIM DATA SHEET NO. 67 (300C/10
3
/sec)
Condition: Test condition:

Heat treatment: Hot rolled, and tempered Temperature:_300C


-3
Hardness:____________________________________ Rising load strain rate: 10 /sec
Product form: 1500 mm x 3000 mm x 50 mm plate Tensile hold time: None
Falling load strain rate: 10-3/sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 188,000 MPa (27.3 x 10 psi) Strength coefficient, H’: 741 MPa (107,500 psi)
Yield strength, YS: 173 MPa (25,100 psi) Strain hardening exponent, n’: 0.082
Ultimate strength, UTS: 440 MPa (63,800 psi)
Reduction in area, %RA: 66% Fatigue strength coefficient, σf’ : 727.1 MPa (105,500 psi)
True fracture strength, σf: 926 MPa (134,300 psi) Fatigue strength exponent, b : -00468
True fracture ductility, εf: 1.08 Fatigue ductility coefficient, εf’: 7.6501
Fatigue ductility exponent, c : -0.9681

Comments: Microstructure :

Chemistry: C Mn P S Si Cr Mo N Ferrite grain size : JIS G 0552 (1977) = 7.7


0.21 0.85 0.009 0.001 0.28 0.03 0.01 0.004
Ni Al Cu Inclusions : 0.00 area % inclusions deformed by plastic work
0.22 0.03 0.32 (sulphide, silicate, etc.) + 0.02 % discontinuous arrays (alumina)
inclusions + 0.01 % isolated (granular oxide) inclusions
o
Processing: Al-Si killed, LD converter, Hot rolled, 900 C/75min air Microstructure : Not documented
o
cooled, 625 C/4hr furnace cooled

Stress-strain curve : Strain amplitude vs. reversals to complete separation into two
parts :

Reference: NRIM FDS No. 67, 1991

F-13
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-10
Materials Characterization Sheet
-
Material: SB450 Carbon Steel Data sheet number: SB450 NRIM DATA SHEET NO. 67 (300C/10
3
/sec)
Condition: Test condition:

Heat treatment: Hot rolled, and tempered Temperature:_300C


-3
Hardness:____________________________________ Rising load strain rate: 10 /sec
Product form: 7800 mm x 2500 mm x 100 mm plate Tensile hold time: None
Falling load strain rate: 10-3/sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 188,000 MPa (27.3 x 10 psi) Strength coefficient, H’: 748 MPa (108,500 psi)
Yield strength, YS: 186 MPa (27,000 psi) Strain hardening exponent, n’: 0.100
Ultimate strength, UTS: 491 MPa (71,200 psi)
Reduction in area, %RA: 55% Fatigue strength coefficient, σf’ : 937.3 MPa (135,900 psi)
True fracture strength, σf: 870 MPa (126,200 psi) Fatigue strength exponent, b : -0.0873
True fracture ductility, εf: 0.799 Fatigue ductility coefficient, εf’: 0.9114
Fatigue ductility exponent, c : -0.6670

Comments: Microstructure :

Chemistry: C Mn P S Si Cr Mo N Ferrite grain size : JIS G 0552 (1977) = 6


0.29 0.81 0.013 0.002 0.23 0.01 Tr 0.004
Ni Al Cu Inclusions : 0.00 area % inclusions deformed by plastic work
0.02 0.002 0.01 (sulphide, silicate, etc.) + 0.01 area % discontinuous arrays
(alumina) inclusions + 0.00 area % isolated (granular oxide)
inclusions
o
Processing: Si-killed, LD converter, Hot rolled, 900 C/230min, air
o
cooled, 625 C/4h, furnace cooled Microstructure : Not documented

Stress-strain curve : Strain amplitude vs. reversals to complete separation into two
parts :

Reference: NRIM FDS No. 67, 1991

F-14
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-11
Materials Characterization Sheet
-
Material: SB450 Carbon Steel Weld Metal Data sheet number: SB450 NRIM DATA SHEET NO. 67 (RT/10
3
/sec)

Condition: Test condition:

Heat treatment: As welded. Temperature:_RT


-3
Hardness:____________________________________ Rising load strain rate: 10 /sec
Product form: 1500 mm x 300 mm x 20 mm plate Tensile hold time: None
Falling load strain rate: 10-3/sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 203,000 MPa (29.4 x 10 psi) Strength coefficient, H’: 695 MPa (100,800 psi)
Yield strength, YS: 429 MPa (62,200 psi) Strain hardening exponent, n’: 0.110
Ultimate strength, UTS: 516 MPa (74,800 psi)
Reduction in area, %RA: 73% Fatigue strength coefficient, σf’ :1034 MPa (150,000 psi)
True fracture ductility, εf: 1.309 Fatigue strength exponent, b : -0.1147
Fatigue ductility coefficient, εf’: 1.1214
Fatigue ductility exponent, c : -0.6561

Comments: Microstructure :
Composition of the weld metal.
Ferrite grain size : Not documented
Chemistry: C Mn P S Si Cr Mo Ni Cu
0.112 1.49 0.022 0.004 0.30 0.019 0.002 0.018 0.48 Inclusions : Not documented

Microstructure : Not documented


Welding Process: Submerged arc welding

Stress-strain curve : Strain amplitude vs. reversals to complete separation into two
parts :

Reference: NRIM FDS No. 67, 1991

F-15
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-12
Materials Characterization Sheet
-
Material: SB450 Carbon Steel Weld Metal Data sheet number: SB450 NRIM DATA SHEET NO. 67 (300C/10
3
/sec)
Condition: Test condition:

Heat treatment: As welded. Temperature:_300C


-3
Hardness:____________________________________ Rising load strain rate: 10 /sec
Product form: 1500 mm x 300 mm x 20 mm plate Tensile hold time: None
Falling load strain rate: 10-3/sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 188,000 MPa (27.3 x 10 psi) Strength coefficient, H’: 666 MPa (100,800 psi)
Yield strength, YS: 298 MPa (43,200 psi) Strain hardening exponent, n’: 0.066
Ultimate strength, UTS: 533 MPa (77,300 psi)
Reduction in area, %RA: 58% Fatigue strength coefficient, σf’ : 845.9 MPa (122,700 psi)
True fracture ductility, εf: 0.8675 Fatigue strength exponent, b : -0.0727
Fatigue ductility coefficient, εf’: 0.3458
Fatigue ductility exponent, c : -0.5904

Comments: Microstructure :
Composition of the weld metal.
Ferrite grain size : Not documented
Chemistry: C Mn P S Si Cr Mo Ni Cu
0.112 1.49 0.022 0.004 0.30 0.019 0.002 0.018 0.48 Inclusions : Not documented

Microstructure : Not documented

Welding Process: Submerged arc welding

Stress-strain curve : Strain amplitude vs. reversals to complete separation into two
parts :

Reference: NRIM FDS No. 67, 1991

F-16
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-13
Materials Characterization Sheet
-
Material: SB450 Carbon Steel Weld Metal Data sheet number: SB450 NRIM DATA SHEET NO. 67 (400C/10
3
/sec)

Condition: Test condition:

Heat treatment: As welded. Temperature:_400C


-3
Hardness:____________________________________ Rising load strain rate: 10 /sec
Product form: 1500 mm x 300 mm x 20 mm plate Tensile hold time: None
Falling load strain rate: 10-3/sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 184,000 MPa (26.7 x 10 psi) Strength coefficient, H’: 549 MPa (79,600 psi)
Yield strength, YS: 351 MPa (50,900 psi) Strain hardening exponent, n’: 0.067
Ultimate strength, UTS: 484 MPa (70,200 psi)
Reduction in area, %RA: 70% Fatigue strength coefficient, σf’ : 1142 MPa (165,600 psi)
True fracture ductility, εf: 1.204 Fatigue strength exponent, b : -0.1335
Fatigue ductility coefficient, εf’: 4.8264
Fatigue ductility exponent, c : -0.9255

