Embrittlement of Power Plant Steels
2013 TECHNICAL REPORT
13828389
13828389
Embrittlement of Power
Plant Steels
EPRI Project Manager
J. Parker
3420 Hillview Avenue
Palo Alto, CA 94304-1338
USA
PO Box 10412
Palo Alto, CA 94303-0813
USA
800.313.3774
650.855.2121
askepri@epri.com
www.epri.com
3002001474
Final Report, December 2013
13828389
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.
REFERENCE HEREIN TO ANY SPECIFIC COMMERCIAL PRODUCT, PROCESS, OR SERVICE BY ITS TRADE
NAME, TRADEMARK, MANUFACTURER, OR OTHERWISE, DOES NOT NECESSARILY CONSTITUTE OR
IMPLY ITS ENDORSEMENT, RECOMMENDATION, OR FAVORING BY EPRI.
THE FOLLOWING ORGANIZATION PREPARED THIS REPORT:
Electric Power Research Institute (EPRI)
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, EPRI, and TOGETHER…SHAPING THE FUTURE OF ELECTRICITY are
registered service marks of the Electric Power Research Institute, Inc.
Copyright © 2013 Electric Power Research Institute, Inc. All rights reserved.
13828389
Acknowledgments
The following organization prepared this report:
Electric Power Research Institute (EPRI)
1300 West W.T. Harris Blvd.
Charlotte, NC 28262
Principal Investigator
J. Parker
This report describes research sponsored by EPRI.
This publication is a corporate
document that should be cited in the
literature in the following manner:
Embrittlement of Power Plant Steels.
EPRI, Palo Alto, CA: 2013.
3002001474.
 iii 
13828389
13828389
Product
Description
Plant operators seek to adopt approaches that can minimize costs,
prevent forced outages, and maximize safety and reliability. Rigorous
life assessment methodologies have been developed over the years
and are commonly employed to determine component integrity and
life. Such assessments examine key operational characteristics
including: elevated temperature exposure, cycling operation, loading,
environmental exposure, etc., to determine remaining life. Many of
these characteristics can have a profound influence on component
and alloy embrittlement.
Background
Premature failures in power plant equipment are often traced to low
ductility issues associated with various forms of metallurgical and/or
environmental embrittlement. Failures in critical components such as
rotors, high-energy piping, or pressure vessels can result in large
costs, extended downtime, and possible loss of life. Demonstrated
approaches to assess component embrittlement are highly desirable,
particularly in today’s marketplace where plants are more often seeing
cyclic operation or are nearing end of life.
Objectives
 Provide a general metallurgical background for common power
plant alloys and methods of manufacture
 Describe time-dependent metallurgical mechanisms for fossil
components and relevant alloys that result in low ductility-type
failures
 Document how damage develops in components, and describe
methods to assess damage levels
 Provide guidance regarding typical short- and long-term
solutions to embrittlement issues
Approach
This report was generated via assembly and review of numerous case
histories documenting industry failures associated with
embrittlement issues. Specific embrittlement phenomena
surrounding each failure were identified, and comprehensive
discussions of each type of embrittlement were developed.
Background information, damage mechanisms, case histories, and
v
13828389
solutions to assess all forms of embrittlement were developed.
Component assessment approaches are provided to address both
critical (those components with a significant influence on safety
and/or a large financial impact) and non-critical components.
Results
In engineering alloys, various metallurgical changes such as temper
embrittlement, phase changes, and formation and growth of
precipitates can significantly enhance brittle-type behavior. Also,
various forms of environmental embrittlement including: liquid metal
embrittlement, oxygen embrittlement, hydrogen embrittlement, and
stress corrosion cracking can influence brittle behavior. This
document examines these phenomena and provides specific solutions
to avoid failures in the future. Methods are provided to accurately
assess the current component ductility (or lack thereof) through
metallurgical and mechanical evaluation methods.
Applications, Value, and Use
Volumes of information on various forms of embrittlement have been
generated over the past 50 years by power producers, universities,
vendors, and various research organizations. This report assembles
key aspects of embrittlement information into one concise document
that specifically addresses fossil power plant components and
operation. It provides power producers with background information
on various embrittlement phenomena, discusses mechanisms of
damage, and gives straightforward guidance to assess embrittlement.
No other EPRI document provides power producers with the
knowledge and tools to individually assess different forms of low
ductility failures and characterize embrittlement issues.
Keywords
Cracking
Defect assessment
Embrittlement
Fracture
Metallurgy
 vi 
13828389
Abstract
A key lesson arising from the Task Group on Brittle Failure of Steel
Forgings sums up the need to consider the balance between strength
and toughness when considering component behavior. The
statement made with respect to the need for balance in the
prevention of brittle fracture was:
The fact is that we have been overlooking ductility and
notch toughness to favor strength, and we had better
consider all factors.
Improved understanding of factors that influence low-ductility
fracture is aiding the process of risk reduction. However, many steels
exhibit time-dependent embrittlement due to the presence of socalled “trace elements.” This report summarizes the primary reasons
for brittle behavior and presents solutions to minimize the risks of
catastrophic failure. Specific case studies describe the lessons learned
from previous fracture incidents in boiler and turbine components.
 vii 
13828389
13828389
Executive
Summary
It is generally recognized that different engineering alloys have
different strengths. However, it is not always appreciated that the
fracture behavior of a particular alloy will vary depending on specific
circumstances. The factors involved in establishing fracture behavior
include:

The operating temperature. For example, many steels will exhibit
ductile, high-energy fracture at high temperature and brittle,
low-energy fracture at lower temperatures.

The microstructure of the material, particularly the grain size,
the presence and distribution of alloying elements and secondphase particles, and the level of trace elements. Because many
components in power generating plants operate at temperatures
where the metallurgical condition can change with time in
service, changes in microstructure can lead to increased
susceptibility for brittle (or at least low-ductility failures).

The local stress, which will be affected by the local geometry and
loading, as well as the presence of cracks or notches, which will
act as stress concentrators.
The specific energies associated with the different modes of failure
and the temperature where the transition from brittle to ductile
behavior occurs are obviously critical parameters.
It is generally the case that the risk of sudden brittle fracture
increases for materials operating where the fracture energy is in the
lower shelf regime since under these conditions the material is most
susceptible to brittle failure. However, there are other circumstances
where rapid fracture can occur. For example:

When the environment has introduced or accelerated cracking,
for example, due to intergranular corrosion, stress corrosion, or
liquid metal embrittlement.

When the upper shelf energy, that is, the energy associated with
the higher energy ductile mode, in combination with high
operating stresses leads to the critical crack size being exceeded
either because of pre-existing fabrication flaws and/or in service
cracking.
It should be apparent then that when rapid low-energy fracture
occurs in a service component, several factors must be acting
together, for example, a defect must be present at a location where
 ix 
13828389
the stress is high enough to overcome the material’s fracture
resistance or toughness. In general, experience suggests that the
number of instances where the necessary combination of
circumstances required for rapid brittle fracture to occur is small.
However, when fractures of this type have occurred, the
consequences can be catastrophic. The present guideline document
reviews key information regarding the factors involved in causing and
preventing low-ductility failures.
Specific sections in this report are as follows:

Introduction, covering the background of fracture behavior and
the assessment of a critical defect size

Testing Methods, including mechanical test techniques to
measure materials properties, as well as small specimen and
metallographic approaches that have been developed specifically
to aid with assessment of components

Metallurgy of Steels, which outlines the interrelationship
between microstructure and properties for traditional alloys and
the newer steels being introduced in modern plants

The Influence of Metallurgical Changes, covering the
susceptibility for brittle fracture, with specialist sections
describing:

-
Phase changes
-
The effect of carbides
-
Temper embrittlement
The Influence of the Environment, considering the particular
effects in causing low-ductility failures of:
-
Oxygen embrittlement
-
Liquid metal embrittlement
-
Cracking due to corrosion

Hydrogen Cracking, Creep Deformation, and Fracture, which
summarizes the factors that lead to lead to brittle failures under
conditions of high stress and temperature

Component Assessment, which provides an outline of the key
issues associated with evaluating the serviceability of the plant

Creep Fracture
Each section provides key background information and guidance
regarding the way particular methods can be used to prevent failures.
References to relevant documents are provided to facilitate further
individual study as necessary.
 x 
13828389
Table of Contents
Section 1: Introduction ............................................1-1
1.1 Background ............................................................... 1-1
1.2 Fracture of Materials .................................................. 1-4
1.2.1 Ductile Fracture ................................................. 1-6
1.2.2 Brittle Fracture ................................................... 1-7
1.2.3 The Brittle – Ductile Transition .............................. 1-9
1.3 Crack Propagation ................................................... 1-12
1.4 Fracture Toughness................................................... 1-14
1.5 Summary................................................................. 1-15
1.6 References............................................................... 1-16
Section 2: Testing Methods ......................................2-1
2.1 Introduction ............................................................... 2-1
2.2 Standard Mechanical Tests ......................................... 2-2
2.3 Assessment of Fracture Toughness ................................ 2-4
2.3.1 Charpy Impact Testing ....................................... 2-4
2.3.2 Charpy Correlations with Fracture Toughness...... 2-13
2.4 Small Punch Testing .................................................. 2-19
2.4.1 Description of Small Punch Test and Results
Analysis................................................................... 2-20
2.4.2 Estimation of Tensile Properties .......................... 2-21
2.4.3 Small Punch Test Assessment of FATT ................. 2-23
2.4.4 Small Punch Test Assessment of Fracture
Toughness ................................................................ 2-28
2.4.5 Creep Embrittlement ......................................... 2-30
2.5 Metallographic Techniques ....................................... 2-31
2.5.1 Optical Microscopy ......................................... 2-32
2.5.2 Grain Size Measurements................................. 2-33
2.5.3 Specialist Etching for Phase Identification ........... 2-36
2.5.4 Assessment of Phosphorus Segregation .............. 2-37
2.5.5 Preparation and Etching to Reveal Creep
Microvoids ............................................................... 2-42
2.5.6 Electron Microscopy......................................... 2-45
2.6 Concluding Comments .............................................. 2-50
2.7 References............................................................... 2-51
 xi 
13828389
Section 3: Metallurgy of Steels ................................3-1
3.1 Introduction ............................................................... 3-1
3.2 Background ............................................................... 3-1
3.3 Non-Equilibrium Cooling of Steels ................................ 3-4
3.4 Continuous Cooling Transformation.............................. 3-5
3.5 Effects of Composition ................................................ 3-8
3.6 Classification of Steels .............................................. 3-11
3.7 Power Plant Steels .................................................... 3-12
3.7.1 Ferritic Boiler Steels.......................................... 3-14
3.7.2 Ferritic Turbine Steels ....................................... 3-15
3.7.3 Austenitic Boiler Steels...................................... 3-15
3.8 References............................................................... 3-16
Section 4: The Influence of Metallurgical Changes
on Brittleness ..........................................4-1
4.1 Introduction ............................................................... 4-1
Section 5: Embrittlement Due to Phase Changes .......5-1
5.1 Introduction ............................................................... 5-1
5.2 Graphitization in C – Mn and C – Mo Steels ................ 5-1
5.2.1 Growth Kinetics of Graphitization........................ 5-4
5.2.2 Case Study/Example ......................................... 5-7
5.3 Embrittlement in Stainless Steels ................................... 5-7
5.3.1 Brittleness Due to Secondary Hardening ............... 5-7
5.3.2 475°C Embrittlement .......................................... 5-8
5.3.3 Embrittlement and Grain Size.............................. 5-9
5.3.4 Sigma Phase Formation ...................................... 5-9
5.4 Assessment of Components ....................................... 5-17
5.5 References............................................................... 5-18
Section 6: The Effect of Carbides on Embrittlement....6-1
6.1 Introduction ............................................................... 6-1
6.2 The Effect of Carbon on Fracture Behavior .................... 6-1
6.3 Tempered Martensite Embrittlement (TME) ..................... 6-5
6.4 Thermal Embrittlement ................................................. 6-6
6.5 Carbides in CrMo Low Alloy Steels .............................. 6-7
6.6 Dissimilar Metal Welds ............................................. 6-10
6.7 Sensitization of Austenitic Steels ................................ 6-12
6.8 Assessment of Components ....................................... 6-13
6.9 References............................................................... 6-14
 xii 
13828389
Section 7: Temper Embrittlement of Steels ................7-1
7.1 Introduction ............................................................... 7-1
7.2 Mechanisms Related to Temper Embrittlement ................ 7-2
7.3 Factors Affecting Temper Embrittlement ......................... 7-6
7.4 Relationships to Describe Metallurgical Effects on
Temper Embrittlement ..................................................... 7-12
7.5 Equations Used to Predict Temper Embrittlement .......... 7-14
7.6 Case Studies/Examples ............................................ 7-19
7.6.1 Assessment of Components ............................... 7-19
7.7 References............................................................... 7-20
Section 8: Embrittlement Influenced by the
Environment............................................8-1
8.1 Introduction ............................................................... 8-1
8.2 Oxygen Embrittlement ................................................ 8-2
8.2.1 Introduction ....................................................... 8-2
8.2.2 Mechanisms ...................................................... 8-3
8.3 Liquid Metal Embrittlement .......................................... 8-4
8.3.1 Introduction ....................................................... 8-4
8.3.2 Mechanism of Liquid Metal Embrittlement ............. 8-6
8.3.3 Factors Affecting Liquid Metal Embrittlement ......... 8-8
8.3.4 Case Studies/Examples ...................................... 8-9
8.4 Cracking Due To Corrosion ....................................... 8-11
8.4.1 Introduction ..................................................... 8-11
8.4.2 Mechanism ..................................................... 8-13
8.4.3 Examples of Alloy/Environmental Systems .......... 8-15
8.4.4 Examples of Power Plant Related Damage .......... 8-17
8.5 Assessment of Components ....................................... 8-19
8.6 References............................................................... 8-20
Section 9: Hydrogen Embrittlement ..........................9-1
9.1 Introduction ............................................................... 9-1
9.2 Mechanisms of Hydrogen Damage .............................. 9-2
9.3 Factors Affecting Hydrogen Embrittlement of Ferritic
Type Steels ...................................................................... 9-4
9.4 Damage Development ................................................ 9-7
9.4.1 Hydrogen Cracking of Welds ............................. 9-7
9.4.2 Hydrogen Damage in Boiler Tubing ................... 9-12
9.5 Case Studies/Examples ............................................ 9-15
9.6 Assessment of Components ....................................... 9-15
9.7 References............................................................... 9-16
 xiii 
13828389
Section 10:
Creep Fracture .................................. 10-1
10.1 Introduction ........................................................... 10-1
10.2 Background ........................................................... 10-1
10.3 Mechanisms .......................................................... 10-3
10.4 Factors Affecting Creep Fracture .............................. 10-7
10.5 Creep Damage in 9 to 12% Cr Martensitic Steels ...... 10-8
10.5.1 Introduction ................................................... 10-8
10.5.2 Factors Affecting the Formation of Creep
Cavities ................................................................. 10-10
10.6 Case Studies/Examples ........................................ 10-22
10.6.1 Creep of Thick Section Weldments ................ 10-22
10.6.2 Tubing ........................................................ 10-27
10.6.3 Dissimilar Metal Welds ................................ 10-29
10.7 References........................................................... 10-31
Section 11:
Summary of Component
Assessment Issues ................................. 11-1
11.1 Fracture Assessment Summary ................................. 11-7
11.2 References............................................................. 11-8
Appendix A: Glossary of Metallurgical Terms ......... A-1
Appendix B: Case Study: Embrittlement in Alloy
80A Fasteners .........................................B-1
B.1 Introduction ............................................................... B-1
B.2 Factors Affecting Life .................................................. B-1
B.3 References ................................................................. B-6
Appendix C: Case Study–Brittle Failure of Ferritic
Steel Bolts ............................................... C-1
C.1 Introduction .............................................................. C-1
C.2 Key Issues ................................................................ C-5
C.3 References ............................................................... C-5
Appendix D: Case Study–Review of Cracking,
Eddystone Unit 1 .................................... D-1
D.1 Introduction .............................................................. D-1
D.2 Design and Operation ............................................... D-1
D.3 Summary of Piping Damage....................................... D-5
D.4 Previous Damage ...................................................... D-7
D.5 Concluding Remarks ................................................. D-9
D.6 References ............................................................. D-10
 xiv 
13828389
Appendix E: Case Study–Cracking in a CrMoV
Weld ...................................................... E-1
E.1 Introduction ............................................................... E-1
E.2 Damage Detected ...................................................... E-1
E.3 Metallurgical Evaluation .............................................. E-2
E.4 Concluding Remarks ................................................... E-4
E.5 Reference .................................................................. E-5
Appendix F: Case Study–Gallatin Unit 2, IP-LP
Single Flow Rotor Failure ......................... F-1
F.1 Introduction................................................................ F-1
F.2 Background ............................................................... F-1
F.3 Damage Evaluation .................................................... F-2
F.4 References ................................................................. F-4
Appendix G: Case Study–Hinkley Point Disc and
Rotor Failure .......................................... G-1
G.1 Introduction ............................................................. G-1
G.2 Developments for Improved Rotor Toughness ............... G-3
G.3 References............................................................... G-5
Appendix H: Case Study Failure Due to
Graphitization in a Carbon-½ Mo
Steel Steam Pipe ................................... H-1
H.1 Introduction ............................................................. H-1
H.2 System History ........................................................ H-1
H.3 Results .................................................................... H-3
H.4 Concluding Remarks................................................ H-5
H.5 Reference................................................................. H-5
 xv 
13828389
13828389
List of Figures
Figure 1-1 Brittle fracture of a steel pressure vessel caused
by hydrostatically testing using cold water .......................... 1-4
Figure 1-2 Fracture map for 2 1/4Cr1Mo low alloy steel ........... 1-5
Figure 1-3 Fracture map for 316 stainless steel ......................... 1-5
Figure 1-4 Fractures observed in laboratory tensile tests
showing (a) ductile fracture and (b) brittle fracture ............... 1-7
Figure 1-5 Detail of the fracture surface associated with (a)
ductile fracture and (b) brittle fracture ................................. 1-7
Figure 1-6 Schematic illustration showing how the transition
from brittle to ductile fracture depends on the yield and
fracture stresses................................................................ 1-9
Figure 1-7 Schematic representation of how an embrittling
event will increase the brittle/ductile transition
temperature ................................................................... 1-10
Figure 1-8 Schematic illustration showing how changes in
grain size modify the yield stress and the fracture stress
and hence change the brittle to ductile transition
temperature ................................................................... 1-12
Figure 2-1 Diagram showing the main features and
operation of a Charpy impact test machine ......................... 2-5
Figure 2-2 Dimensions of a standard Charpy impact
specimen, with detail of the specimen support region of
the test machine ............................................................... 2-6
Figure 2-3 Schematic diagram illustrating the variation of
Charpy absorbed energy with test temperature.................... 2-8
Figure 2-4 Charpy fracture energy measurements for
21/4Cr1.6WVNb steel from tests at different
temperatures.................................................................. 2-10
Figure 2-5 Charpy transition curve for low alloy steel with
typical levels of trace elements ......................................... 2-10
 xvii 
13828389
Figure 2-6 Charpy transition curves for 21/4Cr1Mo steel
for normal composition and for an alloy doped with
embrittling trace elements such as P before and after
aging at high temperature ............................................... 2-11
Figure 2-7 Histograms showing the variation in fracture
energy measured using 2 types of testing machine for
multiple tests on 4340 steel for 3 different heat
treatments ..................................................................... 2-12
Figure 2-8 Schematic illustration of a compact tension
specimen used to measure fracture toughness .................... 2-14
Figure 2-9 Correlation between KIc and the upper shelf
Charpy energy using the Rolfe – Novak equation .............. 2-17
Figure 2-10 Correlation between KIc and the upper shelf
Charpy energy using the Iwadate-Karushi-Watanabe
equation ....................................................................... 2-18
Figure 2-11 The master curve relationship between KIc/KIc-US
and excess temperature for CrMo low alloy steels .............. 2-19
Figure 2-12 Typical small sample machined from an inservice component, and miniature specimens shown
before and after laboratory testing ................................... 2-20
Figure 2-13 Schematic cross sectional diagram of the punch
and die test equipment ................................................... 2-21
Figure 2-14 Schematic diagram showing the punch test
apparatus with the borescope system ............................... 2-22
Figure 2-15 Comparison of predicted tensile strengths made
using equation 2-4 with measured values .......................... 2-23
Figure 2-16 Brittle/ductile transition curves for 2¼Cr1Mo
low alloy steel measured using small punch tests, curve
(left), and standard Charpy impact tests, curve (right) ......... 2-24
Figure 2-17 Correlation developed between the transition
temperature measured in small punch tests and the FATT
measured in Charpy tests for CrMoV low alloy steel
forgings ........................................................................ 2-24
Figure 2-18 Correlation developed between the transition
temperature measured in small punch tests and the FATT
measured in Charpy tests for NiCrMoV LP rotor steel
forgings ........................................................................ 2-25
 xviii 
13828389
Figure 2-19 Correlation developed between the transition
temperatures measured in small punch tests and the
FATT measured in Charpy tests for CrMo low alloy
steels. The dashed lines bound the data scatter and the
solid line is the best estimate FATT correlation based on
results for a range of low alloy steels. ............................... 2-26
Figure 2-20 Relationship between FATT measured in Charpy
impact tests and Tsp, the transition temperature measured
using punch tests for CrMoV bolting steels showing the
influence of grain size on the level of embrittlement
occurring ...................................................................... 2-28
Figure 2-21 Small punch test based K1c values compared
with measurements made using standard ASTM
procedures for typical power plant steels. ......................... 2-30
Figure 2-22 Small punch creep tests on new and creep
damaged CrMoV rotor steel. The punch tests accurately
determine the level of damage present ............................. 2-31
Figure 2-23 Ferrite grains revealed in low carbon steel using
a nital etch .................................................................... 2-33
Figure 2-24 Prior austenite grain structure revealed in
bainitic CrMoV low alloy steel using a saturated picric
acid etch ....................................................................... 2-33
Figure 2-25 Standard ASTM grain size charts for the
classification of steels at 100 times .................................. 2-34
Figure 2-26 A service degraded Type 304H stainless steel
tube sample showing stained sigma phase particles with
fully developed microvoids. Arrow in (A) marks sigma.
Arrow in (B) marks a carbide. (MAG: 1000X, Vilella’s
Etch plus (A) NaOH and (B) KOH electrolytic etch). ........... 2-37
Figure 2-27 Schematic illustration of the relationship of the
hardness indent to the etch depth of the grain
boundaries .................................................................... 2-38
Figure 2-28 Example of the iterative polishing process used
to measure the depth of attack at prior austenite grain
boundaries in 17-4PH martensitic stainless steel. The
hardness indent is reduced in size as the material is
polished away, with specific measured depths indicated
by the increasing values of h. .......................................... 2-39
 xix 
13828389
Figure 2-29 Linear relationships between the depth of grain
boundary etch and phosphorus segregation for (a)
NiCrMoV rotor steels and (b) 17-4 PH martensitic
stainless steel ................................................................. 2-40
Figure 2-30 Relationship between the depth of phosphoric
acid etch depth and ∆FATT for CrMoV rotor steels ............. 2-42
Figure 2-31 Micrographs of the same section of service
degraded Type 304H stainless steel tube sample
showing (A) small voids in the as-polished condition,
(B) outlined second phase particles with some
microvoids, and (C) fully developed microvoids (black
cavities). Arrows mark the same location (A) as
polished. (B) 1 minute etch. (C) Multiple 3, 3, and 2
minute etches. (MAG: 500X, Vilella’s etch). ...................... 2-44
Figure 2-32 Energy dispersive spectra from a Type 304H
stainless steel tube showing the composition of the
austenite matrix (a), and a sigma phase particle (b).
Note the high chromium/iron (Cr/Fe) ratio of the sigma
phase compared to the austenite matrix. ........................... 2-47
Figure 2-33 A scanning electron micrograph showing the
brittle intergranular fracture of an ex-service CrMoV bolt
(a) with AES results from a grain boundary facet
showing the high levels of P present which has
embrittled the microstructure (b) ....................................... 2-49
Figure 2-34 Scanning electron micrograph showing detail
of an intergranular fracture surface (a), and an AES
surface analysis showing that the particles highlighted
on this surface contained high levels of Sb and Cr. In
this image the background shows a general level of
iron (b). ........................................................................ 2-50
Figure 3-1 The iron carbon equilibrium diagram, which
shows how the phases present change with temperature
and carbon composition ................................................... 3-2
Figure 3-2 Detail of the iron carbon diagram illustrating
microstructures formed during equilibrium cooling................ 3-3
Figure 3-3 Illustration of the dimensional changes that occur
on heating and cooling through the temperature range
where microstructural transformations take place ................. 3-6
 xx 
13828389
Figure 3-4 CCT diagram for carbon steel (a) and for
2¼Cr1Mo steel (b) ........................................................... 3-7
Figure 3-5 Typical weld microstructures in CrMo low alloy
steel shown in a macrosection (a), with detail of typical
microstructures in the weld metal (b), and heat affected
zone (c) .......................................................................... 3-8
Figure 3-6 Background regarding the development of power
plant steels .................................................................... 3-13
Figure 3-7 Variation in strength and ductility for new 9 and
12%Cr steels as a function of C + N and chromium
equivalent ..................................................................... 3-14
Figure 5-1 The influence of time and temperature on the
formation of graphite (based on 5.1) ................................. 5-2
Figure 5-2 Formation of graphite bands in a reheater tube ......... 5-3
Figure 5-3 Micrograph from a carbon steel weld showing a
moderate level of “eye brow” graphite in a band
adjacent to the HAZ ......................................................... 5-4
Figure 5-4 Power law approximation of the sigmoidal
growth behavior of graphite .............................................. 5-5
Figure 5-5 Time temperature transformation curves for
graphitization in C, C – Si and C – Mo steels ..................... 5-6
Figure 5-6 Brittle behavior in 12% Cr martensitic steels as a
result of secondary hardening............................................ 5-8
Figure 5-7 Increase in the Charpy FATT with increase in
grain size in ferritic stainless steel ...................................... 5-9
Figure 5-8 Iron – chromium-nickel equilibrium phase
diagram (section at 8% nickel). The two phases that are
relevant to austenitic stainless steels are Austenite
(Gamma Iron, γ + Carbon,) and Sigma Phase, σ (a
grain boundary phase comprised of approximately 50%
chromium and 50% iron). The addition of carbon will
expand the region of stability of Gamma Iron, γ-Fe.
Note that even without the benefit of carbon additions
Sigma Phase is an equilibrium phase for chromium
levels above approximately 18%. .................................... 5-10
 xxi 
13828389
Figure 5-9 Time-temperature-transformation curves for Types
304H, 321H, and 347H materials. Note that even the
stabilized grades of material will sensitize and form
sigma phase if they are exposed to prolonged
temperatures approaching 600°C (1112°F). At 650°C
(1202°F) all three alloys will begin to form sigma phase
after approximately 10,000 hrs. ...................................... 5-11
Figure 5-10 Decrease in creep elongation with the presence
of sigma phase .............................................................. 5-12
Figure 5-11 Schaeffler diagram showing how the
microstructure of austenitic steel welds depends on
nickel and chromium equivalent ....................................... 5-13
Figure 5-12 Brittle creep failures due to ferrite/sigma phase ..... 5-15
Figure 5-13 Room temperature Charpy values for E-308
weld metal after aging at 1100°F (593°C) ....................... 5-16
Figure 5-14 Variation in normalised impact value with time
temperature parameter, P, for a range of stainless steel
weld metals ................................................................... 5-17
Figure 6-1 The effect of increasing carbon content on
Charpy impact behavior, FATT from –50°C to +150°C ....... 6-2
Figure 6-2 The influence of carbide thickness on the
ductile/brittle transition temperature in carbon steels ............ 6-3
Figure 6-3 Effect of grain size and carbide thickness on the
temperature where the Charpy fracture energy is 27 J.......... 6-4
Figure 6-4 Increase in the value of FATT from martensitic,
bainitic to pearlitic steels all with a carbon content of
0.25% ............................................................................ 6-5
Figure 6-5 Time temperature transformation diagram
illustrating the thermal treatment likely to produced
tempered martensite embrittlement, line, compared with
thermal treatments likely to produce temper
embrittlement, lines 2 and 3 .............................................. 6-6
Figure 6-6 Typical distribution of carbides in CrMo low alloy
steel after long term service at around 550°C...................... 6-7
Figure 6-7 Reductions in hardness in CrMo steels as a
function of time at temperature........................................... 6-8
Figure 6-8 Change in FATT with mean carbide size for
21/4CrMo steel .............................................................. 6-9
 xxii 
13828389
Figure 6-9 Charpy impact transition curves for 21/4CrMo
steel prior to service, after laboratory aging and after
prolonged service at 550°C ............................................ 6-10
Figure 6-10 The development of carbides at the weld/HAZ
interface in P22 – austenitic stainless steel transition
weld manufactured with a nickel based weld metal.
Type I carbides shown in (a) and (b), with Type II
carbides shown in (c) ..................................................... 6-11
Figure 6-11 Growth behavior of Type I carbides at the
interface of dissimilar metal welds fabricated between 2
1/4CrMo and austenitic stainless steel using a nickel
based filler metal ........................................................... 6-12
Figure 6-12 Temperature – time relationships related to the
formation of grain boundary carbides in austenitic steels
[6.11]. Note that with increased levels of dissolved
carbon the rate and temperature range over which
sensitization occurs increases. ......................................... 6-13
Figure 7-1 Dependence of the grain boundary concentration
of phosphorus on annealing temperature, for Fe-P alloys
with different P levels ........................................................ 7-3
Figure 7-2 Grain boundary concentration of P and C in Fe –
0.17%P alloys with different carbon contents ...................... 7-4
Figure 7-3 Effects of carbon and chromium on the grain
boundary segregation of P after annealing at different
temperatures in the range 400°C to 800°C for Fe – P,
Fe – Cr – P, Fe – C –P and Fe – Cr – C – P alloys with
about the same bulk concentration of P ............................... 7-5
Figure 7-4 C – curve behavior between temperature and
time for 21/4Cr1Mo steel, showing isothermal ΔFATT
contours .......................................................................... 7-6
Figure 7-5 Typical results for 3 rotor steels ................................ 7-7
Figure 7-6 Grain boundary segregation of Sn in Fe – 0.2%
Sn alloy .......................................................................... 7-7
Figure 7-7 Grain boundary segregation in Fe –Sn – C alloys
as a function of the bulk carbon concentration at 550°C ...... 7-8
 xxiii 
13828389
Figure 7-8 Reanalysis of data of Bruscato [7.5] showing that
increases of Mn, Si and P reduced toughness and
increased levels of Mo improved toughness. No
significant trends in toughness were found for the other
elements present .............................................................. 7-9
Figure 7-9 Variation of ΔFATT with time of aging at 850°F
for CrMoV rotor steel ...................................................... 7-10
Figure 7-10 AES measurements show that high levels of S,
P, and Sb segregated to grain boundaries fall rapidly
with distance away from the boundary ............................. 7-11
Figure 7-11 Variation of FATT with prior austenite grain size
at fixed hardness and impurity levels ................................ 7-12
Figure 7-12 Reduction in the level of trace elements with
time for 21/4Cr1Mo steel components ............................ 7-13
Figure 7-13 Correlation between measure values of FATT
with estimates calculated using equation 7-5 for
NiCrMoV steel............................................................... 7-16
Figure 7-14 Variation of post exposure FATT with the
phosphorus content of the 1Cr1Mo1/4V rotor steel ........... 7-17
Figure 8-1 Ductility of alloy IN 903A as a function of
temperature for in-vacuum tests. Samples were tested
after air and vacuum exposures at 1000°C.
Embrittlement remained in the samples exposed to air
after machining the samples to half diameter prior to
testing............................................................................. 8-3
Figure 8-2 Unetched microstructure of nickel samples
following air testing under the same conditions at
800°C. (a) Pure condition unloaded after 500 hours
with minor cavitation, and (b) embrittled condition which
failed after 23 hours. ........................................................ 8-4
Figure 8-3 Example of an intergranular liquid metal fracture
in alloy steel .................................................................... 8-5
Figure 8-4 The effect of temperature on the reduction in area
of Fe-35% Ni alloy samples in the presence of copper ......... 8-6
Figure 8-5 Micrograph showing CrMo steel weld metal with
liquid metal embrittlement due to copper attack at prior
austenite grain boundaries .............................................. 8-10
 xxiv 
13828389
Figure 8-6 Brittle fracture behavior of 12%Cr martensitic
steel that occured under tensile loading at 680°C when
cadmium containing braze was present (a) compared to
ductile behavior under the same conditions without the
braze (b) ....................................................................... 8-11
Figure 8-7 Typical examples of intergranular corrosion
shown by optical metallography and scanning electron
microscopy.................................................................... 8-12
Figure 8-8 Typical micrographs showing stress corrosion
cracking which is (a) intergranular and (b) transgranular .... 8-13
Figure 8-9 Stress corrosion crack velocity as a function of
stress intensity factor ....................................................... 8-15
Figure 8-10 Effect of low concentrations of arsenic,
phosphorus, antimony, and silicon on the time-to-fracture
of copper by SCC .......................................................... 8-16
Figure 8-11 Failure of a stainless steel bellows by SCC (a),
and detail of the microcracking present (b) ....................... 8-17
Figure 9-1 The normal ductility of steel (a), is severely
reduced when hydrogen is present (b). Failure occurred
with the initiation of multiple microcracks (c)........................ 9-1
Figure 9-2 Effect of hydrogen on yield strength and ductility
of Ti6Al4V ...................................................................... 9-3
Figure 9-3 Appearance of 304 stainless steel showing the
intergranular fracture induced by hydrogen ........................ 9-4
Figure 9-4 Influence of local strain and Mn content on the
release of hydrogen ......................................................... 9-5
Figure 9-5 Intergranular fracture in high strength steel
induced by hydrogen and segregation of trace elements.
When compared to Figure 9-3 the grain facets are
relatively clean with little evidence of local dimples. ............. 9-6
Figure 9-6 llustration of severe embrittlement caused by the
presence of hydrogen and how holding at elevated
temperature will restore ductility ......................................... 9-7
Figure 9-7 Susceptibility to cracking in duplex stainless steel
welds as a function of hydrogen content and ferrite
volume fraction ................................................................ 9-8
Figure 9-8 Diffusion coefficient of hydrogen in steels as a
function of temperature ..................................................... 9-9
 xxv 
13828389
Figure 9-9 Micrograph showing a hydrogen induced crack
in a thick section carbon manganese steel weld. The
cracking appeared to initiate from the unfused region at
the root. ........................................................................ 9-10
Figure 9-10 Micrograph showing a hydrogen crack initiated
in the HAZ at the weld root, which extends into the weld
metal ............................................................................ 9-10
Figure 9-11 Schematic diagram illustrating the generation of
hydrogen in an electrochemical cell ................................. 9-13
Figure 9-12 Micrograph showing the fissuring which
develops due to hydrogen attack in carbon steel tubing...... 9-14
Figure 9-13 Micrographs showing increasing levels of
decarburisation and hydrogen damage, samples etched
in 50% solution of hot hydrochloric acid to reveal the
damage ........................................................................ 9-14
Figure 9-14 Hydrogen induced cracking in the HAZ of an
alloy steel weld .............................................................. 9-15
Figure 10-1 Schematic diagram showing the typical creep
strain : time behavior and identifying the three stages of
creep behavior .............................................................. 10-2
Figure 10-2 Time dependent creep failure of a pipe bend.
Note that although the final very rapid fracture event
causes significant opening the damage leading to crack
initiation occurred without obvious deformation ................. 10-3
Figure 10-3 Linear inverse relationship between minimum
creep rate and time to rupture ......................................... 10-4
Figure 10-4 Micrographs showing wedge type cracking
typical of intergranular creep at relatively high stress (a),
and cavitation developed at relatively low stresses (b) ........ 10-5
Figure 10-5 Effect of aluminum on reduction of area for
creep tests at 1100oF on samples of CrMoV rotor steels ..... 10-5
Figure 10-6 Variation in reduction of area with creep
rupture life for CrMoV rotor steel...................................... 10-6
Figure 10-7 Typical micrographs showing intergranular
fracture following the development of grain boundary
creep voids. .................................................................. 10-6
 xxvi 
13828389
Figure 10-8 Variation of rupture life and failure mechanism
with stress and temperature for Type 304 austenitic
stainless steel ................................................................. 10-7
Figure 10-9 Variation in reduction of area with stress and
temperature for CrMoV rotor steel .................................... 10-8
Figure 10-10 Relationships between reduction in area and
creep life for steel grades P91, E911 and P92 tested at
600oC ........................................................................ 10-10
Figure 10-11 Creep strength and ductility for samples at
550oC ........................................................................ 10-11
Figure 10-12 Creep damage detected at different locations
along the gauge length of a sample tested at 550oC ........ 10-12
Figure 10-13 Relationship between the cavity density and
creep strain for tests performed on X20 steel samples ....... 10-13
Figure 10-14 Micrograph showing creep voids developed
in Grade 91 steel (a), an elemental map of the same
area showing local concentrations of oxygen(b)and an
elemental map of the same area showing local
concentrations of silicon (c) ........................................... 10-14
Figure 10-15 Relationships between reduction of area and
creep rupture life for Grade 91 steel samples with
different levels of ‘trace elements’ [10.9]. Some of the
trace elements are not normally controlled in applicable
component specifications even though elements such as
tin (Sn), antimony (Sb) and copper (Cu) can significantly
reduce the creep ductility. ............................................. 10-15
Figure 10-16 Variation in reduction of area for different test
temperatures and creep rupture lives for Grade 92 steel
base metal samples ...................................................... 10-16
Figure 10-17 Typical micrograph showing creep voids in a
Grade 92 steel base metal sample (a) and the number
density of voids present along the gauge length for
samples tested to failure at 9,037, 10,682 and 19,124
hours at 650oC (b) ....................................................... 10-17
Figure 10-18 An example of a single SEM cross-section slice
taken in sample 600-A 6 mm away from fracture surface
(a). A reconstruction of the data showing the individual
creep voids (shown in blue, purple and green) and
associated particle (shown in red) in 3D. ........................ 10-18
 xxvii 
13828389
Figure 10-19 The influence of temperature on dissolution of
BN inclusions............................................................... 10-20
Figure 10-20 Presence of BN inclusions in 9 to 12%Cr steels
as a function of the concentration of boron and nitrogen ..10-21
Figure 10-21 Relationship established between total boron
and boron available for improving creep performance
(as indicated by the amount of soluble boron) for 9% Cr
steels .......................................................................... 10-22
Figure 10-22 An example of Type IIIa cracking developed
in thick section piping welds(a), with detail showing
subsurface crack initiation,(b) ........................................ 10-23
Figure 10-23 An example of Type IV cracking developed in
a thick section piping weld (a) with detail showing sub
surface creep cavitation and crack initiation (b) ............... 10-24
Figure 10-24 An example of a seam welded component that
leaked [10.16] (a), and an example of a seam welded
hot reheat pipe that ruptured in service (b) ...................... 10-24
Figure 10-25 A ‘U’ groove seam weld with detail of
subsurface creep damage. This damage has developed
in the intercritical region of the HAZ which is the
location where Type IV cracking occurs in girth welds ...... 10-25
Figure 10-26 Double vee seam weld in hot reheat piping
showing creep microdamage at the cusp ........................ 10-26
Figure 10-27 Double vee seam welds in hot reheat piping
showing a subcritical post weld heat .............................. 10-26
Figure 10-28 Creep failure of a low alloy steel superheater
tube. Note that the cracking occurred at a location
where wastage flats had accelerated the formation of
grain boundary creep voids. ......................................... 10-28
Figure 10-29 Creep cavities developed in association with
sigma phase in austenitic stainless steel. The cavities
were revealed using repeat polishing and etching as
described in Section 3 of this report. .............................. 10-29
Figure 10-30 General appearance of brittle creep failures in
DMWs. Fracture occurs at or very near to the fusion line
with limited deformation so that the profile of the weld
beads can be seen. ...................................................... 10-30
 xxviii 
13828389
Figure 10-31 Creep cavities developed in DMWs in the
HAZ of austenitic welds (a), and at the fusion line in
nickel based welds (b) .................................................. 10-30
Figure 11-1 Schematic illustration of crack initiation and
growth showing how the critical crack size is
significantly reduced by embrittlement. Line A shows
growth behavior for normal conditions with line B
indicating the more rapid growth, which occurs for
accelerated conditions such when increased stress or
temperature provide a greater driving force for damage. ... 11-3
Figure 11-2 Examples of the Master Curve approach
relating FATT with fracture toughness for (a) 1/2Mo and
11/4Cr1/2Mo steels and (b) 2 1/4Cr1Mo steel .............. 11-7
Figure B-1 Typical intergranular brittle fractures in an
Alloy 80A bolt ................................................................. B-2
Figure B-2 Stress relaxation behavior of Alloy 80A .................... B-3
Figure B-3 Variation of Charpy energy with aging for
Alloy 80A ....................................................................... B-4
Figure B-4 The embrittling effect of P segregation on the
fracture behavior of Alloy 80A .......................................... B-4
Figure B-5 Improvement in fracture resistance with low levels
of Al +Ti.......................................................................... B-5
Figure C-1 Photograph showing damage caused by failure
of 24 low alloy steel bolts ................................................ C-1
Figure C-2 Detail of creep damage found at the first
engaged thread .............................................................. C-2
Figure C-3 Measured hardness values along the length of an
ex-service stud ................................................................ C-3
Figure C-4 Effect of trace element content on reduction in
area for low alloy bolting steels ........................................ C-4
Figure D-1 Operating history of the Eddystone boiler ................ D-2
Figure D-2 Isometric drawing of Eddystone No. 1 main
steam system .................................................................. D-3
Figure D-3 (a) Macrostructure of the failed main steam pipe;
(b) microdamage in the failed main steam pipe .................. D-4
 xxix 
13828389
Figure D-4 Fractograph confirming that extensive
intergranular damage, with evidence of creep voids,
was present at the crack tip .............................................. D-6
Figure D-5 Microstructure of main steam line section where
creep cracking had developed; (a) etched in
hydrochloric and picric acid and (b) electrolytic etch in
KOH to reveal the sigma phase ........................................ D-6
Figure D-6 Schematic diagram of the junction header ............... D-7
Figure D-7 Cross section of the junction header showing the
ID surface cracking revealed by penetrant testing ............... D-8
Figure D-8 Damage developed in the junction header............... D-8
Figure D-9 Microstructure of main steam line section where
creep cracking had developed; (a) etched in
hydrochloric and picric acid and (b) electrolytic etch in
KOH to reveal the sigma phase ........................................ D-9
Figure D-10 Schematic diagram showing estimates of creep
fatigue usage.................................................................. D-9
Figure E-1 Schematic diagram showing the location of the
cracked weld ................................................................... E-2
Figure E-2 Micrograph showing the cracking on the forging
side of the weld ............................................................... E-4
Figure E-3 Detailed micrographs showing the extensive
intergranular creep damage developed in the coarse
grained regions of the HAZ on the failed side of the
weld ............................................................................... E-5
Figure F-1 Catastrophic failure of Gallatin Unit 2 IP-LPSF
rotor ............................................................................... F-2
Figure F-2 Schematic diagram of the reassembled Gallatin
rotor indicating location of primary fracture surface ............. F-2
Figure F-3 Primary fracture surface of the bore near exhaust
end of the IP section of the rotor revealing a large
oxidized region ............................................................... F-3
Figure G-1 Photograph showing damage caused by failure
of the rotor disc .............................................................. G-1
Figure G-2 Section reconstruction showing disc cracking .......... G-2
Figure G-3 Schematic diagram showing regions of
segregation in the disc..................................................... G-2
 xxx 
13828389
Figure G-4 Photographs showing the location of crack
initiation......................................................................... G-3
Figure G-5 Schematic representation of ingot defects .............. G-4
Figure G-6 Brittle fracture of a rotor from a manufacturing
defect ............................................................................ G-5
Figure H-1 Examples of the grain boundary graphite
revealed using optical metallography ................................ H-3
Figure H-2 Scanning electron micrograph showing the local
nature of the graphite formation on grain boundaries .......... H-4
Figure H-3 Variation of Charpy fracture energy with the
level of graphitization present........................................... H-5
 xxxi 
13828389
13828389
List of Tables
Table 1-1 Summary of component problems ............................. 1-2
Table 1-2 Room temperature yield strength and fracture
toughness data for selected engineering alloys .................. 1-15
Table 1-3 Summary of the effects of microstructural variables
on fracture toughness of steels ......................................... 1-16
Table 2-1 Average Charpy fracture energy values obtained
for multiple tests on one batch of 4340 steel ..................... 2-12
Table 2-2 Correlation between impact transition temperature
and fracture toughness.................................................... 2-15
Table 2-3 Correlation between upper shelf impact properties
and fracture toughness.................................................... 2-16
Table 2-4 Empirical constants identified for use in equation
2-5 which correlates FATT measured by Charpy impact
testing with Tsp the transition temperature measured using
punch tests .................................................................... 2-26
Table 2-5 Selected etchants used in the microstructural
characterization of engineering alloys. In most situations
etchants should be prepared when needed. Application
for successful results is largely experienced based so that
specific information regarding etching conditions and
times cannot be given. .................................................... 2-35
Table 5-1 Formulae developed to calculate values of
chromium and nickel equivalent ....................................... 5-14
Table 7-1 Summary of the influence of alloying elements on
microstructure and embrittlement ...................................... 7-12
Table 8-1 Summary of information concerning metal
combinations, the symbol X indicates the liquid metal
that embrittles a specific solid (based on 8.5) ...................... 8-8
Table 8-2 Common alloy/environment systems known to
exhibit stress corrosion cracking ...................................... 8-16
 xxxiii 
13828389
Table 10-1 Typical composition and heat treatments used
for martensitic boiler steels .............................................. 10-9
Table D-1 Chemical composition of material from the
cracked pipe and turbine stop valve .................................. D-5
Table E-1 Compositions of the base and weld metals ................. E-3
 xxxiv 
13828389
Section 1: Introduction
1.1 Background
Embrittlement can be defined as a general set of phenomena whereby materials
suffer a marked decrease in their ability to deform (loss of ductility) or in their
ability to absorb energy during fracture (loss of toughness), with little change in
other mechanical properties, such as strength and hardness. The susceptibility for
brittle behavior can be affected by a variety of external or internal factors, for
example:

The temperature

The stress

Changes in the microstructure of the material, namely, changes in grain size,
or in the presence and distribution of alloying elements and second-phase
particles

The introduction of an environment which is often, but not necessarily,
corrosive in nature

An increasing rate of application of load

The presence of surface notches
A list of some of the problems, which have resulted in significant component
damage in fossil fuelled power plant, is presented in Table 1-1. In some of these
examples the development of time dependent damage resulted in steam leaks and
lost generation. In other examples the failures were of a catastrophic nature
resulting in rapid fracture and, in a few cases, fragmentation and the launch of
projectiles. It should be pointed out that in some examples of brittle failure the
fracture event occurred because time dependent metallurgical factors resulted in
materials embrittlement. In other cases damage initiated and propagated in a
stable manner before the final fracture event.
In view of the seriousness with which the utility industry views failures, in many
situations detailed cause analysis has been performed, and the results reported.
This information has served as the basis for improvements in alloy selection, alloy
design, manufacture and quality control as well as to provide important
 1-1 
13828389
knowledge to aid programmes of component condition assessment and failure
prevention. The information available has been compiled and analyzed in the
present guideline document which seeks to:

Provide general metallurgical background for typical alloys and methods of
manufacture

Describe, for particular components and relevant alloys, time dependent
metallurgical mechanisms that result in failure with low overall ductility

Document how damage develops and describes methods for assessing
damage level

Provide guidance regarding typical short and long-term solutions to
embrittlement issues
The fact that many failures within the utility industry occur on a worldwide basis
has resulted in a very large number of publications being available.
Table 1-1
Summary of component problems (adapted from ref 1.1)
Component/
Description
Country/
Year
Operating/Fabricating
condition not taken into
account
Metallurgical
condition not taken
into account
1. Pipework
(i) Weldments
Reheat cracking
(CrMoV)
UK 1965/85
Inadequate weld procedures
Coarse grains
Weld metal cracking
UK 1965/85
Improper heat treatment
Trace elements
Type IV Cracking
Global 1980s
System stresses
Weak zone in HAZ
(ii) Cracking of seam
welds
USA 1985/90
Double vee preparation leads to
stress concentration
Low creep strength weld
metal
(iii) Bend failures
(CrMoV, 12CrMo)
Germany,
Russia
1985/90
No allowance made for bend
wall reduction
Overestimate of rupture
strength
(iv) Failure of austenitic
pipework
USA 1985
Residual stresses due to thermal
cycling
Sigma phase formation
(v) Distortion of
austenitic pipework
UK 1975
Thermal cycling
Thermal stresses exceed
yield
(vi) Failure of cold bent
pipework
UK 1965/86
Residual stresses and bend
system stresses
Strain hardening due to
bending creep in service
(vii) Dissimilar metal
weld failures
UK 1975/85
Stresses due to mismatch of
parent and weld metal
Brittle interfaces,
cavitation near or at
interfaces
 1-2 
13828389
Table 1-1 (continued)
Summary of component problems (adapted from ref 1.1)
Component/
Description
Country/
Year
Operating/Fabricating
condition not taken account
Metallurgical
condition not taken
account
2. Bolting
(i) Ferritic (CrMoV(Nb))
Europe
1965/79
Superimposed bending stresses
due to thermal expansion
Course grained structures,
temper embrittlement
(ii) Nimonic 80A
UK, Germany
mid 1980s
Increased stresses due to lattice
ordering and contraction
Embrittlement due to
ordering and segregation
(i) Distortion of CrMoV
rotors
UK 1975
Incorrect heat treatment
Variations in creep
strength
(ii) Cracking in heat
release grooves
(iii) Bore cracking
USA, UK,
Japan
1975/80
Increased stress due to groove
High stresses
Low ductility
microstructure
3. Rotors
Inclusions, brittle
microstructures
Global 1960’s
4. Chests/Casings
Global 1980s
Thermal or Residual stresses
associated with weld repair
Low ductility
microstructure
(i) Catastrophic failure
UK 1969
Excessive temperature
Low rupture ductility
(ii) Stub weld cracking
Global
1970/80
Excessive temperature, joint
geometry
Weld structures /system
stresses
(iii) Ligament cracking
(iv) Nozzle cracking
Global
1988/90
Thermal stresses due to cycling
Local stress concentrations
Oxide cracking
Severe stress, temperature and
environmental conditions in fossil
boilers
Incorrect material, heat
treatments, tubing thinning
etc.
5. Headers
Global 1980’s
6. Boiler tubes
Global
1950/90
In the present document, key references are provided which allow individual
follow up as required. Furthermore, wherever possible specific issues associated
with embrittlement mechanisms are highlighted with Case Studies. Several of
these case studies (which are described in detail in the Appendix) have been
selected because they provide direct evidence that although the numbers of brittle
failures in large components is small, when they do occur, the consequences can
indeed be catastrophic.
 1-3 
13828389
1.2 Fracture of Materials
Several different types of failure can occur depending on the material used and
the stress and temperature conditions imposed. Thus, it is not always possible to
say that a material is either ductile or brittle because the fracture behavior often
depends on the service conditions, being brittle under some conditions and
ductile under others; for example, welded pressure vessels which can operate
satisfactorily at warm temperatures have been known to fail catastrophically when
hydrostatically tested using cold water, for example, Figure 1-1.
Figure 1-1
Brittle fracture of a steel pressure vessel caused by hydrostatically testing using
cold water
To permit the visualization of how the fracture behavior varies with stress and
temperature, fracture maps have been developed [1.2]. These maps are typically
assembled using tensile and creep data with the temperature, T, represented as a
fraction of the absolute melting point, Tm, and with the stress, σ, represented as
a fraction of the temperature corrected elastic modulus, E. For the selected metal
or alloy, the conditions where particular types of fracture should then occur are
provided. The behavior for 2 1/4Cr1Mo low alloy steel and 316 stainless steel are
presented in Figures 1-2 and 1-3 respectively.
 1-4 
13828389
Figure 1-2
Fracture map for 2 1/4Cr1Mo low alloy steel
Figure 1-3
Fracture map for 316 stainless steel
 1-5 
13828389
In the case of the low alloy steel, the fracture map provides information about
ductile fracture, brittle cleavage fracture, transgranular creep failure and
intergranular creep failure. Since the stainless steel exhibits a face centered cubic
microstructure at all temperatures, there is no brittle to ductile transition.
However, brittle fractures may occur under creep conditions or if metallurgical
transformations result in the formation of sigma phase. While these maps
provide a guide regarding behavior, generally application for prediction of
commercial alloys is limited since a single map represents one metallurgical
condition. To obtain a proper appreciation of fracture behavior requires a specific
understanding of the compositional and microstructural factors, which control
failure for the particular operating conditions. Background describing how
metallurgical and loading factors influence fracture behavior is provided in the
following, with detailed consideration of particular conditions, which promote
low ductility brittle failures provided in subsequent chapters of this guideline
document.
1.2.1 Ductile Fracture
Possibly the simplest type of failure process is found during tensile testing of
ductile face centered cubic (FCC) single crystals, when the generation and
movement of dislocations can occur on a large number of independent slip
systems. The material eventually necks down to a point (100% reduction in area)
as slip occurs on several slip systems.
This type of failure is rare with polycrystalline samples of even ductile FCC
materials being found only during deformation at high temperatures when
continued recrystallization can avoid build-up of stress concentrations. Instead,
'ductile failure' of polycrystals usually takes place with 'reductions in area’, which
are well below 100% (that is, the material does not neck to a point). Even so,
ductile failure is normally associated with mechanical instability (that is, the
formation of a neck at some position along the specimen gauge length). The
stresses within the necked region then cause the formation of small holes or
'voids'.
The voids formed in the center of the necked region nucleate at inclusions or
other ‘hard’ particles. The importance of inclusions is illustrated by the fact that
the reduction in area at fracture for commercial aluminum is about 30%
compared with about 90% for superpurity aluminum. The voids are formed
either by cracking the inclusions or by decohesion at the particle/matrix interface.
The material between the voids then gradually necks down to a point, giving
fracture.
As the voids link up to form cracks in center of neck, eventually the stress on the
unfractured section of the specimen becomes so great that final failure is by shear,
giving 'cup and cone' or 'double cup and cone' ductile fractures. A typical ductile
cup and cone fracture in a laboratory tensile specimen is shown in Figure 1-4a,
with a detailed micrograph showing the multiplicity of local voids shown in
 1-6 
13828389
Figure 1-5a. While most ductile materials fail in a 'transgranular' manner
(through the grains), ductile 'intergranular' failures may also be observed in cases
where inclusions or precipitates favor void nucleation, link-up and cracking along
grain boundaries.
(a)
(b)
Figure 1-4
Fractures observed in laboratory tensile tests showing (a) ductile fracture and
(b) brittle fracture
(a)
(b)
Figure 1-5
Detail of the fracture surface associated with (a) ductile fracture and (b) brittle
fracture
1.2.2 Brittle Fracture
The most common mode of brittle fracture involves transgranular cleavage. The
cracks propagate along specific crystallographic planes, which present low energy
fracture paths. Within an individual grain the fracture appears relatively flat;
however, because different grains will have different orientations the cracks
change direction at grain boundaries. A typical brittle fracture in a tensile sample
is shown in Figure 1-4b, with details of transgranular cleavage shown in Figure
1-5b. Brittle fractures may also occur in an inter-granular manner. The tendency
 1-7 
13828389
for brittle grain boundary failures is again normally the result of metallurgical
changes. Moreover, with the exception of certain FCC metals and alloys, almost
all crystalline solids can fail in a brittle way by 'cleavage' if the temperatures are
sufficiently low.
In the absence of a pre-existing flaw, cleavage cracking usually involves a
nucleation and growth stage. Thus, for example, cracks may nucleate where a slip
band intersects a grain boundary. In cases when it is difficult for deformation to
continue in the neighboring grain many dislocations can ‘pile-up’ at this location.
A sufficient density of these micro defects at one location can result in the
formation of a crack. Once a crack exists, the stress concentration at the crack tip
is high and the crack may propagate along well-defined transgranular 'cleavage
planes' or along grain boundaries if this path is easier.
This illustrates why decreasing the grain size improves resistance to brittle
fracture:

Grain boundaries may hinder crack propagation

The larger the grain size, the longer the slip band length and the greater the
number of dislocations, which can form within a ‘pile-up’
It should then be obvious that, unless pre-existing flaws exist, the stress to cause
brittle failure is not less than the yield stress (that is, brittle fracture occurs at
yielding since the slip causes cracks to nucleate and propagate to cause fracture
with no 'apparent' deformation). Thus, the tendency for brittle fracture is
established by consideration of the material yield stress and the fracture stress. It
has been established that the material yield behavior is given by the Hall-Petch
equation, that is
σy = σi + kyd (-1/2)
Eq. 1-1
Where σy is the yield stress, σi is the friction stress, ky is the strengthening
coefficient and d the grain size. The brittle fracture stress, σf, is typically
considered to be directly proportional to the surface energy to form a crack and
the shear modulus and inversely proportional to the square root of the grain size.
Schematic representations of how the yield stress and the fracture stress vary with
temperature for a typical ferrous alloy are shown in Figure 1-6.
 1-8 
13828389
Figure 1-6
Schematic illustration showing how the transition from brittle to ductile fracture
depends on the yield and fracture stresses
At low temperatures, the fracture stress is below yield but since some
deformation is needed to initiate a crack, brittle fracture occurs coincident with
the yield point. At higher temperatures, the fracture stress is significantly above
the yield stress and significant deformation will take place before ductile fracture
occurs. There will be a transition region between these extremes where a mixture
of brittle and ductile behavior is found.
1.2.3 The Brittle – Ductile Transition
In body-centered cubic metals (for example, iron, tungsten) and hexagonal closepacked metals (for example, zinc, magnesium), a critical temperature exists below
which the metal exhibits limited toughness. Fracture is usually brittle in nature,
occurring either through the crystal lattice (cleavage) or along the grain
boundaries (intergranular fracture). In simple terms, low-temperature
embrittlement results from a competition between deformation and brittle
fracture, with the latter becoming preferred at a critical temperature.
 1-9 
13828389
A number of metallurgical factors will influence the critical temperature at which
brittle fracture takes place. In the simplest case, a time/temperature dependent
microstructural change, such as the segregation of an embrittling element such as
phosphorus, will significantly reduce the fracture stress. Since this segregation
will have little effect on the yield strength, the transition from brittle to ductile
fracture will take place at a higher temperature. This effect is shown
schematically in Figure 1-7.
Figure 1-7
Schematic representation of how an embrittling event will increase the
brittle/ductile transition temperature
Grain size changes will influence both the yield and fracture behavior. Thus,
decreasing the grain size will lead to an increase in the strength. However, fine
grain sizes will promote an even greater increase in the fracture stress so that the
transition from brittle to ductile behavior occurs at a higher temperature for
coarse grained material. This effect is illustrated schematically in Figure 1-8.
Consideration of these effects indicates why the segregation of trace elements to
 1-10 
13828389
grain boundaries has a significantly greater embrittling effect for coarse grained
material. In this case, the reduced grain boundary area will increase the tendency
for the level of the trace element at the boundary to reduce the fracture stress so
there is increased tendency for brittle fracture.
The discussion presented thus far has demonstrated how the change from brittle
to ductile fracture is influence by metallurgical factors. The tendency for brittle
type behavior will also be increased by

Mechanical constraint (by increasing the hydrostatic stresses)

The presence of notches (through a stress concentration factor)

Rapid strain rates (by reducing the time for dislocation rearrangement)
Thus, it is normal to evaluate fracture behavior using an impact test on a notched
bar or by performing fracture toughness testing. Charpy impact testing at a range
of temperatures should permit evaluation of the change from brittle to ductile
behavior. As detailed in Section 2 of this guideline, the data typically recorded
from a set of impact tests are the 50% Fracture Appearance Transition
Temperature and the Upper Shelf Energy. This information can then be used
with appropriate empirical correlations to provide estimates of fracture
toughness.
 1-11 
13828389
Figure 1-8
Schematic illustration showing how changes in grain size modify the yield stress
and the fracture stress and hence change the brittle to ductile transition
temperature
1.3 Crack Propagation
Structures and components in service rarely fail in a completely ductile manner,
since it should be easy to design against excessive elastic deflections and general
plastic yielding. So, ductile failures are found more commonly during
manufacture when the forming processes have not been optimized correctly, for
example, cracking during rolling. Consequently, ductile failures during service are
generally a consequence of poor design or incorrect material specification.
Correctly designed components and structures can fail catastrophically by fast
fracture when cracks and other flaws introduced during manufacture propagate
rapidly (for example, cracks present after welding can suddenly become unstable
causing failure of pressure vessels and pipe work). This type of fast fracture can
then occur at ‘average’ stresses well below the general yield stress of the material
because the flaws introduce 'stress concentrations' (that is, the stress at the crack
tip can exceed the fracture stress so the crack propagates suddenly).
 1-12 
13828389
The conditions for fast fracture can be calculated approximately as follows.
Consider the formation of a crack of length, c, in a plate of thickness, I, under a
stress, σ. Energy must be supplied to create the crack surfaces, that is 2clγ where
γ is the surface energy/unit area of the material. However, energy is stored in a
material under stress (the area under a stress/strain curve, equating to half stress ×
strain) and this energy is released at fracture. So, on forming the crack, assume
energy is released from a region of radius c around the crack.
Energy released = (σε / 2)(πc2 l/2)
But since ε = σ / E
Energy released =(σ2 / 2E) (πc2 l/2)
Energy balance = ΔE = (σ2 /2E)(πc2 l/2) - 2clγ
The crack increases in length when dΔE / dc = 0
Therefore, 0 = (σ2 /2E) (πcl) - 2lγ
Rearranging gives Fracture Stress, σf = (4E γ / π c) 1/2
More accurate calculation gives
σf = (Eγ / π c)1/2
(for brittle materials)
Eq. 1-2
This is called the Griffith criterion.
Thus, the stress to cause rapid failure is inversely proportional to the square root
of the crack length (that is, the stress needed to propagate the crack decreases as
the crack grows so, at σf, crack propagation should be catastrophic). The Griffith
criterion proves to be reasonable for brittle materials, like glass, but incorrect values
are obtained for ductile materials.
This influence can be checked experimentally by introducing a crack of known
length (for example, by machining) and, since E and c are then known,
determination of the stress to cause fracture, σf, allows γ to be calculated. When
this procedure is adopted for ductile materials, the calculated value of γ is far
higher than the actual surface energy. One reason for this is that, with ductile
materials, plastic deformation ahead of the crack tip absorbs energy.
This has led to the Orowan modification to the Griffith criterion, namely,
σf = ( EGc / πc ) ½
(for ductile materials)
Eq. 1-3
Where Gc is the toughness (or the critical strain energy release rate, units of
kJmol) which includes the energy to create the new crack surfaces, to cause plastic
deformation ahead of the crack tip, etc. Thus, when Gc is high, the material is
tough and large amounts of energy are required for failure.
 1-13 
13828389
1.4 Fracture Toughness
By rearranging equation 1-3, the condition for fast fracture can be represented as:
σf (πc) 1/2 = (EGc)1/2
Eq. 1-4
The left hand side of this equation states that fast fracture occurs when

A crack grows to length c in a material under a stress, σf

A stress σf , is applied to a material containing a crack of length, c
The right hand side then contains terms that depend only on the material
properties (that is, E is Young's modulus and Gc is the energy required to create a
crack of unit area). Thus, the product σf (πc)1/2 is a material constant (that is,
there is a critical combination of stress and crack length that leads to fast
fracture). This is usually called the 'stress intensity factor', called Kc having
units of MNm -3/2 (ksi∙in.1/2). Fast fracture then occurs when
K=Kc = ~ (EGc) ½ = Yσf (π a)1/2
Eq. 1-5
Where Y is a geometry factor and Kc is the 'critical stress intensity factor' or the
'fracture toughness'. For tensile, or mode 1, loading the stress intensity factor is
used with the additional subscript 1 so that the critical value is given by K1c . This
is analogous to saying that plastic deformation occurs at a critical value of stress,
that is the yield stress. Thus, K1c is a materials parameter that is obtained by
testing under specific controlled conditions. Information regarding values of K1c
can also be obtained from appropriate references (for example, 1.3 to 1.8). As a
guide, some typical values of room temperature toughness are given in Table 1-2.
The tendency for rapid fracture thus depends on three factors:

The fracture resistance or toughness of the material

The crack size

The stress
The risk of Brittle Fracture must be assessed during the design process. In
general, this involves ensuring that any material or fabrication defect that could
lead to failure is of a size sufficient to be readily identified by post manufacture
quality assurance inspection. However, the risk of brittle failure is present not
only early in life. Defects present following fabrication may propagate during
service to reach a critical size or service exposure may lead to microstructural
changes that reduce fracture resistance [1.3]. The metallurgical effects that
modify toughness are summarized in Table 1-3. These embrittling phenomena
are discussed in detail in later sections.
 1-14 
13828389
Table 1-2
Room temperature yield strength and fracture toughness data for selected
engineering alloys
Alloy
Yield Strength
Fracture Toughness, K1c
MPa
ksi
MPa∙m1/2
Ksi∙in1/2
Aluminum (2024-T3)
345
50
44
40
Titanium (Ti-6Al-4V)
910
132
55
50
Steel 4340 (temper 205°C)
1640
238
50
45.8
Steel 4340 (temper 425°C)
1380
200
75
68
Steel 4340 (temper 540°C)
1172
170
110
100
This type of reduction may be reflected by a decrease in the Fracture Appearance
Transition Temperature (FATT) and/or lowering of the upper shelf energy.
While these data do not provide a direct measure of Fracture Toughness,
correlations have been developed which permit fracture toughness to be
estimated from Charpy results, see Section 2 of the guideline. Data of this type
have shown that in service embrittlement can reduce the fracture energy at room
temperature from around 80 ft.lb to less than 5 ft.lb (approximately 120 J to
< 10 J). For operation under similar stresses the embrittlement noted will reduce
the critical crack size for rapid brittle fracture from around 10 inches to
significantly less than ½ inch.
1.5 Summary
It is apparent that time/ temperature dependent embrittlement significantly
increases the risk of in-service components. An integrated approach to structural
integrity assessment thus requires that periodic fitness for service evaluations
include:

Inspections to characterize the size and location of defects and cracks

Metallurgical analysis to assess the extent of degradation in materials
properties
This guideline document:

Provides information regarding the different degradation mechanisms which
have been identified in power plant alloys
Details approaches for identifying and quantifying changes in materials
microstructure and properties which increase the risk of brittle fractures
 1-15 
13828389
Table 1-3
Summary of the effects of microstructural variables on fracture toughness of steels
[1.3]
Microstructural Parameter
Effect on Toughness
Grain size
Decrease in grain size increases KIc in
austenitic and ferritic steels
Unalloyed retained austenite
Marginal increase in KIc by crack blunting
Alloyed retained austenite
Significant increase in KIc by transformationinduced toughening
Interlath and intralath carbides
Decrease KIc by increasing the tendency to
cleave
Impurities (P, S, As, Sn)
Decrease KIc by temper embrittlement
Sulfide inclusions and coarse
carbides
Decrease KIc by promoting crack or void
nucleation
High carbon content (>0.25%)
Decrease KIc by easily nucleating cleavage
Twinned martensite
Decrease KIc due to brittleness
Martensite content in quenched
steels
Increase KIc
Ferrite and pearlite in quenched
steels
Decrease KIc of martensitic steels
1.6 References
1.1
R. D. Townsend, “A review of service problems during high temperature
operation” Proc Materials Congress ’98–Frontiers in Materials Science
and Technology, Materials for High Temperature Power Generation
and Process Plant Applications, (Ed A.Strang) Institute of Materials,
2000, pp. 199–223.
1.2
M. F. Ashby, C. Ghandi, and D. M. R. Taplin, “Fracture Mechanism
Maps and their Construction for F.C.C. Metals and Alloys,” Acta
Metall., Vol. 27, 1979.
1.3
K. S. Ravichandran and A. K. Vasudevan, Fracture Resistance of
Structural Alloys, Fatigue and Fracture, Vol. 19, ASM Handbook, 1996,
p. 381–392.
1.4
C. P. Cherepanov, Mechanics of Brittle Fracture, Magraw-Hill, 1979,
Structural Alloys Handbook, Volume 1, Metals and Ceramics
Information Center, 1987.
1.5
Damage Tolerance Handbook, Volume 2, MCIC – HB –018, Metals
and Ceramics Information Center, 1983.
 1-16 
13828389
1.6
T. L. Anderson, Probabilistic Establishment of Fracture Toughness
Distributions for Fitness-for-Service Evaluations, Fitness-for Service and
Decisions for Petroleum and Chemical Equipment, PVP Vol. 315,
American Society of Mechanical Engineers, 1995, p. 485–490.
1.7
Fracture toughness data for carbon and Cr-Mo steels compared with
lower-bound curves developed for ASME Section III, Appendix G.
1.8
R. Viswanathan, “Damage Mechanisms and Life assessment of High
Temperature Components,” 1989, ASM International.
 1-17 
13828389
13828389
Section 2: Testing Methods
2.1 Introduction
As indicated in Section 1, a key factor when assessing the risk of brittle fracture
is knowledge of the material properties. In general, the most accurate
measurements of appropriate properties are obtained by undertaking laboratory
tests using procedures specified by applicable standards. Many of these methods
have been developed to support performance evaluation of new alloys and to
qualify specific manufacturing process variables. Thus, the usefulness of these
techniques in assessing the properties of a component in service is limited by the
ability to obtain sufficient material to allow fabrication of relatively large
specimens. In the majority of cases removal of large sections necessitates weld
repair and there are concerns regarding:

The effect removal of material has on a location already identified as being at
risk of failure

The influence of weld repair on subsequent performance

The time and cost involved
However, there are situations where material removal does facilitate carrying out
standard tests. In general, where significant damage has been identified and
repair is inevitable then it is normally the case that detailed evaluation should be
carried out to:

Establish the damage mechanism

Perform a root cause analysis
It should be emphasized that a root cause analysis is significantly more than
simply identifying the mechanism. Thus, root cause analysis should establish the
reasons for the accelerated damage so that appropriate remedial action can be
undertaken to minimize the risk of future problems. Examples of root cause
evaluations are given in the Case Studies within these guidelines with relevant
information also contained in the following EPRI documents:

EPRI Life Assessment of Boiler Pressure Parts (TR-103377, Vol. 1–5)

EPRI Condition Assessment Guidelines for Fossil Fuel Power Plant Components
(GS-6724)

EPRI Boiler Tube Failure Metallurgical Guide (TR-102433, Vol. 1–2)
 2-1 
13828389

EPRI Boiler Tube Failure: Theory and Practice (TR-105261, Vol. 1–3)

EPRI Turbine Steam Path Damage: Theory and Practice (TR-108943,
Vol. 1–2)

EPRI Remaining Life Assessment of Austenitic Stainless Steel Superheater and
Reheater Tubing (1004517)
A detailed root cause analysis typically involves metallographic evaluation
supplemented, where necessary, by mechanical tests. The present document
therefore summarizes aspects of these approaches, with particular emphasis on
obtaining key information in the assessment of brittle behavior.
Important alternatives to removal of large sections are often available through the
ability to perform in-situ metallographic preparation followed by replication or
extraction of small sections of material. In many cases removal of small samples
can be undertaken with little or no post sampling action. Component sections
can be used for laboratory metallographic assessment and there are an increasing
number of test techniques available to measure specific properties using miniature
specimens. In general, in both cases the parameter measured under laboratory
conditions must be correlated using a pre-established relationship to give an
estimate of bulk performance.
The present guidelines provide information regarding:

Standard techniques for measurement of properties, with particular reference
to evaluation of fracture toughness

Small specimen approaches for estimation of strength, creep properties and
fracture resistance

Metallographic techniques for assessing damage mechanisms and
characterization of microstructure
In these sections, emphasis is given to methodologies which are particularly
important to the assessment of embrittlement and low ductility fracture.
2.2 Standard Mechanical Tests
The following techniques have been established to measure materials properties.
1. Strength properties
-
Hardness tests
-
Tensile tests (slow or high rates of loading)
-
Bend testing
-
Compression tests
-
Shear and torsion tests
 2-2 
13828389
2. Fracture properties
-
Charpy and instrumented Charpy tests
-
Drop-weight tests
-
Drop-weight tear tests
-
Fracture mechanics tests
o
o
o
o
Static fracture initiation tests (many variations)
Static resistance curve tests
Dynamic initiation or resistance curve tests
Crack arrest tests
3. Fatigue properties
-
Fatigue endurance tests (many variations)
-
Fatigue crack growth tests
4. Influence of environment
-
Stress corrosion tests
-
Hydrogen embrittlement tests
5. High temperature properties (that is, tested above about 30% of the melting
point)
-
Strength, fracture and fatigue tests as above at elevated temperatures
-
Creep tests
-
Creep crack growth tests
-
Low cycle fatigue tests (creep/fatigue interaction)
It is beyond the scope of the present guideline document to provide details on all
the above testing techniques. In view of the direct relevance to assessment of
brittle fracture, summary information is presented regarding methods used to
obtain fracture toughness. For additional detail regarding methods applicable
ASTM standards (for example [2.1]) and advanced reference texts (for example,
ASM Metals Handbook, Vol. 8 and Vol. 19 [2.2, 2.3]) should be studied, as
necessary to obtain specific comments on the specimen geometry, the conduct of
the tests as well as to provide information regarding data analysis and
interpretation.
 2-3 
13828389
2.3 Assessment of Fracture Toughness
Ferritic steels undergo a transition from brittle to ductile fracture. A number of
approaches have been developed to assess this transition behavior. These include:

A drop weight test which defines a nil ductility temperature (NDT).

The explosive bulge test is used to assess the highest temperature where
extensive deformation occurs without brittle cracks. This is referred to as the
fracture transition plastic (FTP). The temperature below which the cracking
extends into the elastically loaded region is referred to as the fracture
transition elastic (FTE).
However, by far the most commonly used method is the Charpy impact test.
2.3.1 Charpy Impact Testing
Test Technique
The Charpy machine is shown schematically in Figure 2-1. The machine is
designed so that the total available striking energy is 300 J (220 ft · lb). The
notched test specimen, supported at both ends, is impacted by a single blow of
the pendulum applied at the middle of the specimen on the unnotched side. For
very tough materials the energy required to break the sample may be so great that
no failure occurs. However, normally the specimen breaks at the notch and the
pendulum passes between the two parts of the anvil. The height of fall, h, minus
the height of rise, h', gives the amount of energy absorption involved in
deforming and breaking the specimen, Figure 2-1. Thus, it should be apparent
that a brittle material, which breaks easily, absorbs very little energy from the
hammer so that the pendulum swings through to a great height after fracture.
Frictional and other losses amounting to 1.5 or 3 J (1 or 2 ft · lb) should be added
to the measured energy value.
The instrument used should be regularly calibrated to ensure that the measured
value of the energy absorbed by the test specimen is accurate. Standard ASTM
E 23 [2.4] provides requirements for test specimen, anvil supports and striker
dimensions and tolerances, the pendulum action of the test machine, the actual
testing procedure and machine verification, and the determination of fracture
appearance and lateral expansion.
 2-4 
13828389
Figure 2-1
Diagram showing the main features and operation of a Charpy impact test
machine
The Charpy specimen is 55 mm in length with a 10 mm square section, Figure
2-2. A 2 mm (0.079 inch) deep, 45° notch with a 0.25 mm (0.01 inch) radius is
machined in the center of the gauge length. The required test temperature is
obtained by appropriate pre-test treatment (for example, in a controlled
temperature bath or oven) before rapid transfer to the testing machine. The test
must be carried out rapidly so that changes in specimen temperature do not
occur.
 2-5 
13828389
Figure 2-2
Dimensions of a standard Charpy impact specimen, with detail of the specimen
support region of the test machine
The Charpy V-notch impact test has limitations due to:

The fact that the machined notch will be relatively blunt compared to an inservice crack

The specimen dimensions are generally small compared to the size of a
component

The tests typically give a total energy measurement that is, there is no
separation of initiation and propagation components of energy
Despite these limitations the test is used widely because it is inexpensive and
simple to perform. The large amount of data generated using this method has
been shown to reasonably describe service performance and thus demonstrated its
usefulness in assessing brittle behavior. Thus, the Charpy V-notch test
commonly is used as a screening test for evaluating notch toughness changes
influenced by

Chemical composition (alloying and impurity elements, including gases)

Microstructural factors (such as the phases present, the grain size)
 2-6 
13828389

Mechanical properties generally considered are yield and flow properties and
hardness

Heat treatment effects, including the influence of service exposure
The Charpy V-notch impact test has limitations due to:

The fact that the machined notch will be relatively blunt compared to an inservice crack

The specimen dimensions are generally small compared to the size of a
component

The tests typically give a total energy measurement that is, there is no
separation of initiation and propagation components of energy
Despite these limitations the test is used widely because it is inexpensive and
simple to perform. The large amount of data generated using this method has
been shown to reasonably describe service performance and thus demonstrated its
usefulness in assessing brittle behavior. Thus, the Charpy V-notch test
commonly is used as a screening test for evaluating notch toughness changes
influenced by

Chemical composition (alloying and impurity elements, including gases)

Microstructural factors (such as the phases present, the grain size)

Mechanical properties generally considered are yield and flow properties and
hardness

Heat treatment effects, including the influence of service exposure
Data Analysis
For each test, the measured Charpy impact value is recorded noting the test
temperature. Tests at a given, pre-selected temperature may be performed to
compare the behavior of different alloys or the same alloy in different heat
treatment conditions. For materials which exhibit a brittle to ductile transition,
a set of tests over an appropriate range of temperatures results in curve of the
form shown schematically in Figure 2-3. This curve permits evaluation of factors
such as:

The lower shelf energy

The upper shelf energy

The transition temperature
 2-7 
13828389
Figure 2-3
Schematic diagram illustrating the variation of Charpy absorbed energy with test
temperature [2.5]
Other quantitative parameters, such as fracture appearance (percent fibrous
fracture) and degree of ductility/deformation (lateral expansion or notch root
contraction), are often measured by examination of the broken specimens.
Samples tested within the lower shelf should exhibit 100% brittle fracture; those
in the upper shelf region should exhibit 100% ductile or fibrous fracture with
those in the transition region exhibiting mixed behavior. Thus, plotting the
percentage ductile, or shear fracture, for the different test temperatures will
produce a curve with similar form to that obtained for absorbed energy,
Figure 2-3 [2.5].
 2-8 
13828389
As shown in Figure 2-3, undertaking a set of tests over an appropriate
temperature range allows the conditions for the transition from brittle to ductile
behavior to be established. However, when comparing results regarding this
transition it should be noted that different definitions of transition are allowed.
These definitions include:

The temperature where there is 50% brittle and 50% ductile fracture, that is,
50% FATT. This is shown as position T2 in Figure 2-3. (In some cases this
position is identified as the temperature corresponding to the energy which is
50% of the difference between 0 and 100% ductile fracture).

The temperature where a particular fracture energy is measured, for example,
40 ft∙lb (54 J). The energy value selected is usually determined by correlations
with other types of test or is based on service performance. This is shown as
position T1 in Figure 2-3.

The lowest temperature where the sample exhibits 100% ductile fracture.
This is shown as position T3 in Figure 2-3.
The different definitions of transition temperature can be confusing. However, it
should be realized that a specific definition is selected which is appropriate for a
particular application. Thus, in severe situations, which require the maximum
toughness, the transition T3 may be appropriate. In contrast, for more normal
toughness requirements, the 50% FATT will likely be more appropriate since
above this value crack propagation will involve a significant amount of ductile
fracture.
A typical Charpy transition curve is shown in Figure 2-4 [2.6]. This curve shows
data for a 21/4Cr1.6WVNb low alloy steel which has been fabricated with low
levels of trace elements. This steel demonstrates good toughness down to
relatively low temperatures. Moreover, in common with results for steels with
relatively low carbon levels the transition behavior is marked by a steep, well
defined curve. Thus, because there are relatively few sites available for initiation
of ductile fracture the change from fully brittle behavior to fully ductile behavior
occurs over a narrow range of temperature.
Charpy impact transition curves for selected low alloy steels are shown in Figures
2-5 and 2-6. As shown in Figure 2-5, rapid cooling after tempering provides
good toughness. However, if this same alloy is slowly cooled embrittlement
occurs, that is, the value of FATT is increased from about –70°C to about +12°C.
In Figure 2-6 similar behavior is shown for steels cooled at similar rates, but with
different levels of P present. The curve showing the results for high purity steel
exhibits excellent toughness even after slow cooling from tempering temperature.
However, for steel with similar levels of alloying elements but deliberately doped
with P, slow cooling increases brittle behavior. When this steel is aged at around
1000°F (540°C) the level of embrittlement increases even more. These results
illustrate a phenomenon known as temper embrittlement, which is described
more fully in Section 7 of this guide.
 2-9 
13828389
Figure 2-4
Charpy fracture energy measurements for 21/4Cr1.6WVNb steel from tests at
different temperatures [2.6]
Figure 2-5
Charpy transition curve for low alloy steel with typical levels of trace elements
[2.7]
 2-10 
13828389
Figure 2-6
Charpy transition curves for 21/4Cr1Mo steel for normal composition and for an
alloy doped with embrittling trace elements such as P before and after aging at
high temperature
Reproducibility
It is important to consider scatter of results. Consideration of the effects of data
scatter are particularly important when conducting post exposure impact tests
since in many cases the amount of material available is limited so that only a
small number of standard tests can be performed. The following example
illustrates the variability observed in a comprehensive study examining the
reproducibility of measured room temperature fracture energies [2.8].
 2-11 
13828389
Figure 2-7
Histograms showing the variation in fracture energy measured using 2 types of
testing machine for multiple tests on 4340 steel for 3 different heat treatments
A total of 1200 specimens from a single heat of 4340 steel (approximate
composition 0.4C, 2Ni, 0.8Cr, 0.25Mo) were divided into three groups and heat
treated to three different ranges of hardness: 43 to 46, 32.5 to 36.5, and 26 to 29
HRC. A total of 200 specimens at each hardness level were impact tested in each
of two Charpy machines manufactured by two companies. The average impact
energy values and distribution of results are shown in Table 2-1 with histograms
of actual test results presented in Figure 2-7.
Table 2-1
Average Charpy fracture energy values obtained for multiple tests on one batch of
4340 steel
Machine
Charpy Test Average, ft · lbf
43-46 HRC,
32.5-36.5 HRC,
26-29 HRC
A
12.7
48.6
78.4
B
12.6
49.1
77.9
 2-12 
13828389
These results demonstrate that accurate results can be obtained under carefully
controlled testing conditions. However, experience suggests that even when the
preparation and testing of impact specimens are closely controlled, a considerable
spread of test results can still occur. When the effects of these variables are added
to the inherent scatter that occurs among different heats of steel, the distribution
of test results is broadened appreciably. Thus, great care must be exercised when
judging notch toughness on the basis of one or two tests for a specific set of
conditions.
2.3.2 Charpy Correlations with Fracture Toughness
The critical plane-strain stress-intensity value, KIc, ahead of an atomically sharp
crack at the moment of unstable crack propagation can be used directly in design
applications; KIc is related to the applied stress, flaw size, and component
geometry. To determine KIc in the laboratory, a specimen of suitable size and
shape, in which a fatigue pre-crack of known dimensions is present, is loaded
monotonically and a load versus load line deflection curve, similar to a stress
strain curve, is developed. Upon reaching a critical load, Pc, instability sets in,
and the rapid crack extension is shown as a sudden change in the slope of the
plot. KIc is then calculated from the critical load by applying known relationships.
For example, for the most common specimen geometries, that is compact tension
(see Figure 2-8) and single edged notched specimens, the following equation is
used [2.5]:
KIc = Pc (a) ½ f(a/W)
Eq. 2-1
BW
Where a is the crack length, B is the specimen thickness and W is the specimen
width. Appropriate values of the function of crack length to specimen width,
f (a/W), are available in the literature [2.9, 2.10, 2.11], and full details of the
standard method for determination of Fracture Toughness is given in ASTM
Standard E399 [2.12].
 2-13 
13828389
Figure 2-8
Schematic illustration of a compact tension specimen used to measure fracture
toughness
Direct measurement of K1c using standard techniques requires significant
amounts of material. The Charpy impact test is also easier to perform and
significant amounts of data from this test method have been produced. A number
of empirical approaches have therefore been developed to correlate the Charpy
impact energy with KIc to allow a quantitative assessment of critical flaw size and
permissible stress levels. In general, these assessments seek to perform evaluations
for a range of operating conditions so that the necessary information is assessed
for:

Transient conditions, which may be performed at relatively low temperature
so that the extent of brittle behavior is most relevant

Operating conditions, which in many cases are within the ductile fracture
region so that the upper shelf energy is the critical parameter
 2-14 
13828389
Some of the more common correlations are listed in Tables 2-2 and 2-3 [2.5].
Note that some of the correlations attempt to eliminate the effects of variations
in loading rate between the two tests and so the dynamic fracture toughness, KId,
is correlated with Charpy energy. Many of these correlations:

Are dimensionally incompatible

Ignore differences between the two measures of toughness (in particular,
loading rate and notch acuity)

Are valid only for limited types of materials and ranges of data
Some of the correlations listed provide a useful guide to fracture toughness. The
accuracy of the correlations using the Rolfe – Novak and Iwadate equations are
shown in Figures 2-9 and 2-10 respectively. It is apparent that in both cases the
correlations between Charpy data and KIc are reasonable and are used as an
important aid in component assessment.
Table 2-2
Correlation between impact transition temperature and fracture toughness
Correlation
Comment
Barsom-Rolf [2.13]
3/2
2
KIC E = 2(CVN) ………………………………………..…………..
σy = 269 to 1696 MPa Static test
2
KIC E = 5(PCVN) ………………………………………………………
Pre-cracked Charpy test
Sailors-Corten [2.14]
1/ 2
2
KIC E = 8(CVN) or KIC = 15.5(CVN) ……………………………
KId = 15.873(CVN)
3/8
……………………………………………….
Static test
Dynamic (high strain-rate) test
Marandet-Sanz [2.15]
1/ 2
KIC = 20(CVN)
………………………………………..…..….
TKIC = 16.2 + 1.37T28 …………………………………………
TKIC at KIC = 100MPa m
T28 at CVN = 28J
Begley-Logsdon [2.16]
KIC at FATT = 1/2 ( KIC from Rolf-Novak relationship + 0.5σ y )…
σy = 269 to 1696MPa
Iwadate-Watanabe-Tanaka [2.17]
KIC KIC−US = 0.0807 + 1.962 exp[0.0287(T − FATT )] ….............
KIC KIC−US = 0.623 + o.406 exp[−0.00286(T − FATT )] …...........
 2-15 
13828389
For −40°C > (T-FATT)
For 350°C > (T-FATT) > −40°C
Table 2-3
Correlation between upper shelf impact properties and fracture toughness
Correlation
Comment
Rolf-Novak [2.18]
(
)
(KIC σ y ) 2 = 5[ CVN σ y − 0.05] ………………………...........
σy = 269 to 1696 MPa
Wullaert-Server [2.19]
1/ 2
K Jd = 20(DVN)
…………………………………………..........
σy = 345 to 483 MPa
1/ 2
Dynamic J-integral initiation
All loading rates with appropriate σy
K JC = 2.1(σ y CVN)
or (K JC σ y ) 2 = 4.41(CVN σ y ) .....….
Lawrence Livermore Laboratory [2.5]
(K JC E) 2 = CVN(9.66 + 0.04σ y ) ………………………..........
1/ 2
K JC = (EJIC )
and K JC = (EJId )1/ 2
Ault-Wald-Bertolo [2.20]
(KIC σ y ) 2 = 1.37(CVN σ y ) − 0.045 ………………………....
High strength, low toughness steels
Iwadate-Karushi-Watanabe [2.21]
(KIC σ y ) 2 = 0.6478(CVN σ y − 0.0098 ) ……………….........
 2-16 
13828389
Pressure vessel steels
Figure 2-9
Correlation between KIc and the upper shelf Charpy energy using the Rolfe –
Novak equation [2.22]
 2-17 
13828389
Figure 2-10
Correlation between KIc and the upper shelf Charpy energy using the IwadateKarushi-Watanabe equation [2.21]
Similar success has been achieved with the correlations in the transition region.
In particular the ‘Master Curve’ approach of Iwadate, Watanabe and Tanaka
[2.21] has shown an excellent correlation between excess temperature (that is,
test temperature minus FATT) and the value of KIc at any temperature
normalized with respect to the upper shelf energy, see Figure 2-11. The 99%
confidence limit curve was reported to result in the following expressions:
KIc
= 0.0807 + 1.962 × exp[0.0287 (T-FATT)]
Eq. 2-2
KIc-US
For −40°C (−40°F) > T – FATT, and
KIc
KIc-US
= 0.623 +0.406 × exp[0.00286 (T-FATT)]
Eq. 2-3
For 350°C (660°F) > (T – FATT) > –40°C (−40°F). Using the above correlations
KIc at any temperature can be estimated as follows. The ratio of KIc/KIc-US is first
determined using equations 2-2 or 2-3 as appropriate. Then KIc-US can be
estimated based on σy and the Charpy energy using the Rolfe – Novak or Iwadate
relationships. The value of KIc at the desired temperature is thus determined.
 2-18 
13828389
Figure 2-11
The master curve relationship between KIc/KIc-US and excess temperature for CrMo
low alloy steels [2.21]
2.4 Small Punch Testing
The ability to remove material samples without adversely affecting further
operation is a major advantage to programs of condition assessment [2.23].
When crack-like indications are identified during routine inspections laboratory
examination permits detailed evaluation and accurate dispositioning.
To extend the benefits of sample removal, a range of approaches have been
developed so that bulk material properties can be obtained from tests on small
specimens. In view of the potential benefits of this technology research and
development activities have been on going in the USA, in Europe and the Pacific
Rim (for example, [2.24 to 2.28]). In general, these involve disc shaped samples,
typically less than 10mm in diameter and about 0.5 mm in thickness, Figure
2-12, which are subjected to punch loading using a ball or hemispherical
indenter. Work at relatively low temperatures can be performed to measure
tensile properties and fracture toughness, with testing at elevated temperatures
conducted to establish creep strength and ductility. The ability to establish actual
component properties in this way permits plant assessments to be made with
confidence.
 2-19 
13828389
Figure 2-12
Typical small sample machined from an in-service component, and miniature
specimens shown before and after laboratory testing
2.4.1 Description of Small Punch Test and Results Analysis
The small punch test is essentially a punch-and-die loading test method wherein
a relatively small, flat (often disk-shaped) specimen is punched with a ball, or
hemispherical head, punch. Small punch test specimens have varied in size
between 3 and 10 mm (0.12 and 0.40 in.) in diameter and between 0.1 and
2.0 mm (0.004 and 0.079 in.) in thickness. Figure 2-13 is a schematic crosssectional view of the punch-and-die test device developed in EPRI supported
studies (for example, [2.24, 2.25, 2.26]). The key dimensions involved in this
work are:

The specimen measured 6.35 mm (0.25 in.) diameter by 0.5 mm (0.020 in.)
thickness

The punch hemispherical head diameter was 2.5 mm (0.1 in.)