Comments: Microstructure :
Composition of the weld metal.
Ferrite grain size : Not documented
Chemistry: C Mn P S Si Cr Mo Ni Cu
0.112 1.49 0.022 0.004 0.30 0.019 0.002 0.018 0.48 Inclusions : Not documented

Microstructure : Not documented

Welding Process: Submerged arc welding

Stress-strain curve : Strain amplitude vs. reversals to complete separation into two
parts :

Reference: NRIM FDS No. 67, 1991

F-17
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-14
Materials Characterization Sheet

-
Material: Grade 22 SCMV 4 (2.25CR – 1Mo) Data sheet number: SCMV 4 NRIM DATA SHEET NO. 62 (RT/10
3
/sec)
Condition: Test condition:

Heat treatment: Hot rolled, normalized and tempered Temperature:_RT


-3
Hardness:____________________________________ Rising load strain rate: 10 /sec
Product form: 2400 mm x 9000mm x 37 mm plate Tensile hold time: None
Falling load strain rate: 10-3/sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 208,000 MPa (30.2 x 10 psi) Strength coefficient, H’: 796 MPa (115,400 psi)
Yield strength, YS: 532 MPa (77,200 psi) Strain hardening exponent, n’: 0.100
Ultimate strength, UTS: 684 MPa (99,200 psi)
Reduction in area, %RA: 80% Fatigue strength coefficient, σf’ : 871.01 MPa (126,300 psi)
True fracture strength, σf: 1648 MPa (239,000 psi) Fatigue strength exponent, b : -0.0639
True fracture ductility, εf: 1.609 Fatigue ductility coefficient, εf’:0.8944
Strength coeff., H: 875 MPa (127,000 psi) Fatigue ductility exponent, c : -0.7061
Strain hard. Exponent, n: 0.079

Comments: Microstructure :

Chemistry: C Mn P S Si Cr Mo Sn Nb N Austenite grain size : JIS G 0551(1977) = 5.5


0.13 0.50 0.007 0.009 0.02 2.43 0.96 0.005 0.069 0.071
Ni Al Cu Inclusions : 0.008 area% inclusions deformed by plastic work (sulphide,
0.16 0.012 0.07 silicate, etc.) + no discontinuous arrays (alumina) inclusions + 0.021
area% isolated (granular oxide) inclusions

Processing: Mn-Al killed, Basic electric arc furnace, Hot rolled Microstructure : Not documented
o o o
920 C/1.3hr, air cooled + 670 C/2.3hr, air cooled + 650 C/10.3hr,
furnace cooled

Stress-strain curve : Strain amplitude vs. reversals to cracking curve :

Reference: NRIM FDS No. 62, 1989

F-18
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-15
Materials Characterization Sheet
-
Material: Grade 22 SCMV 4 (2.25CR – 1Mo) Data sheet number: SCMV 4 NRIM DATA SHEET NO. 62 (500C/10
3
/sec)
Condition: Test condition:

Heat treatment: Hot rolled, normalized and tempered Temperature:_500C


-3
Hardness:____________________________________ Rising load strain rate: 10 /sec
Product form: 2400 mm x 9000mm x 37 mm plate Tensile hold time: None
Falling load strain rate: 10-3/sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 174,000 MPa (25.2 x 10 psi) Strength coefficient, H’: 652 MPa (115,400 psi)
Yield strength, YS: 388 MPa (56,300 psi) Strain hardening exponent, n’: 0.105
Ultimate strength, UTS: 498 MPa (72,200 psi)
Reduction in area, %RA: 83% Fatigue strength coefficient, σf’ : 611 MPa (88,500 psi)
True fracture strength, σf: 1,276 MPa (185,100 psi) Fatigue strength exponent, b : -0.0554
True fracture ductility, εf: 1.772 Fatigue ductility coefficient, εf’:1.1292
Strength coeff., H: 819 MPa (119,000 psi) Fatigue ductility exponent, c : -0.7799
Strain hard. Exponent, n: 0.120

Comments: Microstructure :

Chemistry: C Mn P S Si Cr Mo Sn Nb Austenite grain size : JIS G 0551(1977) = 5.5


0.13 0.50 0.007 0.009 0.02 2.43 0.96 0.005 0.069
N Ni Al Cu Inclusions : 0.008 area% inclusions deformed by plastic work (sulphide,
0.071 0.16 0.012 0.07 silicate, etc.) + no discontinuous arrays (alumina) inclusions + 0.021
area% isolated (granular oxide) inclusions

Processing: Mn-Al killed, Basic electric arc furnace, Hot rolled Microstructure : Not documented
o o o
920 C/1.3hr, air cooled + 670 C/2.3hr, air cooled + 650 C/10.3hr,
furnace cooled

Stress-strain curve : Strain amplitude vs. reversals to cracking curve :

Reference: NRIM FDS No. 62, 1989

F-19
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-16
Materials Characterization Sheet
Material: Grade 22 SCMV 4 (2.25CR – 1Mo) Data sheet number: SCMV 4 NRIM DATA SHEET NO. 62
-3
(600C/10 /sec)
Condition: Test condition:

Heat treatment: Hot rolled, normalized and tempered Temperature:_600C


-3
Hardness:____________________________________ Rising load strain rate: 10 /sec
Product form: 2400 mm x 9000mm x 37 mm plate Tensile hold time: None
Falling load strain rate: 10-3/sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 158,000 MPa (22.9 x 10 psi) Strength coefficient, H’: 428 MPa (62,100 psi)
Yield strength, YS: 313 MPa (45,400 psi) Strain hardening exponent, n’: 0.082
Ultimate strength, UTS: 404 MPa (58,600 psi)
Reduction in area, %RA: 88% Fatigue strength coefficient, σf’ : 456 MPa (66,100 psi)
True fracture strength, σf: 1,062 MPa (154,000 psi) Fatigue strength exponent, b : -0.0566
True fracture ductility, εf: 2.12 Fatigue ductility coefficient, εf’:3.1437
Strength coeff., H: 817 MPa (118,500 psi) Fatigue ductility exponent, c : -0.9629
Strain hard. Exponent, n: 0.082

Comments: Microstructure :

Chemistry: C Mn P S Si Cr Mo Sn Nb N Austenite grain size : JIS G 0551(1977) = 5.5


0.13 0.50 0.007 0.009 0.02 2.43 0.96 0.005 0.069 0.071
Ni Al Cu Inclusions : 0.008 area% inclusions deformed by plastic work
0.16 0.012 0.07 (sulphide, silicate, etc.) + no discontinuous arrays (alumina)
inclusions + 0.021 area% isolated (granular oxide) inclusions

Processing: Mn-Al killed, Basic electric arc furnace, Hot rolled Microstructure : Not documented
o o o
920 C/1.3hr, air cooled + 670 C/2.3hr, air cooled + 650 C/10.3hr, furnace
cooled

Stress-strain curve : Strain amplitude vs. reversals to cracking curve :

Reference: NRIM FDS No. 62, 1989

F-20
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-17
Materials Characterization Sheet
-3
Material: ASTM A387 Grade 91 (9Cr-1Mo-V-Nb) Data sheet number: AES-SA387 P91(RT/10 /sec)

Condition: Test condition:

Heat treatment: Hot rolled, normalized and tempered Temperature:_RT


-3
Hardness:____________________________________ Rising load strain rate: 10 /sec
Product form: 2150 mm x 9477mm x 25 mm plate Tensile hold time: None
Falling load strain rate: 10-3/sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 215,000 MPa (31.1x 10 psi) Strength coefficient, H’: 975 MPa (141,400 psi)
Yield strength, YS: 531 MPa (77,000 psi) Strain hardening exponent, n’: 0.117
Ultimate strength, UTS: 682 MPa (98,900 psi)
Reduction in area, %RA: 76% Fatigue strength coefficient, σf’ : 1052 MPa (152,500 psi)
True fracture strength, σf: 1328 MPa (192,600 psi) Fatigue strength exponent, b : -0.0766
True fracture ductility, εf: 1.427 Fatigue ductility coefficient, εf’: 0.5771
Strength coeff., H: 710 MPa (103,000 psi) Fatigue ductility exponent, c : -0.6098
Strain hard. Exponent, n: 0.047