The receiving die diameter was 3.8 mm (0.15 in.)
During the test the punch advances at a constant displacement rate (typically
−0.25 mm/min or 0.010 in./min), deforming the specimen against the receiving
die, while the load is recorded as a function of the punch displacement.
 2-20 
13828389
Figure 2-13
Schematic cross sectional diagram of the punch and die test equipment [2.24]
Experience has shown that while under very brittle conditions crack initiation
leads to sample fracture, when significant ductility is present the load can
continue to increase for significant displacements prior to final failure. This
situation will lead to overestimates of the strength and fracture behavior. To
provide greater accuracy of results, EPRI funded research has included the
application of a borescope system so that there is visual evidence of the
load/displacement conditions at which cracking initiates. A schematic diagram
showing the punch test apparatus with the borescope system is shown in
Figure 2-14.
2.4.2 Estimation of Tensile Properties
The ability of a material to withstand deformation and fracture is critical to
assessment of structural integrity. These properties are important both in
establishing the maximum loads which can be applied for single, short term
applications of stress and for consideration of component performance under
multiple, cyclic loadings. Test data, produced on miniature disc specimens of the
type shown in Figure 2-12, has shown that results are in excellent agreement
with measurements made from standard samples. Tests were carried out over a
range of specimen thicknesses for different indenter dimensions. Results were
 2-21 
13828389
found to be reproducible with loads measured exhibiting sensible trends with
thickness and indenter size. These data, together with results from a range
of other alloys and pure metals, were analyzed [2.27] to calculate the tensile
stress, σUTS, as
σUTS =
LU
Eq. 2-4
t (0.14D.82Cl+2.17dr + 0.56)
Where Lu is the measured ultimate load in N, D is the punch diameter, Cl is
the punch/die clearance and dr is the displacement to failure all given in mm.
Predictions of tensile strength made using equation 2-4 based on the results
from small sample punch tests for both pure aluminium and copper as well as
21/4Cr1Mo low alloy steel illustrate the accuracy of this approach, Figure 2-15.
Figure 2-14
Schematic diagram showing the punch test apparatus with the borescope system
[2.26]
 2-22 
13828389
Figure 2-15
Comparison of predicted tensile strengths made using equation 2-4 with measured
values
The above experimental based approaches can be applied to assess key strength
parameters. However, using finite element analysis techniques methods have
been developed to permit the results of the punch tests to be used to compute full
stress strain curves, which are equivalent of those recorded using standard large
specimen methods. This information is of direct benefit in assessment of
performance since estimates of strength are one of the inputs required for
methods of calculating fracture toughness based on Charpy impact data,
Table 2-2.
2.4.3 Small Punch Test Assessment of FATT
The measurement of the small punch transition temperature, Tsp, involves a
similar approach to that used to determine Charpy FATT. Thus, a series of tests
are performed over a range of temperatures. Because of the lower constraint in a
small disc specimen compared to a notched Charpy bar, the small punch test
transition typically occurs between liquid nitrogen temperature (−196°C) and
room temperature. The total absorbed energy to the first peak load (peak load
defined as load followed by a load drop in excess of 10% of peak), measured as
the area under the small punch load-displacement curve, is then calculated. The
measured values of energy are then plotted against the test temperature, and Tsp
determined as the temperature at which the energy level is midway between the
upper-shelf and lower-shelf energy levels. The change in energy for punch and
Charpy tests on 2¼Cr1Mo piping steel illustrate typical behavior, Figure 2-16.
 2-23 
13828389
Figure 2-16
Brittle/ductile transition curves for 2¼Cr1Mo low alloy steel measured using small
punch tests, curve (left), and standard Charpy impact tests, curve (right) [2.28]
For individual materials the available Tsp data are then plotted as a function of
the known values of FATT to develop a correlation curve. Information has been
obtained on CrMoV rotor and bolting materials, Figure 2-17, on NiCrMoV LP
rotor steels, Figure 2-18, and on CrMo piping and pressure vessel steels,
Figure 2-19 [2.25].
Figure 2-17
Correlation developed between the transition temperature measured in small
punch tests and the FATT measured in Charpy tests for CrMoV low alloy steel
forgings [2.25]
 2-24 
13828389
Figure 2-18
Correlation developed between the transition temperature measured in small
punch tests and the FATT measured in Charpy tests for NiCrMoV LP rotor steel
forgings [2.25]
 2-25 
13828389
Figure 2-19
Correlation developed between the transition temperatures measured in small
punch tests and the FATT measured in Charpy tests for CrMo low alloy steels. The
dashed lines bound the data scatter and the solid line is the best estimate FATT
correlation based on results for a range of low alloy steels [2.25].
These data indicate that the correlations between the Charpy and punch test data
can be described using relationships of the form:
Eq. 2-5
FATT = A + B Tsp
Where A and B are empirical constants. Mean values for A and B are
summarized in Table 2-4. These values are based on evaluation of data for
CrMoV, NiCrMoV, and CrMo low alloy steels.
Table 2-4
Empirical constants identified for use in equation 2-5 which correlates FATT
measured by Charpy impact testing with Tsp the transition temperature measured
using punch tests
Material
For Data in °C
For Data in °F
A
B
A
B
CrMoV
458
2.54
775
2.54
NiCrMoV
364
2.31
613
2.31
CrMo
507
2.86
853
2.86
 2-26 
13828389
There is evidence that the embrittlement, which occurs during exposure to
elevated temperature, is related to grain size, d. Based on a comprehensive study
[2.29] of the embrittlement and fracture behavior of ex-service CrMoV bolts, it
has been suggested that grain size should be included in the correlation between
FATT and Tsp using the expression:
FATT = 1.35 Tsp – 26.6 (d)- 0.5 + 326
Eq. 2-6
Evidence for using this expression is shown in Figure 2-20, which indicates the
most brittle behavior found in ex-service CrMoV bolts occurred in samples with
the largest grain size. In general the embrittlement found in these bolts was the
result of grain boundary segregation of phosphorus, although in some cases
evidence of segregation of Sn and Sb was identified. In these cases the
appropriate data points have been identified in Figure 2-20.
It is apparent that the small punch technique offers the potential to measure
transitions in fracture behavior. Components such as turbine rotors, where in
service embrittlement is a concern, run/replacement decisions require knowledge
of actual fracture properties from the most susceptible locations. Using the latest
sampling methods it is possible to remove small material samples from selected
regions of the rotor bore. These specimens can be used to manufacture miniature
disc specimens of the type shown in Figure 2-12. Small punch testing then offers
an effective method to measure the actual ductile/brittle transition behavior.
Using the established correlations shown in equation 2-5, with the appropriate
constant in Table 2-3, the Charpy FATT can be determined. This knowledge
can then be used as described in Section 2.3.2 to estimate a value of K1c. In
combination with analysis to calculate component stresses and inspection data to
identify and characterize any defects present an accurate K1c value allows
estimates of the risk of brittle fracture to be made with confidence and avoids the
overly conservative assumptions which must frequently be made in the absence of
actual data.
 2-27 
13828389
Figure 2-20
Relationship between FATT measured in Charpy impact tests and Tsp, the transition
temperature measured using punch tests for CrMoV bolting steels showing the
influence of grain size on the level of embrittlement occurring [2.29]
2.4.4 Small Punch Test Assessment of Fracture Toughness
As indicated above the ability to remove small samples in an effectively non
destructive manner allowing punch testing programs to be carried out is a
significant benefit. Established technologies provide approaches to first estimate
Charpy FATT and then estimate K1c. Recent work is seeking to measure an
accurate K1c directly from the small specimen punch tests [2.28]. This approach
is outlined below.
For determination of fracture toughness (K1c, J1c) at room temperature, two
repeat small punch tests are conducted at room temperature for each material
investigated. Each test involved development of the load-displacement curve, and
identification of crack initiation with respect to the point on the loaddisplacement curve where the initiation occurs and with respect to where on the
test specimen the crack initiates. For identifying crack initiation, a fiberscopecharged coupled device (CCD) camera-video recorder combination system is
used. A schematic of the test setup used in all of the EPRI supported fossil power
plant research is shown in Figure 2-13.
 2-28 
13828389
The test data are then analyzed using a procedure (described in [2.30]) which
involves computing the critical strain energy density at the location of crack
initiation on the small punch specimen, using finite element analysis. This strain
energy density is then computed, also by finite element stress analysis, at the
crack tip of a plane-strain compact tension specimen, "analytically" loaded.
Initiation toughness is next estimated via a handbook J-integral solution at the
load level for which the crack-tip energy density just equals the critical small
punch-measured strain energy density.
The above procedure requires determination of the stress-strain constitutive
behavior of the material. The constitutive behavior is assumed to be RambergOsgood, power law hardening, and the power law constants are determined from
the observed load-displacement behavior by an optimal fitting technique detailed
elsewhere [2.27]. In effect, the procedure produces an estimate of the (tensile)
stress-strain behavior of the material at the test temperature.
As shown in Figure 2-21, the determinations of K1c measured directly from the
punch tests appear to be within ± 25% of the standard ASTM mean values
(where the ASTM values clearly measure initiation toughness). This excellent
agreement suggests that the small punch technology has the potential for direct
measurement of fracture toughness. If this approach becomes established, direct
measurement of actual properties will be possible for a wide range of plant
components. In that case the empirical correlations between fracture toughness,
Charpy and small punch transition will no longer be required and assessment of
embrittlement and the associated run/replace decisions will be easier.
 2-29 
13828389
Figure 2-21
Small punch test based K1c values compared with measurements made using
standard ASTM procedures for typical power plant steels [2.28].
2.4.5 Creep Embrittlement
Performing punch tests under controlled high temperature conditions has
permitted detail of materials creep behavior to be established. Work in this
area has predominantly been undertaken in Europe with information obtained
on a range of pure metals and engineering alloys including aluminum, copper,
low alloy steels, austenitic steels and Grade 91martensitic steel (for example,
2-30, 2-31).
In addition data have been reported examining the capability of this technology
to monitor the levels of prior creep damage under conditions where low ductility
failure is known to occur (that is, when fracture takes place as a consequence of
the nucleation and growth of creep cavities). Using a procedure similar to the
well-established iso-stress temperature acceleration method, a series of postexposure punch creep tests were performed on a CrMoV rotor steel at 190N and
temperatures of 665°C, 645°C, 625°C and 605°C, giving failure lives of 14 to
325 hours [2.31]. As shown in Figure 2-22, a reasonable straight line can
describe the results. Furthermore, the slope of this line is the same as that
 2-30 
13828389
describing results on new material, indicating that the rate controlling the
processes were similar. Extrapolation of the data to 585°C indicates an estimated
life under these conditions of about 900 hours. This is in reasonable agreement
with the test result obtained under these conditions.
Figure 2-22
Small punch creep tests on new and creep damaged CrMoV rotor steel. The punch
tests accurately determine the level of damage present [2.31]
The small punch test technology therefore provides reasonable:

Measurement of the creep strength of components

Estimates of the levels of in-service creep damage
These capabilities offer advantages in component assessment since the over
conservatism associated with basing performance estimates on minimum creep
properties can be avoided.
2.5 Metallographic Techniques
Metallographic techniques for characterization of the microstructure and
assessment of evidence of damage are well established. Techniques, which are
particularly relevant to brittle behavior, are summarized in the following
paragraphs, these include:

Optical metallography, with specific information presented describing
-
Grain size measurements
-
Phase identification
-
Assessment of phosphorus segregation
-
Evaluation of creep microvoids
 2-31 
13828389

Electron microscopy, with information presented describing
-
Scanning electron microscopy, with particular reference to fractographic
examination
-
Auger electron spectroscopy, with particular reference to compositional
analysis
Background regarding metallographic techniques is available in the EPRI Boiler
tube Metallurgical Guide and in ASM reference documents [2.32]. A list of
etchants commonly used in the metallographic preparation and evaluation of
selected alloys is presented in Table 2-5.
2.5.1 Optical Microscopy
In order to observe the microstructure, a metal sample is ground and then
polished to a plane and mirror-like finish. The prepared surface is then
chemically attacked with a selected solution, normally a dilute acid, for a short
period, a process called "etching." The choice of solution will influence the
microstructural feature attacked. Thus, for example, in carbon and low alloy
steels with a ferrite/carbide microstructure, a dilute solution of nitric acid in
methanol (known as nital) will show the grain structure. In this case the grainboundary atoms are more easily and rapidly dissolved or "corroded" than the
atoms within the grains. A small groove is left at the grain boundaries. Since a
groove will not reflect light in the same way as the flat, polished grains; the grain
boundaries appear as black lines and the ferrite grains appear light, Figure 2-23.
When low alloy steels are cooled rapidly from the normalizing temperature,
bainite, or, at the fastest rates, martensite will be formed. These metastable
microstructures form directly from the original austenite structure and for these
microstructures it is frequently necessary to measure the prior austenite grain size.
Typically a nital etch will not provide sufficient contrast to identify these grain
boundaries and etching with a saturated picric acid solution is preferred. A
typical bainitic microstructure for CrMoV low alloy steel is shown in
Figure 2-24.
 2-32 
13828389
Figure 2-23
Ferrite grains revealed in low carbon steel using a nital etch
Figure 2-24
Prior austenite grain structure revealed in bainitic CrMoV low alloy steel using a
saturated picric acid etch
2.5.2 Grain Size Measurements
Grain size can vary greatly depending on the alloy and heat treatment. For
reference, a grain diameter is about 0.001 inch across. Thus, there may be a
billion (109) grains per cubic inch of alloy. Within any one grain there are a very
large number of individual atoms. The diameter of an iron atom is about 10-8
(0.00000001) inch. So across a one-mil (0.001 inch) grain there are 100,000 (105)
iron atoms, with a grain boundary being about 2–10 atomic dimensions thick.
 2-33 
13828389
The ASTM grain-size number is one standard for determining the average grain
size. The ASTM grain size number "N" is defined by:
n = 2N-1
Eq. 2-7
Where "n" is the number of grains per square inch when viewed at a
magnification of l00x. The usual range of N is from 1-9. Typical ASTM grain
size charts are shown in Figure 2-25. Note that with this method as the grains
get smaller, the grain-size number gets larger.
Figure 2-25
Standard ASTM grain size charts for the classification of steels at 100 times
 2-34 
13828389
Table 2-5
Selected etchants used in the microstructural characterization of engineering
alloys. In most situations etchants should be prepared when needed. Application
for successful results is largely experienced based so that specific information
regarding etching conditions and times cannot be given.
Etchant
Composition
Comment
Carbon and Alloy Steels
Nital
2 ml HNO3 and 98 ml
ethanol
Good general-purpose etchant
to reveal microstructure.
Picral
4 gm picric acid, 100 ml
ethanol with 17%zephiran
chloride as a wetting
agent
Provides superior resolution of
fine carbides
Vilella’s reagent
5 ml HCl, 1g picric acid
and 100 ml ethanol
Reveals prior austenite grain
structure in bainitic and
martensitic microstructures
Picric Acid
1 g sodium
tridecylbenzene in 100 ml
saturated picric acid
Reveals prior austenite grain
structure in bainitic and
martensitic microstructures
Stainless Steels
Vilella’s reagent
5 ml HCl, 1g picric acid
and 100 ml ethanol
Outlines second-phase
particles (carbides, σ phase,
δ-ferrite), etches martensite
Glyceregia
3 parts glycerol, 2-5 parts
HCl, 1 part HNO3
Popular etch for all stainless
grades. Higher HCl content
reduces pitting tendency. Use
fresh, never store.
Electrolytic etch at
1.5-3 V dc for 3 s
56 g KOH and 100 mL
H2O
Reveals σ phase (red-brown)
and ferrite (bluish). Chi phase
colored same as sigma
Copper Alloys
Ammonium
Hydroxide/Hydrogen
Peroxide
20 mL NH4OH, 0-20 mL
H2O, 8-20 mL 3% H2O2
Widely used to reveal general
microstructure
Nickel based superalloys
Glyceregia
3 parts glycerol, 2-5 parts
HCl, 1 part HNO3
Widely used to reveal general
microstructure
The shape of individual grains is typically an irregular polyhedron and the grains
are packed together to fill the available space. Although grains are never spherical
the characteristic dimension is referred to as a "diameter." At equilibrium the
shape tends to minimize the grain-boundary surface area for a given volume of
metal within a grain. Thus, this attempt to minimize the surface-to-volume ratio
 2-35 
13828389
is the driving force for grain growth, (that is, within a given volume a few large
grains will have lower energy than a large number of small grains). Grains are
described as equiaxed when the characteristic dimensions are the same in all
directions. Grains are described as elongated when the characteristic dimensions
are not the same, but one direction is much longer than the others.
The size of the grains within a particular alloy can significantly affect the
properties, particularly yield strength and fracture behavior. Generally, improved
strength and toughness occur in fine grained materials because the small available
slip distance reduces the build-up of lattice defects (know as dislocations) at grain
boundaries. Moreover, fine grained material has significantly more grain
boundary area compared to coarse grained material, so that fine grain sizes also
help to minimize build up of trace elements. Thus, grain size will be important in
the assessment of embrittlement. However, care must be exercised when making
measurements of grain size. As with any metallographic technique it must be
remembered that the section prepared will give a 2-dimensional view of the
original 3-dimensional grain structure. Because the section chosen will be a
random plane through the section, and it is impossible for this section to
intersect the maximum dimension of each grain, the average measured grain size
will be less than the actual average grain size. Depending on the method used to
determine the measured average it is generally the case that the measured value
will be around 25 to 50% less than the actual value.
2.5.3 Specialist Etching for Phase Identification
Because engineering alloys frequently contain a range of inclusion types and
multiple phases, a number of different etching techniques have been developed to
preferentially attack specific constituents, thus aiding in identification. Full
details of the etchants and techniques available are given in reference 2.32 but the
benefits of this approach are illustrated here with reference to austenitic stainless
steel. This information has been summarized from the EPRI report Remaining
Life Assessment of Austenitic Stainless Steel Superheater and Reheater Tubes [2.33].
These techniques provide a metallographic method to differentiate between
sigma phase and carbide particles. This is important in assessment of
embrittlementp; further details of embrittlement due to sigma phase formation
are presented in Section 5 of this guideline.
A two-step etching technique is described below in which the sample is first
etched using Vilella’s reagent to outline the second phase particles. The sample is
next electrolytically etched using concentrated sodium or potassium hydroxide
(NaOH or KOH) to stain the sigma phase. The sample must not be wiped
following this etch because it will remove the stain.
Identification of sigma phase in Type 304H stainless steel is illustrated below.
The microstructure as delineated by alternate polish-etch sequence using
Villella’s reagent contains both large and small outlined second phase particles
(Figure 2-26). Electrolytic etching using concentrated NaOH (1.5 volt for
 2-36 
13828389
20 seconds) and KOH (1.5 volt for 10 seconds) stained the large particles but had
no effect on some of the smaller particles (Figure 2-26). The large particles,
therefore, are identified as sigma and the smaller particles that remain clear are
identified as carbides.
(A)
(B)
Figure 2-26
A service degraded Type 304H stainless steel tube sample showing stained sigma
phase particles with fully developed microvoids. Arrow in (A) marks sigma. Arrow
in (B) marks a carbide. (MAG: 1000X, Vilella’s Etch plus (A) NaOH and (B) KOH
electrolytic etch) [2.33].
2.5.4 Assessment of Phosphorus Segregation
The segregation of phosphorus to prior austenite grain boundaries is well
established as a cause of embrittlement in bainitic and martensitic steels, details
are provided in Section 7 of these guidelines. Since the degree of embrittlement
is related to the amount of P in the boundaries a practical metallographic method
of assessing grain boundary P is of direct benefit to programmes monitoring the
risk of brittle fracture in service, particularly for turbine rotors and fasteners. The
key stages involved with this metallographic procedure are given based on
information provided in publications on NiCrMoV rotor steels [2.34], CrMoV
bolts [2.35] and 17-4PH martensitic stainless steel [2.36].
 2-37 
13828389
Samples are initially prepared following standard metallographic techniques to a
1 μm diamond finish. Etching is then performed using picric acid based reagents
at a controlled temperature, time and a 1 cm2 sample size. The conditions used
are as follows:

Low alloy steels, a saturated solution containing 10 grams/litre of
sodium tridecyyl benzene sulfonate was applied at room temperature
for 2.5 hr [2.34]

Stainless steel, the reagent involved was ethanol, picric acid (60 g/l) and
benzalkonium chloride (20 g/l) as a wetting agent for 1 hour [2.36]
The depth of the grain boundary etch is then determined by marking the surface
with a appropriate hardness indent and performing an iterative polish with 3 μm
diamond paste until the required depth has been achieved. This has been taken as
the depth at which 90% or 100% of the prior austenite boundaries are removed.
Based on a Vickers hardness indent the maximum grain boundary depth, h, is
given by:
h = di – df/2√2 tan (68°)
Eq. 2-8
Where di and df are the initial and final diagonal lengths of the Vickers indent
respectively. These dimensions are shown schematically in Figure 2-27, with the
etched microstructure and selected polishing stages shown in Figure 2-28.
Figure 2-27
Schematic illustration of the relationship of the hardness indent to the etch depth of
the grain boundaries
 2-38 
13828389
Figure 2-28
Example of the iterative polishing process used to measure the depth of attack at
prior austenite grain boundaries in 17-4PH martensitic stainless steel. The hardness
indent is reduced in size as the material is polished away, with specific measured
depths indicated by the increasing values of h [2.36].
Based on comprehensive studies where Auger Electron Spectroscopy was
used to measure actual P concentrations [2.33, 2.35] it has been shown that
this approach provides an accurate method for monitoring segregation,
Figures 2-29 a and b.
 2-39 
13828389
(a)
(b)
Figure 2-29
Linear relationships between the depth of grain boundary etch and phosphorus
segregation for (a) NiCrMoV rotor steels [2.33] and (b) 17-4 PH martensitic
stainless steel [2.35]
 2-40 
13828389
Clearly the excellent agreement shown in Figure 2-29 indicates that this
metallographic technique is of significant benefit in determining P segregation.
However, two additional factors have been identified to aid assessment of
embrittlement in service components [2.34]. These are:

A three-stage replication process has been developed which allows the depth
of etching to be measured from the prepared surface of a component. In this
case the depth of grain boundary attack is measured in a scanning electron
microscope with the aid of reference microspheres of known size. Assessment
has shown that the depth measurements from the replicas are in close
agreement with data from specimens measured using the hardness
indentation followed by iterative polishing.

Measurements of ΔFATT have been made on samples of CrMoV rotor
steels also used for P segregation measurements. These data have shown that
there is a relationship between the depth of etch penetration and the increase
in fracture transition temperature. The agreement found for a number of
commercial heats of this steel is shown in Figure 2-30. The change in FATT
with aging could be reasonably described by the expression:
ΔFATT = 9h90 – 34
Eq. 2-9
Where h90 is the 90% of the full etch penetration depth. The scatter observed in
this figure arises from the variations inherent in the experimental measurements.
While a reduced level of variation would be preferred, the ability to estimate the
ΔFATT of commercial rotor steels within an accuracy of 20°C using equation
2-9 is of benefit.
 2-41 
13828389
Figure 2-30
Relationship between the depth of phosphoric acid etch depth and ∆FATT for
CrMoV rotor steels [2.34]
Thus, this approach provides a metallographic technique, which can be used
to make quantitative assessment of the level of embrittlement due to phosphorus
segregation. Undertaking these measurements provides key information
regarding the fracture behavior of in service rotors and other similar components.
As discussed in Section 7 of these guidelines, temper embrittlement resulting
from phosphorus segregation is a major factor in increasing the risk of brittle
fracture for a range of power plant low alloy steels.
2.5.5 Preparation and Etching to Reveal Creep Microvoids
Microstructural characterization of creep microvoids requires a distinct
preparation technique compared to normal metallographic preparation that does
not involve microvoids. The special procedure must be capable of removing the
disturbed layer of metal that will otherwise cover and mask microvoids. It must
also leave second phase particles intact since their dissolution would produce false
positive indications of creep damage.
 2-42 
13828389
The final and unique steps in this procedure consist of alternate polishing and
etching. The need for careful preparation, including repeat polishing and etching
stages, has been shown for both low alloy and stainless steels. The specific
details given below describe the method that has been found to give satisfactory
results for service degraded Types 304H, 321H, and 347H stainless steel boiler
tubes [2.33]:

Rough grind on water lubricated 80 and 120 grit abrasive belts

Hand grind on successively finer water lubricated 240, 320, 400, and 600 grit
papers

Rough polish with a napless nylon cloth impregnated with 6 micron
diamond paste

Etch with Vilella’s reagent for 4 minutes

Polish with a napless nylon cloth impregnated with 1 micron diamond paste

Etch with Vilella’s reagent for 3 to 4 minutes

Final polish with a low nap cloth impregnated with 0.05 micron alumina
paste

Etch with Vilella’s reagent for 2 to 4 minutes
Etching times are given as a range to accommodate the different etching
characteristics of the various stainless steel alloys. The minimum etching times
are usually adequate for Type 304H. Longer times may be required for Types
321H and 347H.
Since etching will be accelerated by heat generated during polishing that is
retained in the sample, the sample should be cooled under running water before
etching. A convenient method for applying Vilella's reagent is to drip it onto the
polished surface using a disposable pipette. Swabbing is not recommended
because it can cause stains.
The importance of proper metallographic preparation to reveal microvoids is
illustrated in the following example. The as-polished surface of a Type 304H
service exposed superheater tube reveals few small isolated voids in the
microstructure (Figure 2- 31A). Microvoids are apparent at this stage of
preparation only if they are filled or lined with oxides.
 2-43 
13828389
(A)
(B)
(C)
Figure 2-31
Micrographs of the same section of service degraded Type 304H stainless steel
tube sample showing (A) small voids in the as-polished condition, (B) outlined
second phase particles with some microvoids, and (C) fully developed microvoids
(black cavities). Arrows mark the same location (A) as polished. (B) 1 minute etch.
(C) Multiple 3, 3, and 2 minute etches. (MAG: 500X, Vilella’s etch) [2.33].
 2-44 
13828389
Second phase particles are revealed by a one-minute etch with Vilella’s reagent
(Figure 2- 31B). This single etch, however, does not reveal the full extent of
the microvoids. Upon completion of the recommended procedure, the number
and size of the microvoids reveals severe creep damage in this tube sample
(Figure 2- 31C). Similar examples illustrating the necessity of proper
metallographic preparation have also been shown for Type 321 stainless steel.
In both of these examples, creep damage is sufficiently severe to result in the
alignment of microvoids normal to the principal stress. These microvoids formed
preferentially at the interface to second phase particles.
2.5.6 Electron Microscopy
A range of sophisticated high resolution techniques are available based on
instruments, which generate and focus a beam of electrons. In the present
guideline information is provided regarding Scanning Electron Microscopy and
Auger Electron Spectroscopy as techniques using these items of equipment are
most commonly used in the assessment of fractures.
Scanning Electron Microscopy
In the Scanning Electron Microscope (SEM), an electron beam is generated,
accelerated by a high voltage and focused into a fine beam by a series of
electromagnetic lenses. The electron beam is rastered across the specimen surface
and the intensity of the secondary electrons produced from the specimen surface
is monitored. These signals are collected by a detector, amplified and displayed
on a cathode ray tube. A topographical image of the ‘surface’ of the specimen is
therefore displayed on the screen.
As the electron beam enters the material, secondary electrons, back scattered
electrons and X-rays, will be produced from a teardrop shape within the
specimen. Typically, the image displayed is that arising from the secondary
electrons which are ejected from the surface (typical depth ~ 10 nm). For specific
applications, imaging can be carried out using the high energy back-scattered
electrons. These back-scattered electrons can be produced from a depth of
~ 1 µm into the sample. The X-rays emitted can be collected by a suitable
detector and, with the aid of suitable software, analyzed to provide information
regarding the composition of the material. This elemental analysis typically
involves an energy dispersive spectroscopy (EDS) attachment to a scanning
electron microscope. It should be noted that, because the X-rays are generated
from a teardrop shape volume within the sample, the results obtained always
involve an average from the volume of material.
The accelerating voltage that is selected influences the depth and diameter of the
teardrop. It is generally the case that the higher the accelerating voltage, the
deeper the beam penetration into the specimen. The typical accelerating voltage
that is used for examination of metallurgical fractures is 20 kV.
 2-45 
13828389
Compositional Analysis
Energy dispersive spectroscopy (EDS) is useful in the identification of inclusions
and the phases present. For example, the compositions of typical inclusions such
as manganese sulfides or alumina can be readily differentiated from the elements
present. Assessment of different phases requires knowledge of the expected
background matrix composition as well as the compositions of the possible
phases. The required information is illustrated with reference to EDS
confirmation of the metallographic identification of sigma and carbide phases
described earlier using specialist etching. EDS spectra obtained from the Type
304H austenite matrix and a large second phase particle (metallographically
identified as sigma as in Figure 2-26a) are shown in Figure 2-32.
 2-46 
13828389
Figure 2-32
Energy dispersive spectra from a Type 304H stainless steel tube showing the
composition of the austenite matrix (a), and a sigma phase particle (b). Note the
high chromium/iron (Cr/Fe) ratio of the sigma phase compared to the austenite
matrix [2.33].
 2-47 
13828389
Semi-quantitative analysis of these spectra indicates the following approximate
compositions for the matrix and sigma phase [2.33]. It is important to note that
the higher chromium/iron ratio in the second phase compared to the matrix is
consistent with sigma phase.
Element
Iron
Chromium
Nickel
Silicon
Matrix
73%
18%
8%
0.6%
Sigma
61%
32%
4%
0.6%
The use of this form of compositional analysis provides an important tool for
identification of inclusions and phases. Accurate metallurgical characterization is
important since:


Inclusions can:
-
Aid the nucleation of creep voids and thus promote brittle intergranular
fracture
-
Act as initiation sites for fracture
-
In extreme cases, link rapidly to form cracks
The presence of specific phases can promote brittle fracture, for example:
-
Graphite in carbon and C-Mo steels operating at temperatures up to
about 1000°F
-
Sigma phase in stainless steels
Details regarding embrittlement from phase changes are presented in Section 5
of these guidelines.
Auger Electron Spectroscopy
In cases where embrittlement is caused by segregation of trace elements such as
phosphorus to grain boundaries, for example, as in Temper Embrittlement
described in Section 7, the averaging effect of compositional analysis from X-rays
generated using the SEM severely limits analytical capabilities. To fully
characterize and quantify grain boundary segregation requires the use of Auger
Electron Spectroscopy (AES).
AES is an extremely surface sensitive technique that allows the surface layers of
a material to be examined with excellent depth resolution. This is because only
Auger electrons from the outermost atom layers of a solid survive to be ejected
and measured, without being inelastically scattered. Indeed, since the 1001000eV electrons analyzed in AES have mean free paths (l) ranging from 1 to
3 nm it is possible to detect a single monolayer of segregation. Furthermore,
Auger analysis combined with Argon ion sputtering enables depth composition
profiles to be generated, thereby providing information as to the depth of any
surface layers. The use of in-situ fracture samples, combined with the ability of
AES to detect sub-monolayers with high spatial resolution, has increased the
 2-48 
13828389
understanding of interfacial fracture in metals and alloys and in particular the
effect of grain boundary segregants. Typical AES results from an intergranular
brittle fracture of a CrMoV low alloy steel are shown in Figure 2-33. The AES
has clearly identified that the grain boundaries contain high levels of phosphorus
that had resulted in significant embrittlement. Auger analysis has also been
widely used for the analysis of passive films and surface oxides on atomized
powders.
Figure 2-33
A scanning electron micrograph showing the brittle intergranular fracture of an exservice CrMoV bolt (a) with AES results from a grain boundary facet showing the
high levels of P present which has embrittled the microstructure (b) [2.34]
Auger analysis can also be used to map the compositions of elements on a
fracture surface. The example shown in Figure 2-34 compares a detailed scanning
electron micrograph of an area of a grain boundary facet on an intergranular
fracture surface with a compositional map of the same area produced by AES.
The compositional map showed small particles rich in Cr and Sb were associated
with the fracture of the rotor steel.
 2-49 
13828389
Figure 2-34
Scanning electron micrograph showing detail of an intergranular fracture
surface (a), and an AES surface analysis showing that the particles highlighted on
this surface contained high levels of Sb and Cr. In this image the background
shows a general level of iron (b).
2.6 Concluding Comments
A key factor when assessing the risk of brittle fracture is knowledge of the
material properties. In general, the most accurate measurements of appropriate
properties are obtained by undertaking laboratory tests using procedures specified
by applicable standards. However, assessment of the future serviceability of an
existing major component must frequently be carried out when it is not possible
to obtain the amount of material necessary to perform standard tests. To aid
engineering judgment in these circumstances, methods have been developed to
measure properties using:

Miniature specimen techniques

Metallographic examination
Small specimen techniques based on the punch test methodologies are playing an
increasing role in evaluations of structural integrity. Traditional metallographic
methods for characterization of microstructure are well established and continue
to provide key information in component assessment in general and the analysis
of failures in particular. However, it is critical that these traditional methods are
properly applied to fully characterize composition and microstructure. The
specialist metallographic techniques described here then offer particular
advantages to assessment of brittle behavior. The development of techniques,
such as the picric acid method for assessment of P segregation and specialist
etching for phase identification in stainless steels, has been possible because
direct links between key microstructural features and critical properties have been
established. When recommended mechanical tests cannot be carried out, the
application of detailed microstructural analysis can frequently provide component
specific knowledge vital to assessment of component behavior.
 2-50 
13828389
2.7 References
2.1
American Society for Testing and Materials: Annual Books of ASTM
Standards, Section 3, Metal Test Method and Analytical Procedures,
Vol. 03.01 Metals-Mechanical Testing; elevated and low temperature
tests, metallography.
2.2
ASM Metals Handbook, Ninth edition, Vol. 8: “Mechanical Testing,”
American Society for Metals, 1985.
2.3
ASM handbook, Vol 19: “Fatigue and Fracture,” ASM International,
1996.
2.4
“Standard Methods for Notched Bar Impact Testing of Metallic
Materials,” E 23, Annual Book of ASTM Standards, Vol. 03.01, ASTM,
Philadelphia, 1984, p. 210–233.
2.5
R. Viswanathan, “Damage Mechanisms and Life assessment of High
Temperature Components,” 1989, ASM International.
2.6
F. Masuyama, T. Yokoyama, Y. Sawaragi, and A. Asada, Development
of Tungsten Strengthened Low Alloy Steel with Improved Weldability,
Service Experience and Reliability Improvement: Nuclear, Fossil, and
Petrochemical Plants, PVP Vol 288, American Society of Mechanical
Engineers, 1994, p. 141–146.
2.7
J. R. Low, Jr., The Effect of Quench-Aging on the Notch Sensitivity of
Steel, in Welding Research Council Research Report, Vol. 17, 1952,
p. 253s–256s.
2.8
D. E. Driscoll, Reproducibility of Charpy Impact Test, in Symposium
on Impact Testing, STP 176, American Society for Testing and
Materials, 1956, p. 70–75.
2.9
G. C. Sih, “Handbook of Stress Intensity Factors for Research
Engineers,” Institute of Fracture and Solid Mechanics, Lehigh
University, 1973.
2.10
H. Tada, P. Paris, and G. Irwin, “The Stress Analysis of Cracks
Handbook,” Del Research Corp., Hellertown. PA, 1971.
2.11
D. P. Rooke and D. J. Cartwright, “Compendium of Stress Intensity
Factors,” HMSO London, 1976.
2.12
“Standard Method for Plane Strain Fracture Toughness of Metallic
Materials,” ASTM E399, American Society for Testing and Materials,
Philadelphia, 1985.
2.13
J. M. Barsom and S. T. Rolfe, Correlations Between KIc and Charpy
V-Notch Test Results in the Transition Temperature Range, in Impact
Testing of Materials, STP 466, ASTM, Philadelphia, 1979, p. 281–302.
 2-51 
13828389
2.14
R. H. Sailors and H. T. Corten, Relationship Between Material Fracture
Toughness Using Fracture Mechanics and Transition Temperature
Tests, in Fracture Toughness, Proceedings of the 1971 National
Symposium on Fracture Mechanics, STP 514, Part II, ASTM,
Philadelphia, 1972, p. 164–191.
2.15
B. Marandet and G. Sanz, Evaluation of the Toughness of Thick
Medium-Strength Steels by Using Linear Elastic Fracture Mechanics
and Correlations Between KIc and Charpy V-Notch, in Flaw Growth and
Fracture, STP 631, ASTM, Philadelphia, 1977, p. 72–95.
2.16
J. A. Begley and W. A. Logsdon, “Correlation of Fracture Toughness
and Charpy Properties for Rotor Steels,” WRL Scientific Paper 71-1E7MSLRF-P1, Westinghouse Research Laboratory, Pittsburgh, July 1971.
2.17
T. Iwadate, J. Watanabe, and Y. Tanaka, “Prediction of the remaining
life of high temperature/pressure reactors made of CrMo steels, Trans.”
ASME, J. Pressure Vessel Tech., Vol. 107, Aug 1985, pp. 230-238.
2.18
S. T. Rolfe and S. R. Novak, Slow-Bend KIc Testing of MediumStrength High-Toughness Steels, in Review of Developments in PlaneStrain Fracture Toughness Testing, STP 463, ASTM, Philadelphia,
1970, p. 124–159.
2.19
R. A. Wullaert, Fracture Toughness Predictions from Charpy V-Notch
Data, in What Does the Charpy Test Really Tell Us?, Proceedings of the
American Institute of Mining, Metallurgical and Petroleum Engineers,
Denver, American Society for Metals, 1978.
2.20
R. T. Ault, G. M. Wald, and R. B. Bertola, “Development of an
improved Ultra High Strength Steel for forged aircraft components,”
AFML TR 7127, Air Force Materials Laboratory, Wright Patterson Air
Force Base, OH, 1971.
2.21
T. Iwadate, T. Karushi and J. Watanabe, “Prediction of Fracture
Toughness K1c of 21/4Cr1Mo Pressure Vessel Steel from Charpy V
notch Test Results,” in Flaw Growth and Fracture, STP 631, American
Society for Testing and Materials, Philadelphia, 1977, pp. 493–506.
2.22
S. T. Rolfe and J. M. Barson, “Fracture and Fatigue Control in
Structures–Applications of Fracture Mechanics,” Prentice-Hall, 1977.
2.23
L. H. Bisbee, J. D. Parker, and D. Mercaldi, “SSAM-a system for nondestructive material removal,” Condition Monitoring and Diagnostic
Engineering Management, Adam Hilger, 1991, pp. 520–524.
2.24
J. R. Foulds, C. W. Jewitt, and R. Viswanathan, “Miniature Specimen
Test Techniques for FATT,” Int Conf on Power Generation, Americam
Society of Mechanical Engineers, 1991.
2.25
J. Foulds and R. Viswanathan, “Small Punch Testing for Determining
the Material Toughness of Low Alloy Steel Components in Service,”
Journal of Engineering Materials and Technology, Vol. 116, 1994,
pp. 457–464.
 2-52 
13828389
2.26
J. Foulds and R. Viswanathan, “Determination of the Toughness of
In-service Steam Turbine discs using Small Punch Testing,” Journal
of Materials Engineering and Performance, Vol. 15, No5, 2001,
pp. 614–619.
2.27
S. D. Norris and J. D. Parker, “Deformation Processes During Disc
Bend Loading,” Materials Science and Technology, Vol. 12, 1996,
pp. 163–170.
2.28
S. D. Norris and J. D. Parker, “The effect of Microstructure on Fracture
Mechanisms of 2.25Cr1Mo low alloy steel, Part A: the influence of
inclusions and Part B: the influence of Carbides, Int. J. Pressure Vessels
and Piping, Vol. 67, 1996, pp. 317–339.
2.29
J. H. Bulloch, “Some Comments Concerning thye Chronic Problem of
Reverse Temper Embrittlement (RTE) in Low Alloy Steels” Key
Engineering Materials, Vols. 118–119, 1996, pp. 69–84.
2.30
J. D. Parker and J. D. James, “Creep Behavior of Miniature Disc
Specimens of Low Alloy Steel,” Int Conf Developments in Progressing
Technology, American Society for Mechanical Engineers, Vol. 279,
1994, pp. 167–172.
2.31
J. D. Parker, G. C. Stratford, N. Shaw, G. Spink, and H. Metcalfe, “The
Application of Miniature Disc Testing for the Assessment of Creep
Damage in CrMoV Rotor Steels,” BALTICA IV–Plant Maintenance,
1998, pp. 477–488.
2.32
Metals Handbook, Ninth edition, Vol. 9: “Metallography and
Microstructures,” American Society for Metals, 1985.
2.33
Remaining life Assessment of Austenitic Stainless Steel Superheater and
Reheater Tubes. EPRI, Palo Alto, CA: 2002. 1004517.
2.34
R. Viswanathan, S. M. Bruemmer, and R. H. Richman, “Etching
Technique for Assessing Toughness Degradation of In-Service
Components,” J of Engineering Materials and Technology, Vol. 110,
1988, pp. 313–318.
2.35
J. J. Hickey, “Investigation of Semi-nondestructive and Nondestructive
Techniques for Assessment of In-service Toughness Degradation in
CrMoV Steels,” Masters Thesis, 1992, University of Limerick.
2.36
F. Christien, R. Le Gall, and G. Saindreman, “Phosphorus Grain
boundary Segregation in Steel 17-4 PH,” Scripta Materialia, 48, 2003,
pp. 11–16.
 2-53 
13828389
13828389
Section 3: Metallurgy of Steels
3.1 Introduction
This section presents general background regarding the microstructures formed
in iron–carbon alloys, known as steels, under equilibrium conditions. Then,
further more detailed information is provided regarding:

Structures formed under non-equilibrium conditions

Continuous cooling transformations and the diagrams available to select
cooling rates necessary to achieve a particular microstructure

The effects of different elements commonly found in steels on microstructure
and properties

The general classification of types of steel depending on composition

The composition and microstructure of selected steels commonly used in
power plant applications
Appendix A contains a Glossary of Metallurgical Terms. This provides brief
statements defining generally used terminology relating to steels manufacture,
microstructure and properties. A large number of publications are available
describing Ferrous Metallurgy in detail (for example, references 3.1, 3.2, 3.3,
3.4). These should be studied as necessary.
3.2 Background
Steels can be defined as iron-carbon alloys containing less than 2.0 wt% C, with
alloys containing higher carbon levels defined as cast irons or pig irons (usually
with about 2 to 3.5 wt% C). The microstructures of steels can then be
understood from the iron-iron carbide (Fe-Fe3C) equilibrium diagram,
Figure 3-1:

From the melting point (1807K or 1534°C) to 1672K (1399°C), δ-iron is
present with a body centered cubic (bcc) structure

From 1663K to 1183K (1390°C to 910°C), γ-iron, called 'austenite', is
present. This has a face centered cubic (FCC) structure

From 0 to 1183K (910°C), α-iron, called 'ferrite', is present. This has a bcc
structure
 3-1 
13828389
Carbon dissolves interstitially in iron but the interstices in FCC austenite (γ) are
larger than those in bcc ferrite (α) so the maximum solubility of carbon in
austenite is 2.0% at the eutectic temperature of 1403K (1130°C), whereas a
maximum of 0.02%C dissolves in ferrite at 996K (723°C). This is called the
'eutectoid temperature'. Above these solubility limits, carbon usually exists as iron
carbide (Fe3C, called 'cementite').
Figure 3-1
The iron carbon equilibrium diagram, which shows how the phases present
change with temperature and carbon composition
Since steels contain less than 2.0%C, the microstructures of steels cooled at
equilibrium rates can be understood using only the 'eutectoid region' of the
Fe/Fe3C diagram. This eutectoid region is similar to a simple eutectic except that
at the eutectoid point (0.8%C), the solid austenite changes to a two-phase
eutectoid of α-iron and Fe3C (whereas, with a eutectic, a liquid changes to a two
phase solid). The eutectoid structure is called 'pearlite' and consists of alternate
fine platelets of ferrite and cementite.
 3-2 
13828389
The microstructures present for equilibrium cooling are illustrated for an 0.4%
carbon steel in Figure 3-2. At point c, the temperature is above about 800°C and
the microstructure is fully austenitic. On cooling just below the AR3 temperature,
ferrite grains nucleate at austenite grain boundaries. Continued cooling to a point
just above AR1 increases the amount of ferrite, then on cooling below the AR1
the remaining austenite transforms to pearlite.
Figure 3-2
Detail of the iron carbon diagram illustrating microstructures formed during
equilibrium cooling
At a given carbon level the proportion of each equilibrium phase present can be
calculated using the Lever rule. As an example, for an 0.4 wt% C steel, at a
temperature given by point c, the wt fraction of austenite = 0.4 - 0.02 = 0.49
0.8 - 0.02
At temperature point, d, just below the eutectoid temperature, austenite has
transformed to pearlite, therefore the wt fraction of pearlite = 0.49 and the wt
fraction of ferrite = 0.51.
 3-3 
13828389
Similarly, for a 1 wt% C steel cementite forms on cooling to temperature c and
on cooling to d, the wt fraction of pearlite = 6.67 −1.0 = 0.97
6.67 − 0.8
and the wt fraction of cementite = 1.0 − 0.8 = 0.03.
6.67 − 0.8
3.3 Non-Equilibrium Cooling of Steels
It is the existence of the austenite-to-ferrite transformation, combined with the
marked change in carbon solubility, which accounts for the enormous range of
microstructures and properties possible in steels. Equilibrium slow cooling from
the austenitic range gives microstructures containing either ferrite or cementite,
together with pearlite.
Since ferrite is soft and ductile, whereas cementite is hard and brittle, the
hardness and strength increase (but the ductility and toughness decrease) with
increasing carbon content as the proportion of cementite increases. Unlike slow
cooling from the austenite range (called 'annealing'), air-cooling is called
'normalizing'. Normalizing allows less than the equilibrium proportions of ferrite
or cementite to separate from the austenite, resulting in a higher proportion of
pearlite with decreased platelet size and spacing. With even more severe cooling
procedures, the eutectoid transformation can be suppressed, that is, quenching
from the austenite range can produce two entirely different types of structure,
namely:

Bainite is formed when austenite is quenched to temperatures around
180–425°C (356°F to 800°F) and held at the quench temperature for
some time (or, frequently, by quenching into oil). Bainite consists of a fine
submicroscopic dispersion of Fe3C particles in a highly strained α matrix.

Martensite is formed following rapid quenching to lower temperatures than
those involved in forming bainite. In this case, the severe quench retains
carbon in solid solution in a distorted body-centered tetragonal iron lattice.
Martensite is hard and brittle, but the toughness can be improved, with a
corresponding reduction in hardness by tempering, that is, heating
martensite to 200–720°C (392°F to 1330°F) allows transformation to a
structure consisting of very small Fe3C particles or precipitates in ferrite. The
size and spacing of these cementite particles increases with increasing
tempering times and temperatures, giving a progressively softer but less
brittle product. Heat-treatment of steels by quenching and tempering
therefore offers a means of optimizing strength and toughness.
The transformation of austenite to pearlite (at the eutectoid point, 0.8%C)
and the austenite to bainite transformation (at any carbon content) occur by
diffusion-controlled processes of nucleation and growth of the new phases so
that, on quenching austenite to temperatures which allow these transformations,
some time elapses before the transformations begin and further time is needed
before the transformations are complete. In contrast, the transformation of
austenite to martensite is diffusionless and almost instantaneous.
 3-4 
13828389
3.4 Continuous Cooling Transformation
Continuous cooling transformations are distinct from isothermal transformations
in that the time at temperature is dictated by the cooling rate that a sample
has experienced. Welding is a typical process where an understanding of
transformation behavior is important since the practicalities of welding mean
that isothermal transformations cannot be applied and weld and heat affected
zone (HAZ) microstructures are thus a function of the time at peak temperature
and the subsequent cooling rate. The use of preheating can assist in the control
of cooling rate and the required conditions are specified in applicable codes.
Continuous cooling transformation (CCT) diagrams show the transformations
that take place in a continuously cooled sample of material. These diagrams can
predict the microstructure of a sample at any cooling rate covered by the diagram.
CCT diagrams are generated from controlled heating and cooling experiments
carried out while monitoring the physical dimensions of the sample (that is,
diameter or length). Because there will be a change in dimensions when
transformation from BCC ferrite type microstructures to austenite occurs, the
heating curve shows inflexion points allowing the AC1 and AC3 temperatures to
be determined. Similarly the dimensional changes on cooling allow the specific
transformation temperatures to be determined so that by conducting a set of
experiments on a selected alloy the transformation behavior can be established.
The effect of heating and cooling cycles on specimen dimensions is illustrated in
Figure 3-3. The CCT curves measured for carbon and 2¼Cr1Mo steel are shown
in Figures 3-4 and 3-5. The additional alloying elements present in 2¼Cr1Mo
steel increase hardenability allowing bainite and martensite to form at slower
cooling rates than for carbon steel.
 3-5 
13828389
Figure 3-3
Illustration of the dimensional changes that occur on heating and cooling through
the temperature range where microstructural transformations take place
Continuous cooling transformation curves can be used to develop required
microstructures through control of the applied thermal cycles for a number of
different manufacturing processes. However, it is important that the thermal
cycle used to produce the CCT curve is relevant to the particular process. One
important factor to be considered is the prior austenite grain size, that is, the
grain size at the time of first transformation from austenite. The prior austenite
grain size has been shown to affect the transformation start temperature or AR3
for hot rolling and annealing processes. For example, a ten-fold increase in the
prior austenite grain size, from 20 to 200 microns, decreased the AR3
temperature by around 50°C at a cooling rate of 0.5°C/s. However, this effect
had reduced to a difference of only a few degrees at a cooling rate of 30°C/s [3.5].
 3-6 
13828389
(a)
(b)
Figure 3-4
CCT diagram for carbon steel (a) and for 2¼Cr1Mo steel (b)
The austenite grain size can be controlled in two ways:

Through a change in the temperatures used

Through a change in the time held at a single temperature
A variation in the peak temperature between 955 and 1390°C for CrMoV piping
steel resulted in an increase the grain size from 20 microns to 200 microns. For a
given temperature the grain size increases parabolically with time so for any
time/temperature combination the grain size at a given time, t, is given by the
equation [3.6]:
Dt2 - Do2 = Constant [exp {- Qc / RT} t]
Eq. 3-1
Where Do and Dt are the initial and final grain size, T is the temperature, R is
the gas constant and Qc is the activation energy.
 3-7 
13828389
The influence of thermal cycles in modifying microstructure is most commonly
noted in weldments. Variations in peak temperature and cooling rate result in
a range of different grain sizes and transformation microstructures within the
weld metal and the HAZ. A macrograph of a typical CrMo low alloy steel
weld is shown in Figure 3-5a, with detail of typical weld metal and HAZ
microstructures presented in Figures 3-5b and 3-5c, respectively. As shown,
within the weld and HAZ the cooling rate is relatively rapid so that the
predominant microstructure is bainite. However, because of the differing thermal
histories a wide range of prior austenite grain sizes is present. Relatively slow
cooling rates result in a predominantly ferritic microstructure. However, for
sections which have been cooled rapidly, mostly bainitic structures will be
present.
Figure 3-5
Typical weld microstructures in CrMo low alloy steel shown in a macrosection (a),
with detail of typical microstructures in the weld metal (b), and heat affected
zone (c)
3.5 Effects of Composition
The microstructure and hence properties of steels are determined primarily by
composition and heat treatment. The general effects of individual elements are
summarized below. However, applicable references should be studied for detailed
information regarding the microstructure and properties of a specific alloy or
alloy type.
ALUMINUM – Al, is used to deoxidize steel and control grain size. Grain size
control is affected by forming a fine dispersion with nitrogen and oxygen, which
restricts austenite grain growth. Aluminum is also an extremely effective nitride
former in nitriding steels.
 3-8 
13828389
ANTIMONY – Sb and ARSENIC – As, are trace elements, which are believed
to reduce ductility through temper embrittlement.
BORON – B, is added between 0.0005–0.003% to significantly increase the
hardenability, especially for low carbon alloys. It does not affect the strength of
ferrite, therefore not sacrificing ductility, formability or machinability in the
annealed state.
CALCIUM – Ca, is used to control the shape, size and distribution of oxide and
sulfide inclusions. Benefits of having fine, well distributed inclusions include
improved toughness and machinability.
CARBON – C, is one of the most important alloying elements. It is essential for
the formation of cementite, pearlite, bainite, and iron-carbon martensite.
Compared to steels with similar microstructures, strength, hardness,
hardenability, and ductile-to-brittle transition temperature are increased with
increasing carbon content. Toughness and ductility of pearlitic steels are
decreased with increasing carbon content. The significant increase in
hardenability with increasing carbon content results in decreased weldability.
CHROMIUM – Cr, influences hardenability and is a carbide former and
stabilizer. It is used in low alloy steels to increase 1) resistance to corrosion and
oxidation, 2) high temperature strength, 3) hardenability, and 4) abrasion
resistance in high carbon alloys. Straight chromium steels are susceptible to
temper embrittlement and can be brittle so that steels for elevated temperature
service tend to contain both chromium and molybdenum. At composition levels
of around 9 to 13% Cr the increased hardenability is such that for normal cooling
rates martensite is formed. In the absence of nickel, high chromium steels, above
about 18%Cr, are fully ferritic and are used where high resistance to corrosion is
required.
COPPER – Cu, is detrimental to hot workability and subsequent surface quality
and it may reduce creep ductility. It is used in certain steels to improve resistance
to atmospheric corrosion.
LEAD – Pb, improves machinability. It does not dissolve in steel but is present
as metallic globules. Environmental concerns are resulting in a decreased usage of
lead in the steel industry.
MANGANESE – Mn, is important because it 1) controls transformation
kinetics on cooling from austenite, 2) deoxidizes the melt, and 3) facilitates hot
working of the steel by reducing the susceptibility to hot shortness caused by the
presence of free sulfur, that is it combines with sulfur to form MnS inclusions.
Manganese increases the tendency for trace elements to cause temper
embrittlement.
 3-9 
13828389
MOLYBDENUM – Mo, increases hardenability of steels and helps maintain a
specified hardness. Even in small amounts (0.1 to 0.5%), molybdenum increases
high temperature tensile and creep strengths and acts to reduce the effects of
trace elements, such as P, in causing temper embrittlement.
NICKEL – Ni, is used in low alloy steels to reduce the sensitivity of the steel to
variations in heat treatment and distortion and cracking on quenching. It also
improves low temperature toughness and hardenability. In stainless steels, at
levels above about 8% Ni, the austenite is stabilized to room temperature.
NIOBIUM – Nb (Columbium – Cb), forms stable carbides that increase
strength at elevated temperatures, and, by providing a finer grain size, lowers the
fracture transition temperature. Niobium is added to austenitic stainless steels to
form carbides in ‘stabilized’ grades to reduce risk of sensitization.
NITROGEN – N, increases the strength, hardness and machinability of steel,
but it decreases the ductility and toughness. In aluminium killed steels, nitrogen
combines with the aluminium to provide grain size control, thereby improving
both toughness and strength. Nitrogen can reduce the effect of boron on the
hardenability of steels.
PHOSPHORUS – P, is generally restricted to below 0.04 weight percent to
minimize its detrimental effect on ductility and toughness. Certain steels may
contain higher levels to enhance machinability, strength and/or atmospheric
corrosion resistance.
SILICON – Si, is one of the principal deoxidizers with the amount used
dependent on the deoxidization practice. Silicon can increase high temperature
strength and reduce the amount of surface scale formed during exposure to high
temperature but has been shown to increase temper embrittlement caused by
segregation of trace elements in low alloy steels.
SULFUR – S, is detrimental to fracture strength so that Mn must be added to
form inclusions. These manganese sulfide stringers can reduce transverse strength
and impact resistance and fine inclusions on grain boundaries can facilitate the
formation of creep cavities.
TIN – Sn, is a trace element which is believed to increase temper embrittlement
and has been shown to reduce creep ductility and accelerate creep damage
development.
TITANIUM – Ti, is added to steels containing boron because it combines
preferentially with oxygen and nitrogen, thus allowing the boron to increase
hardenability. Titanium, as titanium nitride, also provides grain size control at
elevated temperatures.
 3-10 
13828389
TELLURIUM – Te, may be added to modify sulfide type inclusion size,
morphology and distribution. The resulting sulfide type inclusions are finer and
remain ellipsoidal in shape following hot working, thereby improving transverse
fracture properties.
TUNGSTEN – W, increases hardenability and forms carbides, acts in a similar
manner to Mo.
VANADIUM – V, additions up to 0.05% increase hardenability whereas larger
amounts tend to reduce hardenability because of extensive carbide formation.
The presence of vanadium carbides or carbonitrides improves elevated-strength,
provides resistance to tempering and hydrogen attack as well as inhibits grain
growth during heat treatment. Proper application leads to improved strength and
toughness of hardened and tempered steels, however, excessive strengthening can
lead to reheat cracking associated with welds.
3.6 Classification of Steels
In addition to carbon, all commercial steels contain varying amounts of Mn, Si,
S, P, gases and other trace impurities that are present from the methods used
during steel production. Consequently, steels are defined as plain carbon even
when they contain one or more percent manganese, up to 0.3%wt Si, 0.06%wt P
and S, etc. However, the structure and properties of plain carbon steels depend
not only on composition, but also on the heat-treatment and on the hot and cold
working operations prior to or after heat-treatment. Consequently, specifying
only the composition is insufficient to provide an adequate description of
properties. Although there is no universally agreed system of steel specifications,
for most general purposes,

Plain carbon steels are normally grouped as low-carbon or mild steels
(< 0.25%wt C)

Medium-carbon steels (0.25 to 0.6%wt C)

High-carbon or plain carbon tool steels (0.6%wt C and above)
The low carbon grades are used for sheet and strip manufacture for cans,
pressings, etc., where ductility and toughness combined with reasonable strength
is required. The stronger mild steels (0.2%wt C) are then used as weldable
structural steels. Medium carbon steels are chosen for casting, forging, etc. of
axles, gears, wire ropes and springs. The hardness and strength of high carbon
steels then result in their selection for tools, ball bearings, and dies.
With plain carbon steels, the critical cooling rate needed to form martensite
depends on the C content and, with less than 0.3%wt C, even water quenching
austenite does not produce a fully hard martensitic product. Even with higher C
levels, the critical cooling rates are high, which limits hardenability, that is,
because of the relatively low thermal conductivities of steels; there is a maximum
section thickness that can be hardened from surface to center. With plain carbon
 3-11 
13828389
steel, this hardening depth is not more than 2-3 cm (about 1 inch) and, with
rapid quenching, the volume change accompanying the austenite-martensite
transformation can cause severe distortion and cracking of higher carbon steels.
Although plain carbon steels are perfectly satisfactory for most applications, there
are distinct limits to the combination of strength and toughness attainable, to the
section sizes that can be fully hardened, and to the temperatures at which even
high carbon steels can operate without softening. These property limitations can
be alleviated, and new property ranges introduced by alloying. Once again, no
classification system is universally agreed for alloy steels but, discounting steels
for electrical and magnetic applications and certain other special products, it is
possible to consider three broad categories of alloy or special steels.

'Low alloy structural steels' are grades where strength is a major criterion for
selection. The amount of any alloying element present is usually less than
2%, except for nickel that can be up to 4%wt. The elements (Ni, Cr, Mo, V)
improve strength and toughness, hardenability, etc., so that stronger but
lighter components and structures can be made without sacrificing some of
the most desirable features of plain carbon steels such as easy workability,
weldability and cost.

'Tool and die steels' must maintain strength and hardness at temperature,
so elements such as Cr, W, Ta, V are added to provide hard stable carbide
dispersions in the steel; for example, a typical high speed tool steel can
contain 18%wt W, 4%wt Cr, l%wt V, 5%wt Co and O.75%wt C. The low
diffusion rates of elements such as tungsten, chromium and vanadium in
the iron alloy matrix minimize strength loss at high temperatures.

'Corrosion and heat-resistant steels' usually rely on additions of chromium
to provide protection; for example, chromium oxidizes giving a protective
coating on the steel. This category is dominated by the stainless steels, for
example, austenitic stainless steels which combine corrosion resistance
and ductility usually contain 16-23%wt Cr, 6-22%wt Ni and 0.03-0.2%wt C.
(It is the Nil which stabilizes the FCC structure so that austenite is present at
room temperature) and martensitic stainless steels suitable for valves, turbine
blades and bolts, etc., which combine hardness and corrosion resistance, with
composition in the range 12–18%wt Cr and
0.15–1.2%wt C).
3.7 Power Plant Steels
Carbon and low alloy steels are the most commonly used steels for power plant
applications. The strength of steel is affected by the typical strengthening
mechanisms—namely:

Grain refinement

Solid-solution hardening

Precipitation hardening
 3-12 
13828389
Of these various strengthening mechanisms, the refinement of grain size is perhaps
the most unique because it is the only strengthening mechanism that also increases
toughness.
While carbon steels can be used in applications where operating temperatures do
not exceed about 800°F, alloy steels must be used at higher temperatures or in
situations where additional corrosion resistance is required. The high
temperature strength of chromium-molybdenum steels is mainly derived from a
complex combination of solid solution and precipitation effects. These steels will
experience a progressive change in the type and size of the precipitates present.
Detailed metallurgical analysis [3.7, 3.8] has shown that the precipitates present
in 21/4Cr1Mo steel changes with time at high temperature, with the most
recently reported sequence being:
M3C → M3C + M2C → M3C + M2C + M7C3 → M2C + M7C3 + M6C + M23C6
Where ‘M’ denotes the metal element (this is normally Fe, Cr or Mo but
complex combinations of elements can be involved). The above sequence
indicates that the number of carbide types present increases with time in elevated
temperature service. However, it should be emphasized that the creep strength
will decrease, because the carbides present after long times are less effective in
terms of strengthening, that is dislocation slip processes become easier. The
changes taking place will depend on time and temperature, indeed this
microstructural instability limits the useful operating temperature for low alloy
steels to less than about 575°C (1065°F).
Figure 3-6
Background regarding the development of power plant steels
 3-13 
13828389
CrMo based steels have provided excellent service in a range of high temperature
applications. However, to realize the benefits of more efficient operation at
higher temperatures and pressures, alloys with greater strength and ductility
have been developed (for example, see references 3.8, 3.9), Figure 3-6. These
steels seek to optimize performance through careful control of composition and
heat treatment. It is apparent that these new generation steels have been
successful in increasing strength. It should also be emphasized that high strength
and ductility are only achieved through careful control of composition and heat
treatment, Figure 3-7.
Figure 3-7
Variation in strength and ductility for new 9 and 12%Cr steels as a function of
C + N and chromium equivalent
3.7.1 Ferritic Boiler Steels
Ferritic boiler steels are typically based on 2%Cr, 9%Cr or 12 % Cr [3.9]. The
high strength 9-12 % Cr steels exhibit relatively good corrosion resistance and
can be used as low-cost alternatives to l8% Cr-8% Ni steels. Furthermore, in
comparison with the conventional 2.25CrlMo steels, pipe wall thickness can be
reduced and oxidation and corrosion resistances can also be enhanced. Alloy
9Cr2Mo is a low carbon steel which has been used successfully in superheater
and reheater tubes and piping. The creep rupture strength is between those of
2.25CrlMo steels and TP304H. Low C 9CrlMoVNb, 9Cr2MoVNb and
9CrlMoVNb (ASME T9l) are modified 9%Cr steels with high temperature
strength being enhanced by adding carbonitride-forming elements such as V
and Nb. Of these, modified 9Cr grade 91steel has a high allowable stress and
has already been used extensively worldwide not only for superheater tubes but
 3-14 
13828389
also for thick walled components such as headers and main steam pipes. The
emergence of this material made it possible to use ferritic steels for fabrication
of major pressure parts for ultra-supercritical pressure power plants using
temperatures up to 593°C. Furthermore, 9%Cr steels [9CrO.5Mo1.8WVNb
(ASME T92) and 9Cr1Mo1WVNb (ASME T911)] having a higher allowable
stress than that of the T91 have been developed. These were obtained based on
steels with Mo content replaced by addition of W Mo was decreased to 0.5% and
1.8% of W added to T91 in the case of T92, while 1 % W was added to T91 in
the case of T911.
Of 12%Cr steels, 12Cr1MoV (DIN X20CrMoV121) is extensively used for
superheater tubes, steam pipes, etc. in Europe, and has extensive service
experience. However, because this steel has a carbon content as high as 0.2%,
weldability is found to be somewhat poor, and because high temperature strength
is not satisfactorily high, this material is hardly used in Japan or in the US.
However, improved 12%Cr steels for boiler application, for example,
12Cr1Mo1WVNb and 12Cr0.4Mo2WCuVNb (ASME T122), have been
developed with improved performance.
3.7.2 Ferritic Turbine Steels
Traditionally steels for HP and IP rotor applications are based on 1Cr1Mo0.25V
or 3Ni0.75Cr0.25V. A range of new steels has recently been developed for
turbine components. Because, for turbine steels, emphasis is placed on strength at
ordinary and intermediate temperatures the carbon contents are generally higher
and the tempering temperatures are generally lower than for boiler steels.
3.7.3 Austenitic Boiler Steels
Chemical compositions of austenitic heat resistant steels are typically based on
18% Cr-8% Ni. These steels are used for the highest temperature boiler
components; various improvements have been made to enhance corrosion
resistance while maintaining high creep strength. Furthermore, new steels with
Cr content of 20% or more have been developed for the purpose of improving
creep strength and corrosion resistance. 18% Cr-8% Ni steels such as TP304H,
TP32IH, TP316H and TP347H are still used for fossil-fired power plants
operating under conventional steam conditions. TP347H, which has the highest
allowable stress among these four types of steels, has been produced with a finegrained structure (grain size No.8 and finer) for improved steam oxidation
resistance and creep strengthening; this alloy is designated as TP347HFG in
ASME. This steel is very useful for improved performance in superheater tubes
for ultra-supercritical pressure power plants operating at temperatures up
to 593°C.
 3-15 
13828389
3.8 References
3.1
A. K. Sinah, Ferrous Physical Metallurgy, 1989, Butterworths.
3.2
H. K. D. H. Badheshia, “Bainite in Steels,” 1992, Institute of Materials.
3.3
Metals Handbook,Vol. 1 Properties and Selection: Iron and Steels, 1979,
American Society for Metals.
3.4
R. W. K. Honeycombe, Steels—Microstructure and Properties, American
Society for Metals, 1982.
3.5
P. J. Alberry, and W. K. C. Jones, ’Diagram for the prediction of weld
heat affected zone microstructure,’ Metals Technology, Vol. 4, No. 4,
1977, pp. 360 to 364.
3.6
J. Nutting, “The Structural Stability of Low Alloy Steels for Power Plant
Applications,” Conf Proc “Advanced Heat Resistant Steels for Power
Generation” 1999, Inst of Materials, London, pp. 12 to 30.
3.7
N. Fujita and H. K. D. H. Badheshia, “Modelling simultaneous alloy
carbide sequences in power plant steels,’ ISIJ Int, Vol. 42, No. 7, 2002,
pp. 760 to 769.
3.8
Materials for Ultra Supercritical Fossil Power Plants. EPRI, Palo Alto, CA:
2000. TR-114750.
3.9
F. Masuyama, Review, “History of Power Plants and Progress in Heat
Resistant Steels,” ISIJ International, Vol. 41, No. 6, 2001, pp. 612
to 625.
 3-16 
13828389
Section 4: The Influence of Metallurgical
Changes on Brittleness
4.1 Introduction
In engineering alloys embrittlement can be the result of metallurgical changes,
which occur as a result of specific thermal exposure. Not all such changes can
introduce embrittlement but in some cases the tendency for brittle behavior can
be marked. The metallurgical changes which can enhance brittle behavior
include:

Phase Changes. In many cases the initial processing route will result in a
metastable microstructure. Exposure to elevated temperatures during service
then provides the thermal activation for changes towards the low energy
equilibrium constituents.

The Formation and Growth of Precipitates. A significant factor in the
improved strength of most alloys compared to the base pure metal comes
from the presence of a fine dispersion of precipitates throughout the
microstructure. During operation at high temperature there will be changes
to the size and type of precipitates present.

Temper Embrittlement. Some elements, which are present in alloys as trace
impurities, show a tendency to segregate to grain boundaries. This
segregation, which will depend on composition as well as time and
temperature, can lead to the local concentration at the boundary becoming
significantly greater than the nominal average for the alloy.
All these metallurgical effects are influenced by composition and temperature
history. Because the fundamental processes controlling microstructural
development will be related to thermo-dynamics, the key process governing
microstructural changes will be diffusion. The generalized equation which relates
the Diffusion Coefficient, D, and temperature, T, is as follows:
D = Constant ∙ exp – {Q / RT}
Eq. 4-1
Where Q is the activation energy of the process (units Jmol-1), and R is the
universal gas constant (8.31 Jmol-1K-1). Processes which are controlled by
diffusion will thus take place more rapidly as the temperature increases.
Moreover, in a given alloy, different diffusional processes can occur at the same
time, so that it is possible, indeed in some cases likely, that all of the above
 4-1 
13828389
metallurgical changes may be taking place simultaneously. Thus, while the
following sections each present evidence regarding the primary metallurgical
factor leading to enhanced brittle behavior, it should be recognized that
secondary influences may also be contributing.
The diversity of the alloys used in engineering components and the importance
of understanding, and wherever possible quantifying, the factors affecting
changes in microstructure, as well as the attendant influence these changes
exhibit on critical properties such as strength and ductility, have resulted in the
publication of a very large volume of information. The number of available
papers in most areas is far too extensive for detailed review. The present guideline
document thus focuses on providing key information, supported with visual aids
to illustrate and enhance descriptions, with appropriate references, which
facilitate access to the original sources.
The next three sections cover issues associated with embrittlement due to
metallurgical changes:

Section 5, Phase Changes

Section 6, Influence of Carbides

Section 7, Temper Embrittlement
 4-2 
13828389
Section 5: Embrittlement Due to Phase
Changes
5.1 Introduction
Heat treatments are frequently selected to produce non-equilibrium
microstructures. Subsequent thermal exposure will then provide the driving force
for phase transformation to minimum energy states. In some cases the
development of new phases during service will promote brittle behavior. The
following section provides information regarding embrittlement due to the
formation of new phases, which occurs in some carbon and low alloy steels and in
stainless steels. Detailed information is provided on:

Graphitization in C – Mn and C – Mo steels including a case study

Embrittlement in Stainless Steels including
-
Secondary hardening
-
475°C (885°F) Embrittlement
-
Embrittlement and Grain Size
-
Sigma Phase Formation
-
Weldments
Finally, a guideline list of actions that allow assessment of the influence of phase
changes on service performance is provided.
5.2 Graphitization in C – Mn and C – Mo Steels
Graphitization occurs when iron carbide decomposes into the true equilibrium
structure of ferrite and graphite. The formation of graphite particles or nodules,
if dispersed throughout the metal, are not considered a problem; however, if they
form a continuous zone the resulting embrittled material can fail catastrophically
by brittle fracture.
Carbide spheroidization is also a mechanism of pearlite decomposition. Of the
two, graphitization is less common, but because it results in embrittled material,
it is more serious when it does occur. Because of the difference in activation
energies of the two processes, it has generally been considered that graphitization
is preferred at temperatures below about 550°C (1020°F), Figure 5-1. However,
 5-1 
13828389
recently it has been observed from field experience with degraded materials that
the graphitization-to-spheroidization temperature may differ from the accepted
value, be dependent on steel composition and microstructure, and occur in a
manner which, to date, is not completely predictable.
Figure 5-1
The influence of time and temperature on the formation of graphite (based on 5.1)
Pearlite decomposition to ferrite and graphite has been found when the steel has
been heated briefly above the A1 temperature, approximately 725°C (1340°F).
Such a temperature regime occurs during the welding process, which is why
traditionally graphitization damage is mostly associated with the heat affected
zones of welds, usually at a characteristic distance from the weld. Since this
susceptibility is related to the location of a particular isotherm associated with
welding the graphite can form in a particular plane leading to severe
embrittlement. However, field investigations have recently identified
graphitization that has occurred in base metal removed from the influence of
welds [5.2]. This phenomenon referred to as "non-weld-related graphitization"
seems to be associated with locations that have been subjected to large plastic
deformations. This form of graphite formation also occurs in bands, Figure 5-2,
and has resulted in brittle failure of boiler tubes, Figure 5-3.
The propensity to graphitization damage has also been considered to be
dependent on the steel-making practice used. Aluminium-killed steels, once in
common usage have been shown to be more susceptible than those deoxidised
with silicon or titanium, unless the aluminium content is restricted to <0.025%.
 5-2 
13828389
Figure 5-2
Formation of graphite bands in a reheater tube [5.2]
 5-3 
13828389
Figure 5-3
Micrograph from a carbon steel weld showing a moderate level of “eye brow”
graphite in a band adjacent to the HAZ [5.2]
5.2.1 Growth Kinetics of Graphitization
The kinetics of graphite growth have been described through consideration of an
incubation period, and a growth period that is approximated by an equation of
the form [5.2]:
y = A exp (-Q' / RT) tgm
Eq. 5-1
Where
y = fraction of transformation (0 to 1.0)
A = constant; Q' is approximately equal to Q, the activation energy for
the controlling process
T = exposure temperature in absolute units
R = Universal Gas Constant
tg = exposure time following incubation
m = time dependence power
 5-4 
13828389
Figure 5-4
Power law approximation of the sigmoidal growth behavior of graphite [5.2]
The two regimes of interest, namely the initial incubation period and subsequent
growth region, are shown in Figure 5-4. An analysis of available data led to a
best-fit determination for predicting fractional transformation in weld HAZ
graphitization [5.2] of:
y = 2.07 × 108 exp (-20,000 / T) tg 0.53
Eq. 5-2
Where
T = exposure temperature in K
tg = growth period following incubation
The incubation period, ti, was also derived from available data and found to be:
ti = 226.25 exp (3693 / T)
 5-5 
13828389
Eq. 5-3
Best estimate time-temperature-transformation curves were developed. These
curves have subsequently been reviewed in the light of additional service
experience. The latest transformation curves are defined by the following
equations, and are shown in Figure 5-5:

Start/Low Risk:
t(operating hours) = 0.56296 exp (8322.3/T(K))

Moderate:
t(operating hours) = 0.56296 exp (8322.3/T(K)) + exp [−6.95846 +
12348/T(K)]

Significant:
t(operating hours) = 0.56296 exp (8322.3/T(K)) + exp [−6.29847
+12348/T(K)]
Figure 5-5
Time temperature transformation curves for graphitization in C, C – Si and C – Mo
steels [5.2]
 5-6 
13828389
5.2.2 Case Study/Example [5.3]
A tube failure occurred in a reheater tube section after about 15 years operation
while spreading the pendants for installation of new support clips. The tube was
specified as 21/4 inch outside diameter by 0.165-inch wall thickness SA-209T1A steel. Chemical analysis confirmed that the composition agreed with the
specification at the time of manufacture showing 0.19% C, 0.47% Mn, 0.29% Si,
0.47% Mo and 0.22% Cr. The fracture, shown in Figure 5-3, followed a spiral
path at about 60° to the tube axis. Continuous bands of graphite were identified
by metallographic examination, Figure 5-2, indicating that operation up to about
950°F had occurred. The formation of graphite in specific bands is believed to
have taken place because of the local strains that were introduced during tube
manufacture. It is now widely recommended that for operation above 800°F
tubes should contain at least 0.6% Cr.
A second detailed case study is presented in Appendix H.
5.3 Embrittlement in Stainless Steels
Stainless steels describe a range of alloys containing more than about 12% Cr,
which are used in applications where resistance to corrosion and or oxidation is
required. However, these alloys generally contain additional elements, which are
added to control microstructure as well as mechanical, fracture and creep
properties. The primary microstructural groups are:

Martensitic steels, containing 12-17% Cr, 0-4% Ni, 0.1-1% C and
sometimes Mo, V, Ni, Al and Cu

Ferritic steels, containing 15-30% Cr, low carbon, no nickel and often some
Mo, Ni, or Ti

Austenitic steels, containing 18-25% Cr with 8-20% Ni and low carbon
content; other alloying additions include Mo, Ni, or Ti
Both the martensitic and ferritic types will show brittle to ductile transition
behavior with increasing temperature. Moreover, because of the relatively high
alloy contents these steels can develop a range of precipitate types and may
exhibit different phases. The following summary highlights issues where
compositional and thermal effects can lead to brittle behavior.
5.3.1 Brittleness Due to Secondary Hardening
The chromium levels in martensitic steels are such that air cooling leads to
martensitic structures. To provide a balance between strength and toughness
these steels are normally tempered prior to service. As shown in Figure 5-6,
a distinct minimum occurs in the room temperature impact energy when
tempering is carried out at around 500°C. This minimum is the result of the
formation of large numbers of precipitates, which cause a significant increase in
the brittle to ductile transition temperature. Application of higher tempering
temperatures coarsens the precipitates restoring the toughness.
 5-7 
13828389
Figure 5-6
Brittle behavior in 12% Cr martensitic steels as a result of secondary hardening
[5.4]
5.3.2 475°C Embrittlement
Martensitic and ferritic stainless steels containing more than about 12% Cr
become embrittled with extended exposure to temperatures between about 400
and 510°C (750 and 950°F), with the maximum embrittlement at about 475°C
(885°F). Therefore, this problem is referred to as 475°C embrittlement. Aging at
475°C (885°F) increases strength and hardness, decreases ductility and toughness,
and changes electrical and magnetic properties and corrosion resistance. The time
at the aging temperature intensifies these changes.
Detailed metallurgical studies have shown that these reductions in toughness
are the result of the formation of Cr rich second phase precipitates [5.5]. This
phase, typically identified as α', has a bcc structure and is formed by spinodal
decomposition on {100} planes. This effect, and the associated embrittlement,
are greater at higher Cr levels. For example tests have shown that the room
temperature Charpy energy is reduced from a typical value of 64 J (47 ft · lb)
to around 1.4 J (1 ft · lb). Chromium nitrides also precipitate during aging and
are observed at grain boundaries, dislocations, and inclusions. The reaction is
reversible; heating above the embrittlement range dissolves α'. In duplex stainless
steels, the embrittlement temperature range appears to be broader, with
additional phases precipitating in the upper portion of the range.
 5-8 
13828389
5.3.3 Embrittlement and Grain Size
It is generally accepted that fine grained material exhibits greater ductility the
coarse grained structures. This effect is a particular issue with ferritic stainless
steels because unless additions are made to limit grain growth, for example,
second phase particles of Ti(CN) or Nb(CN), very large grain sizes can occur in
weld HAZ’s or even during service at temperatures above about 600°C. Coarse
grained material will exhibit a significantly higher FATT compared with refined
structures, Figure 5-7.
Figure 5-7
Increase in the Charpy FATT with increase in grain size in ferritic stainless steel
[5.5]
5.3.4 Sigma Phase Formation
Sigma phase (σ-phase) is a hard (>HRC 60), brittle, non-magnetic phase that
forms during service in austenitic stainless steels and nickel based superalloys.
The development of σ-phase has been the subject of significant study and several
reviews are available which consider the factors affecting the development [5.9].
The approximate composition of σ-phase is FeCr (Figure 5-8), although it is
often listed as Fe(Cr,Ni,Mo) because the actual composition depends on the
specific alloy system. Sigma phase usually forms when the material is exposed to
temperatures in the range of approximately 550°C to 900°C (1022°F to 1652°F).
Sigma phase is unstable and redissolves if heated to a temperature of about 870°C
(1600°F), the exact dissolution temperature depends on the composition. For
Type 304H material σ-phase formation requires more than 1000 hours of
exposure in this critical temperature range (Figure 5-9). The specialist etching
techniques used to reveal σ-phase are described in Section 2 of this report.
 5-9 
13828389
Figure 5-8
Iron – chromium-nickel equilibrium phase diagram (section at 8% nickel). The two
phases that are relevant to austenitic stainless steels are Austenite (Gamma Iron,
γ + Carbon,) and Sigma Phase, σ (a grain boundary phase comprised of
approximately 50% chromium and 50% iron). The addition of carbon will expand
the region of stability of Gamma Iron, γ-Fe. Note that even without the benefit of
carbon additions Sigma Phase is an equilibrium phase for chromium levels above
approximately 18% [5.10].
 5-10 
13828389
In general, the tendency to form sigma phase will be increased by:

Higher levels of chromium; thus when δ-ferrite is present the rate of sigma
formation is increased, and elements such as Mo, Si, W, V, Nb, which act to
stabilize ferrite

Prior cold work
Elements such as C, N and Ni will reduce the susceptibility for the formation of
sigma.
Figure 5-9
Time-temperature-transformation curves for Types 304H, 321H, and 347H
materials. Note that even the stabilized grades of material will sensitize and form
sigma phase if they are exposed to prolonged temperatures approaching 600°C
(1112°F). At 650°C (1202°F) all three alloys will begin to form sigma phase after
approximately 10,000 hrs [5.11].
 5-11 
13828389
Sigma phase can be extremely deleterious to material performance. The
formation of even a few volume fraction percentage points of σ-phase can reduce
the creep rupture ductility, the corrosion resistance and the toughness of the
material. Significant levels of embrittlement have been noted with the presence
of even small amounts of σ-phase. Thus, for example, data for 304 stainless steel,
Figure 5-10, shows that the total creep elongation is reduced by about 50% with
only 7% sigma phase, reductions in ductility of nearly 80% were found at about
14% sigma.
Figure 5-10
Decrease in creep elongation with the presence of sigma phase [5.12]
Weldments
It is well known that the hot cracking resistance of austenitic weldments can be
improved by ensuring primary δ-ferrite solidification or the presence of residual
δ-ferrite at room temperature. In the late 1940s, Schaeffler [5.13, 5.14]
developed a method for predicting the room temperature microstructure based on
chromium and nickel equivalents, Figure 5-11. The Schaeffler diagram was
originally developed for predicting the microstructure in weldments (that is,
martensitic, austenitic, and/or ferritic) produced by joining dissimilar steel
grades. The Schaeffler diagram given in Figure 5-11 contains a number of
rectangular zones, which show the compositional range of different weld metal
grades suitable for welding austenitic steels.
It should be emphasized that the Schaeffler diagram was derived empirically from
Shielded Metal Arc weld deposits produced under specific conditions, and
consequently it will not give an exact description of the weld metal
microstructure under all welding conditions, especially with the newer high
energy density welding techniques employing lasers, etc. However, the diagram
 5-12 
13828389
gives a reasonable estimate of the weld metal microstructure and has been widely
used by welding metallurgists. Considerable effort has been directed to
developing improved chromium and nickel equivalents for predicting δ-ferrite
content [5.15 to 5.18]. These equations are summarized in Table 5-1.
Figure 5-11
Schaeffler diagram showing how the microstructure of austenitic steel welds
depends on nickel and chromium equivalent
Since when δ-ferrite is present the higher Cr content of this phase favors sigma
formation, approximate estimates for the amount of sigma likely to be developed
in weld metal can be obtained from knowledge of the ferrite content. Three
approaches for estimating the δ-ferrite content can be obtained from the
appropriate Cr and Ni equivalents given in Table 5-1:
From Schaeffler: %Ferrite = - 39.1 + 43.5 (Cr equiv – 5.8) / (Ni equiv + 2)
Eq. 5-4
From Hull:
%Ferrite = -13.77 + 2.88Cr equiv – 3.125Ni equiv
Eq. 5-5
From DeLong:
%Ferrite = - 30.65 + 3.49(Cr + Mo + Si + 0.5Nb)
– 2.5(Ni + 30C + 30N + 0.5Mn)
 5-13 
13828389
Eq. 5-6
Statistical studies have shown that the Schaeffler diagram tends consistently to
overestimate the δ-ferrite content. Delong et al. [5.15] have suggested an
improved Schaeffler diagram which takes into account the austenitizing effect of
nitrogen and is particularly suitable for gas-tungsten arc (GTA) and gas-metal
arc (GMA) weld metals. Delong et al [5.15] have shown that nitrogen incursion
into the weld pool can appreciably reduce the δ-ferrite content of the weld
deposit. It is claimed that the Delong diagram predicts the ferrite number within
± 3 FN (ferrite number). The ferrite number has been adopted by the
International Welding Institute as the preferred unit of measurement of δ-ferrite
content. This number is normally measured using a magnetic device calibrated by
standard ferrite specimens.
Table 5-1
Formulae developed to calculate values of chromium and nickel equivalent
Workers
Ref
Source
Cr Equivalent
Ni Equivalent
Schaeffler
14
Weld
%Cr + %Mo + 1.5%Si + 0.5%Nb
%Ni + 0.5%Mn + 30%C
Delong et al.
15
Weld
%Cr + %Mo + 1.5%Si + 0.5%Nb
%Ni + 0.5%Mn + 30%C +
30%N
Guiraldenq
16
Casting
%Cr + 2%Mo + 1.5%Si + %Nb +
4%Ti
%Ni + 30%C + 30%N
Hull
17
Casting
%Cr + 1.21 %Mo + 0.48%Si +
0.14%Nb + 2.2%Ti + 2.27%V +
2.48%AI + O. 72%W + 0.21 %Ta
%Ni + 0.11%Mn-0.0086
(%Mn)2 + 24.5%C +
18.4%N + 0.44%Cu +
0.41%Co
Hammar and
Svennson
18
Thermal
analysis
%Cr + 1.37%Mo + 1.5%Si + 2%Nb
+ 3%Ti
%Ni + 0.31 %Mn + 22%C +
14.2%N + %Cu
While differences in the amount of ferrite, and hence sigma, influence ductility
in the short term, longer exposures can lead to similarly brittle fractures for
samples containing different amounts of ferrite, for example, Figure 5-12.
 5-14 
13828389
Figure 5-12
Brittle creep failures due to ferrite/sigma phase [5.19]
Effect on Impact Properties
The influence on the amount of ferrite present on room temperature Charpy
impact energy has been demonstrated in a comprehensive study of the behavior
of E308-16 welds [5.19, 5.20]. These welds all had compositions within the
allowable range but the compositions were selected to give the following:

Ferrite number about 3, designated extra low ferrite

Ferrite number about 6.5, designated low ferrite

Ferrite number about 9.9, designated medium ferrite

Ferrite number about 13.7, designated high ferrite
Sections of each weld were aged at 593°C (1100°F) and it was apparent that the
high ferrite weld showed a significant reduction in room temperature Charpy
energy after 1000 hrs aging, Figure 5-13. The medium ferrite weld reached a
minimum impact energy value after about 5000 hrs aging. In both cases the
minimum values were measured with aging for greater periods up to 10000 hrs.
In the case of the low ferrite and extra low ferrite some reduction in impact
energy was noted with aging up to about 2000 hrs but with longer aging times
the values increased. This behavior was attributed to the formation of precipitates
without sigma in the extra and low ferrite welds, with significant sigma
formation in the medium and high ferrite welds.
 5-15 
13828389
Figure 5-13
Room temperature Charpy values for E-308 weld metal after aging at 1100°F
(593°C) [5.19]
Specific reductions in impact energy will be dependent on composition. Thus, for
example, the formation of intermetallic sigma resulted in a 50-90% reduction in
impact properties for 18Cr-12Ni-2-3Mo weld metal (with ~ 5 FN).
An empirical analysis of a wide range of weld metal data has produced a LarsonMiller type parametric model, which describes the trend in the changes in impact
properties of austenitic steels with both time and temperature. The data
produced were presented as a fraction of the unaged impact value using the
expression:
P = T (5.81 + log t) × 10-3
Eq. 5-7
Where P is the Larson-Miller Parameter, t is the time in h, and T is the
temperature in °C. As shown in Figure 5-14 [5.12], reductions in impact value
approaching 90% are indicated for high degrees of tempering. Moreover, the
parametric expression provides a reasonable description of the data. Although
this approach has been applied to other stainless weldments, including 19Crl2Ni-3Mo, l7Cr-12Ni-3Mo, l7Cr-8Ni-3Mo, and l7Cr-8Ni-2Mo weld metals,
 5-16 
13828389
care must be exercised since the general applicability for broad ranges of
composition and δ-ferrite levels must lead to inaccuracies. However, careful
analysis of data within a particular composition and ferrite number may allow
useful estimates to be made.
Figure 5-14
Variation in normalised impact value with time temperature parameter, P, for a
range of stainless steel weld metals [5.12]
5.4 Assessment of Components
The following steps should be considered to assess embrittlement: due to phase
changes:
1. Review available documentation to identify component compositions. If no
reliable information from manufacturing records exists the material chemistry
should be measured using material filings or by local excavation. A portable
X-ray/spectrographic unit may be used for direct measurement provided
accurate results concerning trace elements can be obtained.
2. For Carbon and C-Mo steel tubes Figure 5-5 can be used for assessing the
potential for graphite formation. If there is a risk of graphite being present,
samples should be taken for metallographic assessment.
3. For stainless steels the quantitative chemistry results should be used to
estimate the Ni and Cr equivalents from the predictive formulae outlined in
Table 5-1.
 5-17 
13828389
4. For any locations that appear to be high risk, perform metallography and
EDS. This should establish microstructural features, for example, grain size,
the constituent phases present which can be confirmed by assessing any local
variations in composition (see Section 2).
5. If results from items 2 or 3 suggest a potential problem, perform small punch
testing to estimate toughness (if comparative S.P. data is available).
6. An alternative to item 4 is to perform a series of Charpy impact tests to
determine FATT. However, this option requires that sufficient material
exists for testing. Comparison of the original and ex-service data will provide
evidence of a substantial shift in FATT.
7. Evaluate the component using detailed non destructive test methods to
characterize the size and location of any flaws or defects present.
8. Depending on the shift (or the present value of FATT), an assessment of
fracture toughness may be warranted over operating temperature range. This
can also be performed with Cv vs. K1c correlations.
9. Evaluate the stresses in the location, using appropriate analysis, to compare
K1c vs. Kmat.
5.5 References
5.1
W. L. Hemingway, The Study of Graphitization, Edwards Valve Co.,
1952.
5.2
J. R. Foulds and R. Viswanathan, “Graphitization of Steels in Elevated
temperature Service,” Microstructures and Mechanical properties of
Aging Materials, The Minerals, Metals and Materials Society, 1993,
pp. 61–69.
5.3
Graphitization in C and C-Mo Steels. EPRI, Palo Alto, CA: 2010.
1019783.
5.4
K. J. Irvine and F. B. Pickering, “The Physical Metallurgy of 12%
Chromium Steels,” J of The Iron and Steel Institute, Vol. 195, Part 4,
1960, pp. 386–405.
5.5
F. B. Pickering, “Physical metallurgy of Stainless Steel developments,”
International Metals Reviews, Review 211, Vol. 21, 1976, pp. 227–268
5.6
E. O. Hall and S. H. Algie, The Sigma Phase, Met. Rev., Vol. 11, 1966,
p. 61–88.
5.7
G. Matern et al., The Formation of Sigma Phase in Austenitic Ferrite
Stainless Steels and Its Influence on Mechanical Properties, Mem. Sci.
Rev. Met., BISI 13972, 1974, p. 841–851.
5.8
W. J. Boesch and J. S. Slaney, Preventing Sigma Phase Embrittlement in
Nickel Base Superalloys, Met. Prog., Vol. 86, July 1964, p. 109–111.
5.9
J. R. Mihalisin et al., Sigma—Its Occurrence, Effect, and Control in
Nickel-Base Superalloys, Trans. AIME, Vol. 242, Dec 1968,
p. 2399–2414.
 5-18 
13828389
5.10
The Making, Shaping and Treating of Steel, United States Steel, 1964,
p. 1115.
5.11
Remaining Life Assessment of Austenitic Stainless Steel Superheater and
Reheater Tubes. EPRI, Palo Alto, CA: 2002. 1004517.
5.12
J. J. Smith and R. A. Farrar, “Influence of microstructure and
composition on mechanical properties of some AISI 300 series weld
metals,” International Materials Reviews, Vol. 38, No. 1, 1993,
pp. 25–51.
5.13
A. L. Schaeffler. Selection of austenitic electrodes for welding dissimilar
metals. Weld J., Vol. 26, 1947, pp. 603s–620s.
5.14
A. L. Schaeffler. Constitution diagram for stainless steel weld metal.
Met. Prog., Vol. 56, 1949, p. 680–688.
5.15
W. T. Delong, G. A.Ostrom and E. R. Szumachowski, Measurement
and calculation of ferrite in stainless steel weld metal. Weld J., Vol. 35,
1956, pp. 526s–532s.
5.16
P. Guiraldenq. Measure des coefficients d'autodiffusion intergranulaire du fer
en phase γ et comparision avec l'autodiffusion aux joints de grains du fer α
Mem. Sci. Rev. Metall., Vol. 164, 1967, pp. 415–417.
5.17
F. C. Hull, Effects of composition on embrittlement of austenitic
stainless steels. Weld J., Vol. 52, 1973, pp. 193s–203s.
5.18
O. Hammar and U. Svensson, in Solidification and Casting of Metals, The
Metals Society, London, 1979, pp. 401–410.
5.19
D. Hauser and J. A. VanEcho, “Effect of Delta Ferrite Content of
E308-16 Stainless Steel Weld Metal: II Mechanical Property and
Metallographic Studies,” Winter Annual Meeting ASME, 1978,
pp. 17–46.
5.20
D. P. Edmonds, D. M. Vandergriff, and R. J. Gray, “Effect of Delta
Ferrite Content of E308-16 Stainless Steel Weld Metal: III
Supplemental Studies,” Winter Annual Meeting ASME, 1978,
pp. 47–62.
 5-19 
13828389
13828389
Section 6: The Effect of Carbides on
Embrittlement
6.1 Introduction
Carbon is a key alloying element in steels, and the formation of carbide
precipitates is used in alloy steels to improve strength particularly at high
temperatures. However, the improvements in strength can be associated with
reductions in ductility and in extreme situations lead to embrittlement. This
section highlights critical areas where brittle behavior can occur; these include:

The effect of carbon on fracture behavior

Tempered Martensite Embrittlement

Thermal Embrittlement of Maraging Steels

Carbides in CrMo low alloy steels

Dissimilar metal welds

Sensitization of austenitic stainless steels
Finally, a guideline list of actions that allow assessment of the influence of
carbides on service performance is provided
6.2 The Effect of Carbon on Fracture Behavior
It is well established that increasing levels of carbon up to the eutectoid limit,
that is, up to about 0.8% C, will promote brittle behavior. Thus, as illustrated in
Figure 6-1 [6.1], the Charpy test FATT will be moved to higher temperature
and the upper shelf energy is decreased with increase in carbon. This effect is due
to the fact that as the carbon content increases the volume fraction of pearlite will
increase. The pearlitic structure consists of alternating plates of iron carbide, or
cementite, and ferrite.
Thermal effects influence the thickness and spacing of these plates, or lamella,
however, the general trend is similar in all cases. For example, low-carbon fully
ferritic steel has a room temperature Charpy V-notch impact energy of about
200 J (150 ft · lbf). In contrast, fully pearlitic steel, 0.8% C, has a roomtemperature impact energy of less than 10 J (7 ft · lbf). It is thus apparent that
although fully pearlitic steels have high strength, high hardness, and good wear
resistance, they also have poor ductility and toughness.
 6-1 
13828389
These effects on fracture behavior illustrate the basic principle of brittle fracture
that cracking can initiate from localized strain or slip causing cracks to initiate at
brittle second phase particles. The size of the crack will be dependent on the size
of the second phase particle. For small particles the size of the crack will be subcritical and the crack formed will be blunted by plastic deformation at the tip.
However, when the particles are of sufficient size the crack formed can
immediately cause brittle cleavage fracture. Thus, as the size of the particles
increases brittle behavior would be expected to occur at higher temperatures.
Based on optical metallographic study of coarse filamentary carbides in carbon
steels it has been shown that carbides of about 2 μm in thickness will promote
brittle behavior, with increases in FATT continuing until a thickness of about
5 μm is reached, Figure 6-2 [6.2].
Figure 6-1
The effect of increasing carbon content on Charpy impact behavior, FATT from
–50°C to +150°C [6.1]
 6-2 
13828389
Figure 6-2
The influence of carbide thickness on the ductile/brittle transition temperature in
carbon steels [6.2]
The transition temperature (that is, the temperature at which a material changes
from ductile fracture to brittle fracture) for fully pearlitic steel can be
approximated from the following relationship:
FATT = 217.84 - 0.83 (dc-1/2) - 2.98(d -1/2)
Eq. 6-1
Where FATT is the transition temperature (in °C), dc is the pearlite colony size
(in mm), and d is the prior austenite grain size (in mm). Measurement of pearlite
colony size and the prior austenite grain size requires the application of specialist
metallographic techniques. An alternative approach [6.3] for estimating the
FATT in pearlitic steels is based on the equation:
FATT = −19 + 44(Si) + 700Nf + 2.2(Pl) - 11.5(d -1/2)
Eq. 6-2
Where Pl is the fraction of pearlite formed. It can be seen in both these
relationships that grain size is an important parameter in improving toughness.
The effect of carbide size and grain size on the 27 J Charpy impact transition
temperature is shown in Figure 6-3. It should be noted that in this figure the
grain size is plotted according to the Petch convention (as d -1/2, thus, the fine
grain sizes show lower values of transition temperature). At all values of grain
size, carbides less than about 0.5 μm in size have no significant effect on brittle
behavior. For thicknesses above about 1 μm there is an effective increase in the
transition temperature of about + 70°C. This rapid change indicates the
sensitivity of carbide size and illustrates the need for accurate data in attempting
to quantify the effects on fracture behavior.
 6-3 
13828389
Figure 6-3
Effect of grain size and carbide thickness on the temperature where the Charpy
fracture energy is 27 J [6.4]
In general, with steels heat treated to similar strengths it is found that the FATT
increases in the order martensite > bainite > ferrite and pearlite. An example of
this trend is given in Figure 6-4. Moreover in low-carbon bainitic steels, upper
bainite has inferior toughness to lower bainite. This trend is a direct consequence
of size and distribution of the carbides present since typically, the finest carbides
are developed in tempered martensitic microstructure. This is further evidence
that even when cracking occurs in fine carbides, the very small defect formed is
blunted by local deformation whereas with carbides above a particular size the
crack is able to propagate and brittle fracture occurs. For any given
microstructure a finer prior austenite grain size will lower FATT, Figure 6-3,
thus it is generally the case that lower normalizing temperatures will be beneficial
to improving fracture resistance.
 6-4 
13828389
Figure 6-4
Increase in the value of FATT from martensitic, bainitic to pearlitic steels all with a
carbon content of 0.25% [6.5]
6.3 Tempered Martensite Embrittlement (TME)
The development of a uniform, fine dispersion of carbides is a fundamental
reason why martensitic structures have excellent strength and toughness.
However, because newly formed martensite is very hard and brittle a tempering
treatment is normally required to ensure reasonable toughness. Steels with
particularly high tensile strength are susceptible to TME when this tempering is
carried out between about 250 and 400°C (480 and 760°F). Charpy impact data
reveal a decrease in impact energy in the embrittlement range. The ductile-tobrittle transition temperature will also increase with tempering in this range.
Within the lower shelf regime, TME produces a change in the fracture mode
from either predominantly transgranular cleavage to intergranular fracture along
the prior-austenite grain boundaries. This form of embrittlement has been found
to occur over a significant range of carbon contents, including in very low carbon
steels, although it appears that at very low carbon levels the segregation of
residual impurity elements has also been suggested as a key factor in promoting
brittle fracture. It has been suggested that embrittlement is primarily a
consequence of:

The formation of iron carbide (cementite) on prior austenite grain
boundaries

The decomposition of interlath retained austenite into cementite films

The segregation of impurities, such as phosphorus, to the prior-austenite
grain boundaries, that is, temper embrittlement
 6-5 
13828389
The thermal treatment typical of tempered martensite embrittlement is shown as
line 1 in Figure 6-5; this behavior should be contrasted with the heat treatment,
which results in temper embrittlement, lines 2 and 3.
Figure 6-5
Time temperature transformation diagram illustrating the thermal treatment likely to
produced tempered martensite embrittlement, line, compared with thermal
treatments likely to produce temper embrittlement, lines 2 and 3 [6.6]
6.4 Thermal Embrittlement
Maraging steels typically contain high nickel (around 18%), plus Co, Mo and Ti
to form precipitates and low carbon (0.03% max), and are solution treated at
around 820°C followed by air cooling. This treatment results in a martensitic
structure with fine precipitates, which exhibit tensile strength around 1100MPa
with reduction of area around 30%. However, these alloys are susceptible to
brittle intergranular fracture when held at temperatures above about 1095°C
(2000°F), followed by slow cooling or by interrupted cooling with holding in the
range of 815 to 980°C (1500 to 1800°F). Embrittlement has been attributed to
precipitation of TiC and Ti(C, N) on the austenite grain boundaries during
cooling through the critical temperature range. The severity of the embrittlement
increases with:

Decreasing cooling rate through the range 815 to 980°C

Increases in the concentration of titanium, carbon and nitrogen
 6-6 
13828389
6.5 Carbides in CrMo Low Alloy Steels
The type of precipitates formed will depend on the composition, temperature
history during fabrication, as well as the time and temperature of in service
exposure. Indeed, even though the preferred precipitates in steels are
predominantly carbides, with nitrides and carbo-nitrides also present in many
modern advanced steel, different carbide types will be present depending on
service conditions. It is generally agreed that the sequence of precipitation will be:
M3C → M3C + M2C → M3C + M2C + M7C3 → M2C + M7C3 + M6C + M23C6
Figure 6-6
Typical distribution of carbides in CrMo low alloy steel after long term service at
around 550°C
In addition to the changes in carbide type, there will be growth of preferred
carbides. This growth will be driven by the reduction of surface energy, which
occurs when a large number of small precipitates are replaced by a smaller
number of large precipitates. These changes will occur by diffusion and, since
diffusion along grain boundaries will tend to be faster than diffusion through the
grains, there will be a tendency with increased aging for the largest precipitates to
form at the boundaries. A typical electron micrograph showing the precipitate
distribution in 21/4 Cr1Mo low alloy steel after long term service at around
550°C is presented in Figure 6-6. This micrograph shows that the largest
precipitates are present on the boundaries and the growth of these precipitates
has resulted in dissolution of neighboring precipitates. Thus, precipitate free
zones are developed so that the distribution of precipitate sizes is non uniform.
Since the strength of these alloys is largely a function of the ability of the
 6-7 
13828389
precipitates to impede dislocations as growth takes place there is a consistent
reduction in strength. The trends in behavior are well established and data of the
type shown in Figure 6-7 have been compiled to monitor the effects of time
dependent aging by monitoring hardness.
Figure 6-7
Reductions in hardness in CrMo steels as a function of time at temperature [6.7]
The relationship between time and temperature can be derived directly from the
rate equation given earlier, that is,
1/t = constant. Exp – {Q/RT}
Eq. 6-3
taking logs and rearranging gives,
T {log t + C} = Q/2.3R
Eq. 6-4
Thus, Q/2.3R is equivalent to a Holloman-Jaffe parameter for hardness changes
[6.8] (and the Larson-Miller parameter linking creep life to time and
temperature).
 6-8 
13828389
The influence of carbides on the FATT of 21/4Cr1Mo steel has been evaluated
in a series of aging experiments [6.9]. The particular alloy selected was very low
in trace elements and a step cool heat treatment typical of the type used to
evaluate temper embrittlement revealed that relatively low temperature exposure
did not change FATT. In contrast, significant reductions in FATT were found
after aging at 550°C, 600°C, and 625°C.
Considering both aging time and temperature it was found that a reasonable
description of the fracture behavior was obtained using the equation:
ΔFATT = A × T (log t + 8) + B
Eq. 6-5
Where A and B are constants. The change in FATT was directly related to the
increase in the average size of the grain boundary carbides. Figure 6-8. In this
case a change in the average size of 0.4 μm increased the value of FATT by 60°C.
The complete transition curves for both the virgin steel (that is, at
implementation into service) and for samples heat treated under laboratory
conditions to increase the size of the carbides are shown in Figure 6-9. This
figure also includes data from an ex-service sample which had experienced
88 000 h at 540°C. The points representing samples of steel aged at 600°C for
10,000 hours simulate the change in FATT measured after prolonged service.
Comparison of the data presented in Figure 6-2 for C/Mn steel and Figure 6-8
for 21/4Cr1Mo steel indicates very similar trends in behavior with increasing
carbide size. However, there is approximately an order of magnitude difference in
the average carbide size, that is, in Figure 6-2 the sizes range from 1 to 6 μm
whereas those in Figure 6-8 range from about 0.1 to 0.6 μm.
Figure 6-8
Change in FATT with mean carbide size for 21/4CrMo steel [6.9]
 6-9 
13828389
Figure 6-9
Charpy impact transition curves for 21/4CrMo steel prior to service, after
laboratory aging and after prolonged service at 550°C [6.9]
This difference may be a consequence of the fact that the earlier results were
obtained from optical metallography and the carbide measurements in the CrMo
steel were made using high resolution electron microscopy. This influence
demonstrates the importance of ensuring the resolution applied to quantitative
metallographic measurements must be selected to match that used in the studies
which developed the relationships.
6.6 Dissimilar Metal Welds
Transition joints between 21/4Cr1Mo low alloy steel and austenitic stainless steel
are frequently manufactured using nickel based filler. Experience shows that
these joints are susceptible to low ductility failure as a results of creep or creep
fatigue damage initiated at carbides developed at the weld interface with the low
alloy steel. Thus, in this case carbide development leads to brittle type failures by
promoting the nucleation and growth of voids.
 6-10 
13828389
Figure 6-10
The development of carbides at the weld/HAZ interface in P22 – austenitic
stainless steel transition weld manufactured with a nickel based weld metal. Type I
carbides shown in (a) and (b), with Type II carbides shown in (c) [6.10]
Comprehensive research programmes have documented the microstructure
present after fabrication and monitored the changes, which take place during
service. This work has shown that initially there are two different distributions of
carbides, namely:

Type I, a relatively narrow line of carbides, Figure 6-10 a and b.