Comments: Microstructure :

Chemistry: C Mn P S Si Cr Mo Ferrite grain size : JIS G 0552(1977) = 11.6


0.08 0.49 0.005 0.004 0.34 8.34 0.89
V Nb N Ni Al Cu Inclusions : 0.013 area% inclusions deformed by plastic
0.23 0.069 0.059 0.09 0.008 0.02 work (sulphide, silicate, etc.) + no discontinuous arrays
(alumina) inclusions + 0.028 area% isolated (granular
oxide) inclusions
Processing: S-Al killed, Basic electric arc furnace, Hot
o o
rolled 1050 C/10min, air cooled + 770 C/1hr, air cooled + Microstructure : Not documented
o
740 C/1hr, furnace cooled

Stress-strain curve : Strain amplitude vs. reversals to cracking curve :

Reference: NRIM FDS No. 78, 1993

F-21
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-18
Materials Characterization Sheet
-3
Material: ASTM A387 Grade 91 (9Cr-1Mo-V-Nb) Data sheet number: AES-SA387 P91(550C/10 /sec)

Condition: Test condition:

Heat treatment: Hot rolled, normalized and tempered Temperature:_550C


-3
Hardness:____________________________________ Rising load strain rate: 10 /sec
Product form: 2150 mm x 9477mm x 25 mm plate Tensile hold time: None
Falling load strain rate: 10-3/sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 172,000 MPa (24.9x 10 psi) Strength coefficient, H’: 609 MPa (88,300 psi)
Yield strength, YS: 346 MPa (50,200 psi) Strain hardening exponent, n’: 0.142
Reduction in area, %RA: 87%
True fracture strength, σf: 1010 MPa (146,500 psi) Fatigue strength coefficient, σf’ : 647.8 MPa (93,950 psi)
True fracture ductility, εf: 2.040 Fatigue strength exponent, b : -0.0979
Strength coeff., H: 482 MPa (69,900 psi) Fatigue ductility coefficient, εf’: 0.1790
Strain hard. Exponent, n: 0.054 Fatigue ductility exponent, c : -0.5140

Comments: Microstructure :

Chemistry: C Mn P S Si Cr Mo V Nb N Ferrite grain size : JIS G 0552(1977) = 11.6


0.08 0.49 0.005 0.004 0.34 8.34 0.89 0.23 0.069 0.059
Ni Al Cu Inclusions : 0.013 area% inclusions deformed by plastic work
0.09 0.008 0.02 (sulphide, silicate, etc.) + no discontinuous arrays (alumina)
inclusions + 0.028 area% isolated (granular oxide) inclusions

Processing: S-Al killed, Basic electric arc furnace, Hot rolled Microstructure : Not documented
o o o
1050 C/10min, air cooled + 770 C/1hr, air cooled + 740 C/1hr, furnace
cooled

Stress-strain curve : Strain amplitude vs. reversals to cracking curve :

Reference: NRIM FDS No. 78, 1993

F-22
EPRI Proprietary Licensed Material

Monotonic and Cyclic Properties of HRSG Pressure Part Materials

Table F-19
Materials Characterization Sheet
-3
Material: ASTM A387 Grade 91 (9Cr-1Mo-V-Nb) Data sheet number: AES-SA387 P91(600C/10 /sec)

Condition: Test condition:

Heat treatment: Hot rolled, normalized and tempered Temperature:_600C


-3
Hardness:____________________________________ Rising load strain rate: 10 /sec
Product form: 2150 mm x 9477mm x 25 mm plate Tensile hold time: None
Falling load strain rate: 10-3/sec
Compressive hold time:None
Waveform: Symmetrical sawtooth

Monotonic properties: Cyclic Properties:


6
Modulus of elasticity, E: 158,000 MPa (22.9x 10 psi) Strength coefficient, H’: 443 MPa (64,300 psi)
Yield strength, YS: 330 MPa (47,900 psi) Strain hardening exponent, n’: 0.121
Ultimate strength, UTS: 413 MPa (59,900 psi)
Reduction in area, %RA: 91% Fatigue strength coefficient, σf’ : 374.4 MPa (54,300 psi)
True fracture strength, σf: 743 MPa (107,800 psi) Fatigue strength exponent, b : -0.0589
True fracture ductility, εf: 2.408 Fatigue ductility coefficient, εf’: 0.5306
Strength coeff., H: 330 MPa (47,900 psi) Fatigue ductility exponent, c : -0.6370
Strain hard. Exponent, n: 0.042

Comments: Microstructure :

Chemistry: C Mn P S Si Cr Mo V Nb Ferrite grain size : JIS G 0552(1977) = 11.6


0.08 0.49 0.005 0.004 0.34 8.34 0.89 0.23 0.059
N Ni Al Cu Inclusions : 0.013 area% inclusions deformed by plastic work
0.069 0.09 0.008 0.02 (sulphide, silicate, etc.) + no discontinuous arrays (alumina) inclusions
+ 0.028 area% isolated (granular oxide) inclusions

Processing: S-Al killed, Basic electric arc furnace, Hot rolled Microstructure : Not documented
o o o
1050 C/10min, air cooled + 770 C/1hr, air cooled + 740 C/1hr,
furnace cooled

Stress-strain curve : Strain amplitude vs. reversals to cracking curve :

Reference: NRIM FDS No. 78, 1993

F-23
Export Control Restrictions The Electric Power Research Institute (EPRI)
Access to and use of EPRI Intellectual Property is granted with The Electric Power Research Institute (EPRI), with major locations in
the specific understanding and requirement that responsibility Palo Alto, California, and Charlotte, North Carolina, was established
for ensuring full compliance with all applicable U.S. and in 1973 as an independent, nonprofit center for public interest
foreign export laws and regulations is being undertaken by energy and environmental research. EPRI brings together members,
you and your company. This includes an obligation to ensure participants, the Institute’s scientists and engineers, and other leading
that any individual receiving access hereunder who is not a experts to work collaboratively on solutions to the challenges of electric
U.S. citizen or permanent U.S. resident is permitted access power. These solutions span nearly every area of electricity generation,
under applicable U.S. and foreign export laws and delivery, and use, including health, safety, and environment. EPRI’s
regulations. In the event you are uncertain whether you or members represent over 90% of the electricity generated in the
your company may lawfully obtain access to this EPRI United States. International participation represents nearly 15% of
Intellectual Property, you acknowledge that it is your EPRI’s total research, development, and demonstration program.
obligation to consult with your company’s legal counsel to
determine whether this access is lawful. Although EPRI may Together...Shaping the Future of Electricity
make available on a case-by-case basis an informal
assessment of the applicable U.S. export classification for
specific EPRI Intellectual Property, you and your company
acknowledge that this assessment is solely for informational
purposes and not for reliance purposes. You and your
company acknowledge that it is still the obligation of you and
your company to make your own assessment of the applicable
U.S. export classification and ensure compliance accordingly.
You and your company understand and acknowledge your
obligations to make a prompt report to EPRI and the
appropriate authorities regarding any access to or use of EPRI
Intellectual Property hereunder that may be in violation of
applicable U.S. or foreign export laws or regulations.

Programs:
© 2006 Electric Power Research Institute (EPRI), Inc. All rights reserved.
Heat Recovery Steam Generator (HRSG) Dependability
Electric Power Research Institute and EPRI are registered service marks of
the Electric Power Research Institute, Inc.

Printed on recycled paper in the United States of America 1010440

ELECTRIC POWER RESEARCH INSTITUTE


3420 Hillview Avenue, Palo Alto, California 94304-1395 • PO Box 10412, Palo Alto, California 94303-0813 USA
800.313.3774 • 650.855.2121 • askepri@epri.com • www.epri.com

You might also like