Type II fine carbides which form in a relatively wide band extending from
the interface for about 50 μm to 100 μm, Figure 6-10 c. It appears that the
regions of Type II carbides form from initially martensitic regions.
Detailed examination has shown that the Type I carbides grow in size with time
at temperature until creep voids initiate. In view of the importance of these
carbides in the initiation of fracture significant efforts have been expended to
characterize the growth behavior with time and temperature. These studies have
demonstrated that growth can be described by equations of the form:
M3 = C t exp – (Q/RT)
Eq. 6-6
Where M is the major axis dimension and the activation energy Q takes a value
of approximately 272 kJ mol-1. As shown in Figure 6-11 this expression provides
a good description of measured values.
 6-11 
13828389
Figure 6-11
Growth behavior of Type I carbides at the interface of dissimilar metal welds
fabricated between 2 1/4CrMo and austenitic stainless steel using a nickel based
filler metal [6.10]
6.7 Sensitization of Austenitic Steels
When austenitic stainless steels are exposed to temperatures within the range of
430°C (805°F) to 900°C (1650°F), chromium carbides will form on the grain
boundaries (Figure 6-12). This condition will also result in a chromium-depleted
region along the grain boundaries. In this “sensitized” condition, the material will
have increased susceptibility to intergranular corrosion, intergranular stress
corrosion cracking, and creep cavitation. All of the common austenitic stainless
steel tubing alloys will sensitize in service. The stabilized grades such as 321H
and 347H will sensitize somewhat slower but after a few years of typical SH/RH
service they will also sensitize. The same is true of the lower carbon grades of
these alloys.
 6-12 
13828389
Figure 6-12
Temperature – time relationships related to the formation of grain boundary
carbides in austenitic steels [6.11]. Note that with increased levels of dissolved
carbon the rate and temperature range over which sensitization occurs increases.
6.8 Assessment of Components
The following steps should be considered to assess Carbide Embrittlement:
1. Review available documentation to assess time and temperature of operation.
2. Using the methods outlined as appropriate to the particular alloy and
component under assessment evaluate the risk of embrittlement due to
carbide formation.
3. For any locations that appear to be high risk, perform metallography and
EDS. This should establish microstructural features, for example, grain size,
the size, type and distribution of carbides present and any local variations in
composition.
4. If results from items 2 or 3 suggest a potential problem, perform small punch
testing to estimate toughness (if comparative S.P. data are available).
5. An alternative to item 4 is to perform a series of Charpy impact tests to
determine FATT. However, this option requires that sufficient material
exists for testing.
 6-13 
13828389
6. Evaluate the component using detailed non destructive test methods to
characterize the size and location of any flaws or defects present.
7. Depending on the shift (or the present value of FATT), an assessment of
fracture toughness may be warranted over operating temperature range. This
can also be performed with Cv vs. K1c correlations.
8. Evaluate the stresses in the location, using appropriate analysis, to compare
K1c vs. Kmat.
6.9 References
6.1
K. W. Burns and F. B. Pickering, “Deformation and Fracture of FerritePearlite Structures,” J. Iron Steel Inst., Vol. 202 (No. 11), p. 899–906.
6.2
T. Gladman, B. Holmes, and I. D. McIvor, “Effect of Second Phase
Particles on the Mechanical Properties of Steels,” The Iron and Steel
Institute, London, 1971, p. 68.
6.3
F. B. Pickering and T. Gladman, “Metallurgical Developments in carbon
Steels, Iron and Steel Institute,” Special Report No. 81, 1963, p. 10.
6.4
B. Mintz, W. B. Morrison, and R. C. Cochrane. “Advances in the
Physical Metallurgy and Applications of Steels,” The Metal Society,
London, 1982, p. 222.
6.5
G. J. Roe and B. L. Bramfitt, “Notch toughness of Steels,” ASM
International.
6.6
Fracture Mechanics Properties of Carbon and Alloy Steels, Fatigue and
Fracture, ASM International.
6.7
R. Viswanathan, J. R. Foulds, and D. I. Roberts, “Methods for
Estimating the Temperature of Reheater and Superheater Tubes in
Fossil Boilers,” Conf. Proc. “Boiler Tube Failures in Fossil Power
Plants,” EPRI, 1987, pp. 3-35 to 3-53.
6.8
J. H. Hollomon and L. D. Jaffe, Time-Temperature Relations in
Tempering Steel, Trans. AIME, Vol. 162, 1945, p. 223–249.
6.9
S. Wignarajah, I. Masumoto, and T. Hara, “Evaluation and Simulation
of the Microstructural Changes and Embrittlement in 21/4Cr1Mo Steel
due to long term service,” ISIJ International, 1990, Vol. 30, pp. 58–63
6.10
J. D. Parker and G. C. Stratford, “Characterization of Microstrucures in
Nickel based Transition Joints,” J. of Materials Science, Vol. 35, No. 16,
2000, pp. 4099–4107.
6.11
S. M. Bruemmer, “Quantitative Modeling of Sensitization Development
in Austenitic Stainless Steel,” Corrosion, 1990, pp. 698–709.
 6-14 
13828389
Section 7: Temper Embrittlement of Steels
7.1 Introduction
Temper embrittlement is a major cause of degradation of toughness of ferritic
steels. Numerous components become candidates for retirement if they are
severely embrittled since under these conditions the critical crack size can become
very small. The problem is encountered as a result of exposure of a range of alloy
steels in the temperature range 345 to 540°C (650 to 1000°F). Slow cooling
following tempering or post weld heat treatment, or service exposure in this
temperature range can lead to embrittlement. Following the introduction this
section covers:

Mechanisms related to temper embrittlement

Factors affecting temper embrittlement

Relationships to describe metallurgical effects on temper embrittlement

Approaches to assess temper embrittlement in service components
Temper embrittlement may be avoided by heat-treating above the susceptible
temperature range followed by rapid cooling. Unfortunately, in the case of
massive components no rate of cooling is fast enough and some embrittlement
may be inevitable. Thus, in components such as LP rotors, generator rotors and
retaining rings even though normal operation occurs at relatively low temperature
some temper embrittlement may be present as a result of slow cooling following
heat treatment.
Clearly there will be a significant number of components where normal operating
temperatures are in the critical range so that temper embrittlement can occur
during service. This group will involve components of all section size including
boiler headers, steam pipes, turbine casings, pressure vessels, blades, fasteners,
HP-IP rotors, and combustion turbine disks. It should be emphasized that the
larger section components within this group may suffer embrittlement both
during the heat treatment cycle and in service.
This problem has been identified in a wide range of alloys including low alloy
steels, higher strength alloy steels and stainless steels and it is traditionally of
greater risk with components manufactured using older methodologies. This
increased susceptibility is related to higher normalizing temperatures, since these
higher temperatures result in larger grain sizes, and when steel making practices
 7-1 
13828389
lead to higher levels of impurities, particularly involving elements such as P, Sn,
Sb and As. Temper embrittlement occurs when these trace elements diffuse to
grain boundaries so that with respect to the behavior observed during Charpy
impact testing:

Intergranular fracture rather than cleavage occurs in the brittle lower shelf
region

The brittle to ductile transition takes place at a higher temperature, that is,
there is an increase in FATT, which under extreme situations may be as
much as 300°C
The increased risk of rapid brittle fracture is generally not a major concern during
steady operation at highest temperatures since it is normal to have highest
fracture toughness at highest temperature (that is, under normal operation an
assessment of the risk of fast fracture is typically related to the upper shelf
energy). However, problems have been encountered during hydrostatic testing of
pipes and vessels or during transients since high stresses can be present at
relatively low temperatures when there is still a low toughness. The risk of failure
should also consider behavior during steady operation since stable crack growth
in service may lead to instability at subsequent transients. When undertaking
assessment at the higher temperatures it is important to consider whether
embrittlement has resulted in a reduction in upper shelf energy.
7.2 Mechanisms Related to Temper Embrittlement
In a material, when the grain boundary energy of a system is reduced by the
presence of an alloying element, the concentration of that element in the
boundary will be higher than that in the matrix. This will occur because it is
energetically favorable for the element to be situated in the grain boundary since
the relatively disordered state compared to the lattice will offer low energy sites.
In grain boundary segregation theory, grain boundary solution concentrations.
Xb, are expressed as a fraction of a monolayer. One monolayer, that is, Xb = 1,
means that the atoms in the boundary could be arranged to form a single close
packed layer of atoms. For low mole fractions of solute, concentration Xb is
approximately given by the Langmuir-McLean equation [7.1]:
Xb = Xo. exp ( ΔGb / RT)
Eq. 7-1
Where ΔGb is the free energy released per mole when a solute atom is released
from the matrix to the boundary. The ΔGb is usually positive and roughly
increases as the size misfit between the solute and the matrix increases and as the
solute-solute bond strength decreases. That is, there will be a greater tendency
for segregation with solute atoms that are larger than the matrix and for solute
atoms that are not strongly bonded to each other.
Several detailed studies have been carried out to measure the appropriate free
energy values in specific system, The greater accuracy associated with these
investigations has been possible through application of advanced quantitative
analytical techniques including Auger electron spectroscopy, low energy electron
 7-2 
13828389
diffraction and x-ray photoelectron spectroscopy. A number of examples taken
from the recent review paper by Grabke [7.2] are particularly relevant to
understanding segregation and hence temper embrittlement. In specialist ironphosphorus alloys containing between 0.003 and 0.33 wt %P it was found that
the grain boundary segregation of P decreased with increasing temperature and
decreasing bulk concentration, Figure 7-1. This is in direct agreement with
equation 7-1.
In a similar manner, segregation effects in iron-carbon-phosphorus alloys were
studied. It was found that with increasing free carbon content the grain boundary
concentration of phosphorus decreased, Figure 7-2. The plateau shown at a
carbon level of 55 ppm occurs because that is the solubility limit at 600°C, (that
is, higher levels of carbon would result in the formation of cementite so there
would be no further increase in the grain boundary concentration).
Detailed analysis showed that the ΔG values for grain boundary segregation of
carbon and phosphorus at 550°C were –72kJ/mol and –49kJ/mol respectively.
Since the more negative the value the greater is the driving force, segregation of
carbon to grain boundaries is energetically more favorable compared to
phosphorus.
Figure 7-1
Dependence of the grain boundary concentration of phosphorus on annealing
temperature, for Fe-P alloys with different P levels [7.2]
 7-3 
13828389
Figure 7-2
Grain boundary concentration of P and C in Fe – 0.17%P alloys with different
carbon contents [7.2]
In carbon steels with carbon contents greater than 0.02% there will always be
sufficient free carbon to minimize phosphorus segregation. However, when
alloying elements, which have a strong tendency to form carbides (for example,
Cr or V), are added this situation changes. This change is a direct result of the
strong carbide forming element removing carbon from solution so that there are
insufficient carbon atoms available to fill the grain boundary sites. Thus,
phosphorus segregation can occur. These effects are illustrated in Figure 7-3.
Clearly, similar concentrations of grain boundary phosphorus are shown in the
Fe-P and Fe-P-2%Cr alloys. However, when carbon is added to Fe-P, the level
of P segregation is dramatically reduced, then with the addition of 2% chromium,
that is, an Fe-C-P-Cr alloy, the segregation of P is nearly at the same level as in
the alloys without carbon.
 7-4 
13828389
Figure 7-3
Effects of carbon and chromium on the grain boundary segregation of P after
annealing at different temperatures in the range 400°C to 800°C for Fe – P, Fe –
Cr – P, Fe – C –P and Fe – Cr – C – P alloys with about the same bulk
concentration of P [7.2]
From equation 7-1, it can be seen that the driving force for segregation decreases
as temperature increases (that is, at a high temperature the solute will be
dissipated throughout the matrix). For lower temperatures, Xb will increase
towards unity and Xb reaches saturation for very low temperatures. However, at
low temperatures diffusion rates will be slow so that segregation cannot occur in
reasonable times. Time-Temperature Relationships for Temper Embrittlement
thus follow C-curve behavior, Figure 7-4. At high temperatures, the kinetics of
impurity diffusion to grain boundaries are rapid, but the tendency to segregate is
low because the matrix solubility for the element increases with temperature.
Hence, embrittlement occurs rapidly but to a small degree. At low temperatures,
the tendency to segregate is high, but the diffusion kinetics are not rapid enough
to reach maximum embrittlement. The optimum combination of thermodynamic
and kinetic factors favoring embrittlement occurs at some intermediate
temperature, called the "knee" of the C-curve.
For commercial steels of interest, the knee occurs in the temperature range from
455 to 510°C (850 to 950°F) but can be shifted up or down depending on the
composition, grain size, and microstructure of the steel.
 7-5 
13828389
Figure 7-4
C – curve behavior between temperature and time for 21/4Cr1Mo steel, showing
isothermal ΔFATT contours [7.3]
7.3 Factors Affecting Temper Embrittlement
The following points have been identified based on review and analysis of general
body of literature. All generally held factors have been included even if most
recent work has led to some shifts in emphasis.
No significant segregation of impurity elements has been found in carbon (C)
steels; presumably this is because there is sufficient free C in solution that
potential grain boundary sites are taken up by carbon. Embrittlement can occur
when carbon is tied up as carbides, for example, due to the addition of chromium
(Cr), vanadium (V) or niobium (Nb).
In typical commercial alloy steels there will be little free carbon so that
phosphorus (P) will produce embrittlement due to segregation during slow
cooling from high temperatures involved with normalizing, during controlled
step cooling and on holding in the temperature range 345 to 540°C (650 to
1000°F). Typical results for 3 rotor steels are shown in Figure 7-5.
 7-6 
13828389
Figure 7-5
Typical results for 3 rotor steels

Tin (Sn), Antimony (Sb) and Arsenic (As) have also been reported as leading
to embrittlement. However, in some cases the embrittling effect of these
elements has been shown in special alloys produced with no P. When P
is present, it appears to exhibit the dominant effect since in several
investigations where the combination of impurities is present; the effect of
P is shown to be dominant and systematic.
Figure 7-6
Grain boundary segregation of Sn in Fe – 0.2% Sn alloy [7.2]
 7-7 
13828389
Recent experiments have demonstrated that the tendency for tin to segregate to
grain boundaries is significantly lower than that of phosphorus, Figure 7-6, and
that this tendency will be further lowered by the presence of carbon atoms,
Figure 7-7. However, at temperatures above about 500°C the rates of tin
diffusion will be such that because of the favorable energy for tin diffusion to free
surfaces, there will be a tendency for tin to diffuse to any creep cavities. This
diffusion will then accelerate tertiary creep processes leading to rapid rates of
creep microcrack formation and growth in alloys with significant levels of tin.
Figure 7-7
Grain boundary segregation in Fe –Sn – C alloys as a function of the bulk carbon
concentration at 550°C [7.2]

Manganese (Mn) and silicon (Si) in combination appear to affect the level of
embrittlement when P is present. Thus, although it is clear that
embrittlement can be noted in special alloys produced without Mn and Si, in
commercial alloys a systematic effect of greater embrittlement due to P at
higher levels of the sum of Mn and Si is identified. Indeed, there is evidence
to suggest that there will be increased embrittlement with each of the
elements individually, since Mn is believed to reduce the grain boundary
fracture strength and Si is believed to promote the segregation of P. A recent
reevaluation of embrittlement data from a range of 2CrMo weld metals has
demonstrated this effect. It should be noted that although these welds also
contained Sb, Sn and As the neural network analytical techniques applied
suggested that these elements exhibited no significant effect and that the
variations in brittleness could be described on the basis of the negative effects
of P, Mn, Si and the positive effect of Mo, Figure 7-8.
 7-8 
13828389

The presence of molybdenum (Mo) acts to slow down the embrittlement due
to P. Specifically, even apparent changes of Mo in P22 type alloys where the
variations were between 0.9% and 1.27% have been shown to reduce
embrittlement in exposures of a given time, Figure 7-8. This effect is
believed to arise either because Mo and P atoms tend to associate so the
latter is prevented from segregating to prior austenite grain boundaries or
because Mo will increase the coherency of the grain boundary structure.
Figure 7-8
Reanalysis of data of Bruscato [7.5] showing that increases of Mn, Si and P
reduced toughness and increased levels of Mo improved toughness. No significant
trends in toughness were found for the other elements present [7.6]

The rate of embrittlement with time appears to follow a parabolic
relationship such that a plateau is reached. This has been seen in studies of
rotor steels where FATT increases as the time of exposure at 850°F up to
about 40,000 h, Figure 7-9. Indeed, in this work, continued ageing actually
lead to a reduction in FATT. A similar effect has been reported in the ageing
of 2¼Cr1Mo steel.

The segregation effects are reversible and embrittlement can be removed by
high temperature heat treatment followed by rapid cooling.
 7-9 
13828389

The temperature of exposure is important. Studies looking at the effect of
step cooling with significant hold times in the range 345 to 540°C (650 to
1000°F), isothermal heat treatment or evaluation of surface exposed
components indicate that maximum embrittlement occurs between about
700-800°F. This effect is due to the fact that the driving force for
embrittlement is based on the tendency for segregation to follow a ‘C’ curve.
This behavior is illustrated in Figure 7-4.
Figure 7-9
Variation of ΔFATT with time of aging at 850°F for CrMoV rotor steel [7.7]

The segregation effect results in steep gradients of the impurity at the grain
boundary, Figure 7-10. While the specific models of grain boundary
structure vary (and it is observed that even in embrittled material there will
be different levels of segregation at different boundaries), it is generally
agreed that the transition from the crystallography from one grain to the next
occurs over a distance equivalent to 2-3 atomic planes. The fact that peak
segregation has been shown experimentally over this type of distance, and
then fall rapidly indicates that energy considerations limit segregation to
within the peak regions of grain boundary structure. There appears to be a
trend in decreasing susceptibility to embrittlement from martensitic to
bainitic to ferritic structure. It should be emphasized that the embrittlement
in martensitic and bainitic structures occurs as a consequence of trace
element concentration at prior austenite grain boundaries.

Grain size appears to be an important factor. Since the available grain
boundary area will decrease as grain size increases, it is reasonable that for a
given trace element the ability to reach a critical value at any boundary will
depend on grain size and the overall concentrations of the trace element,
Figure 7-11. Further issues that should be important include the fact that
grain size will lead to an increase in yield strength and fine-grained structures
generally exhibit improved fracture resistance.
 7-10 
13828389

Some studies have indicated that the strength of the material will be
important, thus may be considered directly from measurements of yield or
tensile strength or inferred from measurements of hardness.
Figure 7-10
AES measurements show that high levels of S, P, and Sb segregated to grain
boundaries fall rapidly with distance away from the boundary [7.8]
 7-11 
13828389
Figure 7-11
Variation of FATT with prior austenite grain size at fixed hardness and impurity
levels [7.4]
7.4 Relationships to Describe Metallurgical Effects on Temper
Embrittlement
A very large number of investigations studying the effects of composition,
microstructure and mechanical properties have been carried out. A summary of
the compositional effects associated with the major alloying elements normally
found in power plant steels is provided in Table 7-1.
Table 7-1
Summary of the influence of alloying elements on microstructure and embrittlement
[7.9]
 7-12 
13828389
In addition, a number of studies have assessed the levels of embrittlement
associated with specific metallurgical factors and suggested a range of different
expressions which can be used to evaluate the potential for embrittlement. As
noted earlier, many of these studies focused on particular alloys or examined
embrittlement in specialist alloy systems. Thus, whilst each expression has some
merit in describing the behavior of the data generated within a given set of
results, the general applicability to predict embrittlement for other alloys has in
several cases not been demonstrated. Despite these reservations it must be
emphasized that the general appreciation of embrittlement phenomena, which
has come from these efforts, has resulted in significant improvements in the
levels of trace elements in modern steels and has thus the research involved has
played a key role in reducing the risks from embrittlement due to grain boundary
segregation in steels presently available, for example Figure 7-12.
Figure 7-12
Reduction in the level of trace elements with time for 21/4Cr1Mo steel
components [7.17]
 7-13 
13828389
7.5 Equations Used to Predict Temper Embrittlement
Work in this area has been undertaken based on the influence of composition on
susceptibility to embrittlement for fixed embrittlement conditions. These studies
have:

Considered special alloys produced for specific laboratory study or examined
the following commercial steel grades, chromium-molybdenum-vanadium
and nickel-chromium-molybdenum-vanadium, which are used in the
fabrication of rotors and 21/4Cr-1Mo, which is used for pressure vessels and
piping.

Evaluated embrittlement effects using different thermal exposures including
slow or step cooling through the susceptible temperature range, isothermal
treatments under controlled conditions or by evaluation of ex-service
components.

Adopted empirical approaches to describe embrittlement effects using
selected variables usually through the application of regression analysis
techniques.

Monitored embrittlement effects through Charpy impact tests, although in
some studies the full transition curve has been developed in others changes
were monitored through variations in room temperature impact energy.

Considered the changes in fracture behavior in terms of temper
embrittlement, that is, as a function of grain boundary segregation. In some
cases changes in the overall fracture behavior may have also involved the
formation of grain boundary carbides and been influenced by variations in
metallurgical parameters such as grain size, the type size and distribution of
inclusions.
The equations available are summarized below, for further details the specific
references are provided:

Vacuum carbon deoxidized nickel-chromium-molybdenum-vanadium rotor steels
were isothermally embrittled at 400°C (750°F) for 10,000 h (107), the shift
in FATT (ΔFATT) in degrees Celcius was correlated [7.10] to the impurity
content and molybdenum concentration (all in weight percent) by the
equation:
ΔFATT = 7524P + 7194Sn + 1166As − 52Mo
− 450,000(P × Sn)
Eq. 7-2
While no significant influence was found for antimony, phosphorus, tin,
and arsenic increased embrittlement, and molybdenum decreased it. A
phosphorus-tin interaction that decreased embrittlement was also observed.
 7-14 
13828389

A correlation between the 50% FATT and impurity content (J factor) for
both nickel-chromium-molybdenum-vanadium and 21/4Cr-1Mo steels has been
demonstrated [7.11]. The J factor equation is:
J = (Mn + Si)(P + Sn) × 104
Eq. 7-3
Where all concentrations are in weight percent. While this expression has found
significant acceptance and use in commercial applications it should be
emphasized that it cannot have universal applicability since it requires calculation
of the product of Mn and Si and P and Sn. Clearly, laboratory studies have
shown that significant embrittlement can occur even when no Mn and Si are
present yet for these situations equation 7-3 gives a J factor of 0, that is, no
embrittlement will be predicted.
Despite this limitation additional work has suggested that the J factor can be
used directly to estimate ΔFATT as:
∆FATT (°C) = 0.38 (J) – 45
Eq. 7-4
This indicates that improved ΔFATT will be obtained for decreasing values of J.

A detailed correlation has been provided for nickel-chromium steels doped with
manganese, phosphorus, and tin [7.12]. The equation combines the grainboundary phosphorus and tin concentrations, the prior-austenite grain size,
and the hardness level. This equation was extended to a nickel-chromiummolybdenum-vanadium steel containing 0.02Si 0.32Mn, 0.019P and 0.021S
heat treated to a bainitic microstructure with hardnesses of 20 and 30 HRC,
ASTM grain sizes of No. 3 and No. 7, and isothermal embrittlement at
480°C (895°F) for 6000 h [7.12]. The resulting equation was:
∆FATT = 4.8P + 24.5Sn + 13.75(7−GS) + 2(Rc − 20)
+ 0.33(Rc−20)(P + Sn) + 0.036(7 − GS) (Rc-20)(P + Sn)
Eq. 7-5
Where ∆FATT is given in °C, concentrations of P and Sn are expressed as the
global average of the respective peak height with respect to iron; Rc is the
Rockwell ‘C’ hardness and GS is the ASTM grain size number. The complicated
nature of this expression is a potential limitation to its application but as shown
in Figure 7-13 the calculated values of FATT are in reasonable agreement with
actual measurements.
 7-15 
13828389
Figure 7-13
Correlation between measure values of FATT with estimates calculated using
equation 7-5 for NiCrMoV steel [7.12]

A study on ultra low carbon steels used to manufacture sheet has suggested the
following expression [7.13], however, no specific experimental details were
provided:
∆FATT (°C) = 0.28P + 0.38Sb + 0.16Sn + 0.48As
– (0.85Be + 21C + 20B)
Eq. 7-6
In this equation the element compositions are given in ppm weight.
It was also emphasized that the coefficients depend in detail on overall
composition and processing history (that is, microstructure and grain size). It
should be noted that this equation takes into account the potential benefit
from elements with small atomic radii which if available in solution can
segregate to grain boundaries in preference to the deleterious trace elements.

A very comprehensive investigation has been reported assessing
embrittlement effects in 1Cr1Mo1/4V rotor steels [7.7]. This work compiled
and analyzed relevant data from:
-
18 core bars which were aged at 850°F for various times up to
90 000 hours
-
3 high purity rotor steels
-
Published and unpublished reports describing the design, operation as
well as the microstructure and properties of 19 retired rotors
 7-16 
13828389
It was found that the embrittlement effects could be reasonably described by the
equation
Post exposure FATT (°C) = 100(F) + 1.8P
Eq. 7-7
Where P is in ppm. The results presented are shown in Figure 7-14, and it is
apparent that despite the different potential influences the simple expression
between FATT and P provides a very reasonable description of all the results.
In Figure 7-14 data from grade C and grade D rotors have been identified
since these grades were subject to different normalizing treatments and hence
exhibited different values of prior austenite grain size. The highest degrees
of embrittlement were found with the coarse grained C grade rotors. This
observation again reinforces the fact that fine grained material provides improved
performance.
Figure 7-14
Variation of post exposure FATT with the phosphorus content of the 1Cr1Mo1/4V
rotor steel [7.7]

The embrittlement behavior of shielded metal arc weld deposits of 21/4 Cr1Mo
steel has been evaluated. Realistic comparisons of behavior were obtained
since after fabrication and PWHT, one half of each weld was subjected to a
step cooling procedure, which involved significant hold times at progressively
lower temperatures in the range 1000°F to 725°F. The energy measured in
 7-17 
13828389
room temperature Charpy tests was then assessed for both batches of
material and based on the results approximate zones could be identified when
the levels of Mn + Si were plotted against an X Factor [7.5] This factor was
given by the expression:
X = 10P + 5Sb + 4Sn + As
Eq. 7-8
100
Where the value for each element is the composition expressed as ppm.
Thus, this work identified the fact that embrittlement was influenced by
alloying and impurity elements, this influence has been emphasized in the
recent reevaluation of these data, Figure 7-8 [7.6]. No transition curves were
measured so that detailed effects on FATT could not be considered.

A further expression for an embrittlement factor which attempts to utilize
the X parameter in combination with the influences of alloying elements in a
way which overcomes problems associated with the ‘J’ factor described earlier
has been suggested [7.14]. This work examined the behavior of 21/4 Cr 1Mo
and 3Cr1Mo steels and suggested an embrittlement factor, EF, as:
EF = %Si + %Mn + %Cu + %Ni x Y
Eq. 7-9
Where Y= 1/100 (10P + 5Sn + Sb + As) with all these given in ppm

Evidence of unexpectedly rapid fracture in a range of CrMoV piping welds has
been assessed [7.15]. Further details of one of these failures is provided in the
case study; see Appendix E. However, based on programmes of
metallographic investigation and post exposure testing it was identified that
trace high levels of tin were present in these welds and a creep embrittlement
factor based on the composition of impurities was suggested as:
Creep Embrittlement Factor =
P + 3.57Sn + 8.16Sb + 2.43As

Eq. 7-10
Carbon Manganese steels will typically exhibit microstructures of ferrite and
pearlite. In evaluations of the behavior of these steels it has been observed
that Mn will affect the transformation characteristics during cooling, that is
the transformation temperature will be depressed so that there will be
significant refinement of the ferrite grains produced. Thus, in these steels
Mn will have a number of effects on microstructure and thus mechanical and
fracture behavior including effects on grain size as well as pearlite volume
fraction and intermellar spacing. For C/Mn steels, with carbon levels up to
about 0.2%, that is about 25-30% pearlite the FATT has been related to
composition and microstructure using the equation [7.16]:
50% FATT (°C) = 19 + 44 (%Si) + 700 (Nf)1/2
+ 2.2 (%pearlite) – 11.5d-1/2
 7-18 
13828389
Eq. 7-11
Where Nf is the free nitrogen content in wt% and d is the mean linear ferrite
grain size in mm. This equation was developed using multiple regression analysis
and for low values of nitrogen, max N 0.013wt%, it was not possible to
differentiate between linear and root functions. The significant influence of
pearlite in this equation arises because there are lots more iron carbide bonds that
have low fracture resistance. With very high fractions of pearlite the fracture path
can be transgranular and very flat. The influence of elements on FATT was also
assessed, giving an overall ranking as:

1% Si FATT+ 44°C

0.01%N FATT + 70°C

1wt %Sn FATT + 136°C

1wt%P FATT + 459°C
Thus, this work again shows that P has a major influence on brittle fracture.
7.6 Case Studies/Examples
7.6.1 Assessment of Components
The following steps should be considered to assess Temper Embrittlement:
1. Review available documentation to identify component compositions. If no
reliable information from manufacturing records exists the material chemistry
should be measured using material filings or by local excavation. A portable
X-ray/spectrographic unit may be used for direct measurement provided
accurate results concerning trace elements can be obtained.
2. Using the quantitative chemistry results, estimate the FATT from the
predictive formulae outlined in the Section 7.5.
3. For any locations that appear to be high risk, perform metallography and
EDS. This should establish microstructural features, for example, grain size,
and any local variations in composition.
4. If results from items 2 or 3 suggest a potential problem, perform small punch
testing to estimate toughness (if comparative S.P. data are available).
5. An alternative to item 4 is to perform a series of Charpy impact tests to
determine FATT. However, this option requires that sufficient material
exists for testing. If this is a rotor/disc, the OEM can often provide original
toughness property curves (and original FATT) to compare room
temperature Charpy results (a couple of test specimens). Comparison of the
original and ex-service data will provide evidence of a substantial shift in
FATT.
 7-19 
13828389
6. Evaluate the component using detailed non destructive test methods to
characterize the size and location of any flaws or defects present.
7. Depending on the shift (or the present value of FATT), an assessment of
fracture toughness may be warranted over operating temperature range. This
can also be performed with Cv vs. K1c correlations.
8. Evaluate the stresses in the location, using appropriate analysis, to compare
K1c vs. Kmat.
7.7 References
7.1
D. McLean, Grain Boundaries in Metals, Oxford Press, 1957.
7.2
H. J. Grabke, Review “Surface and Grain boundary Segregation on and
in Iron and Steels,” ISIJ International, Vol. 29, No. 7, 1989,
pp. 529–538.
7.3
I. Masaoka, I. Takase, S. Ikeda, and R. Sasaki, Investigation on the
Hydrogen Attack of Welded Joints for 1/2Mo Steels (Report 1), J. Japan
Weld Soc., Vol. 46, No. 11, 1977, p. 818.
7.4
C. J. McMahon Jr, “Problems of Alloy Design in Pressure Vessel Steels,”
in Fundamental Aspects of Structural Alloy Design, McGraw-Hill, New
York, 1977, pp. 295–322.
7.5
R. Bruscato, “Temper Embrittlement and Creep Embrittlement of
21/4Cr1Mo Shielded Metal Arc Weld deposits,” Welding Research
Supplement, April 1970, pp. 148s–156s.
7.6
S. H. Lalam, H. K. D. H. Badheshia, and D. J. C. MacKay, “Bruscato
Factor in Temper Embrittlement of Welds,” Science and Technology of
Welding and Joining, Vol. 5, No. 5, 2000, pp. 338–340.
7.7
R. Viswanathan and S. Gehl, “A Method for Estimation of the Fracture
Toughness of CrMoV Rotor Steels Based on Composition,” J of
Engineering Materials and Technology, Vol. 113, 1991, pp. 263–270.
7.8
P. W. Palmberg and H. L. Marcus, “An Auger Spectroscopic Analysis of
the Extent of Grain Boundary Segregation,” Trans. ASM, Vol. 62, 1969,
p. 1016–1018.
7.9
The Elimination of Impurity Induced Embrittlement in Steels, Part I. EPRI,
Palo Alto, CA: 1980. NP-1501.
7.10
D. L. Newhouse et al., Temper Embrittlement Study of NickelChromium-Molybdenum-Vanadium Rotor Steels, I: Effects of Residual
Elements, in Temper Embrittlement of Alloy Steels, STP 499,
American Society for Testing and Materials, 1972, pp. 3–36.
7.11
J. Watanabe and Y. Murakami, “Prevention of Temper Embrittlement
of Chromium-Molybdenum Steel Vessels by Use of Low-Silicon Forged
Steels,” Proc. API Refin. Dept., Vol. 60, 1981, pp. 216–224.
7.12
Impurity Induced Embrittlement of Rotor Steels, Vol. 1 Temper
Embrittlement. EPRI, Palo Alto, CA: 1983. CS-3248.
 7-20 
13828389
7.13
J. C. Herman and V. Leroy, “Influence of Residual Elements on Steel
Processing and Mechanical Properties,” Metal Working and Steel
Processing, Cleveland, OH, 1996.
7.14
M. Katsumata and S. Kinoshita, “Microfractographic studies of temper
embrittled steels,” Iron Steel Inst. Jpn., 1977, No. 12, pp. 693–700.
7.15
B. L. King, “Intergranular embrittlement in CrMoV steels: An
assessment of the effects of residual impurity elements on high
temperature ductility and crack growth,” Phil. Trans. Roy. Soc. (1980)A
295, pp. 235–251.
7.16
F. B. Pickering and T. Gladman, “Metallurgical Developments in
Carbon Steels” Iron and Steel Institute, Special Report No. 81, 1963,
p. 10.
7.17
T. Iwadate, “Pressurization Temperature of Pressure Vessels made of
CrMo steels,” PVP-Vol. 288 Service Experience and Reliability
Improvement: Nuclear, Fossil and Petrochemical Plants, ASME 1994,
pp. 156–163.
 7-21 
13828389
13828389
Section 8: Embrittlement Influenced by the
Environment
8.1 Introduction
Metals can fracture catastrophically when exposed to a variety of environments.
These environments can range from liquid metals to aqueous and nonaqueous
solutions to gases such as hydrogen. The phenomenology of these processes and
some of the corrective procedures used or envisaged are described in three main
sub-sections. These are:
1. Oxygen Embrittlement, which has been shown to occur in several metals, for
example, iron, copper and nickel, and some alloys, for example, IN903A and
a number of nickel based or cobalt based superalloys. Following very high
temperature exposure the susceptibility for brittle behavior is increased by
diffusion of oxygen along grain boundaries. Although not generally a
problem in fossil boilers concerns have been expressed in some combustion
turbine applications.
2. Liquid metal embrittlement can cause cracking and fracture in stressed parts
of many metals and alloys. Not all combinations of solid and liquid metals
produce embrittlement. For example, aluminum is embrittled by liquid
gallium, sodium, and tin, and steel has been reported to be embrittled by
liquid cadmium, lithium copper, brass, aluminum bronze, antimony, and
tellurium. Both aluminum and steel are embrittled by liquid indium, zinc,
and mercury. The melting temperature and chemical reactivity of a liquid
metal are not deciding factors as to whether it will cause embrittlement or
not. Instead, most instances of embrittlement are accompanied by low
solubility and absence of intermetallic-compound formation.
3. Cracking due to corrosion relates to instances where the environment leads
to brittle crack development. Information is provided regarding:
-
Intergranular corrosion which refers to the phenomenon of localized
attack at and adjacent to grain boundaries, with relatively little corrosion
of the grains. Sensitization of stainless steels is a classic example of
intergranular corrosion.
-
Stress-corrosion cracking, which is a mechanical-environmental failure
process in which mechanical stress and chemical attack combine to
initiate and propagate fracture in a metal part. The synergistic action of
sustained tensile stress and a specific corrosive environment produce
 8-1 
13828389
stress corrosion cracking. Thus, it should be emphasized that this
synergistic behavior causes failure to occur more rapidly than it would if
the separate effects of the stress and the corrosive environment were
simply added together. Indeed, failure by stress corrosion cracking is
frequently caused by simultaneous exposure to a seemingly mild chemical
environment and to a tensile stress well below the yield strength of the
material. Under such conditions, fine cracks can penetrate significant
distances into the component while the surface exhibits only insignificant
evidence of corrosion. Therefore, there may be no external indications of
an impending failure.
Each sub-section covers:

Background information

Mechanisms of damage

Case studies or examples
The final discussion outlines actions to be taken for component assessment.
8.2 Oxygen Embrittlement
8.2.1 Introduction
The degradation of properties, particularly ductility due to exposure to oxygen
has been recognized for some time. This phenomenon has been shown to occur
in several metals, for example, iron, copper and nickel, and some alloys, for
example, IN903A and a number of nickel based or cobalt based superalloys. The
following summary has been prepared based on a review of this subject prepared
by Woodford and Bricknell [8.1].
Relatively short-term prior exposure in air at very high temperature could lead to
profound embrittlement at intermediate temperatures. This effect was explained
on the basis of intergranular diffusion of oxygen that penetrated rapidly along
grain boundaries. The embrittlement was demonstrated using measurements of
tensile or creep ductility at intermediate temperatures in iron-based, nickelbased, and cobalt-based alloys. An example of the results for the Fe-Ni-Co alloy,
IN903A, is shown in Figure 8-1. Post-exposure tests on cast alloys also showed
that this embrittlement could lead to a reduction in rupture life of several orders
of magnitude.
 8-2 
13828389
Figure 8-1
Ductility of alloy IN 903A as a function of temperature for in-vacuum tests.
Samples were tested after air and vacuum exposures at 1000°C. Embrittlement
remained in the samples exposed to air after machining the samples to half
diameter prior to testing [8.1].
8.2.2 Mechanisms
In alloys based on iron, nickel and cobalt, the low solubility for oxygen provides
the potential, based on the concept of a grain boundary enrichment factor [8.1]
(which is the ratio of the interfacial concentration in fraction of a monolayer and
the bulk solute concentration), to result in enhanced oxygen levels at grain
boundaries.
Using model alloys based on nickel, three embrittling reactions involving oxygen
were confirmed: These involved:

Reaction with carbon to form carbon dioxide gas bubbles, these bubbles act
as pre-existing grain boundary voids, which resulted in very rapid creep
failure with extensive grain boundary cracking compared to unaffected
material, Figure 8-2

Reaction with manganese sulfides on grain boundaries to release sulfur; it is
widely accepted that sulfur in elemental form will produce severe
embrittlement

Reaction with oxide formers to form fine oxides that act to provide multiple
sites for the nucleation of creep voids and pin grain boundaries
 8-3 
13828389
These phenomena are believed to be the same processes that serve to embrittle
the region ahead of a crack tip. Thus, oxygen attack may occur dynamically to
account for the accelerated advance of a crack in air tests compared with inert
environment tests, and it may occur during higher-temperature exposure with or
without an applied stress to set up an embrittlement situation. Thermal fatigue in
combustion turbines is a particularly challenging situation for oxygen attack since
maximum strains develop at intermediate temperatures in the cycle, but holding
may be at the maximum temperature.
Figure 8-2
Unetched microstructure of nickel samples following air testing under the same
conditions at 800°C. (a) Pure condition unloaded after 500 hours with minor
cavitation, and (b) embrittled condition which failed after 23 hours [8.1].
8.3 Liquid Metal Embrittlement
8.3.1 Introduction
Liquid – metal embrittlement, which occurs at temperature above the melting
point of the embrittling metal or alloy, leads to rapid (normally intergranular)
fracture. An example is shown in Figure 8-3. A similar phenomenon, namely
solid-metal induced embrittlement has been noted at temperatures near to but
below the melting point. Both types of embrittlement appear to be influenced by
similar factors, namely:

Intimate contact at the atomic scale between the stressed solid and the
embrittling element

The presence of tensile stress sufficient to cause, at the very least, local plastic
deformation

Crack nucleation at the solid/embrittling element interface from a
discontinuity such as a grain boundary
Metallurgical factors that have been associated with increased brittleness in
metals and alloys also appear to increase the tendency for liquid metal
embrittlement. Thus, coarse grain size, rapid strain rate, high yield strength, and
the presence of notches or stress raisers, all appear to increase embrittlement.
 8-4 
13828389
Figure 8-3
Example of an intergranular liquid metal fracture in alloy steel
In general, the susceptibility to embrittlement is stress and temperature sensitive
and does not occur below a specific threshold stress value. This threshold stress
may be that needed to permit local plastic strain and, in alloys that form a
protective oxide, cause cracking of the surface oxide thus allowing the liquid to
access the metal. It has also been noted that in many solid/liquid systems there is
an embrittlement temperature zone, where the lower bound relates to a
temperature near the melting point with the upper bound at some more elevated
temperature; for example, see Figure 8-4. However, the damage observed does
not appear to be systematically explained by temperature or time. Indeed, in
many cases once the conditions for embrittlement have been established very
rapid rates of crack propagation and failure have been noted.
 8-5 
13828389
Figure 8-4
The effect of temperature on the reduction in area of Fe-35% Ni alloy samples in
the presence of copper [8.2]
The embrittlement process is not associated with corrosion, dissolution, or any
diffusion-controlled intergranular penetration, but is considered to be a special
case of brittle fracture. Thus, in most cases of liquid metal embrittlement, except
at grain boundaries little or no penetration of liquid metal into the solid metal is
observed. The embrittlement of the solid metal coated with liquid metal or
immersed in the liquid does not depend on the time of exposure to the liquid
metal or on whether the liquid is pure or presaturated with the solid. It has been
noted that in very many cases when embrittlement occurs the solid has little or
no solubility in the liquid and forms no intermetallic compound to constitute an
embrittlement couple. However, exceptions to this empirical rule have been
noted.
The amount of the embrittling liquid required to generate damage is small. Thus,
crack initiation can occur with very limited levels of liquid. Propagation by
continued LME will be dependent on a continued supply of liquid to crack tip.
Because the amount needed is that just sufficient to wet the new grain boundary
surfaces, in many cases failure occurs by this mechanism alone. In circumstances
where insufficient liquid is present, rapid failure can still occur if the initial crack
exceeds the critical value for extension by brittle fracture.
8.3.2 Mechanism of Liquid Metal Embrittlement
The established theoretical models for intergranular brittle fracture appear to
apply to crack formation and growth under conditions where liquid metal
embrittlement develops. Thus, the initiation phase requires that the local grainboundary stresses generated by impinging slip bands of dislocations produced by
plastic deformation result in microcrack formation. It appears that the interfacial
 8-6 
13828389
energy of the solid metal-liquid metal pair is sufficiently low for very rapid macro
cracking [8.3, 8.4]. Laboratory measurements indicate that the surface energy
associated with crack formation in the presence of some liquids is only 1/100th
that for a particular alloy when fracture occurs in air. Since the fracture stress, σf,
is given by:
σf = {AGγ / (1- ν) L} 1/2
Eq. 8-1
Reductions in the surface energy term γ will lead to a significant decrease in the
stress needed for brittle fracture.
The other significant issues are thus to consider which combinations of elements
will lead to liquid metal embrittlement and identification of any specific features,
such as grain boundary structure, which can affect susceptibility. Thus, the key
problem seems to be how the crack starts rather than how it spreads. It has been
suggested [8.4] that the elements likely to result in embrittlement can be defined
by a susceptibility factor, η, where:
η = γSL/ γSV or (2γSL – γGB) / (2γSV – γGB)
Eq. 8-2
Where γSL is the solid – liquid surface energy, γSV is the energy of creating new
surface in the solid alone and γGB is the grain boundary energy. Using this
criterion a value of η less than 0.5 has been found to differentiate between
embrittling and non-embrittling elements in zinc. Since damage is intergranular
features associated with initiation must be related to the properties of grain
boundaries. There is evidence to suggest that prior austenite boundaries in steels
are more susceptible than ferrite boundaries however the fact that damage has
been found in alloys with different crystal structures suggests that there must be a
general grain boundary factor, which promotes this form of embrittlement. In a
typical polycrystalline material there will be variability in the misorientation
across different boundaries. On this basis there will be some boundaries with the
necessary structure to promote complete wetting so that the liquid metal
penetrates certain grain boundaries spontaneously, possibly even with a saving of
energy. That this is feasible is shown by the system solid Al-liquid Ga, since
when aluminium is dipped into liquid gallium at room temperature it
spontaneously separates along grain boundaries. Evidently liquid gallium can
penetrate many grain boundaries of aluminium. In the more usual case where
general spontaneous penetration does not occur variations in grain boundary
structure, and the associated differences in wetting ability, will favor isolated
penetration at selected boundaries.
Each penetration is a crack that can spread as fast as the liquid can extend into it.
With such a relatively long crack, and on average a small new surface energy,
fracture will spread at low stress, therefore with little plastic deformation.
Moreover, the crack length can be taken as proportional to the grain diameter,
so that the fracture stress will be proportional to (grain diameter)-1/2. Both these
expectations have been realized in experiments on brass immersed in liquid
mercury [8.4]. In this case the effective surface energy is very low so is thus not
 8-7 
13828389
surprising that a metal which is highly ductile in air may become brittle without
limit on immersion in an unsuitable environment. Further, evidence for the
general applicability of this mechanism has been obtained from fracture energies
measured by fracture-mechanics experimental procedures which show that levels
of Kc that agree with brittle behavior.
8.3.3 Factors Affecting Liquid Metal Embrittlement
Mechanisms of embrittlement have been developed to explain much of the
physical evidence related to the causes of damage. However, understanding of
the process is still limited so that a full appreciation of the fundamental factors
involved has not been completely established. Present knowledge cannot predict
the risk of cracking nor define the liquid metal that will probably embrittle
engineering alloys. The number of critical combinations of liquid and solid that
have known cracking potential is small compared with the number for which
there is no experience with cracking. As a guide, Table 8-1 summarizes solid
liquid couples, which have been found by experiment, or in some cases to cause
liquid metal embrittlement.
Table 8-1
Summary of information concerning metal combinations, the symbol X indicates
the liquid metal that embrittles a specific solid (based on 8.5)
 8-8 
13828389
From Table 8-1, it is apparent that some liquid metals have a more general
tendency for embrittlement than others. It should also be apparent that this
problem has been noted in both ferrous and non-ferrous alloys. Four general
categories have been found for observed service failures:

The embrittling liquid may be present following fabrication because the
material selection process did not identify that operating conditions would be
in a susceptible range. This may be found in steels where the lead added to
aid machinability will produce cracking during subsequent forming
operations. Embrittlement of steel by lead occurs from the melting point of
lead up to about 600°C. Above this temperature high ductility failures are
observed.

Surface metal coatings, which have been introduced to prevent corrosion, can
lead to problems either because coating process is not properly controlled or
because the coated product is used under service conditions, which promote
embrittlement. Many instances of problems have been found with galvanized
(zinc coated) or Cadmium-plated components.

Problems have been seen with alloys used for soldering or brazing, for example,
in silver-brazed titanium parts at temperatures above 315°C (600°F).
Welding problems associated with liquid metal embrittlement are less
common but have been reported.

The susceptible combination of metals can occur accidentally. This is most
common due to supply and installation of material different from that
specified. However, in rare occasions the contact arises from one failure,
which can then promote secondary, but potentially more catastrophic,
fracture. The most infamous incident of this type occurred at the
Flixborough petrochemical facility where due to an initial failure liquid zinc
came in contact with stainless steel process piping with disastrous
consequences.
8.3.4 Case Studies/Examples
The most common causes of liquid metal embrittlement in power plant steels are
from:

Exposure to Cu, Cd, Pb and Zn in carbon and ferritic type

Exposure to Zn in austenitic stainless steels
Two case studies are presented which illustrate embrittlement issues.
Embrittlement by Copper in Low Alloy Steel Welds
Through wall cracking was found in CrMo low alloy steel repair welds. The
primary cause of these cracks was determined to be liquid metal embrittlement
due to the presence of copper. Copper deposits were present in the larger cracks
and copper was observed along the grain boundaries of the finer, tighter
intergranular cracks, Figure 8-5.
 8-9 
13828389
Figure 8-5
Micrograph showing CrMo steel weld metal with liquid metal embrittlement due to
copper attack at prior austenite grain boundaries [8.6]
High hardness values were measured indicating that the repair weld had not
been adequately tempered and was likely to exhibit high values of residual
stress. Copper causes LME in carbon and low alloy steels, and the susceptible
temperature of between 1800 and 2190°F would have been present in these
materials. No evidence of damage was detected in the HAZ or parent metal. It
was apparent that the presence of copper was fundamental to the cause of these
cracks but no reason could be cited for the contamination.
Embrittlement of Martensitic 12Cr Turbine Blades
It is common to use brazing to attach erosion shields to the leading edge of LP
blades. The majority of these joints are sound and no problems are encountered.
However, some joints have been found to develop cracks due to liquid metal
embrittlement. Detailed laboratory evaluation confirmed the mechanism as LME
and has also indicated that brittle fracture occurs in these steels at temperatures
around 680°C with a silver based braze alloy containing cadmium, Figure 8-6.
For the same combination of materials no brittle fractures occurred during
exposure at 850°C.
 8-10 
13828389
Figure 8-6
Brittle fracture behavior of 12%Cr martensitic steel that occured under tensile
loading at 680°C when cadmium containing braze was present (a) compared to
ductile behavior under the same conditions without the braze (b) [8.7]
8.4 Cracking Due To Corrosion
8.4.1 Introduction
Low ductility crack formation can occur in the presence of a corrosive medium
due to Intergranular Corrosion. This problem has been well documented in
austenitic stainless steels [8.7, 8.8] and arises due to exposure to temperatures in
the range 425 to 815°C (−800 to 1500°F). This exposure may be the result of:

Incorrect initial heat treatment

Temperature effects introduced by welding thermal cycles

As a result of in-service operation
During exposure to these conditions chromium carbides will form at grain
boundaries, in this condition the material is frequently described as “sensitized”.
Because of variations in diffusion rate the matrix adjacent to the boundary will be
depleted in chromium. The chromium content can fall from the average 18% to
significantly less than 10% Cr in a band about 10 μm wide on both sides of the
grain boundary. This depleted zone will have markedly different corrosion
properties from the adjacent high-chromium matrix. For instance, when the
chromium content falls much below 12%, then the corrosion rate in acidic
solutions rises markedly in a given oxidizing potential range, so that preferential
corrosion occurs. This behavior will be aggravated by the fact that the narrow
depleted zone will have a lower corrosion potential than the larger area of
passivated high-chromium alloy connected to it. Thus, accelerated corrosion can
also occur due to galvanic effects, Sensitized stainless steel is particularly
susceptible to attack by chlorides. Typical micrographs illustrate the general
nature of the corrosion attack, Figure 8-7. Because the metallurgical conditions
for attack occur to some degree on most boundaries many boundaries exposed to
the acid will be damaged.
 8-11 
13828389
Figure 8-7
Typical examples of intergranular corrosion shown by optical metallography and
scanning electron microscopy
Similar intergranular attack phenomena are seen in other systems, where the
localized attack is associated with either active depleted zones (for example, the
copper-depleted zones in Al-Cu or Al-Zn-Mg-Cu alloys or the molybdenumdepleted zones in Ni-Cr-Mo alloys) or with active precipitates (for example,
Mg2Al3 in the Al-Mg alloys or MgZn2 in Al-Zn-Mg alloys).
When a component exposed to a suitable environment is also under significant
stress then localized damage can occur due to stress corrosion cracking (SCC).
This form of cracking occurs as a result of the combined action of:

Susceptible material

Corrosive environment

Tensile stress
All of these factors must be present for damage to initiate, although it should be
emphasized that stresses may be present in the form of applied or residual stresses
and locations where damage by SCC accumulates are often associated with a
stress riser or concentration. It should also be pointed out that once initiated a
corrosive environment in the presence of a tensile stress can propagate the crack.
SCC may be intergranular following a relatively straight path or transgranular
with extensive crack branching, Figure 8-8.
 8-12 
13828389
Figure 8-8
Typical micrographs showing stress corrosion cracking which is (a) intergranular
and (b) transgranular
Comparison of micrographs in Figures 8-7 and 8-8 indicates that in the absence
of significant stress, intergranular corrosion has resulted in damage over a
relatively broad front with many boundaries affected. The intergranular stress
corrosion cracking occurs in a more focused manner and a defect of significant is
likely to be formed. While the change from transgranular to intergranular SCC
in stainless steels is related to “sensitization” in other alloys, including those of
aluminum, magnesium and copper, damage can be intergranular without visible
microstructural changes.
The available information about SCC is extensive (see, for example, references
8.9 to 8.13). After a review of the basic mechanism, the next two portions of this
section briefly review susceptible materials in general, and boiler tubing and
environmental effects in particular. Stresses may be present in the form of applied
or residual stresses and locations where damage by SCC accumulates are often
associated with a stress riser or concentration.
8.4.2 Mechanism
The mechanism of stress corrosion cracking is not fully determined and
continues to be one of the most researched topics in corrosion; a basic review of
some aspects of the mechanism follows.
 8-13 
13828389
The development of stress corrosion cracking requires that a susceptible material
be subjected to both stress and environmental effects. SCC is an electrochemical
phenomenon, an aqueous solution is required, however local, for its occurrence.
Stress only will result in fatigue (cyclic stresses) or overload failures;
environmental attack only will result in pitting or generalized corrosion. SCC
often occurs in ductile materials that form a passivation layer on the surface and
therefore are resistant to general corrosion [8.10].
For purposes of introducing the influences on SCC, the discussion of mechanism
is considered as transgranular attack and crack growth, and intergranular growth.
It should be remembered, however, that these forms of damage can occur
together either sequentially or simultaneously, and there is evidence that the two
propagation modes can be represented as a continuum.
For transgranular SCC, initiation occurs at the surface, from pits, discontinuities,
scratches, or from local modification of the protective oxide. A rupture of the
protective surface film allows an initial corrosive attack. Film rupture is
determined by a variety of local factors, including local stress, strain rate, film
thickness, film rupture strength, substructure of moving dislocations, and
microstructure [8.10]. A further influence on the rate of damage accumulation is
the repassivation rate of the crack walls controlled by the material and such
environmental parameters as potential, pH, solution, and composition [8.10].
Subsequently, the geometric effects of stress concentration at the root of the
initial surface defect lead to crack formation and advance. Successive rupture and
reformation of the passive film occurs at the crack tip as a result of tensile
stresses.
In the case of intergranular SCC, preferential attack occurs at the grain
boundaries and may be further influenced by segregation of impurity elements.
In general, stress corrosion cracks will propagate slowly, and propagation rates
can vary from 0.1 to 10-9 mm/s (0.004 to 4.0 × 10-11 in./s) [8.12, 8.13].
Cracking continues until the stress exceeds the fracture strength of the remaining
non-cracked cross section. Crack initiation and propagation are usually divided
into three stages:

Crack initiation and stage I propagation (increasing growth rate with
increasing stress intensity)

Steady-state crack propagation (stage II: growth constant over intermediate
stress intensities

Crack propagation to final fracture (stage III: rapid increase in crack growth
rate at high stress intensities
The threshold stress intensity necessary to produce SCC is called KISCC. An
example of the cracking behavior is shown in Figure 8-9 for AISI 304L exposed
to magnesium chloride at 130°C (265°F) [8.13]. The threshold stress intensity
level, KISCC, was about 8 MPa∙m1/2 (7.3 ksi∙in1/2). Single cracks were observed in
the stage I region, and branching cracks were observed in the stage II region.
 8-14 
13828389
Figure 8-9
Stress corrosion crack velocity as a function of stress intensity factor [8.13]
8.4.3 Examples of Alloy/Environmental Systems
Stress corrosion damage can occur in many different alloys provided the
conditions for susceptibility are present. Examples of the most common instances
for power plant related alloys are shown in Table 8-2 [8.14]. In general, the most
common environment which leads to the formation of stress corrosion cracking
is that involving chloride and halide ions. These may be present in a wide variety
of water and process streams. In ferritic steels hydroxides can produce damage
and in copper alloys ammoniacal solutions induce cracking.
 8-15 
13828389
Table 8-2
Common alloy/environment systems known to exhibit stress corrosion cracking
Figure 8-10
Effect of low concentrations of arsenic, phosphorus, antimony, and silicon on the
time-to-fracture of copper by SCC [8.14]
 8-16 
13828389
Very high-purity copper is almost immune to SCC in most environments and in
the practical range of service stresses. However, intergranular cracking of copper
has been observed under some conditions, apparently as a result of segregation of
trace impurities at the grain boundaries. The resistance of copper to SCC is
greatly reduced by the presence of low concentrations of arsenic, phosphorus,
antimony, and silicon as alloying elements. Time-to-fracture for copper
containing various concentrations of these alloying elements when stressed at an
applied tensile stress of 69 MPa (10 ksi) is shown in Figure 8-10. As the
concentration of each alloying element is increased, time-to-failure at first
decreases, reaching a minimum between about 0.1 and 1%, then increases.
8.4.4 Examples of Power Plant Related Damage
The range of possible instances of stress corrosion damage in power plant
components is extensive, and problems can be encountered due to issues
associated with operation or because contamination is introduced accidentally.
A simple example of accidental SCC is shown in Figure 8-11. Transgranular
failure of a stainless steel bellows occurred as a result of rainwater ingress. The
plant involved was near the sea and the water contained high enough levels of
chlorides to induce failure, the stresses necessary were present as a result of the
cold working involved with bellows manufacture.
Figure 8-11
Failure of a stainless steel bellows by SCC (a), and detail of the microcracking
present (b)
Further information regarding SCC in boiler tubing follows based on
information provided in the EPRI BTF manual [8.8].
Austenitic Stainless Steels
When austenitic stainless steels are exposed to temperatures in the range 425 to
815°C (-800 to 1500°F) chromium will diffuse to grain boundaries to form
chromium carbides. The depletion of chromium from the matrix nearby the grain
boundary reduces the local corrosion resistance of the material. Stainless steel in
 8-17 
13828389
this condition is termed "sensitized". Sensitized stainless steel is particularly
susceptible to attack by chlorides. Since stainless steel used in SH/ RH tubing is
routinely exposed to these temperature levels, the material will develop a
susceptibility to SCC. Two approaches to limit the degree of sensitization are the
use of:

Low carbon grades

Stabilized grades of stainless steels
In the low carbon grades, the objective is to have insufficient carbon present to
form carbides. Unfortunately, as the low carbon levels can significantly reduce
creep strengths, these materials are generally not suitable for SH/RH tubing
applications. The second approach is to utilize grades, such as Type 321H and
347H that contain elements (titanium and niobium respectively) which are
stronger carbide formers than chromium; thus the material maintains its
corrosion resistance in these "stabilized" grades. However, both Types 321 H
and 347H will sensitize in SH/RH applications.
Ferritic Materials
Attack of ferritic materials by SCC is very uncommon. It was a significant
problem in power plants in the early 1900s because of caustic contamination and
buildup associated with crevices of riveted or welded joints. SCC has also been
observed in rolled tubes in drums and headers. There have been some
occurrences in T1A reheater tubes, manifested as axial cracks, originating on the
outside surface of tubes and intergranular in nature. Most typically the caustic
was introduced in the desuperheating or attemperator spray.
The attack of ferritic materials by caustic in tubing and turbine materials is one of
the most important features to consider for any units that are considering change
over to NaOH treatment, or indeed that operate with phosphate treatment in the
free hydroxide zone. It is one reason that the amount of free NaOH is usually
limited to 1 ppm in the boiler water.
 8-18 
13828389
Sources of Corrosive Environments in Boiler Tubes that Can Lead to SCC
Steamside Sources
Steamside contributors to SCC are usually chlorides or hydroxides; ferritic
materials are susceptible to NaOH, stainless steels to either NaOH or chlorides.
There are two primary sources for these contributing species:
1. Contamination from chemical cleaning, for example, due to:
-
Poor back-fill procedures that fail to protect SH circuits from the
carryover of solvents, such as HCI, during cleaning of waterwalls.
-
Problems that develop during cleaning of SH/RH circuits such as a
breakdown of inhibitors caused by excessive temperatures, or leaving
acids in circuits because of improper flushing of chemicals. Furthermore,
the inhibitors used sometimes contain sulfur, which, if it deposits, can
lead to a problem with intergranular attack.
2. Carryover of volatile chemicals from the boiler. For example:
-
Na2SO4 can be mechanically carried over in steam and will subsequently
combine with moisture from condensate to cause pitting, usually in the
RH. If there is carryover of NaOH in units' under either caustic
treatment or phosphate treatment problems may be encountered due to
operation with excessive levels of NaOH.
-
There is also emerging information that high levels of organics
(hundreds of ppm) can result in SCC especially when they are oxidized
by high dissolved oxygen levels [8.15].
Fireside Sources
Fireside corrosives are either polythionic acids, or less commonly, nitrates and
sulfates. Polythionic acids are H2SOx compounds where X is equal to 3, 4, or 5.
They form in oil-fired units from reactions between sulfur corrosion products,
SO2, moisture and air.
8.5 Assessment of Components
The following steps should be considered to assess Environmental enhanced
Embrittlement:
It should be recognised that for problems to occur an environmental factor
which is aggressive to the alloy is involved. In the case of stress corrosion
cracking the presence of tensile stress is also a requirement. Thus, the
problems may be a consequence of inadequate environmental control,
incorrect material selection (or heat treatment) or both. Since these
 8-19 
13828389
conditions are both unexpected the best approach to these issues is to ensure
sufficient control that problems do occur. However, in the event of problems
it is necessary to take appropriate steps to evaluate the cause so that effective
remedial action can be initiated. Potential actions are:
1. Compositional analysis of any deposits or scale present. This analysis
should identify if aggressive elements are present.
2. Review operating relating to the quality of water/steam.
3. Metallographic evaluation to assess microstructure and, if possible
composition. These checks should assess the specific type of alloy (for
example, differentiate between normal, H and L grades of austenitic
steel) and the levels of thermal degradation present.
4. In cases where stress corrosion is an issue, review of potential sources of
stress should be performed. A reasonable visual inspection should be able
to identify if there are problems with supports and so on.
5. An additional consideration is nondestructive inspection of susceptible
locations to map the extent of any damage.
Review of the results from these actions should permit the most effective actions
to be established.
8.6 References
8.1
D. A. Woodford and R. H. Bricknell, Environmental Embrittlement of
High Temperature Alloys by Oxygen, Embrittlement of Engineering
Alloys, C. L. Briant and S. K. Banerji, Ed., Academic Press, 1983,
p. 157.
8.2
H. G. Suzuki, “Strain Rate Dependence of Cu Embrittlement in Steels,”
ISIJ International, Vol. 37, No. 3, 1997, pp. 250-254.
8.3
C. F. Old and P. Trevena, “A Suggested Method for the Prediction of
Liquid Metal Embrittlement,” Third Int Conf on “The Mechanical
Behavior of Materials,” 1979, Pergamon Press, pp. 397–407.
8.4
M. G. Nicholas and C. F. Old, “Review Liqid Metal Embrittlement,” J
of Materials Science, Vol. 14, 1979, pp. 1–18.
8.5
W. M. Robertson, “Propagation of a Crack Filled with Liquid Metal,”
Trans. Met. Soc., AIME 236, 1966, p. 1478.
8.6
Photomicrograph supplied by W. Weiss, Structural Integrity
Associates, Inc.
8.7
Photographs supplied by T. Baker, University of Wales, Swansea.
8.8
Boiler Tube Failures: Theory and Practice. Volume 3: Steam Touched Tubes.
EPRI, Palo Alto, CA: 1996. TR-105261.
8.9
Remaining Life Assessment of Austenitic Steel Superheater and Reheater
Tubes. EPRI, Palo Alto, CA: 2002. 1004517.
8.10
H. L. Logan, The Stress Corrosion of Metals. Wiley, New York, 1966.
 8-20 
13828389
8.11
D. D. Macdonald and G. A. Cragnolino, “Corrosion of Steam Cycle
Materials,” Chapter 9 in The ASME Handbook on Water Technology
for Thermal Power Systems, American Society of Mecanical Engineers,
New York, 1989.
8.12
R. W. Staehle, A. J. Forty, and D. Van Rooyen, “Fundamental Aspects
of Stress Corrosionn Cracking,” National Association of Corrosion
Engineers, Houston, TX, 1969.
8.13
R. W. Staehle, J. Hockmann, R. D. McCright, and J. E. Slater, “Stress
Corrosion Cracking and Hydrogen Embrittlement of Iron Based
Alloys,” National Association of Corrosion Engineers, Houston, TX,
1977.
8.14
Environmentally Induced Cracking, ASM Handbook, Ninth Addition
Volume 13: Corrosion, ASM International, 1987.
8.15
Mechanisms of Environmental Cracking Peculiar to the Power Generation
Industry. EPRI, Palo Alto, CA: 1982. NP-2589.
8.16
F. P. Ford, “Stress Corrosion Cracking of Iron-Base Alloys in Aqueous
Environments,” in Embrittlement of Engineering Alloys, Vol. 25, Treatise
on Material Science and Technology, Academic Press, 1983, pp. 235–274.
8.17
O. I. Martynova and A. B. Vainman, “Some Problems of Oxygenated
Treatment use in Supercritical Units,” Thermal Engineering, Vol. 41,
No. 7, 1994, pp. 497–503.
 8-21 
13828389
13828389
Section 9: Hydrogen Embrittlement
9.1 Introduction
Various forms of hydrogen-related problems have been observed; these include:

Blistering, porosity, or cracking during processing due to the lack of
solubility during cooling of supersaturated material

Adsorption or absorption of hydrogen at the surface of metals in a hydrogenrich environment producing embrittlement or cracking

Embrittlement due to hydride formation

Embrittlement due to the interaction of hydrogen with impurities or alloying
elements
Hydrogen is known to cause problems in many metals and alloys, most notably in
steels, aluminium, nickel, and titanium alloys [9.1 to 9.4]. When embrittlement
occurs it can be particularly severe, with ductility reduced by over 90% compared
to behavior when no hydrogen is present. A simple example of the embrittlement
effect is given in Figure 9-1, which shows that bending of a steel sample is not
possible when hydrogen is present.
Figure 9-1
The normal ductility of steel (a), is severely reduced when hydrogen is present (b).
Failure occurred with the initiation of multiple microcracks (c) [9.5]
Sources of hydrogen can be extremely varied making this type of embrittlement
hard to control. For example, hydrogen can be:

Retained internally during the melting, casting and pickling of alloys

The result of inadequate control during welding, including from the
dissociation of moisture, grease or other contaminant
 9-1 
13828389

Discharged at cathodic areas in electrolytic cells, including cells set up as the
result of localised corrosion (it is apparent that, at both the anode and the
cathode, events which may lead to embrittlement occur)

Introduced to the metal during chemical milling or plating operations

Present from an external molecular gas environment
This section describes:

Mechanisms of hydrogen damage

Factors affecting hydrogen embrittlement

Damage development, with particular reference to:
-
Hydrogen cracking of welds
-
Hydrogen damage in boiler tubing
Finally an example case study and guidelines for assessment of components are
provided.
9.2 Mechanisms of Hydrogen Damage
The effects of hydrogen will vary depending on the particular alloy. Thus,
consideration of the mechanisms associated with hydrogen must relate to the
alloy system involved:

Aluminium alloy welds will introduce porosity due to the rapid change in
solid solubility on cooling.

Titanium alloys can produce hydride formation, however, these problems are
generally limited to conditions where the normally protective surface oxide is
broken. As shown in Figure 9-2, this brittleness occurs dramatically with the
increase in atomic hydrogen.
 9-2 
13828389
Figure 9-2
Effect of hydrogen on yield strength and ductility of Ti6Al4V [9.6]

Austenitic stainless steels will reduce ductility, apparently by reacting with
carbides to form methane and thus promoting the development and inter
linkage of fine grain boundary voids on carbide particles. In results on 304
stainless steel, the behavior of plane samples loaded in tension indicated [9.7]
that at 25 ppm hydrogen the reduction in tensile strength was 10% with a
decrease in ductility of 20%. At 60 ppm hydrogen the strength and ductility
values were further reduced, with measured reductions of 23% and 38%
compared to normal values. Even greater brittleness would be expected in the
presence of notches. Moreover, in the presence of hydrogen, failures were
initiated by intergranular cracks, with transitions to the normal ductile mode
only observed in regions away from the highest hydrogen levels, Figure 9-3.
The intergranular nature of the cracking is because in austenitic material the
diffusion of hydrogen is significantly faster along grain boundaries than it is
within the grains.
 9-3 
13828389
Figure 9-3
Appearance of 304 stainless steel showing the intergranular fracture induced by
hydrogen [9.7]

Alloy steels with a ferritic type microstructure will promote brittle fracture
either by reducing the cohesive strength between atoms or due to a reduction
in the surface energy of a crack. However, even in this one alloy group a wide
range of effects have been noted so that the influences of hydrogen on
behavior have been summarized below.
9.3 Factors Affecting Hydrogen Embrittlement of Ferritic Type
Steels
The severe embrittlement is a consequence of the very low solid solubility of
hydrogen in steels and the other alloys susceptible to damage. The influence of
hydrogen on fracture appearance is however, complex and variations in behavior
depend on both mechanical and metallurgical issues. The critical factors are
summarized as:

Notches or surface flaws generally enhance embrittlement, both by acting as
a stress raiser and by modifying the local stress state to increase the level of
hydrostatic stresses. In the presence of hydrogen the strain concentration
effect will be particularly important since recent information suggests that
small differences in local strain can lead to a significant increase in the level
of hydrogen available, Figure 9-4. The high sensitivity data shown also
indicate that for a given level of plastic strain an increased Mn content will
also lead to higher effective hydrogen levels. These results were consistent
with general trends indicating that Mn promoted hydrogen embrittlement.
 9-4 
13828389
Figure 9-4
Influence of local strain and Mn content on the release of hydrogen [9.4]

Loading rate, very rapid rates of loading will suppress hydrogen
embrittlement. For embrittlement to occur, hydrogen must diffuse through
the crystal lattice to maintain an appropriate concentration at the tip of the
moving crack. The temperature largely determines the rate of diffusion. If
the crack is moving too quickly, as under impact conditions, hydrogen cannot
keep pace, and the severity of embrittlement decreases. For normal rates of
loading, the maximum effect of hydrogen is unfortunately at, or near,
ambient temperatures.

Strength and hardness are critical with the susceptibility for embrittlement
increasing with increasing strength. Thus, for example, review of the
behavior of alloy steel fasteners indicates that for a given alloy failures
attributed to hydrogen embrittlement were only observed at hardness values
above 35 HRC. Similarly, a recent assessment of Cr and CrMo steels
revealed that hydrogen embrittlement was noted in samples tempered at
400°C but no embrittlement occurred when tempering was performed at
700°C. An effect of composition was also indicated in this work with the
susceptibility of embrittlement greater in the Cr steel compared to the CrMo
steel. The fracture appearance is also influenced by the strength of the
material. In general, as the strength level increases, fractures are more
intergranular.
Impurities also influence fracture mode. For example, at low impurity levels,
hydrogen-induced cracking was shown to occur by cleavage at very high
stress intensities. At high impurity levels (for example, P, Sn, Sb and As), the
fracture path was intergranular, (for example, Figure 9-5), and the stress
intensity required for crack growth decreased. In a similar study, tempering
 9-5 
13828389
between 350 and 450°C (660 and 840°F) produced entirely intergranular
fractures. Therefore, in the presence of both hydrogen and trace elements the
fracture mode will be influenced by the cooperative actions of temper
embrittlement and hydrogen embrittlement.
Figure 9-5
Intergranular fracture in high strength steel induced by hydrogen and segregation
of trace elements. When compared to Figure 9-3 the grain facets are relatively
clean with little evidence of local dimples.

Grain size appears to exhibit a secondary influence on hydrogen
embrittlement, although there will be some effect on the ease of crack
propagation.

Effects of hydrogen can be mitigated by post exposure heat treatment. As
illustrated in Figure 9-6 holding at elevated temperature will allow hydrogen
to diffuse from the material restoring ductility.

Depending on the particular alloy composition, microstructure and strength
level there appears to be a critical level of hydrogen needed to induce
embrittlement. However, the presence of hydrogen has been found to be
particularly deleterious in tempered martensitic ultra high strength steels
where values of K1c at room temperature were reduced from around
135 MPa∙m1/2 to 12 MPa∙m1/2 for concentrations of diffusible hydrogen of
less than 8 wppm.
Studies have shown that cracks propagate discontinuously, suggesting that the
crack growth rate is controlled by the diffusion of hydrogen to the triaxially
stressed region ahead of the crack tip. The fracture appearance is influenced by
the strength of the material. As the strength level increases, in general fractures
are more intergranular. In some cases, the hydrogen induced crack may propagate
by other damage mechanisms if it is not of sufficient size to lead directly to
cleavage fracture.
 9-6 
13828389
Figure 9-6
llustration of severe embrittlement caused by the presence of hydrogen and how
holding at elevated temperature will restore ductility [9.6]
9.4 Damage Development
9.4.1 Hydrogen Cracking of Welds
A major source of under bead cracking in welds deposited in carbon, low-alloy
and other hardenable steels is cold cracking (also called delayed hydrogen
cracking). Cold cracks may form within minutes, hours, or days after welding and
can result in catastrophic failures of weld structures. Factors required for cold
cracking to occur are:
1. A crack-sensitive microstructure, usually martensitic
2. Sufficient hydrogen concentration in the weld
3. Rigid tensile restraint
4. A temperature between approximately 150°C and ambient
Elimination of one or more of these factors greatly reduces crack susceptibility.
This is illustrated in data produced examining hydrogen cracking in duel phase,
ferrite-austenite stainless steel, Figure 9-7. The results obtained show a clear
boundary of susceptibility to cracking based on the level of hydrogen present and
the amount of ferrite in the weld.
 9-7 
13828389
Figure 9-7
Susceptibility to cracking in duplex stainless steel welds as a function of hydrogen
content and ferrite volume fraction [9.8]
Hydrogen can get dissolved in the molten weld metal, as the weld pool cools it
becomes supersaturated and some hydrogen diffuses into the HAZ. With rapid
cooling there is insufficient time to allow the hydrogen to diffuse away so the
hydrogen will segregate at pores, discontinuities, inclusions, and other
microscopic flaws. These flaws are effective traps and can severely reduce the
diffusion coefficient of hydrogen, see Figure 9-8.
This phenomenon is known to be diffusion-controlled, time-dependent, and
either transgranular or intergranular. Several theories explain why cold cracking is
time-dependent. Generally, a pre-existing microcrack or discontinuity acts as a
stress-concentration site. When a tensile stress is applied, hydrogen diffuses at
room temperature to the regions of greatest tensile strain. After the concentration
of hydrogen at or near the tip of the discontinuity has accumulated to a critical
value, which depends on the magnitude of externally applied tensile stresses or
residual stresses, the hydrogen is believed to cause severe reduction in the
cohesive bonding energy between iron atoms ahead of the discontinuity, and
cracking initiates. Propagation of the crack occurs in discrete bursts or steps,
which are repeated as fresh hydrogen diffuses ahead of the crack tip. At low stress
intensity values, cracking is likely to follow an intergranular path between prior
 9-8 
13828389
austenite grains that have transformed to martensite, while at high stress
intensities, the fracture could be transgranular. In C:Mn or C:Mn lightly alloyed
steels, weld metal microcracking is often angled at approximately 45° to the weld
surface and is frequently referred to as ‘chevron cracking’.
Figure 9-8
Diffusion coefficient of hydrogen in steels as a function of temperature
An example of an in-service, hydrogen induced crack is shown in Figure 9-9.
This defect was identified in carbon steel weld joining an economizer feed nozzle
to the steam drum. The defect was identified after around 12,500 hours
operation and detailed laboratory examination revealed that the weld metal
contained many hydrogen induced micro defects, which had interlinked in a
zigzag pattern by a local ductile fracture mechanism to form the macro defect. It
was also apparent that this defect extended from an unfused region at the root of
the weld.
In welding, the combination of tensile shrinkage stresses and hydrogen
contamination may cause micro cracks to occur in both the weld fusion zone and
HAZ. In fact, cold cracking occurs more commonly in the HAZ because the
 9-9 
13828389
hydrogen contamination entering the molten weld pool diffuses rapidly into the
HAZ and most steel filler metals have less carbon than the base metal for good
weldability, making the HAZ microstructure more susceptible. An example of a
hydrogen induced HAZ crack is shown in Figure 9-10.
Figure 9-9
Micrograph showing a hydrogen induced crack in a thick section carbon
manganese steel weld. The cracking appeared to initiate from the unfused region
at the root.
Figure 9-10
Micrograph showing a hydrogen crack initiated in the HAZ at the weld root,
which extends into the weld metal [9.10]
 9-10 
13828389
The cold cracking susceptibility of a given composition of steel is related to the
Dearden and O'Neill equation for carbon equivalent, CE:
CE = %C + %Mn + %Cu+%Ni + %Cr+%V+%Mo
6
15
Eq. 9-1
5
This formula was derived for plain-carbon and low-alloy steels containing 0.12%
C or more. Weight percentages are used in the calculation. For low-carbon steels
in the range from 0.07 to 0.22 % C, the Ito and Bessyo equation can be used:
CE = C + Si + Mn + Cu + Cr + Ni + Mo + V + 5B
30
20
60
15
Eq. 9-2
10
It is generally agreed that a value of CE > 0.35 to 0.40 (depending on plate
thickness and the degree of restraint) indicates that a steel will be susceptible to
cold cracking in the HAZ unless steps are taken to reduce the amount of
hydrogen contamination in the molten weld pool. This involves slow cooling,
through the correct choice of preheat and heat input, allowing gas to escape or
preferably keeping welds free from hydrogen by using low hydrogen electrodes
(plus baking to drive off moisture) and by proper cleanliness. PWHT should also
be employed to relieve local residual stresses.
A special form of hydrogen related cracking, known as Lamellar Tearing, occurs
in the base metal or HAZ of restrained welded joints, as a result of inadequate
ductility in the through-thickness direction of the steel parent plate.
Susceptibility to lamellar tearing depends on the joint geometry, oxide and
sulphide inclusion content, and the extent to which these inclusions are elongated
or flattened in the rolling direction to form parallel planes of weakness. The
reduction of area values from tensile tests on steel plate taken in the throughthickness direction indicates susceptibility to lamellar tearing. Steel plates
exhibiting reduction of area values less than 10% are likely to be sensitive to
cracking. Hydrogen is preferentially trapped at inclusions and tends to accelerate
the occurrence of lamellar tearing. Therefore, welds on steels deposited by lowhydrogen processes are more resistant to lamellar tearing than similar welds
deposited with methods other than low hydrogen welding.
Preheating and buttering can be effective in reducing the susceptibility of steel to
lamellar tearing. Preheating decreases both the magnitude of residual tensile
stresses acting in the through-thickness direction and the severity of embrittling
effects caused by hydrogen. Because lamellar tearing usually occurs within 0.1 to
0.2 in. from the weld interface in the HAZ, susceptible plates can be buttered
with crack resistant weld metal. When joining two buttered plates, the major
tensile stresses act on the relatively immune buttered regions of the joint.
However, the most economical preventative measure, when possible, is to
assemble the plates to be welded with the rolling direction of the plate
perpendicular to the weld axis.
 9-11 
13828389
9.4.2 Hydrogen Damage in Boiler Tubing
The following information is based on information from the EPRI report, Boiler
Tube Failures: Theory and Practice [9.11].
Introduction
When breakdown of normal operating conditions results in both deposits and
concentration of impurities hydrogen damage can occur in carbon and low alloy
steel boiler tubing. Under normal conditions, boiler water flow through the tube
is continuous and a magnetite layer is formed that protects the tube from
chemical attack. The manner in which magnetite scale is modified or affected by
various contaminant species will determine how the various damage types are
manifested. Hydrogen damage requires a locally acidic environment that affects
both the mechanism of magnetite growth and its rate.
Potter and Mann [9.12] found that protective two-layer magnetite scale grows
parabolically on mild steel in moderate solutions of sodium hydroxide. This has
subsequently been found for the other cycle chemistries as well. These two layers
grow essentially stress-free. The inner layer of the two layers was found to be
porous by Field, et al. [9.13]. In contrast, in the presence of acidic solutions,
Potter and Mann [9.14] found that the oxidation rate of mild steel becomes
linear (non-protective); in this mode the oxide layers are not grown in a stressfree configuration and the total oxide consists of multi-layers of magnetite. That
is, the oxide growth process is affected both mechanically and chemically for the
case of acidic contamination [9.15].
The Development of Hydrogen Damage
When boiler water flow conditions depart from design, local areas of high steam
quality, deposition and concentration of acidic (chloride) contaminants can occur
and change the normal protective oxide. A local flow disrupter causes the normal
nucleate boiling process to be disrupted and a local steam blanket or bubble is
formed. This area is intermittently dried and then rinsed, and it may be at a
slightly higher temperature than the surrounding area that is cooled by the
flowing water. This process will cause dissolved or suspended solids to begin to
be deposited just downstream of the flow disruption and on the hot side of the
tube. These deposits might consist of feed-water corrosion products such as
copper, nickel, and iron oxides. Moreover, once present these deposits can further
cause flow disturbance, lead to poor heat transfer, and eventually destroy the
protective nature of the magnetite.
The local condition in the tube is now conducive to hydrogen damage; it contains
deposits, a source of acid contamination, and a means of concentrating both. A
cross-section through the forming deposits would show an upper layer of the
 9-12 
13828389
deposited feedwater corrosion products (Fe, Cu, Ni, etc.) with the underlying,
distinctive multi-laminated growth of Fe3O4 and a layer of FeCI2 “influence.”
Unfortunately, although the oxide continues to grow according to the equation:
3 Fe + 4 H2O - Fe3O 4 + 4 H2
Eq. 9-3
the mechanism of oxide growth changes due to the presence of the concentrated
chloride. Firstly, the normal counter-flux mechanism is modified so that
although oxygen ions (O2-) continue to diffuse inward, there is no counter
diffusion of Fe2+ outward. Oxide, instead of forming at two locations and in a
stress-free state, now forms only at the interface of the boiler tube metal and
inner oxide and its formation is now under stress. At a critical stress level the
oxide layer will delaminate. Thus, there is a repetitive cycle of linear growth then
delamination.
Figure 9-11
Schematic diagram illustrating the generation of hydrogen in an electrochemical
cell
Secondly, all hydrogen generation now occurs at the tube metal surface. Figure
9-11 shows the resulting electrochemical cell with diffusion of hydrogen into the
tube steel. The hydrogen atoms react with iron carbide (Fe3C) in the pearlite
component of the steel to form methane according to:
Fe3C + 4 H = 3 Fe + CH4
Eq. 9-4
As methane is a fairly large molecule, it does not easily diffuse through the
material, pressure builds up and microfissuring begins at the grain boundaries.
 9-13 
13828389
Figure 9-12
Micrograph showing the fissuring which develops due to hydrogen attack in
carbon steel tubing
Figure 9-12 shows the distinctive microstructure that results. Usually, but not
always, the microcracking is accompanied by a general, localized decarburization
of the pearlite. Thus, CrMo steels are more resistant to damage because the
carbides are stabilized by the alloying additions. However, even in these steels
damage can develop in extreme situations of high hydrogen concentrations. The
decrease in local carbon leads to lower strength, so that microcracked material is
now susceptible to failure. In general, the region of damage increases in size with
time as the level of degradation increases, Figure 9-13.
Figure 9-13
Micrographs showing increasing levels of decarburisation and hydrogen damage,
samples etched in 50% solution of hot hydrochloric acid to reveal the damage
Masterson, et al. [9.15] emphasized that in order to give comparable corrosion
rates, sodium hydroxide must concentrate by a factor of ten to one hundred times
more strongly than acid chloride. This explains the seriousness of ingress of acid
such as from a breakdown in the water treatment or makeup system. Acidic
contamination can lead to very rapid corrosion rates (> 10 mm/year) which
contrasts with caustic contamination which shows lesser but still significant rates
of attack (up to 2 mm/year).
 9-14 
13828389
9.5 Case Studies/Examples
Cracking was detected at the root of an alloy steel weld. Detailed laboratory
investigation revealed that the crack had initiated in the HAZ at the root with
through wall propagation with the coarse grained HAZ, Figure 9-14. Hardness
measurements revealed that the weld had not been properly tempered and the
peak hardness in the HAZ was an average hardness of 45 HRC. The brittle
nature of the failure and high hardness indicated that damage was due to
hydrogen embrittlement. Prevention of the problem required that future welding
was undertaken with precautions to prevent hydrogen contamination, for
example, selecting electrodes with a low hydrogen flux, using appropriate preweld
baking and cleaning of the weld preparation, and ensuring that the weld and
HAZ were properly tempered either by carrying out a PWHT or through the
implementation of a temper bead procedure.
Figure 9-14
Hydrogen induced cracking in the HAZ of an alloy steel weld
9.6 Assessment of Components
Historically, hydrogen-embrittlement effects have been evaluated by reversedbend tests, single-bend tests, and fatigue tests. Reduction-of-area and elongation
values determined by standard tensile tests also show the effect of hydrogen
embrittlement. Impact tests are generally not a good method of detecting
hydrogen embrittlement.
 9-15 
13828389
The following steps should be considered to assess Hydrogen Embrittlement:
1. In the event of a cracking event, a sample should be removed and detailed
metallographic investigation performed.
2. If hydrogen embrittlement is associated with the cracking then the source of
this hydrogen must be identified. If the crack is found in a recently
manufactured weld then the procedure, consumables , and so on used in
fabrication should be rechecked. All similar welds produced at the same time
should be inspected by nondestructive techniques suitable for the
identification of hydrogen microcracking.
3. In both cases it is important that the correct cause of the problem is
identified and the necessary remedial action taken.
Consideration should also be given to carrying out nondestructive examination at
high risk locations to identify and map the extent of the problem.
9.7 References
9.1
I. M. Bernstein and A. W. Thompson, Effect of Metallurgical Variables
on Environmental Fracture of Steels, Int. Met. Rev., Vol. 21, Dec 1976,
pp. 269–287.
9.2
J. P. Hirth, Effects of Hydrogen on the Properties of Iron and Steel,
Metall. Trans., Vol. 11A, June 1980, pp. 861–890.
9.3
I. M. Bernstein and A. W. Thompson, Ed., Hydrogen in Metals,
American Society for Metals, 1974.
9.4
M. Nagumo, “Review ‘Function of Hydrogen in Embrittlement of High
Strength Steels,” ISIJ International, Vol. 4, No. 6, 2001, pp. 5900–598.
9.5
L. E. Probert and J. J. Rollinson, “Hydrogen embrittlement of High
Tensile Steels During Chemical and Electrochemical Processing,”
Electroplating and Metal Finishing, Vol. 14, 1961, p. 396.
9.6
“Electrodeposition,” The Materials Science of Coatings and Substrates,
1993.
9.7
M. Au, “Mechanical Behavior and fractography of 304 Stainless Steel
with high Hydrogen Concentration,” WSRC-TR-2002-00558.
9.8
A. J. Leonard, R. N. Gunn and T. G. Gooch, “Hydrogen Cracking of
Ferritic-Austenitic Stainless Steel Weld Metal,” Presented at 'Stainless
Steel World Duplex America 2000', 2000.
9.9
I. L. Mogford and A. T. Price, “Application of fracture Mechanics to
predict weld performance,” Int Conf on Welding Research Related to
Power Plant, Central Electricity Generating Board, 197 P H M Hart 2,
pp. 172–191.
 9-16 
13828389
9.10
P.H.M. Hart, “Hydrogen cracking-its causes, costs and future
occurrence,” Weld Metal Hydrogen Cracking in Pipeline Girth Welds,
Proc. 1st International Conference, Wollongong, Australia, 1–2 March
1999. Published by Welding Technology Institute of Australia (WTIA),
Silverwater, NSW, Australia, 1999.
9.11
Boiler Tube Failures: Theory and Practice, Volume 2: Water Touched Tubes.
EPRI, Palo Alto, CA: 1996.TR-105261.
9.12
E. C. Potter and G.M.Mann, Proc First Int Congress on Metall.
Corrosion, Butterworths, 1961, p. 417.
9.13
E. M. Field, R.C.Stanley, A. M. Adams and D. R. Holmes, “The
Growth, Structure and Breakdown of Magnetite Films on Mild Steel,”
Proc. Second Int Conf on Metallic Corrosion, New York, 1963, p. 829.
9.14
E. C. Potter and G. M. Mann, Proc Second Int. Conf. Metall.
Corrosion, 1963, p. 872.
9.15
H. G. Masterson, J. E. Castle,and G. W. Mann, “Waterside Corrosion
of Power Station Boiler Tubes,” Chemistry and Industry, 1969,
pp. 1261–1266.
 9-17 
13828389
13828389
Section 10: Creep Fracture
10.1 Introduction
Creep cracking and failures have occurred in a range of steels particularly in
tubing, headers and piping and the associated welds. However, creep processes
are also relevant to HP / IP rotors as well as valve bodies, casings and fasteners.
Low ductility creep failures involve the nucleation, growth and propagation of
grain boundary cracks. The susceptibility for fracture is increased by the presence
of inclusions and trace elements, and in view of the high temperatures involved
temper embrittlement and carbide embrittlement may occur with the
development of creep damage. The present section:

Provides general background regarding creep damage development

Outlines damage mechanisms

Discusses factors affecting creep damage, and

Presents examples of creep cracking
10.2 Background
Creep processes take place at temperatures above around 0.4 Tm, where Tm is the
absolute melting point, and lead to time dependent deformation and fracture
[10.1]. The elevated temperatures involved are such that diffusion can take place
and the details of these diffusional processes are critical to an appreciation of
creep behavior in a particular metal or alloy. In creep resistant engineering alloys,
such as low alloy steels, the movement of individual lattice point defects, known
as vacancies, and the atomic diffusion of alloying elements are both involved in
damage development.
 10-1 
13828389
Figure 10-1
Schematic diagram showing the typical creep strain : time behavior and
identifying the three stages of creep behavior
Except at temperatures very near the melting point, creep deformation with time
follows the trend of an initially decreasing creep rate in a primary stage, a period
of approximately steady state deformation or secondary creep followed by an
increasing creep rate leading to crack formation and fracture. This period of
increasing creep rate is typically referred to as tertiary creep, Figure 10-1.
Invariably, under the action of an applied stress, strain develops due to the
generation and movement of lattice imperfections or dislocations. During
primary creep, the rate of deformation decreases as a consequence of an increase
in the number of dislocations present. Eventually, the ability of diffusion
controlled processes to reduce the number of dislocations present will
approximately balance the increase due to further strain, and an approximately
steady deformation rate is noted. Under these conditions the rate of recovery
(that is, the rate which dislocations are removed) is approximately equal to the
rate of work hardening, with the specific details depending on the stress,
temperature and microstructure. In general, the rate of deformation, έs, is related
to stress, σ, and temperature, T, by the expression:
έs = A σn exp (− Qc/RT)
Eq. 10-1
Where A is a material dependent constant, Qc is the activation energy, R is the
gas constant and n is the stress exponent (typically greater than or equal to 4).
 10-2 
13828389
Figure 10-2
Time dependent creep failure of a pipe bend. Note that although the final very
rapid fracture event causes significant opening the damage leading to crack
initiation occurred without obvious deformation
In engineering alloys evidence of damage development is typically revealed by the
increase in the rate of creep deformation during tertiary creep. This acceleration
in rate may be due to either or both of the following effects:

Microstructural instability, as the precipitates present are modified by aging
there is reduced ability to limit dislocation movement

Nucleation and growth of grain boundary voids or cavities, which eventually
link to form microcracks
It is the nucleation and growth of grain boundary voids that result in low ductility
or burst fracture, for example, as shown for the piping component in Figure 10-2.
10.3 Mechanisms
Generally accepted mechanisms of creep deformation involve the diffusion
controlled generation and movement of dislocations. Thus, engineering alloys
with improved creep resistance primarily use alloy additions to introduce solid
solution strengthening and, more importantly, precipitate strengthening to limit
the movement of the dislocations. In steels, elements such as carbon, nitrogen,
 10-3 
13828389
chromium, molybdenum, vanadium and niobium are all used to improve creep
resistance. In most cases increasing the strength will also lead to an increase in
creep life, indeed for many metals and alloys these two factors are directly
related as:
έs × t f = constant
Eq. 10-2
Where the constant is approximately 0.2 to 0.4%. This equation has been shown
to provide a reasonable relationship in many alloys covering lives from a few
hours to many years, for example, Figure 10-3. However, factors such as
inclusions may influence the fracture behavior without modifying the strength, so
that it is important to consider creep ductility when evaluating behavior.
The general relationship between creep rate and rupture holds for different
failure regimes. Thus, at relatively high stresses, and thus short lives,
intergranular fracture is normally initiated by wedge cracking predominantly at
triple points, that is, at the intersection of 3 individual grains, Figure 10-4a.
Under these conditions significant movement of individual grains by sliding or
rotation can occur and if the local stress developed is sufficient de-cohesion of the
boundaries takes place. This type of damage contrasts to the grain boundary
cavities normally associated with longer term fracture observed at lower stress
levels, for example, Figure 10-4b. In both cases, the highest levels of damage
occur on grain boundaries approximately perpendicular to the applied stress.
Figure 10-3
Linear inverse relationship between minimum creep rate and time to rupture
 10-4 
13828389
Figure 10-4
Micrographs showing wedge type cracking typical of intergranular creep at
relatively high stress (a), and cavitation developed at relatively low stresses (b)
It has been demonstrated that creep cavities nucleate at grain boundaries as a
result of grain boundary sliding which opens voids at particles, precipitates or
other stress concentration at the boundary. Thus, greater numbers of creep
cavities have been found in alloys that contain high densities of small grain
boundary inclusions. The relatively large numbers of voids formed will lead to a
dramatic reduction in creep strain at failure. This effect is demonstrated for
CrMoV rotor steel where increases in the amount of aluminium, which resulted
in the formation of increased numbers of inclusions, reduced the reduction in
area at fracture from over 80% to less than 20%, Figure 10-5.
Figure 10-5
Effect of aluminum on reduction of area for creep tests at 1100oF on samples of
CrMoV rotor steels
 10-5 
13828389
Figure 10-6
Variation in reduction of area with creep rupture life for CrMoV rotor steel
The numbers of voids formed, and the rate that these voids grow and eventually
link up to form microcracks, will therefore control the creep fracture process. It
has been proposed that voids increase in size either directly through stress
directed diffusion or as the result of deformation controlled ‘constrained’ cavity
growth. Specifics of the cavity nucleation and growth rates will again depend on
stress, temperature and microstructure, for example, grain size, as well as the
presence of trace elements, with conditions typical of those found in service
leading to brittle intergranular fracture, Figure 10-6. Since the rate of
deformation is inversely proportional to the time to rupture, an equation similar
to equation 10-1 can be used to describe creep life. In general, significant
numbers of creep voids are present prior to the development of microcracking;
measurements from plant welds indicate typical numbers of voids in low alloy
steels before cracking develops are around 1000 per mm2. Scanning electron
micrographs showing typical intergranular creep fracture due the development of
grain boundary voids are presented in Figure 10-7.
Figure 10-7
Typical micrographs showing intergranular fracture following the development of
grain boundary creep voids.
 10-6 
13828389
10.4 Factors Affecting Creep Fracture
For most engineering alloys stress and temperature will have a significant
influence on creep life. Thus, typical values suggest that:

An increase in stress from 6 ksi to 7 ksi (~16%) will reduce life by about a
factor of 2

An increase in temperature from 1000°f to 1040°f (~4%) will reduce life by
about a factor of 4
This general behavior is illustrated in Figure 10-8, where the variation in the
creep life with applied stress in controlled laboratory tests on Type 304 austenitic
stainless steel are presented for different temperatures. This figure also identifies
detail regarding the type of fracture. It is apparent that over the range of
conditions involved the mode of failure changes from ductile transgranular
rupture at the shortest times to brittle intergranular fracture at lives approaching
those expected under service conditions. At the longest test lives the development
of creep cavities leading to the brittle fractures is aided by the formation of sigma
phase at the grain boundaries. This phase change results in greater tendency for
brittle behavior and reduced creep lives.
Figure 10-8
Variation of rupture life and failure mechanism with stress and temperature for
Type 304 austenitic stainless steel [10.2]
Changes in mode from the predominantly transgranular mode in short times to
brittle intergranular fracture in long times are typical in creep resistant alloys.
Indeed, this change in behavior can take place in a relatively narrow band of
conditions, for example, Figure 10-9. In these data for CrMoV HP rotor steel, at
lives greater than about 5000 hrs the reduction in cross sectional area at fracture
is less than 5%.In low alloy steels, exposure to high temperatures will lead to
coarsening of the precipitates present. These coarsening processes will in most
cases lead to softening of the material and an increase in deformation. Thus,
 10-7 
13828389
under conditions where significant softening occurs the rupture ductility will
again increase. For a set of data at different stresses at one suitable elevated
temperature the creep ductility will exhibit a minimum corresponding to the
conditions where grain boundary cavitation occurs and low ductility intergranular
fracture takes place. Unfortunately, for many power plant applications, this
minimum occurs in the regime of normal operation. Trace elements can also
influence the formation of creep damage. For example increased levels of
aluminium will reduce creep ductility in low alloy steels, Figure 10-5. This
reduction in ductility occurs because the distribution of small particles present aid
in the nucleation of creep cavities. A high density of manganese sulfide inclusions
has also been shown to promote cavity formation and hence reduce ductility in
low alloy steels. Similarly, it has been shown that the presence of tin will
accelerate creep cavitation and crack growth in CrMoV piping steels, see
Appendix E. Moreover, for low alloy steel components operating at high
temperature the presence of elements such as phosphorus may lead to temper
embrittlement.
Figure 10-9
Variation in reduction of area with stress and temperature for CrMoV rotor steel
[10.1]
10.5 Creep Damage in 9 to 12% Cr Martensitic Steels
10.5.1 Introduction
Creep strength enhanced ferritic (CSEF) steels typically contain 9 to 12% Cr,
Table 10-1. These steels are used in a range of boiler applications because of their
combination of properties which include; high thermal conductivity, low thermal
expansion coefficient, low susceptibility to thermal fatigue, good corrosion and
oxidation resistance, and relatively good creep resistance. These properties derive
from the microstructure, which, when properly processed, exhibits a tempered
martensitic matrix containing a substructure with a high dislocation density and a
 10-8 
13828389
fine dispersion of second phase precipitates. It is interesting to note that the
typical alloy compositions for these steels frequently do not include
recommendations for trace or ‘other’ elements even though it is well established
that these elements lead to embrittlement in low alloy steels.
Table 10-1
Typical composition and heat treatments used for martensitic boiler steels
Detailed research examining the microstructure of 9 to 12% Cr tempered
martensite ferritic steels has provided key information concerning the formation
of new phases and the coarsening of carbides during long term creep. However,
the microstructures of CSEF steels evolve during service at elevated temperatures
and pressures and creep strain can enhance the changes which take place. Indeed,
a number of microstructural degradation mechanisms have been identified which
are thought to be responsible for the loss of long term creep strength. These
include; the precipitation of new phases (for example, Laves and Z phases), the
dissolution of fine M2X and MX carbonitrides, the recovery of the dislocation
sub-structure and the development of creep voids in the microstructure.
The complex nature of the long term creep behavior is emphasized by
consideration of data compilations considering creep ductility. Published
information showing the variation of the reduction of area (R of A) at fracture
after creep testing at 600oC is shown in Figure 10-10 for Grade 91, Grade E911
and Grade 92 steels. In short term tests, the samples fail with high ductility due
to local deformation and ‘necking’. While even in long durations some high
ductility type failures are reported, it is apparent that as lives increase there is a
tendency for some tests to fail in brittle manner. This means that for all three
steels there is a very large variation in fracture behavior in tests at 600oC of
durations greater than 10,000 hours. This is clearly significant since 600oC is a
typical in-service temperature for these steels. Indeed, in the development of steel
P92 a target was set for the minimum reduction of area to be at least 40% at
10,000 hours and 600°C. Greater understanding of the reasons for the wide
variation in fracture characteristics of these martensitic steels is clearly important.
 10-9 
13828389
Figure 10-10
Relationships between reduction in area and creep life for steel grades P91, E911
and P92 tested at 600oC [10.3]
The complexities associated with establishing the factors affecting the creep
behavior of CSEF steels mean that it is very challenging to identify and
understand the influence of individual parameters.
10.5.2 Factors Affecting the Formation of Creep Cavities
There has been some apparently conflicting evidence reported regarding when
creep voids form during creep of martensitic steels. Indeed, several studies have
reported that voids can only be indentified relatively late in creep life. In contrast,
other work has demonstrated that creep voids nucleate relatively early in the
creep life. These apparently conflicting observations are considered with respect
to the creep behavior of selected CSEF steels where the results of long term
creep testing have been published.
Damage in 12% Cr (X20) Steel
The martensitic steel 12% Cr-Mo-V steel has been used in power boiler
applications for around 50 years. Interestingly, even in this steel there are
conflicting observations regarding the formation of creep voids. Thus, reviews of
performance have been published showing that remarkable few components have
been found with creep voids detectable by optical microscopy. In contrast, some
experience has shown that creep cracking in components has occurred and
replacement of components has been required. While these observations have all
been made for steels operating beyond 100,000 hours, it could be that the
variation in observations is linked to the temperature of operation. It is well
known that the accumulation of creep damage is very dependent on the exposure
temperature. However, the information considered here shows that variable
fracture behavior has been identified even in samples exposed at the same creep
temperature.
 10-10 
13828389
The creep behavior for a number of different X20 steels tested at 550oC is
compared in Figure 10-11. These results show data up to and even beyond
100,000 hours duration. While there is no major difference in creep strength for
the different casts there are significant differences in the reduction in area
measured after testing. It is apparent that some casts show reduction in ductility
for test lives of around 10,000 hours.
At the high stress levels, all of the specimens rupture in a ductile manner. These
failures are linked to voids which form and grow plastically at large inclusions.
The inclusions are hard particles within the plastically deforming matrix and at
some critical value of strain void formation occurs. In general, it has been
reported that relatively large oxide and sulfide inclusions are more effective in the
initiation of voids which grow by plastic deformation than small particles. The
generally smaller, second phase particles do not appear to play a role in the
ductile high temperature rupture of 12% Cr-Mo-V steel.
Figure 10-11
Creep strength and ductility for samples at 550oC [10.4]
 10-11 
13828389
\
Figure 10-12
Creep damage detected at different locations along the gauge length of a sample
tested at 550oC [10.4]
At the low stress level, creep cavitation appears to be the dominant damage
mechanism. This change in the rate controlling mechanism results in much lower
strains to rupture, Figure 10-11. Indeed, as shown in Figure 10-12, post test
metallographic examination of samples which failed with low R of A revealed
significant numbers of individual voids. Indeed, not only were creep voids and
micro cracks found at the fracture location but also along the specimen gauge
length even well away from the fracture surface. These observations demonstrate
that, under the testing conditions employed, void nucleation had taken place
generally throughout the whole gauge length. Final link up of damage as
expected focused at a specific location but even here the deformation observed is
mostly associated with developing the strain needed to grow cracks across the
section.
Further study of damage development in long term creep tests of tempered
martensite ferritic steel (German grade: X20) at a stress of 120MPa and a
temperature of 550oC. These test conditions resulted in a creep life of 139,971
hours. Additional tests were performed under the same loading conditions but
these tests were interrupted after 12,456, 51,072 and 81,984 hours, that is, at life
fractions of about 9%, 37% and 59%. It was reported that nucleation of cavities
was found in each of the samples examined. Thus, it appeared that void
nucleation occurs continuously during creep. Detailed study showed that it
appeared that the number of creep voids present, that is, the cavity density, was
 10-12 
13828389
proportional to the creep strain. Indeed, this observation has been reported
previously for creep tests on X20 steel samples [10.5]. Comparison of the results
of the test programmes suggests that, even though the specifics of the steels and
the tests performed were different, a similar relationship reasonably described all
the results, Figure 10-13.
Figure 10-13
Relationship between the cavity density and creep strain for tests performed on
X20 steel samples
As illustrated in Figure 10-13, a characteristic feature of creep cavitation in high
chromium ferritic steels is that cavities are not only found on prior austenite grain
boundaries (PAGBs) but also in the interior of former austenite grains. There are
additional internal surfaces such as sub- grain boundaries and high angle ferrite
boundaries, which have formed during tempering, where segregation can occur
preferentially and which will facilitate diffusion.
However, a quantitative metallographic evaluation of the cavity population
showed that cavities on PAGBs which were oriented at 90o to the maximum
stress direction play the dominant role in the rupture process.
Damage in Grade 91 Steel
There has been a very wide range in the reported reduction in area of creep tests
performed on Grade 91 base metal that is normalized at tempered steel. This
wide range is illustrated with reference to Figure 10-10. As shown, the measured
R of A begins to fall even for test lives of the order of 5,000 hours at 600oC. In
contrast, some tests of close to 100,000 hours are shown with a reported R of A
above 70%. While this range may be in part explained by differences in test
temperatures and specimen dimensions, further evaluation of specific test results
is of value to review potential trends in behavior. It is however, apparent that
under condition of low R of A fracture occurs as a result of the nucleation and
growth of creep voids.
 10-13 
13828389
Previous work has shown that there is a relationship between N:Al ratio and
creep strength in tempered martensitic Grade 91 base metal [10.6]. This trend
has been explained on the basis that in Grade 91 steel the levels of nitrogen are
set so that strengthening will involve the formation of MX type carbides, nitrides
and carbonitrides. When excess aluminium is present in the steel, this combines
with nitrogen to form aluminium nitrides [10.6]. These aluminium nitrides are
relatively large and so do not contribute to creep strength. Thus, since the level of
free nitrogen is reduced by the formation of aluminium nitrides, the volume
fraction of MX precipitates is decreased and the creep strength is decreased.
Previous work has shown that there can be a link to the formation of aluminium
nitrides and increased susceptibility for the nucleation of creep voids. This
observation is consistent with the behavior reported for other boiler steels where
steels with higher inclusion levels exhibited lower creep ductilities than ‘clean’
steels of the same composition [10.7].
Figure 10-14
Micrograph showing creep voids developed in Grade 91 steel (a), an elemental
map of the same area showing local concentrations of oxygen(b)and an elemental
map of the same area showing local concentrations of silicon (c)
Examination of Grade 91 steel after creep testing has been performed to measure
the number density of creep voids and to evaluate factors involved in void
nucleation. It is apparent that the creep behavior of tempered martensitic steels is
influenced by nonmetallic inclusions. In cases where these inclusions exceed a
critical size there is established evidence that voids are nucleated on the ‘hard’
particles [10.8]. Photomicrographs in Figure 10-14 illustrate this behavior. A
back scattered electron micrograph of the creep voids present is shown in Figure
10-14a, with Figures 10-14b and c showing elemental maps for oxygen and
silicon. It is apparent at least some of the creep voids are associated with particles
which have relatively high concentrations of silicon and oxygen.
When considering nucleation of creep cavities the local microstructure and
composition are clearly of particular interest. Elements which have been shown
to decrease the resistance to creep fracture in engineering steels include
phosphorus (P), sulfur (S), Copper (Cu), Tin (Sn), Antimony (Sb), Arsenic (As).
A systematic study evaluating the influence of these elements on the strength and
fracture behavior has been published [10.9]. The reduction in the rupture life
resulting from the higher amount of tramp elements at 650oC does not appear to
 10-14 
13828389
be caused by the increase in the creep rate, but rather by the decline in creep
ductility. This is consistent with the Sb and Sn-doped samples exhibiting
minimal R of A. Consequently, the rupture life associated with the Sb- and Sndoped samples is shorter than that of the P- and S-doped samples, Figure 10-15.
The rupture life associated with the Cu-doped sample was seen to be
independent of the content of Cu. This is because the Cu­doped samples
exhibited poor ductility irrespective of Cu content.
Figure 10-15
Relationships between reduction of area and creep rupture life for Grade 91 steel
samples with different levels of ‘trace elements’ [10.9]. Some of the trace elements
are not normally controlled in applicable component specifications even though
elements such as tin (Sn), antimony (Sb) and copper (Cu) can significantly reduce
the creep ductility.
The results from Grade 91 provide insight into why the observed distribution of
cavities in tempered martensitic base metal is very variable for different casts. It is
apparent that the size and distribution of non-metallic inclusions and the
concentration of trace elements both significantly influence the nucleation of
creep voids. Since these factors will not simply influence the behavior at grain
boundaries, voids nucleate in different positions. In the steels with lowest creep
ductility, void nucleation starts early in life and continues with increasing strain.
Damage in Grade 92 Steel
It is apparent from consideration of Figure 10-10, that variability has also been
noted in the fracture behavior of Grade 92 base metal samples. An alternative
approach to assessment of creep fracture behavior to the typical variation of R of
A with rupture life is shown in Figure 10-16 [10.8]. Here the rupture life
reported for different test temperatures is shown with selected ranges in R of A
 10-15 
13828389
designated by different symbols. The general key to this figure is that samples
with an R of A greater than 50% are shown as an open circle, with tests with an
R of A below 50% are shown as a solid square. The difference between open and
solid symbols facilitates comparison of the results. Greater definition of the
specific ductility ranges are provided through the use of different colors. The tests
included in Figure 10-16 cover data from many different sources yet a general
trend in the rupture behavior is apparent. Thus, for tests at 650oC and durations
near to or above 10,000 hours the ductility is below 50%. In contrast, for tests at
550oC even with durations approaching 100, 000 hours the reported ductilities
are above 50%. Interestingly the results reported for tests at 600oC, that is, near
to the design temperature for many Grade 92 steel components, tests with
durations above 10,000 hours show a mixed behavior. Thus, some steel casts
show relatively low ductility at lives around 10,000 hours yet others show R of A
above 50% even at creep rupture lives very close to 100,000 hours. Review of
background information provided with the creep test data suggests that different
fracture behavior was related to the level of silicon in the steels.
Figure 10-16
Variation in reduction of area for different test temperatures and creep rupture lives
for Grade 92 steel base metal samples [10.8]
Clearly, the fact that very low creep ductilities have been reported in Grade 92
base metal samples in tests performed near typical operating conditions requires
further study. Indeed, the fact that differences in fracture characteristics have
been found for tests in different temperature regimes and for different
compositions suggests that both fabrication and creep testing factors are
important. In view of the metallurgical complexities of advanced tempered
martensitic steels, careful planning, selection of samples for examination followed
 10-16 
13828389
by the application of advanced microscopy are required to establish trends in
behavior. Details from these characterization activities have been presented.
However, selected information is provided here to compliment the present review
of fracture behavior in 9 to 12% Cr steels.
A typical micrograph showing the development of cavities after creep testing of
Grade 92 base metal is shown in Figure 10-17a. Samples were selected for
examination after testing at 550oC, 600oC and 650oC. As indicated in Figure 1016, all the tests at 650oC with duration of above about 10,000 hours showed an R
of A below 50%. Detailed characterization of samples tested to failure at 9,037,
10,682 and 19,124 hours at 650oC showed that a uniformly high number of creep
voids were present along the gauge length, Figure 10-17b. This evidence on
Grade 92 steel supports the earlier results showing a high degree of uniformity in
void density for long term tests on X20 steel, Figure 10-13. For the Grade 92
tests, the size of cavities at fracture was generally in the range from 2.1 to 3 µm.
However, detailed sizing was difficult because the voids were not spherical.
(a)
(b)
Figure 10-17
Typical micrograph showing creep voids in a Grade 92 steel base metal sample
(a) and the number density of voids present along the gauge length for samples
tested to failure at 9,037, 10,682 and 19,124 hours at 650oC (b) [10.10]
Because of the complex shapes of the creep cavities even careful metallographic
preparation and evaluation has difficulties to unambiguously characterize voids
and their relationship to microstructural details. A sophisticated approach
involving serial ion beam sectioning followed by documentation was therefore
applied. This approach allowed both the void shape and the associated particles
to be reconstructed in three dimensions. An example of a reconstruction is shown
in Figure 10-18 [10.10]. In the reconstruction shown, the cavity has a diameter
of around 2 µm. This void was clearly associated with a boron nitride particle of
around 1-1.5 µm. Much smaller second phase particles including Laves phase
were found to decorate the inside of many of the cavities. In some cases, fine
manganese sulfide (MnS) or alumina (Al2O3) particles were found within the
cavities. It is potentially the case that the fine MnS or Al2O3 formed in the steel
at very high temperatures act as sites which promote the nucleation of boron
nitride during subsequent cooling.
 10-17 
13828389
Details of the characterization process and selected results are presented
elsewhere. Further research examining additional Grade 92 base metal samples is
in progress. It is anticipated that a broader pattern of results for steels with
different composition and after a variation of heat treatments will become
available in the future.
Figure 10-18
An example of a single SEM cross-section slice taken in sample 600-A 6 mm away
from fracture surface (a) [10.10]. A reconstruction of the data showing the
individual creep voids (shown in blue, purple and green) and associated particle
(shown in red) in 3D.
Discussion
For typically processed tempered martensitic steel base metal it has been observed
that the long term performance and creep rupture strength is below that
originally expected from simple extrapolation of short term creep data. This
effect has resulted in reductions in some of the values quoted as representing long
term creep life. The reasons for the loss of long term creep rupture strength have
been investigated extensively for a number of 9 to 12 Cr steels. The following
microstructure degradation effects appear to be primarily responsible for the loss
of creep strength:

The formation of new phases which lead to dissolution of fine M2X and MX
carbonitrides

Recovery of the dislocation substructure (increase in subgrain size) and
reduction in the overall dislocation density. This is believed to initiate as the
result of preferential recovery of microstructure in the vicinity of PAGBs

The development of creep voids resulting in a significant loss of creep
ductility
The precipitation of Z phase, M6X carbonitrides and Laves phase during creep
can cause a loss of creep strength at long times. The loss of strength occurs if the
formation of these phases is sufficient to result in a significant reduction in the
fine M2X and MX and or M23C6 precipitates. The size (which determines the
 10-18 
13828389
climb distance required to overcome the particle) and the number density (which
determines the mean particle spacing and thus the particle back stress) are critical
to stabilizing the dislocation substructure and hence, play a major role in
determination of the creep strength. It has been the primary focus of this section
to consider the factors affecting the nucleation and growth of creep voids in
CSEF steels. This focus was established in part because high densities of small
voids can have a particularly significant influence on approaches to manage the
safe life of boiler components and in part because while the creep dependent
microstructural changes in CSEF steels have been widely studied there is less
work reported on creep void development. This is particularly true for the
tempered martensitic microstructures present in base metal.
It is now clearly established that for creep conditions at, or close to, those of
components in power boilers, creep voids can be nucleated early in life. That
there have been different opinions published regarding void nucleation. In
particular, several papers have suggested that creep voids are only formed late in
life. This diversity of findings is due to a number of factors, including the
following:

Selection of test conditions which are not relevant to long term behavior,

Testing steels with a composition and microstructure which are not
susceptible to formation of voids,

Post test examination limited to the surface of the samples (when damage in
CSEF steels is greatest below the surface), and

Using methods of sample preparation and evaluation which do not properly
reveal the creep voids present
In long term creep tests on CSEF steels, it is now established that in most cases
voids initiate relatively early in creep life. These voids growth throughout the
creep life and will be around 1 to 2μm in size at or very close to fracture. This
size of void is important because it is only relatively close to fracture that
individual voids can be relatively identified using optical microscopy. Based on
published information [10.4, 10.5] it appears that long term creep of X20 steel
can results in higher densities of voids than detected in other tempered
martensitic steels. It is generally agreed that the number density of voids increases
with creep strain with the number of voids at or very close to fracture in the range
2000 to 10,000 mm-2.
One key microstructural factor related to the number of creep voids nucleated
appears to be the distribution of non-metallic inclusions above a critical size.
Inclusions may be directly linked to void nucleation or the presence of one type
of inclusions may promote formation of other, even larger particles. It appears
that this seeding of inclusions can be illustrated with reference to the formation
of BN in P92 and P122 steels. Detailed study indicates that the BN inclusions in
P122 are different from those of the P92 steel. In P122 steel it appears that the
BN agglomerate in large colonies, which grow up to about 20 μm in size [10.11].
These colonies consist of many individual inclusions of about 2 or 3 μm in size.
In P92 steel [10.10], coarse size BN type inclusions grown up to 4 μm are
 10-19 
13828389
observed. In both of these commercial heat resistant steels, it appears that
alumina type inclusions, which may be sourced to the furnace refractory in
melting process, are key to the formation of large BN type inclusions.
Figure 10-19
The influence of temperature on dissolution of BN inclusions [10.11]
Under slow cooling following solidification, it appears that the BN inclusions
develop on the alumina or magnesia particles which are formed during
deoxidation or originate from the refractory of the steel making furnace. Once
formed these BN inclusions grow rapidly to over 1μm in size. Microstructural
assessment indicates that subsequent heat treatment at temperature up to about
1150oC, does not dissolve the coarse size BN type inclusions [10.11], Figure 1019. However, raising the heat treatment temperature to 1200 oC, results in the
coarse size BN inclusions dissolving with time. It has been reported that all
coarse size BN inclusions completely disappeared after a short holding time at
1250 oC, Figure 10-19.
The relationship between boron and nitrogen concentration associated with the
formation of BN inclusions in high Cr ferritic heat resistant steels is shown in
Figure 10-20. Chemical analyses of twenty-three steels, including P122 and P92,
with different concentrations of boron and nitrogen were reported. In each steel,
SEM examination was performed to establish, or otherwise, the existence of BN
inclusions. Except for the commercial P92 and P122 steels, manufacture of each
cast involved melting of 50 to 150 kg of steel, hot working at 1200 to 1000oC,
normalizing at 1100oC and holding 0.5 to 1h, tempering at 770 to 800oC and
holding 1 to 4h. In Figure 10-20, solid circles represent casts where coarse size
BN inclusions, that is, over the size of 1μm, were observed. The triangular
 10-20 
13828389
symbols show small BN inclusions, that is, under 0.5μm, and open circles
represent no BN. In this experimental concentration range, BN type inclusions
could not be found by SEM observation in the concentration range less than
0.001%B or 0.015%N.
Figure 10-20
Presence of BN inclusions in 9 to 12%Cr steels as a function of the concentration
of boron and nitrogen [10.11]
The fact that in the majority of steels with boron and nitrogen large sized BN
inclusions are present means that in each case about 80% of added boron forms
BN inclusions. After normalizing at 1100oC and tempering at 800oC only 20% of
added boron remains dissolved in the metal matrix, Figure 10-21. It is the
available (that is, not in BN inclusions) boron which is thought to have the
beneficial effects on creep strength. These benefits are achieved through
improvements in the stability of precipitates and positive influences at grain
boundaries.
 10-21 
13828389
Figure 10-21
Relationship established between total boron and boron available for improving
creep performance (as indicated by the amount of soluble boron) for 9% Cr steels
[10.11]
10.6 Case Studies/Examples
10.6.1 Creep of Thick Section Weldments
Operating experience suggests that damage developed in service is frequently
associated with weldments. This susceptibility is a consequence of the variations
in microstructure, and hence properties, difficulties associated with welding
residual stresses and defects, as well as the fact that welds are frequently located
in areas involving stress concentrations. Lifetime is primarily based on the time
taken for voids to develop and form a local macro crack, however in thick section
components some time will be required for stable creep crack growth. A number
of computer based packages have been developed to estimate creep crack growth
behavior and the final critical defect size for example, the EPRI-developed
BLESS code.
Problems in thick-section welds have been identified as a result of creep, fatigue
and creep/fatigue with the different damage types and locations frequently
described using an internationally accepted classification system. Indeed, with the
increase in operating hours it is clear that increased levels of time dependent
damage will be developed, and thus an increase in the number of thick section
welds needing repair.
Girth Welds
Recent plant experience suggests that in-service damage is typically of a
circumferential character, known either as Type IIIA or Type IV. It generally
appears that Type IIIA damage develops at, or very near to, the fusion line in
welds manufactured between steels where there is a difference in carbon activity
 10-22 
13828389
[10.12]. During operation at elevated temperature and pressure there will be
diffusion of carbon to the steel with the higher alloy content, and local reduction
in carbon level from the steel with the lower alloy content. Since creep strength is
related to carbon level, a weak zone develops with time, normally near to the
weld fusion line. High levels of local deformation lead to cavitation and cracking.
An example of Type IIIA damage is shown in Figure 10-22. While this form of
damage has been predominantly observed to date in ½Cr½Mo¼V steel piping
welds, which were fabricated using 2CrMo electrodes, the potential for this form
of damage exists in other alloy combinations for example,, where new Grade 91
alloy sections have been welded to Grade P22 material.
Figure 10-22
An example of Type IIIa cracking developed in thick section piping welds(a), with
detail showing subsurface crack initiation,(b) [10.12]
Type IV cracking has been identified in a wide range of steels from low alloy to
high alloy [10.13]. It is now generally accepted that cracking develops as a result
of highly localized strain accumulation, which leads to creep cavitation [10.14].
Thus, the specific damage rate is related to the difference in local creep strength,
the susceptibility for cavity initiation and the geometric constraint factor. In
general, the thermal cycles associated with welding result in a region of the heat
affected zone (HAZ) that is partly retransformed or highly tempered. Specific
welding conditions as well as details of the alloy composition and prior thermomechanical treatment will influence the size of the zone present and the creep
strength. However, in-service damage normally occurs towards the edge of HAZ
near the parent, Figure 10-23 [10.14]. The presence of fine inclusions etc. will
influence the tendency for nucleation of damage in low alloy steels [10.7] as well
as in CSEF steels [10.8]. As discussed in section 10.5, these inclusions promote
damage by providing sites for cavity nucleation. Typically, Type IV cracking can
lead to a significant reduction in weldment life, with the normally accepted value
for a weld efficiency factor given as about 50% of the base metal.
 10-23 
13828389
Figure 10-23
An example of Type IV cracking developed in a thick section piping weld (a)
[10.15] with detail showing sub surface creep cavitation and crack initiation (b)
Seam Welds
Failures in hot reheat lines may occur where the fracture faces remain closed
leading to steam leaks, Figure 10-24a, or, fracture may take place rapidly with
significant crack opening, Figure 10-24b.
Figure 10-24
An example of a seam welded component that leaked [10.16] (a), and an
example of a seam welded hot reheat pipe that ruptured in service [10.17] (b)
While failures of thick section components are always serious in view of lost
generation, the potential for catastrophic rupture has additional concerns
regarding safety of personnel. In view of the importance of understanding
damage development in these components the historical information from plant
has been collated and reviewed [for example,10.17, 10.18, 10.19].
 10-24 
13828389
The factors affecting the high temperature behavior of seam-welded components
are:

Weld geometry. Different preparations can be used to fabricate seam welds.
In general, thicker walled main steam components are welded using a ‘U’
groove Figure 10-25; with welds in hot reheat piping manufactured using a
double vee preparation, Figure 10-26. The geometries for seam welded
elbows and fittings similarly vary depending on the wall thickness and the
preference of the manufacturer. Additional geometric complexities can then
be introduced depending on the specific welding process used, the resultant
bead size and shape, and any over welding involved in production of the
capping passes.

Inclusions. Because seam welded piping is typically fabricated from rolled
plate the extent of inclusions present in the parent metal can be variable. In
cases where significant parent plate inclusions are present adjacent to the
weld these can act as preferred sites for cavity nucleation. Similarly, the weld
process and procedure as well as the type of flux used will influence the size,
type and density of inclusions in the weld metal. These again can promote
cavity nucleation and, in extreme cases, facilitate crack propagation.
Figure 10-25
A ‘U’ groove seam weld with detail of subsurface creep damage [10.19]. This
damage has developed in the intercritical region of the HAZ which is the location
where Type IV cracking occurs in girth welds (see Figure 10-23)
 10-25 
13828389

Creep Strength. Typically, the aim of design is to produce a weld with
properties that match those of the parent. This is always a major challenge
since local differences in composition and thermal treatment have a
significant influence of creep strength and ductility. A particular concern
with the performance of seam welds produced in a double vee preparation
occurs when the weld metal has lower creep strength than the parent. In this
case the stress can become concentrated in the cusp region and this location
becomes the preferred site for crack initiation.

Microstructural variations. These may arise from changes in composition
within the weld or from the thermal cycles modifying the structure of the
parent metal or weld beads.
Figure 10-26
Double vee seam weld in hot reheat piping showing creep microdamage at the
cusp
Figure 10-27
Double vee seam welds in hot reheat piping showing a subcritical post weld heat
 10-26 
13828389

Post Weld Heat Treatment. Tempering of thick section welds in low alloy
steels is required to minimize residual stresses and to improve the ductility of
the constituent structures. This is normally achieved by a subcritical heat
treatment at around 700oC. However, in an attempt to mitigate the
variations introduced by welding some components are given a full
renormalizing and tempering treatment, Figure 10-27.
10.6.2 Tubing
Low ductility creep failures are typically encountered in superheater or reheater
tubing and have been noted in both ferritic steel, austenitic stainless steels and
the dissimilar welds used to join these sections together. As indicated earlier, the
creep process is critically dependent on stress, temperature and microstructure. In
the majority of cases failures occur where specific in-service conditions result in
acceleration of damage. Full details regarding failure of tubing is available in the
EPRI Boiler Tube Failure manual [10.20]; thus only a summary of key issues is
presented here.
Superheater/Reheater
In ferritic alloys problems are normally associated with long term overheating.
Since damage initiates through the development of grain boundary voids
fractures are typically “thick lipped” with significant scale. Precursors to failure
may include wastage flats on the outside surface, at the 10 o’clock and 2 o’clock
positions, and the development of a local thickening of the steam side scale. Both
of these factors can lead to an increase in tube metal temperature, and loss of wall
thickness will result in local increases in stress. These effects thus lead to local
increase in the rate of creep damage and accelerated fracture. An example of a
brittle creep fracture in a superheater tube is shown in Figure 10-28.
 10-27 
13828389
Figure 10-28
Creep failure of a low alloy steel superheater tube. Note that the cracking
occurred at a location where wastage flats had accelerated the formation of grain
boundary creep voids.
In austenitic stainless steels damage also initiates as grain boundary cavities.
However, in this case, the initiation process is accelerated by the formation of
sigma phase and chromium carbides on grain boundaries. This illustrated in
Figure 10-29, which shows cavities and microcracks formed on sigma phase.
This sample was prepared using specialist metallographic techniques involving
etching first with Vilella’s reagent followed by an electrolytic etch using KOH.
This secondary etch reveals and identifies the grain boundary phases [10.2].
 10-28 
13828389
Figure 10-29
Creep cavities developed in association with sigma phase in austenitic stainless
steel. The cavities were revealed using repeat polishing and etching as described
in Section 3 of this report.
10.6.3 Dissimilar Metal Welds
Traditionally the DMWs of concern are those that join the ferritic materials to
the austenitic stainless steel. Either fusion or induction welding processes are
normally used. Filler metals are either nickel-based or iron-based austenitic
stainless steels. Welds made by an induction process have properties that are
similar to those for fusion welding with austenitic filler metals; thus the
comments made pertaining to austenitic filler metals will also apply to induction
welds. Differences in thermal expansion and creep behavior of the joined
materials, and local metallurgical changes at the low-alloy steel to weld metal
interface make the DMW more susceptible to failure than like- material welds.
The degradation of DMWs after long- term service includes a number of
observable features, including:
1. Oxidation of the ferritic steel, including oxide notching
2. Softening of the ferritic steel HAZ
3. Migration of carbon from the HAZ into the weld metal, precipitation and
growth of carbides at the weld interface and the HAZ prior-austenite grain
boundaries
4. The formation and growth of creep voids. These processes are strongly
influenced by stress, temperature and time. The times to failure for field
DMWs are strongly influenced by service conditions
 10-29 
13828389
Factors 3 and 4 are of primary concern with respect to low ductility failures.
Typical examples of creep failures in DMWs are shown in Figure 10-30. While
brittle creep failure of these welds is macroscopically similar there are important
detailed differences.
Figure 10-30
General appearance of brittle creep failures in DMWs. Fracture occurs at or very
near to the fusion line with limited deformation so that the profile of the weld
beads can be seen.
In welds manufactured with austenitic steel filler there is a significant difference
in the coefficient of thermal expansion between the weld metal and the low alloy
steel. This difference introduces thermal stresses, which are additional to stresses
from the internal pressure and any systems loading, and creep cavitation and
cracking develops primarily on prior austenite grain boundaries in the HAZ very
near to the fusion line, Figure 10-31a. In the welds fabricated with the nickel
based filler carbides developed at the fusion line provide the preferred sites for
nucleation of cavities, Figure 10-31b.
Figure 10-31
Creep cavities developed in DMWs in the HAZ of austenitic welds (a), and at the
fusion line in nickel based welds (b)
 10-30 
13828389
10.7 References
10.1
R. Viswanathan, “Damage Mechanisms and Life Assessment of High
Temperature Components,” ASM International, 1989.
10.2
Remaining Life Assessment of Austenitic Stainless Steel Superheater and
Reheater Tubes. EPRI, Palo Alto, CA: 2002. 1004517.
10.3
L. Cipolla, A. Di Gianfrancesco, D. Venditti, G. Cumino, and S.
Caminada, “Microstructural Evolution During Long Term Creep Tests
Of 9%Cr Steel Grades.” Proceedings of CREEP8:Eighth International
Conf on Creep and Fatigue at Elevated Temperatures, San Antonio, Texas;
ASME Paper CREEP 2007–26030.
10.4
W. Bendick, B. Hahn, and W. Schendler, “Development of Creep
Damage in Steel Grades X10CrMoVNb9-1 (P/T91) and
X20CrMoV12-1,” Third Int Conf on Advances in Material Technology for
Fossil Power Plants, Institute of Materials, 2002, pp. 33–67
10.5
G. Eggeler, J. C. Earthman, N. Nilsvang and B. Ilschner, “A
Microstructural Study of Creep Rupture in a 12% Cr Steel,” Acta
Metallurgica, 37, 49–60, 1989.
10.6
S.J. Brett, J.S. Bates, and R.C. Thomson, “Aluminium Nitride
Precipitation in Low Strength Grade 91 Power Plant Steels,” Proceedings
of the Fourth International Conference on Advances in Materials Technology
for Fossil Power Plants, October 25 to 28, Hilton Head Island, South
Carolina, EPRI 2005, pp. 1183–1197.
10.7
J. D. Parker and A. W. J. Parsons, “High Temperature Deformation and
Fracture Processes in 2CrMo-0.5CrMoV Weldments” International
Journal of Pressure Vessels and Piping, 1995, 63, pp. 45–54.
10.8
J.D.Parker, “Creep Cavitation in CSEF steels,” International Conf on
Advances in Material Technology for Fossil Power Plants, 2013.
10.9
F. Masuyama and T. Toyoda, “Creep Strength and mechanical
Properties of air melt modified 9Cr-1Mo cast steel,” Int. Symposium on
Improved Technology for Fossil Power Plants, New and Retrofit Applications,
Washington, D.C., March 1–3, 1993.
10.10
Y. Gu, G. D. West, R. C. Thomson and J. D. Parker, “Investigation of
Creep Damage and Cavitation Mechanisms in P92 Steels,” International
Conf on Advances in Material Technology for Fossil Power Plants, 2013.
10.11
K. Sakuraya, H. Okada, and F. Abe, “Coarse Size BN Type Inclusions
Formed in Boron Bearing High Cr Ferritic Heat Resistant Steel,” Tetsu
to Hagane, Vol. 90, pp. 819–826, 2004.
10.12
S. J. Brett, “Cracking experience in steam pipework welds in National
Power,” VGB Conf Materials and Welding Technology in Power Plants,
Essen, (1994).
10.13
Review of Type IV Cracking in Pipe Welds. EPRI, Palo Alto, CA: 1995.
TR-108971.
 10-31 
13828389
10.14
J. D. Parker, “Factors Affecting Type IV cracking,” Proceedings of the
International Conference on Integrity of High-Temperature Welds, I
Mech E, UK (1998) pp. 143–152.
10.15
H. J. Westwood, M. A. Clark and D. Sidey, “Creep failure and damage
in main steam line weldments,” Fourth Int Conf on Creep and Fracture
of Engineering Materials and Structures, (Eds B.Wilshire and
R.W.Evans), Institute of Metals, (1990), pp. 621–634.
10.16
L. H. Walters and P. K. Evans, “Examination of the E.C.Gaston Unit 4
Hot Reheat Piping Seam Weld Failure,” Southern Company services,
(2002) pp. 1–20.
10.17
C. H. Wells and R. Viswanathan, “Life assessment of high energy
piping”, ASME Pressure Vessel and Piping–Decade of Progress, ASME
New York (1993) pp. 181–215.
10.18
R. Viswanathan and J. R. Foulds,” Service Experience, Structural
Integrity, Severe Accidents and Erosion in Nuclear and Fossil Plants”,
ASME PVP- Vol. 303 (1995) pp. 187–205.
10.19
C. D. Lundin, and G. Zhou, “the effect of submerged arc welding and
PWHT on creep damage occurrence in long seam welded Cr-Mo high
energy piping”, Conf Advanced Heat Resistant Steels for Power
Generation (Eds R. Viswanathan and J. Nutting) Inst. of Mats (1998)
pp. 668–680.
10.20
Boiler and Heat Recovery Steam Generator Tube Failures: Theory and
Practice. EPRI, Palo Alto, CA: 2011. 1023063.
 10-32 
13828389
Section 11: Summary of Component
Assessment Issues
Plant operators seek to adopt approaches which:

Minimize costs of maintenance, inspection and repair/refurbishment

Prevent forced outages

Maximize safety and reliability
The operation of power generation facilities must therefore be justified on the
basis of sound cost effective engineering analysis. However, it is neither practical
nor sensible to attempt detailed analysis of every component or sub-component
within a plant. In a balanced approach, the level of analysis performed will vary
for different plant components.
The selection of specific assessment methods should be based in part on the
consequences of local component failure. Established methodologies for
maintenance planning have advocated that components are identified as either
“critical” or “influence”. A “critical” component is defined as one in which failure
would have significant safety and/or financial consequence. Typically such
components are high-energy pressure vessels and piping or turbine generator
rotors. In these cases, initial failure would be life threatening, the cost of
replacement components is high and the long lead times for replacement would
result in high costs for associated loss of output.
Conversely, an “influence” component is usually considered, as one where failure
would not normally be life threatening and a single failure does not have a major
impact on cost. An example of an influence component would be tubing in a
boiler or heat exchanger. Here it is very unlikely that a single failure would lead
to personal injury or major repair costs. However, if failures occur over a period
of time with an increasing rate, eventually the costs associated with repeated
failures will be unacceptably high and remedial action must be taken. In general,
the level of failures considered unacceptable will depend on the economics of a
particular plant. Thus, for a particularly high efficiency unit more than one or
two tube failures per year may be considered excessive whereas for low merit
plant when there is sufficient spare capacity, higher rates of failure may be
tolerated.
 11-1 
13828389
In general, this approach provides for specific actions based on the risk of failure.
Where risk can be defined as the product of failure probability, and failure
severity (that is, risk estimates can be obtained from the likely impact of a given
concern and its likelihood of occurrence):
Risk = impact × likelihood.
Eq. 11-1
The simplest approach to assess risk is to extrapolate from an experience base.
This approach is frequently adopted for tubing failures since there is usually a
statistically significant sample size available and the risk of underestimating
performance and having one or two failures is usually not catastrophic.
For situations where significant records are available documenting the condition
of critical components, it is possible, and in many cases beneficial, to extend the
risk based approach to make failure projections based on historical information
regarding components of the same design. However, where historical data are
limited, it is generally the case that the most effective methodology is based
initially on a relatively simple yet conservative assessment of the plant as a whole.
Firstly, the general condition of all key items should be established, with
inspections and maintenance work focused in the plant areas most in need. A
logical method of condition assessment uses a phased approach with a number of
decision-making levels [11.1, 11.2]. At each level the estimated remaining life is
compared to the desired operating life. If the estimated life is too short, the next
phase of the procedure is conducted. In this way, progressively more rigorous
evaluation procedures are only performed if the desired remaining life is not
shown from the lower level. As the assessment level increases more accurate data
are required and a more accurate estimate of remaining life can be calculated.
However, the more rigorous the assessment, the greater the cost and time
required. Using a phased approach means that the most appropriate inspection
methods are applied to high risk locations at the correct time.
In general, then, assessment of performance is based upon:

Knowledge of current damage level

Predictions of the rate of future damage accumulation

Acceptance of a suitable failure criterion
An extreme failure criterion is one in which the plant will no longer operate
either due to excessive dimensional changes or because of fracture. In situations
where significant deformation is a concern dimensional checks should provide
appropriate warning of the need for action. However, when assessing the risk of
fracture a number of critical issues should be considered, these include:

The damage mechanism and the extent of micro damage

The size, shape and location of any macro defects
 11-2 
13828389

The alloy composition, microstructure and properties

The geometry of the component, including any stress concentrations

Operating conditions both for normal performance as well as
load/temperature information associated with transients
Because of the uncertainties associated with this type of assessment it is generally
the case that following one inspection, a period of further operation is defined
before reinspection is required. This time between reinspections should be
selected on the basis that the risk of failure is acceptably small. In general, the
higher the resolution and accuracy of the inspection methods applied, the lower
will be the risk of failure and the greater will be the time between inspections.
The key aspects involved in the assessment of the risk of fracture are illustrated in
Figure 11-1. In this schematic diagram a significant period of time is shown for
crack initiation. This duration will depend on the factors listed above with
damage mechanisms including creep cavitation, fatigue, corrosion, etc. Following
initiation of a macro defect there will normally be a period of stable crack growth.
Again, the specifics associated with the growth period will depend on a range of
factors, but in all cases stable growth will continue until a critical defect size is
reached.
Figure 11-1
Schematic illustration of crack initiation and growth showing how the critical crack
size is significantly reduced by embrittlement. Line A shows growth behavior for
normal conditions with line B indicating the more rapid growth, which occurs for
accelerated conditions such when increased stress or temperature provide a
greater driving force for damage.
 11-3 
13828389
The critical crack size is dependent upon the stress and the fracture toughness of
the material so each of the following should be considered:

The stress concentration at the crack tip increases as the crack extends and
with decreasing tip radius, (that is the stress increases for longer, sharper
cracks). This increase in stress will decrease the critical defect size even in
material which is microstructurally stable.

Embrittlement effects can reduce the fracture resistance. The degree of
embrittlement increases with time of exposure until a plateau is reached so
that the critical crack size is significantly reduced.
Moreover, factors such as excess cycling, temperature excursions, systems stresses
etc. can accelerate the crack propagation rate (line B compared to line A)
resulting in reaching the critical condition more rapidly. The following trends are
thus apparent:

Appropriate levels of quality assurance are required to ensure that no
significant manufacturing cracks are present. These defects will lead to rapid
failure because the crack initiation phase is not required.

In the absence of pre-existing manufacturing defects, significant operating
periods are associated with crack initiation. During this time the application
of detailed methods of damage detection permit the early stages of
degradation to be identified so that the maximum time is available for
decisions regarding further action.

Additional accelerating stresses or environmental factors will increase the rate
of crack growth and reduce life but will not significantly reduce the critical
crack size.

Embrittlement phenomena will significantly reduce the expected life and will
also reduce the critical crack size and increase the risk of catastrophic fast
fracture.
Embrittlement phenomena lead to an increase in the fracture appearance transition
temperature (FATT), lowering of the upper shelf Charpy energy and a reduction in the
fracture toughness (KIC, JIC) at low and intermediate temperatures.
It is generally the case that the risk of sudden brittle fracture increases for
components operating at temperatures where the fracture energy is in the lower
shelf regime. Under these conditions the material is most susceptible to brittle
behavior. Thus, assessments should consider the risks of failure during transients
and for pressure vessels and piping the temperatures and pressures for hydro
testing. There are other circumstances where rapid fracture can occur, for
example:

When time/cycle dependent cracking has developed as the result of creep,
fatigue or the interaction between creep and fatigue

When the environment has introduced or accelerated cracking, for example,
intergranular corrosion, stress corrosion or liquid metal embrittlement
 11-4 
13828389

When the upper shelf energy in combination with high stresses leads to the
critical crack size being small. This crack size may be small enough to cause
rapid fracture from:
-
A fabrication defect
-
A fabrication defect which has increased in size due to in service cracking
-
In service damage has developed to form a crack
A large body of available FATT data on service exposed steam pipe/header grade
material has been compiled by Marshall, Jaske and Majundar [11.3]. and Liaw et
al. [11.4]. Data on cast steels used for casings have been reported [11.5 to 11.9].
Having determined the FATT by non-destructive or destructive tests,
quantitative use of this information still requires that the FATT be converted to
a fracture toughness parameter (KIC). The current procedure for cold hydro
testing of headers and pipes uses empirical approaches based on FATT and Nil
ductility temperature. There also many published Charpy/Fracture Toughness
correlations which can be used; see Tables 2-1a and 1b in Section 2 of this
guideline. Three correlations have been recommended in Annex J of the recent
standard BS7910, 1999 ('Guidance on methods for assessing the acceptability of
flaws in metallic structures [11.10] (incorporating Amendment 1'). Two of these
are used for materials on the lower shelf/transition of the ductile/brittle transition
curve, while the third is recommended for upper shelf behavior.
Steels on the lower shelf and in the transition region. A simple lower bound
estimate of toughness can be made from the Charpy energy measures at the
temperature of interest. The appropriate expression is:
Kmat = 820 (Cv)1/2 - 1420 + 630
B
Eq. 11-2
1/4
where Kmat is a lower bound estimate of fracture toughness in N/mm-3/2, B is the
thickness (in mm) of the material for which an estimate of Kmat is required, and
Cv is the Charpy energy (in J) for a 10 mm thick specimen tested at the
minimum service temperature.
Ferritic Steels Using the Charpy Energy. The so-called Master Curve approach
can be used to make a preliminary estimate of the fracture toughness of ferritic
steels from Charpy energy. This is a well-validated approach, which is based on a
correlation between the 27 J transition temperature and the temperature at which
a 25mm thick fracture mechanics specimen shows a fracture toughness, Kmat, of
 11-5 
13828389
100 MPa√m. The approach has been extensively validated for a range of parent
steels; see, for example, data for 1¼Cr-½Mo and 21/4Cr1Mo steels, Figure 7-2.
In these figures the ordinate is the fracture toughness normalized by the upper
shelf fracture toughness KIC-US as
(K1c-US ) = 0.6478 (CVN –US - 0.0098 )
σ 0.2
Eq. 11-3
σ 0.2
where CVN-US(J) and σ0.2(MPa) are the impact energy and the 0.2% offset yield
strength at the upper shelf temperature, which is defined at the lowest
temperature at which no evidence of brittle fracture is found. The narrow scatter
of fracture toughness, KIC/KIC-US is observed for the steels shown in Figure 11-2.
Using these master curves, the fracture toughness, KIC transition curves of the
materials can easily be obtained with successful results. There are cases where the
master curve approach overestimates Kmat, for example:

Where Charpy specimens exhibit unusual behavior such as fracture path
deviation

Where splits are present on fracture surface of fracture toughness specimens
due to crystallographic texture

Where microstructure and properties vary through the section thickness,
making it difficult to ensure that the Charpy specimen samples the same
microstructure as that associated with initiation in the fracture toughness
specimen

Where mis-match induced constraint may be a factor (highly over- or undermatched welds)

Where material is cold-worked
The method takes into account the test temperature, (T), 27 J temperature (T27J),
thickness of specimen (B), and desired probability of failure (Pf). Toughness at a
given temperature is given by the equation:
Kmat = 630 + [350 + 2435 exp [0.019(T - T27J - 3)]]
Eq. 11-4
(25 / B )1/4[ln (1 / 1-Pf )]1/4
With units: of Kmat in Nmm-3/2, T and T27 J in °C, and B in mm. A value
Pf = 0.05 (5%) is recommended for initial assessments. Note that this equation
increases without limit as the temperature is increased, and it is important to take
into account the onset of upper shelf behavior, so that the upper shelf toughness
is not overestimated.
 11-6 
13828389
Lower Bound Upper Shelf Toughness. A simple equation, also given in Annex J
of BS7910, can be used to estimate the lower bound upper shelf toughness in
cases where the Charpy test results show 100% shear fracture:
Kmat = 17Cv + 1740
Eq. 11-5
where Cv is the Charpy energy at the temperature of interest.
Figure 11-2
Examples of the Master Curve approach relating FATT with fracture toughness for
(a) 1/2Mo and 11/4Cr1/2Mo steels and (b) 2 1/4Cr1Mo steel
11.1 Fracture Assessment Summary
In ferritic steels, the overall fracture behavior will depend strongly on
temperature. At low temperatures, brittle fracture prevails; once the crack has
started to extend, crack propagation may occur extremely rapidly. At high
temperatures and for materials such as austenitic stainless steels, the fracture
behavior is ductile and crack growth takes place by a stable tearing mechanism.
Whatever the mechanism, for fracture or crack growth to occur, a detrimental
combination of applied stress, crack dimension and the material's fracture
toughness is required. This condition can be expressed as:
KI ≥ Kmat
Eq. 11-6
If the crack driving force (expressed as the applied stress intensity factor, KI) is greater or
equal than the brittle or ductile fracture toughness, Kmat, fracture will occur.
The stress intensity factor characterizes the stress field at the crack tip, and it is
the conditions at the crack tip, which govern the general behavior of a cracked
structure.
 11-7 
13828389
The applied stress intensity factor, KI, is calculated using relations involving the
geometry of the component, the magnitude of the applied stresses and the crack
dimensions. For elastic-plastic conditions, the strain hardening behavior of the
material in question is also important. The stress analysis should consider stress
concentrations, including those, which may arise from deviations from the
intended design, such as misalignment; and welded residual stresses (of up to
yield strength magnitude) must be taken into account.
Kmat is measured using pre-cracked specimens taken from the material, which
represent the region in which the subject crack is located. For example, if the
subject crack is located in weld metal, the fracture toughness specimen will be
notched and fatigue pre-cracked into a test weld representing the structural weld.
The test procedures are described in national and international standards.
Fracture toughness values are sensitive to material microstructure, heat treatment
condition, loading rate and test temperature (particularly in ferritic steels) and, in
certain circumstances, specimen thickness.
11.2 References
11.1
R. Viswanathan, “Damage Mechanisms and Life Assessment of High
Temperature Components,” ASM International, 1989.
11.2
J. D. Parker and D. Sidey, “Residual Life of Boiler Components,”
Materials Forum, Vol. 9, No. 1 and 2, 1986, pp. 78–89.
11.3
Guidelines for the Evaluation of Seam-Welded High-Energy Piping. EPRI,
Palo Alto, CA: 2012. 1025326.
11.4
P. K. Liaw, A. Saxena, and M. G. Burke, “The Microstructure and
Toughness Behavior of Ex-service Cr Mo Steel,” Report of EPRI
Project RP 2253-10, June 1988.
11.5
N. S. Cheruvu. “Degradation of Mechanical Properties of CrMoV and
2¼Cr-1Mo Steel Components after Long Term Service at Elevated
Temperatures,” Met. Trans. A, Vol. 20A, January 1989, pp. 89–97.
11.6
M. Shiga et al. “Mechanical Properties of Casing and Rotor Steels after
Long Term Service in Steam Turbine,” JSME, Vo. 33, No. 366, March
1984, pp. 298–303.
11.7
D. Bishop, Mermac No. 3 Turbine Intermediate Pressure Shell
Cracking,” Metallurgy and Piping Task Force of the EEI Movers
Committee.
11.8
W. A. Logsdon, P. K. Liaw, and A. Saxena, “Residual Life Prediction
and Retirement for Cause Criteria for SSTG Upper Casings–I.
Mechanical and Fracture Mechanics Material Properties Development,”
Eng. Fracture. Mech. Vol. 25, No. 3, 1986, pp. 259–288.
11.9
Coulon et al., “Expectation of Life of Cast Cr_Mo Steam Turbine
Casings after 100,000 hours Operation,” Conference: Mechanical
Behavior of Materials, Vol. 2, Cambridge, England, August 1979.
 11-8 
13828389
11.10
BS 7910:1999: “Guidance on methods for assessing the acceptability of
flaws in metallic structures (incorporating Amendment 1).” London;
British Standards Institution, 1999.
11.11
PD6493: 1991 Guidance on methods for assessing the acceptability of
flaws in fusion welded structures. London; British Standards Institution,
1991.
11.12
INSTA Technical Report, 1991: “Assessment of structures containing
discontinuities,” Materials Standards Institution, Stockholm.
11.13
Wallin, K: “Simple theoretical Charpy V-KIc correlation for irradiation
embrittlement,” in Innovative approaches to irradiation damage and fracture
analysis, ed. D. L. Marriott, T. R. Mayer, WH Barnford, New York:
ASME. PVP-170. 93.100. ISBN 0791803260.
11.14
Sailors, R. H. and Corten, H. T. (1972): “Relationship between material
fracture toughness using fracture mechanics and transition temperature
tests” in Fracture toughness. Propc. national symp. on fracture mechanics,
Urbana, IL, 31 Aug–2 Sept 1971. ASTM STP 514, part 2, 164–191.
 11-9 
13828389
13828389
Appendix A: Glossary of Metallurgical
Terms
This summary listing has been developed to provide definitions used in defining
metallurgical processing and testing terminology. Other reference documents
should be reviewed to provide more complete information.
ANNEALING
A treatment consisting of heating uniformly to a temperature, within or above
the critical range, and cooling at a controlled rate to a temperature under the
critical range. This treatment is used to produce a definite microstructure, usually
one designed for best machinability, and/or to remove stresses, induce softness,
and alter ductility, toughness or other mechanical properties.
ELONGATION
In tensile testing, the increase in gage length, measured after the fracture of a
specimen within the gage length, usually expressed as a percentage of the original
gage length.
JOMINY END-QUENCH TEST
A laboratory procedure for determining the hardenability of steel. Hardenability
is determined by heating a standard specimen above the upper critical
temperature, placing the hot specimen in a fixture so that a stream of cold water
impinges on one end, and, after cooling to room temperature is completed,
measuring the hardness near the surface of the specimen at regularly spaced
intervals along its length. The data are normally plotted as hardness versus
distance from the quenched end.
HARDENABILITY
Describes the ability of steel to form a given microstructure for a given heat
treatment. It is frequently evaluated by performing hardness tests where an
appropriate definition would be the capacity of steel to harden in depth under a
given set of heat treatment conditions. In decreasing order of significance the
influence of elements on hardenability is C, V, Mo, Cr, Mn, Si, Cu and Ni.
Even relatively low levels of Boron, 0.002 to 0.003%, will increase hardenability
provided that Ti is added to react preferentially with oxygen and nitrogen.
 A-1 
13828389
HARDNESS
Resistance to plastic deformation which is usually measured by indentation
testing. This may also refer to resistance to scratching, abrasion, or cutting.
IMPACT TEST
A test to determine the behavior of materials when subjected to high rates of
loading, usually in bending, tension or torsion. The quantity measured is the
energy absorbed in breaking the specimen by a single blow, as in the Charpy or
Izod tests.
INGOT
A simple shape produced by casting that can be used for subsequent hot working
or remelting.
KILLED STEEL
Steel treated with a strong deoxidizer to reduce oxygen in the molten metal to a
level where no reaction occurs between carbon and oxygen during solidification.
LAP
A surface imperfection that appears as a seam or a linear defect. It is caused by
the folding over of hot metal, fins, or sharp corners and then rolling or forging
these into the surface. Laps on tubes can form from seams on piercing mill
billets.
MACHINABILITY
This is a generic term for describing the ability of a material to be machined. To
be meaningful, machinability must be qualified in terms of tool wear, tool life,
chip control, and/or surface finish and integrity. Overall machining performance
is affected by a myriad of variables relating to the machining operation and the
workpiece.
NORMALIZING
A heat treatment consisting of heating a part uniformly to set temperature at
least 100°F above the critical range, followed by holding for a given time and
then cooling in still air to room temperature. The treatment produces
recrystallization and refinement of the grain structure and gives uniformity in
hardness and structure to the product.
PICKLING
An operation that involves removal of surface oxide (scale) developed during
manufacturing by chemical action. Sulfuric acid is typically used for carbon and
low-alloy steels.
 A-2 
13828389
POST WELD HEAT TREATMENT (PWHT)
PWHT is carried out to relieve a proportion of the welding residual stresses
under conditions where there is sufficient ductility to prevent cracking. The
elevated temperatures involved will also temper the microstructure reducing the
hardness and increasing ductility, thus reducing danger of cracking in subsequent
service.
The conditions required depend on the alloy composition and section thickness.
The applicable ASME Code sections contain requirements for PWHT
specifying rate of heating and cooling above 800°F and requiring a holding
temperature (usually one hour per inch of thickness of the material). The holding
temperatures vary with the P-numbers of the material, which in turn are based
on alloy content. As an example, P-1 through P-4 require 1100°F holding
temperature, P-1 being carbon steels, P-3 being carbon steels alloyed in relatively
small percent with molybdenum, manganese and vanadium. P-4 steels are the
nickel steels, chrome-molybdenum and nickel- chrome-molybdenum. P-5, P-6
and P-7 high alloy steels generally require a higher holding temperature ranging
up to 1350°F. Some of the special steels now listed in the Code sections call for
even higher holding temperatures.
PREHEATING
Preheating of the weldment area achieves better weld penetration and slows the
cooling process, thus allowing added relief of stresses and reduced hardening of
the materials.
QUENCHING
A treatment consisting of heating uniformly to a predetermined temperature and
cooling rapidly in air or liquid medium to produce a desired crystalline structure.
REDUCTION OF AREA
The difference, expressed as a percentage of original area, between the original
cross-sectional area of a tensile test specimen and the minimum cross-sectional
area measured after complete separation.
RIMMED STEEL
A low carbon steel having enough iron oxide to give a continuous evolution of
carbon monoxide during solidification giving a rim of material virtually free of
voids.
SEMI-KILLED STEEL
Incompletely deoxidized steel which contains enough dissolved oxygen to react
with the carbon to form carbon monoxide to offset solidification shrinkage.
 A-3 
13828389
SPHEROIDIZE ANNEAL
A special type of annealing that requires an extremely long cycle. This treatment
is used to produce globular carbides and maximum softness for best machinability
in some analyses, or to improve cold formability.
STRESS RELIEVE TEMPER
A thermal treatment to reduce residual stresses and strains so that during
subsequent machining or hardening operations distortion is minimized. This
treatment is usually applied to material that has been quenched and tempered.
Normal practice would be to heat to a temperature 100°F lower than the
tempering temperatures used to establish mechanical properties and hardness.
Ordinarily, no straightening is performed after the stress relieve temper.
TEMPERING
A treatment consisting of heating uniformly to some predetermined temperature
under the critical range, holding at that temperature a designated period of time
and cooling in air or liquid. This treatment is used to produce one or more of the
following end results: A) to soften material for subsequent machining or cold
working, B) to improve ductility and relieve stresses resulting from prior
treatment or cold working, and C) to produce the desired mechanical properties
or structure in the second step of a double treatment.
TENSILE STRENGTH
In tensile testing, the ratio of maximum load to original cross-sectional area.
YIELD POINT
The first stress in a material, usually less than the maximum attainable stress, at
which an increase in strain occurs without an increase in stress. If there is a
decrease in stress after yielding, a distinction may be made between upper and
lower yield points.
YIELD STRENGTH
The value of stress when a material under increasing load exhibits a specified
deviation from linear stress and strain behavior. An offset of 0.2% is commonly
used and the stress is known as the proof stress.
 A-4 
13828389
Appendix B: Case Study: Embrittlement in
Alloy 80A Fasteners
B.1 Introduction
The following summary has been prepared based on information published in
references [1, 2, 3]. Alloy 80A (Ni-20Cr-2.4Ti-l.4Al) has been used extensively
as a bolting material in Europe for valve covers, steam strainers, cylinders, loop
pipe flanges and nozzle plates in sizes up to 1000 mm long and 115 mm
diameter. Steam pressures have been up to 165 bar and the bolts have operated at
temperatures in the range 450-550°C.
Altogether there have been over 14,000 Alloy 80A bolts used in the UK and
some 6500 used elsewhere and operating times now exceeding 175,000 h. In
general, the service experience has been excellent with 74 reported failures in the
UK and less than 0.4% of failures in total. Of these approximately 50% have been
attributed to intergranular fast fracture after in-service embrittlement, 33% to
stress corrosion cracking and the remainder to creep attributed to over-tightening
and high strain fatigue in one specific location. Apart from the early creep
failures, it was generally found that the failed bolts had actually shortened in
length during service and the material had apparently embrittled exhibiting low
impact energies. A further significant aspect of the failures in the UK was that
most of these occurred on CEGB oil fired plant with an operating temperature of
540°C or in joints on coal fired stations that were operating below this
temperature. A typical intergranular brittle failure of an alloy 80A bolt is shown
in Figure B-1.
B.2 Factors Affecting Life
A relevant factor to these observations is that Alloy 80A undergoes ordering
reactions, short range order (SRO) of Ni and Cr atoms at temperatures below
550°C and at temperatures below a critical temperature T c = 525-530°C, the
SRO may transform to long range order (LRO) with the formation of Ni2Cr.
The important aspect about these transformations is that they give rise to lattice
contraction, 0.03% for SRO [3] and −0.11 % for LRO after 30,000 h at 450°C.
These reactions, particularly LRO, are extremely sluggish with an incubation
period of> 10,000 h and may take up to 30,000 h for completion.
 B-1 
13828389
The significance of these reactions to the use of Alloy 80A as a bolting material,
is the way in which they influence the stress relaxation properties of the material.
Thus at temperatures above 550°C Alloy 80A displays normal stress relaxation
behavior as the initial elastic strain, 0.15%, is converted to plastic strain due to
creep (Figure B-2).
Figure B-1
Typical intergranular brittle fractures in an Alloy 80A bolt [2]
At lower temperatures, in the ordering range, there are two competing processes:

Lattice contraction due to ordering which gives an increase in stress during a
'stress relaxation' test

A decrease in stress due to creep, giving rise to the complex behavior shown
in Figure 2 at 500 and 550°C. At 450°C the ordering reactions dominate
giving rise to the large pick-up in stress shown
These observations have been extensively studied by Nath et al. [3]. It should be
noted, however, that the actual increase in stress realized in service may be made
less than that shown in Figure B-2 due to creep of the flange material as observed
by Mayer and Konig in model bolt tests. Nevertheless, it is considered that a
contributory factor to the Alloy 80A bolt failures in service is an increase in stress
on the bolts due to lattice contraction arising from atomic ordering.
In addition to lattice contraction the other common feature to most Alloy 80A
bolt failures has been embrittlement. It is considered [1] that the combination of
these two phenomena is largely responsible for the failures due to intergranular
 B-2 
13828389
fast fracture and that embrittlement also increases the susceptibility to stress
corrosion cracking. Again, work by Nath et al. [3] has found Charpy impact
values in failed ex-service bolts in the range 5-17 J whereas in unfailed bolts the
impact values were in the range 25-48 J (Figure B-3).
Figure B-2
Stress relaxation behavior of Alloy 80A
Fracture surfaces in the former displayed brittle intergranular cleavage with no
evidence of ductile micro void formation on boundary carbides. Materials with
higher impact values displayed transgranular/ductile intergranular fracture with
micro voids associated with grain boundaries. Ageing studies on the more brittle
materials demonstrated that the embrittlement phenomenon was reversible and
Auger Electron Spectroscopy revealed phosphorous segregated to grain
boundaries. In new casts of Alloy 80A the kinetics of embrittlement decreased
significantly with decreasing bulk phosphorus content over the range 50-20 ppm
P (Figure B-4) [3].
 B-3 
13828389
Figure B-3
Variation of Charpy energy with aging for Alloy 80A
Figure B-4
The embrittling effect of P segregation on the fracture behavior of Alloy 80A
 B-4 
13828389
In view of the problems encountered greater control of alloying elements such as
Al and Ti as well ensuring strict control of trace elements is recommended.
Recent studies have reported significant improvements in Charpy Fracture
Energy for 80A with low levels of Al and Ti, for example, Figure B-5.
Figure B-5
Improvement in fracture resistance with low levels of Al +Ti
A variety of laboratory investigations has now established that Alloy 80A is
susceptible to stress corrosion cracking in certain environments. These include
sulphuric and hydrochloric acid solutions and water with circulating SO2 but
seemingly not in non acidic chloride (with or without circulating SO2) nor
concentrated sodium hydroxide solution. Failures in service attributed to SCC
have invariably sulfur contamination as a common feature on fracture surfaces
and particularly those that had been lubricated with molybdenum disulfide,
which laboratory tests now indicate promotes the formation of acidic
environments. Evidence is also available to indicate that embrittlement in service
due to P segregation also increases the susceptibility to SCC in this material.
 B-5 
13828389
B.3 References
1. R. D. Townsend, “Performance of High Temperature Bolting in power
plant,” Proc “Performance of bolting materials in High Temperature
Applications,” Ed. A. Strang, Institute of Materials, 1995, pp. 15–40.
2. P. Vinders, R. Gommans, G. van Oppen, and K Verheesen, “Brittle Fracture
of Alloy 80 A bolts in a steam turbine” Proc “Performance of bolting
materials in High Temperature Applications,” Ed. A. Strang, Institute of
Materials, 1995, pp. 271–283.
3. B. Nath, K. H. Mayers, S. M. Beech, and R. Vanstone “Recent
developments in Alloy 80A for high Temperature Bolting applications” Proc
“Performance of bolting materials in High Temperature Applications,” Ed.
A. Strang, Institute of Materials, 1995, pp. 306–317.
 B-6 
13828389
Appendix C: Case Study–Brittle Failure of
Ferritic Steel Bolts
C.1 Introduction
The information given below is summarized from reference 1.
Figure C-1
Photograph showing damage caused by failure of 24 low alloy steel bolts
A photograph of the damage caused by the failure of 24 Durehete 1055 stud
bolts on a steam chest at a power station in the UK in 1979 is shown in Figure
C-1. The incident was remarkable not only for the violence of the event, the
chest cover and associated valve gear was blown 80 ft through the roof of the
turbine hall, but also because of the large number of design parameters (faults)
and adverse metallurgical parameters which contributed to the failure. At the
time of the failure the unit had operated for 54,000 h and the stud bolts on the
steam chest had been subjected to five tightening operations. According to the
applicable specification the effective bolt lives were thus:
Hours in services plus Tightening Penalty = 54,000 h + (5 × 15,000 h) =
129,000 h
 C-1 
13828389
In this case the Design Factor KD = 1 and the Strain Factor = oh = 1, since each
retightening operation on all bolts had been performed (within the permitted
tolerance) to 0.15% strain. It should be noted that the stud bolts in the chest were
due for replacement at the next outage.
Figure C-2
Detail of creep damage found at the first engaged thread
Visual examination of the 24 fractured stud ends retained in the chest suggested
that the failures were initiated by creep cavitation and cracking in the region of
the first engaged threads (Figure C-2). Most of the fractures were relatively flat
and typical of bolt creep failures. However, six fractures exhibited areas of cup
and cone shear failures but with some creep cracking at the perimeters. It was
noted that the initiation site of failure in all studs occurred on the outside of the
stud pitch circle. From examination of the studs in situ, it was clear that the final
failure event was associated with fracture of only six bolts and that the remainder
had in fact fractured well before then.
Oxide thickness measurements made on the fracture surfaces allowed estimates
of the crack ages. Assuming parabolic oxidation kinetics,
X2 = kt,
where X is the oxide thickness in cm, t the time and k the parabolic rate content
= 1.65 × 10-12 cm2 s-1 [2]. Crack initiation times were determined on some studs
to have occurred between 14,000-15,000 h prior to the final failure which
compared well with the total operating hours (15,143) since the last inspection.
Complete fracture of these same studs was calculated to have occurred at over
1000 h prior to the final failure. In most cases, the oxide was thicker (indicating
older cracks) on the outer portion of the stud fracture surfaces.
Hardness measurements on the nut end of the studs indicated all but two had
been within the required specification range 250-320 HV for D1055 but
longitudinal hardness traverses indicated significant hardness gradients along the
 C-2 
13828389
studs. In some cases the hardness dropped by 30-40 HV points from the nut end
to the fractured end (Figure C-3). These observations were consistent with the
studs having operated with a longitudinal temperature gradient, later determined
in some cases to have been as high as 40-70°C.
Figure C-3
Measured hardness values along the length of an ex-service stud
Since fracture initiated in all studs on the outer stud pitch circle, it was clear that
bending of the studs could have contributed significantly to the failure process.
Bending of studs in joints of this type occurs when the valve cover operates at
significantly lower temperatures than the valve body and hence sustains a smaller
thermal expansion when at service temperatures.
Measurements made on unfractured studs in a similar chest indicated permanent
out of plane deflection of up to 0.075 in. suggesting that the creep strain due to
bending could be as high as 0.12% per tightening and this of course would have
been superimposed on the relaxed creep strain from the original tightening
tensile strain of 0.15%.
On this basis of these observations, it was concluded that the operating condition
that had contributed most significantly to this failure, and not accounted for in
the original design, was the severe temperature gradient in the valve cover chest.
This occurred primarily because of the difficulty (almost inability) to properly lag
the valve cover due to interference by the valve gear mechanism. The presence of
the temperature gradient contributed to failure in two significant ways:
1. By the superimposition of a bending stress in addition to the original tensile
stress thereby increasing (possibly doubling) the total elastic strain relaxed
during each operating period
2. By concentrating the creep strain (acquired during relaxation of the studs in
service) in the hottest region of the studs, the first engaged thread located
just below the flange cover
 C-3 
13828389
Although the lack of proper lagging and consequential temperature gradients in
the valve cover were major contributing factors to this failure it was clear that
other factors were also significant. The oxide dating of cracks had indicated that
the studs had in fact failed in three different batches/phases. Detailed
metallurgical investigations indicated that early cracking occurred in studs with
coarse-grained bainitic structures, of high hardness and significant residual
element content. The effect of trace elements on reduction in area for these steels
is shown in Figure C-4. In contrast, the last studs to fail were fine-grained, much
lower hardness and had much lower residual element contents.
Figure C-4
Effect of trace element content on reduction in area for low alloy bolting steels
Ultimately, this failure incident and the subsequent investigations following it,
led to significant improvements in the understanding of bolt performance, the
need to control operating regimes more carefully and improvements in the
metallurgical design and performance. Improvements to the metallurgical
specification for CrMoV bolts, including

Improvements in heat treatment to control grain size and hardness.

Improved composition control to lower the residual element content and
reduce creep embrittlement. These improvements are directly linked to
higher available creep ductility, see Figure C-4. Significant improvements in
steel making practice were introduced. These included the use of secondary
melting techniques such as Vacuum Arc Remelting (VAR) and Vacuum
Induction Melting (VIM) for primary melting.
 C-4 
13828389
C.2 Key Issues
The following were identified as key issues associated with the fracture event:

Combination of unexpected operation and materials performance

Hardness used to identify variations in operating temperature

Interval between inspections too great to reveal initiation of micro damage,
the development to macro cracking and the crack propagation

Variation in fracture resistance of the individual bolts related to the prior
austenite grain size and the level of residual elements (trends agree with more
recent studies on 1CrMoV fasteners which show that coarse grains and high
residuals lead to embrittlement, and that for similar operating times some
bolts do not show embrittlement)
C.3 References
1. R. D. Townsend, “Performance of High Temperature Bolting in power
plant,” Proc “Performance of bolting materials in High Temperature
Applications,” Ed. A. Strang, Institute of Materials, 1995, pp. 15–40.
2. L. W. Pinder, “Role of Oxide Scale Thickness Measurements in Boiler
Failure Analysis,” Corrosion Science, 21, 1981, pp. 749–763.
 C-5 
13828389
13828389
Appendix D: Case Study–Review of
Cracking, Eddystone Unit 1
D.1 Introduction
Through wall cracking occurred in one of the 8 main steam leads after about
130,520 hours of operation. All major cracking events generate significant
interest within the power industry and this incident was thoroughly investigated
and reported. The following case study has been developed from several key
publications [1, 2] which presented unit information, detailed the damage found
and described the in depth cause analysis. Also briefly described is information
regarding an earlier incidence of cracking which occurred at the junction header
and the turbine stop/control valves.
D.2 Design and Operation
The steam generator of Eddystone No. 1 unit was of the supercritical oncethrough design. The feedwater control was designed on the basis of maintaining
a predetermined temperature at the outlet of the transition section. Spray
desuperheater controlled the temperature at various points in the super-heater.
The steam generator involved four parallel circuits, each of which was separately
controlled for temperature and flow, and each of which comprised up to sixty
tubes in parallel from the economizer inlet to the transition section outlet.
The generated steam passed through eight superheater circuits and in this way
superheating to 1200°F is accomplished. The superheated steam was lead to eight
superheater outlet headers. The main steam system was constructed of eight 316
stainless steel pipes, 232 mm outside diameter and 63.5 mm wall thickness. Main
steam was collected at a junction header and then passed to a super pressure
turbine through four sets of main stop/by-pass valves. The turbine generator was
a cross-compound 3600/1800 rpm set rated at 325 MW, designed for the main
steam conditions, 5000 psig and 1,200°F. Details of the pressure and temperature
history together with the number of starts are presented in Figure D-1, with a
piping isometric shown in Figure D-2.
 D-1 
13828389
It should be noted that the design of the Eddystone boiler is different from that
of more recent supercritical and ultra supercritical designs. In the once-through
design, feedwater is exhausted to the by-pass water separator located downstream
of the boiler stop/by-pass valves. Therefore, the main steam pipe between
superheater outlet header and boiler stop valves can be cooled down rapidly
during shutdown. In modern designs, the location of a water separator prevents
this possibility.
Figure D-1
Operating history of the Eddystone boiler
 D-2 
13828389
Figure D-2
Isometric drawing of Eddystone No. 1 main steam system
 D-3 
13828389
(a)
(b)
Figure D-3
(a) Macrostructure of the failed main steam pipe; (b) microdamage in the failed
main steam pipe
 D-4 
13828389
D.3 Summary of Piping Damage
A steam leak was detected in one main steam lead upstream of the boiler stop
valve on unit start up following a scheduled outage. At the location of the leak,
axial cracking was present in the parent metal of one pipe spool. This was
approximately 1.2 m long on the outside and 36 cm long at the inside. In the
inspection, which followed additional OD, surface connected cracks were found.
In total, cracks were found in 4 of the 14 heats used in pipe fabrication.
Typical photo macrographs of the cracking observed are presented in Figure
D-3. As shown, the primary cracking was connected to the outside surface with
propagation predominantly in a through wall direction. Within the pipe wall the
level of microdamage appeared to increase. A section near the tip of the main
cracks was taken; scanning electron microscopy of the fracture surface revealed
that the damage was intergranular, with individual boundaries showing large
numbers of creep cavities, Figure D-4. These failures were therefore attributed to
long-term creep cracking.
Although the damage mechanism was established from the metallographic
evaluation proper cause analysis requires that the reasons for the cracking are
determined. The creep rupture strength of the cracked pipe was estimated to
have been relatively low, on account of its low carbon content of 0.037%, Table
D-1. Furthermore, the cracking was found to have occurred only in zones where
a large amount of sigma phase was present, which will have promoted the
multiplication of voids that came to be strung together into cracks.
Table D-1
Chemical composition of material from the cracked pipe and turbine stop valve
Elements
(wt %)
Cracked
pipe
Turbine Stop
valve
ASTM Standard
TP 316
C
0.037
0.075
0.08 (max.)
Mn
1.78
1.98
2.00 (max.)
P
0.015
0.017
0.04 (max.)
S
0.025
0.012
0.03 (max.)
Si
0.38
0.55
0.75 (max.)
Cr
17.33
15.01
16.00-18.00
Ni
12.69
13.71
11.00-14.00
Mo
2.34
2.18
2.00-3.00
N
0.034
0.024
Nv
2.8545
2.7858
Ni balance
0.186
4.92
 D-5 
13828389
Figure D-4
Fractograph confirming that extensive intergranular damage, with evidence of
creep voids, was present at the crack tip
Sigma phase precipitation was observed in large amounts in the microstructure of
the heats TP 316 stainless steel that had suffered cracking. Figure D-5 shows the
evidence for the presence of sigma established using specialist etching techniques.
Further evidence was obtained using electron microscopy, which confirmed the
composition to be 49% Fe, 37% Cr, 9% Mo, 4% Ni typical of sigma phase. The
creep cracks proved to have generated only in the heats that had precipitated
sigma phase, indicating that the high temperature strength of TP 316 steel is
impaired by sigma phase precipitation. Indeed, detailed metallographic
preparation established that creep cavities were nucleated on the sigma phase.
Figure D-5
Microstructure of main steam line section where creep cracking had developed;
(a) etched in hydrochloric and picric acid and (b) electrolytic etch in KOH to
reveal the sigma phase
In contrast, downstream of the boiler valves, piping of the same material was
found free from cracking though showing similarly levels of sigma phase
precipitation. It was established that the sigma phase caused a significant
reduction in room temperature Charpy impact energy, the measured values on
ex-service material ranged from 3 to 4.8 kg∙m. However, the differences in
 D-6 
13828389
observed creep cracking are further evidence of the sensitivity of creep to stress.
In the damaged piping, in addition to pressure stresses, high thermal and residual
stresses were also present. Moreover, sigma precipitation was absent from nine of
13 heats, which demonstrates that the formation of this phase is sensitively
influenced by small differences in chemical composition.
D.4 Previous Damage
Cracking at the internal surface of the Junction Header and Turbine stop valve
had been detected early in the life of the unit after about 25,000 hours and 77
starts. Both of these were very thick walled components, the junction header,
shown schematically in Figure D-6, was machined from a forged block of
material and was approximately 7.25 inches in thickness. It was apparent from
post service evaluation that damage was greatest at the position of maximum
thickness, Figure D-7.
Figure D-6
Schematic diagram of the junction header
 D-7 
13828389
Figure D-7
Cross section of the junction header showing the ID surface cracking revealed by
penetrant testing
Cracks extended up to about 20 mm into the header wall, detailed
metallographic evaluation showed that these cracks were very wide at the ID
surface and there was evidence in the propagation stage of interaction of the
defect with a grain boundary phase. Thus, away from the surface the defects
exhibited a tendency to be intergranular, Figure D-8.
Figure D-8
Damage developed in the junction header
Detailed metallographic investigation revealed that sigma phase had developed
on the grain boundaries at this location, Figure D-9. The compositions of both
the steam piping and the junction header indicted that these locations gave
similar high values of chromium equivalent and low values of nickel equivalent.
Thus, significant sigma phase had developed in both locations within the main
steam system, Figures D-5 and D-9. However, because of the differences in
operating conditions different damage developed from different mechanisms
Thus, the damage in the junction header exhibited classical characteristics of
 D-8 
13828389
thermal fatigue–creep interaction, and subsequent modification to the system
allowed pre warming of these thick sections prior to start up so reducing any
thermal transients. This modification to operation was successful in mitigating
the driving force for damage.
Figure D-9
Microstructure of main steam line section where creep cracking had developed;
(a) etched in hydrochloric and picric acid and (b) electrolytic etch in KOH to
reveal the sigma phase
D.5 Concluding Remarks
Damage was due to creep and creep fatigue. Life fraction estimation performed
using detailed creep analysis and the fatigue damage curve for inelastic analysis in
Code Case N-47 demonstrated reasonable agreement with the behavior
observed, Figure D-10.
Figure D-10
Schematic diagram showing estimates of creep fatigue usage
 D-9 
13828389
D.6 References
1. J. F. DeLong, W. F. Siddall, F. V. Ellis, H. Haneda, T. Tsuchiya,
T. Daikoku, F. Masuyama, and K. Setoguchi, “Operation experiences
and reliability evaluation on main steam line pressure parts of Philadelphia
Electric Co., Eddystone No. 1,” Mitsubishi Boiler Bulletin MBB-84112E,
Translated from November 1984 issues of The Thermal and Nuclear
Power 35 (11) (1984), 1225.
2. S. Kihara, M. Nakashiro, R. Ishimoto, I. Kajigaya, J. F. DeLong,
“Investigation of thru-wall cracking in main steam pipe of a super high
temperature and pressure plant,” Ishikawajima-Harima Heavy Industries
Co., Ltd (IHI), an article from IHI Engineering review, Vol. 17, No. 3,
IHI-380-8408, pp. 152–158.
 D-10 
13828389
Appendix E: Case Study–Cracking in a
CrMoV Weld
E.1 Introduction
During the 1970’s many European power stations experienced difficulties
resulting from the formation and growth of cracks in the HAZ’s of welds in the
CrMoV pipework systems [1]. This case study summarizes the information
reported with one particular failure that clearly illustrates the roles of composition
and microstructure on damage development.
E.2 Damage Detected
Following steam leakage from a butt weld joining a closed die forged valve to a
loop pipe adaptor on and HP steam chest (Figure E-1) of a 500 MW turbine
after 11,000 h operation the unit was shut down. The origin of the leak was
shown to be a circumferential crack approximately 615 mm long extending over a
220° arc on the forging side of the weld. The design steam conditions were
16 MPa (2300 psi) and 565°C and the dimensions of the weld were 318 mm OD
and 218 ID. The cracked weld was removed as a complete ring. Access to this
material enabled metallographic examination to be performed and mechanical
properties to be determined.
 E-1 
13828389
Figure E-1
Schematic diagram showing the location of the cracked weld
E.3 Metallurgical Evaluation
The full examination involved measurement of composition, metallographic
characterization of the constituent microstructures and the damage present as
well as hardness and mechanical testing. Only a summary of the key findings is
presented here.
The compositions of the weld metal and the parent from the forging and the
adaptor are given in Table E-1. As was normal practice at the time of fabrication
the weld metal used to join Cr Mo V steel was 2CrMo. In general, the measured
compositions with respect to major alloying elements agreed with specification.
However, it was apparent that the level of tin present in the forging was
sufficiently high to have accelerated creep cavity formation and growth, thus
significantly reducing the expected creep life of this material.
 E-2 
13828389
Table E-1
Compositions of the base and weld metals
Metallographic examination revealed that the cracking had developed in the
HAZ within the region where coarse grained bainite had formed as a
consequence of the welding thermal cycles, Figure E-2.
Detailed examination revealed that many of the prior austenite grain boundaries
adjacent to the macro defect exhibited creep cavities and micro cracks, Figure
E-3. In contrast, no evidence of micro damage was detected in the weld metal or
in the HAZ on the adaptor side of the weld.
Measurements of hardness were in general agreement of values expected for the
specified heat treatment, that is, 675–700°C. However, the creep rupture lives at
565°C of samples removed from the HAZ of the forging were 1-2 orders of
magnitude less than for similar samples from the adaptor HAZ. This difference
could be explained by the large amounts of intergranular damage in coarse
grained regions of the forging HAZ and by differences in the intrinsic rupture
properties of the forging and adaptor materials due to the high tin levels.
 E-3 
13828389
Figure E-2
Micrograph showing the cracking on the forging side of the weld
E.4 Concluding Remarks
The cause of cracking in the HAZ of the forging was attributed to two principal
factors:

The predominantly coarse grained microstructure of the HAZ

The high content of tin in the governor valve forging
These factors combined to give abnormally high susceptibility to the initiation of
creep cavitation during post weld heat treatment and to crack growth in service.
Post weld heat treatment conditions are normally established so that welding
residual stresses will be relaxed under conditions of high ductility. The
susceptibility to reheat cracking is greatly increased when coarse grains are
developed in the HAZ during welding. These coarse grained structures have
been shown to significantly reduce creep ductility compared with the behavior of
fine grained material. Thus, improved performance is achieved by utilizing weld
procedures which involve refinement of the HAZ microstructures produced.
In the present example, the creep rupture behavior was below normal because the
level of tin present accelerated cavity formation and growth thus promoting
intergranular fracture. The creep properties of the forging HAZ were indicative
of a tensile stress of 30-40 MPa acting over the service life of 11,000 h. This was
consistent with the actual stress based on internal pressure and estimated system
stresses.
 E-4 
13828389
Figure E-3
Detailed micrographs showing the extensive intergranular creep damage
developed in the coarse grained regions of the HAZ on the failed side of the weld
E.5 Reference
1. B. Freeman, T. Rowberry and B.L. King, “The role of composition and
microstructure in the failure of a weld on a 500MW turbine,” Conference,
The Institute of Mechanical Engineers, London, UK, 1980, C333/80,
pp. 107–112.
 E-5 
13828389
13828389
Appendix F: Case Study–Gallatin Unit 2,
IP-LP Single Flow Rotor
Failure
F.1 Introduction
The catastrophic failure of the Gallatin rotor involved subcritical crack growth
from a high density of MnS inclusions, with the critical crack size influenced by
the presence of hydrogen and temper embrittlement. The summary presented
here is based on information published in several key references [1–5].
F.2 Background
After approximately 107,000 hours of operation an intermediate-low pressure
single flow turbine rotor failed catastrophically on a 225 MW 1050°F/1050°F
tandem compound 3600 rpm turbine. The rotor was made of CrMoV steel and
was produced with an austenising temperature of 1750°F. This rotor contained
one impulse row and nine reaction rows of blading in the IP section and seven
rows of reaction blading in the LPSF section.
The turbine was being returned to service following a 6-day outage. At a speed of
3300 rpm and 3400 rpm, the rotor burst without warning. There were 23 missiles
of greater than 100 lbs ejected. Fragments of the rotor pierced the turbine casing
and the turbine room’s concrete roof. Most of the turbine damage was confined
to the IP-LPSF turbine section (Figures F-1 and F-2).
 F-1 
13828389
Figure F-1
Catastrophic failure of Gallatin Unit 2 IP-LPSF rotor
Figure F-2
Schematic diagram of the reassembled Gallatin rotor indicating location of
primary fracture surface
F.3 Damage Evaluation
Examination of the rotor fragments revealed that the primary fracture occurred
across a radial-axial plane (that initially divided into several fragments). A large
oxidized region was found on each of the two primary fracture surfaces, one on
each side of the bore near the exhaust end of the IP section, as shown in Figure
F-3. Further examination showed a high concentration of non-metallic
inclusions, identified as manganese sulfides, existed in the oxidized regions. One
such concentration was found in each oxidized region near the bore.
 F-2 
13828389
Figure F-3
Primary fracture surface of the bore near exhaust end of the IP section of the rotor
revealing a large oxidized region
It was concluded that small cracks developed around the nonmetallic inclusions
in the rotor from the combined effect of creep and low-cycle fatigue causing a
link-up between inclusions. During each cold startup, thermal stresses, as a result
of the bore being relatively cold as compared to the rim, caused the cracks in the
inclusion areas to propagate until they finally became critical and rotor bursting
occurred. The presence of hydrogen at the manganese sulphide interfaces and
temper embrittlement, are also believed to have contributed to the failure.
The steps involved in formation of a critical crack were:
1. Crack growth at operating temperature of 770°F (410°C) at an operating
stress of 52 ksi (358 MPa) by a creep mechanism leading to intergranular
cracking
2. Crack growth at a temperature of about 275°F (135°C) during cold starts
where the maximum stress was 75 ksi (517 MPa) resulting from thermal
stress superimposed on the steady stress; in the latter case the subcritical
crack growth mechanism was fatigue at prior austenite boundaries
A number of other points from relevant publications are presented [1–5]:

Creep acting alone cannot explain the failure

Low cycle fatigue (LCF) acting alone cannot explain the failure
 F-3 
13828389
The steps by which the Gallatin rotor failed are as follows:

Small cracks initiate around MnS inclusions by the mechanism of the
interaction of creep and LCF

The small cracks link up, joining together inclusions, to form the primary
flaw (origin)

The linking-up process continues until the flaw reaches critical size

The crack "pops-in" (in a brittle mode) to the large oxidized semi-circular
crack during a cold start

At the next cold start, the rotor fails catastrophically
An ultrasonic inspection had never been made from the bore of this rotor. It is
believed that such an inspection prior to this failure would have detected the
inclusions, and steps could have been taken to prevent this catastrophic failure. In
fact, this was a major reason for the decision to initiate a program in 1974 for
ultrasonic inspection of all TVA turbine rotors from the bore.
A new IP-LPSF rotor had previously been ordered for this turbine because of
excessive blade groove cracking. However, an outage of 30 months was required
to obtain new inner and outer cylinders for the IP and LPSF sections and other
required parts and rebuild the turbine.
Following this IP-LPSF rotor failure, a sister unit was removed from service to
inspect its IP-LPSF rotor. An ultrasonic indication found in it at approximately
the same location as the inclusion of the rotor that failed. The ultrasonic
indication was verified by bottle boring in the indicated region; and, at the
indicated depth of 1.6 inches, a crack was visually seen. Bottle boring continued
until all indications of the crack disappeared at a depth of approximately 2 inches.
This rotor was returned to service for a limited time.
F.4 References
1. H. S. Fox: “Tennessee valley authority’s turbine rotor experience,” EPRI WS79-235 Workshop Proceedings: Rotor Forgings for Turbines and Rotors.
September 1981, pp. 2-60-71, Edited by R.I. Jaffee.
2. S. H. Bush, “Failures in Large Steam Turbine Rotors,” ibid., pp. 1-1 to 1-27
3. J. M. Schmerlin and J. C. Hammon “Investigation of the Tennessee Valley
Authority Gallatin unit 2 Turbine Rotor Burst,” American Power
Conference, Chicago, 1976.
4. L. D. Kramer, D. D. Randolph, and D.A.Weisz, “Analysis of the Tennessee
Valley Authority, Gallatin unit 2 Turbine Rotor Burst” Winter Annual
Meeting of ASME, New York, 1976.
5. R. I. Jaffee, “Metallurgical problems and opportunities in coal powered
steam power plants,” 1977 ASM Campbell Memorial Lecture, Met Trans,
Vol. 10A, 1979, p. 139.
 F-4 
13828389
Appendix G: Case Study–Hinkley Point
Disc and Rotor Failure
G.1 Introduction
Acid open hearth (AOH) and basic open hearth (BOH) steelmaking were still
employed for the manufacture of rotor and disc forgings in the 1950s. Introduced
at the end of the 19th century, the refining was by slag/metal reaction and the
deoxidation products, MnO and SiO2, floated into the slag. In the AOH process
there was no removal of sulfur or phosphorus. However the oxidizing slag in the
BOH process did reduce these elements to some degree.
The effect of high S and P content on the fracture toughness properties of disc
forgings was catastrophically demonstrated in 1969 when a steam turbine disc
in a 60 MW turbine at Hinkley Point A suffered a fast fracture, Figure G-1 and
G-2 [1].
Figure G-1
Photograph showing damage caused by failure of the rotor disc
 G-1 
13828389
Figure G-2
Section reconstruction showing disc cracking
Figure G-3
Schematic diagram showing regions of segregation in the disc
 G-2 
13828389
Subsequent investigation showed that in the high S and P steel, which was of
large grain size and contained considerable chemical segregation, Figure G-3.
The fracture toughness was only about 40 MPa√m. At the service stress level this
degree of toughness was only adequate to tolerate a small defect, Figure G-4.
Figure G-4
Photographs showing the location of crack initiation
G.2 Developments for Improved Rotor Toughness
To maximize toughness, phosphorus together with tramp elements such as
arsenic, antimony, tin etc., have to be reduced to very low levels to minimize
grain boundary temper embrittlement during manufacture and also in service
[2, 3]. Since the tramp elements cannot be removed by refining, there has been
a necessity to exert a high level of control on the raw materials and especially in
the scrap selection to improve the toughness of rotor quality alloy steel forgings.
This philosophy has culminated in the development of Superclean 3.5%
NiCMoV steel which has been shown to be essentially immune to temper
embrittlement
The technology used in casting is very important to ensuring that a large rotor
forging will be of the required high standard of integrity in terms of the
soundness, cleanliness and chemical uniformity. There are many classical cut-ups
of ingots which show the typical unsoundness, due to primary piping and secondary
shrinkage, and the V- and A-segregates as in Figure G-5.
 G-3 
13828389
Traditionally, the factors controlling these effects became generally known and
controlled by experience. As a result, larger ingots were gradually introduced.
However, commercial incentives were such that these developments sometimes
took place too rapidly. In particular, in the late 1960s, developments in Germany
were put in place lo make an advance to a larger ingot of 250 tonnes. This was
achieved by lengthening a standard ingot. The height to diameter (H/D) ratio
was increased to 1.7. The ingot was then forged on a 6000 tonnes press to make
a rotor 1760 mm in diameter and 7.5 m long in which the forging work was
about 3:1.
Figure G-5
Schematic representation of ingot defects
 G-4 
13828389
Some years later, in 1987 after 16 years and almost 58,000 hours service, this
rotor burst during a routine restart of the machine [4]. The investigation showed
that brittle fracture had initiated at a large original defect which comprised planar
MnS inclusions and incompletely forged shrinkage, Figure G-6.
Figure G-6
Brittle fracture of a rotor from a manufacturing defect
G.3 References
1. D. Kladeron., “Steam Turbine Failure at Hinkley Point A,” Proc. ImechE,
1972, 186 (32), 341–377.
2. E. Potthast, K. Langer, and F. Tince., “Manufacture of superclean 3–3.5%
NiCrMoV steels for gas turbine components,” Clean Steel:Superclean Steel,
1995, 59–69.
3. R. Viswanathan., “Application of clean steel/superclean steel technology in
the Electric Power Industry–Overview of EPRI Research and Products,”
Clean Steel:Superclean Steel, 1995, 1–31.
4. J. Ewald et al., “Untersuchungan einer geborstenen Niederdruckwelle,”
VGB-Werkstoffagung, 1989, Vortrag 12.
 G-5 
13828389
13828389
Appendix H: Case Study Failure Due to
Graphitization in a Carbon½ Mo Steel Steam Pipe
H.1 Introduction
The information in this case study is based on information detailed in
reference 1.
Analysis performed on SA-335, Carbon-½ Molybdenum steel pipe confirmed
that failure in cold­formed bends was due to graphitization. However, the
graphite had developed in a manner that has received little attention in the
technical literature. In particular, the graphite developed as many small nodules,
preferentially concentrated within grain boundaries that were oriented normal to
the hoop stress. This suggests that local strain developed as a result of the piping
hoop stress contributed to the development of damage.
H.2 System History
The steam generator began commercial operation around 1970, and the subject
piping was reported to have approximately 275,000 hours of service. Full load
steam capacity of the steam generator was reported as 4,600,000 pounds per hour
at 3,800 psig superheater outlet pressure with 1005°F superheater and reheat
steam temperatures. The subject piping reportedly operated at a temperature of
approximately 830°F within a normal pressure range of approximately 3,400 to
3,800 psig (with a design pressure of 4,180 psig).
The most recent rupture within this superheater inlet piping occurred via a
longitudinal split along the extrados of a 90° bend (with a 24 inch bend radius).
The fracture was OD-initiated and thick-lipped with no evidence of macroductility. At the time of the pipe rupture, there were no recorded pressure or
temperature transients and the PSH inlet piping reportedly was operating well
within design conditions. Prior to a failure in 2011, there were four failure events
in the P 1 piping over a span of approximately 25 years. However, the first three
of these events were not investigated in sufficient detail to identify the failure
mechanisms.
 H-1 
13828389
About two years prior to the 2011 rupture, a number of pipe sections in the
steam generator failed during a significant pressure transient. Metallurgical
testing performed subsequent to that event revealed evidence of some type of
contamination of the grain boundaries in the failure areas, but the specific
contaminant was never identified. Mechanical property testing (in the
longitudinal direction) revealed higher than anticipated tensile and yield strength
values and lower ductility values, and this was attributed to strain age
embrittlement. It was concluded at that time that the failures were due to the
pressure transient and that the material "contamination" was a secondary issue.
Due to the extent of damage identified in the pipe bends after the over-pressure
event, a number of the bends were replaced, and smaller cracks in other bends,
including the bend that ruptured in 2011, were weld-repaired. While there is
ample documentation of graphitization in P 1 material after prolonged exposure
to elevated temperatures, the observed morphology of volumetric grain boundary
graphitization is very unusual.
 H-2 
13828389
H.3 Results
Figure H-1
Examples of the grain boundary graphite revealed using optical metallography
 H-3 
13828389
Examination of samples for grain boundary graphitization was found to be
much easier in the unetched condition, as the damage is more readily visible.
Representative images of the grain boundary graphitization are shown in Figure
H-1. In these micrographs, regions of heavy graphitization, as well as localized
regions showing the onset of damage, are visible. To further evaluate the radial
alignment of the grain boundary damage, metallographic samples were prepared
to allow for examination of the longitudinal-tangential plane and the
longitudinal-radial plane. Examination of these samples revealed that the damage
was generally planar in nature, and similar in dimension in the radial and
longitudinal directions.
In response to this unique form of graphitization, cryo-cracking, scanning
electron microscopy, energy dispersive spectroscopy, and X-ray diffraction
were carried out to confirm that the small nodular "particles" within the grain
boundaries were in fact graphite and not some other type of material
contamination. Examination by scanning electron microscopy confirmed
the local nature of the graphite at specific grain boundaries, Figure H-2.
Figure H-2
Scanning electron micrograph showing the local nature of the graphite formation
on grain boundaries
Mechanical tests were also performed to assess the degree of material
degradation, and the results have been compared to the damage levels observed in
metallographic samples using a five-level damage ranking system developed for
the purpose of surveying the multiple bends involved in the study. Specific results
showing the variation of Charpy impact energy with the level of graphite present
is shown in Figure H-3. It is apparent that fracture resistance is significantly
reduced by the formation of graphite. This significant reduction in fracture
resistance greatly increases the risk of fast fracture.
 H-4 
13828389
H.4 Concluding Remarks
Many of the same factors that in previous studies have been identified as
important influences on the formation of graphite in steel were observed in the
samples examined as part of this failure analysis. These included the original steel
making practice (that is, high levels of aluminum), the original forming process
(that is, cold bending), and the service conditions (that is, time and temperature).
Figure H-3
Variation of Charpy fracture energy with the level of graphitization present
H.5 Reference
1.
C. McDonald, J. Arnold, and J. Henry, Grain Boundary
Graphitization in P1 (C-1/2 Mo) Alloy Pipe, Proceedings of the
ASME 2012 Pressure Vessels & Piping Conference PVP2012,
July 15–19, 2012, Toronto, Ontario, Canada.
 H-5 
13828389
13828389
13828389
Export Control Restrictions
The Electric Power Research Institute, Inc. (EPRI, www.epri.com)
Access to and use of EPRI Intellectual Property is granted with the spe-
conducts research and development relating to the generation, delivery
cific understanding and requirement that responsibility for ensuring full
and use of electricity for the benefit of the public. An independent,
compliance with all applicable U.S. and foreign export laws and regu-
nonprofit organization, EPRI brings together its scientists and engineers
lations is being undertaken by you and your company. This includes
as well as experts from academia and industry to help address challenges
an obligation to ensure that any individual receiving access hereunder
in electricity, including reliability, efficiency, affordability, health, safety
who is not a U.S. citizen or permanent U.S. resident is permitted access
and the environment. EPRI also provides technology, policy and economic
under applicable U.S. and foreign export laws and regulations. In the
analyses to drive long-range research and development planning, and
event you are uncertain whether you or your company may lawfully
supports research in emerging technologies. EPRI’s members represent
obtain access to this EPRI Intellectual Property, you acknowledge that it
approximately 90 percent of the electricity generated and delivered in
is your obligation to consult with your company’s legal counsel to deter-
the United States, and international participation extends to more than
mine whether this access is lawful. Although EPRI may make available
30 countries. EPRI’s principal offices and laboratories are located in
on a case-by-case basis an informal assessment of the applicable U.S.
Palo Alto, Calif.; Charlotte, N.C.; Knoxville, Tenn.; and Lenox, Mass.
export classification for specific EPRI Intellectual Property, you and your
company acknowledge that this assessment is solely for informational
Together...Shaping the Future of Electricity
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:
Fossil Materials and Repair
© 2013 Electric Power Research Institute (EPRI), Inc. All rights reserved. Electric Power
Research Institute, EPRI, and TOGETHER...SHAPING THE FUTURE OF ELECTRICITY are
registered service marks of the Electric Power Research Institute, Inc.
3002001474
Electric Power Research Institute
3420 Hillview Avenue, Palo Alto, California 94304-1338 • PO Box 10412, Palo Alto, California 94303-0813 USA
800.313.3774 • 650.855.2121 • askepri@epri.com • www.epri.com
13828389