Chalk, Kieran (2013) Weld consumables and PWHT for

P92 power plant steel. PhD thesis, University of

Nottingham.

Access from the University of Nottingham repository:

http://eprints.nottingham.ac.uk/14572/1/594738.pdf

Copyright and reuse:

The Nottingham ePrints service makes this work by researchers of the University of
Nottingham available open access under the following conditions.

· Copyright and all moral rights to the version of the paper presented here belong to
the individual author(s) and/or other copyright owners.
· To the extent reasonable and practicable the material made available in Nottingham
ePrints has been checked for eligibility before being made available.
· Copies of full items can be used for personal research or study, educational, or not-
for-profit purposes without prior permission or charge provided that the authors, title
and full bibliographic details are credited, a hyperlink and/or URL is given for the
original metadata page and the content is not changed in any way.
· Quotations or similar reproductions must be sufficiently acknowledged.

Please see our full end user licence at:

http://eprints.nottingham.ac.uk/end_user_agreement.pdf

A note on versions:

The version presented here may differ from the published version or from the version of
record. If you wish to cite this item you are advised to consult the publisher’s version. Please
see the repository url above for details on accessing the published version and note that
access may require a subscription.

For more information, please contact eprints@nottingham.ac.uk

 The University of
Nottingham

Weld Consumables and PWHT for P92

Power Plant Steel

Kieran Chalk, BEng, MSc

Thesis submitted to the University of Nottingham

for the degree of Doctor of Philosophy in Materials
Engineering and Materials Design

2013
Acknowledgements

I would like to thank my supervisors Prof. Philip Shipway and Prof. Tom Hyde for their
help and guidance throughout my studies, Dr David Allen of E.On for his advice and
direction, Dr David Tanner for his support and collaboration in creep testing, Dr Geoff
West of Loughborough University for his help with material characterization, Keith
Dinsdale and Tom Buss for their help and advice. I would like to acknowledge the
support of The Energy Programme, which is a Research Councils UK cross council
initiative led by EPSRC and contributed to by ESRC, NERC, BBSRC and STFC, and
specifically the Supergen initiative (Grants GRIS86334/01 and EP/F029748) and the
following companies; Alstom Power Ltd., Doosan Babcock, E.On, National Physical
Laboratory, Praxair Surface Technologies Ltd, QinetiQ, Rolls-Royce pic, RWE
npower, Siemens Industrial Turbomachinery Ltd. and Tata Steel, for their valuable
contributions to the project.

1
Abstract

P92 steel is a high-alloy steel that has been specifically designed for
operating at high temperatures (600°C - 650°C) and has found wide use in
the power generation industry, particularly since 2005. For the successful
installation and use of this advanced steel, all aspects of its behaviour, in
terms of both metallurgy and in-service behaviour, must be investigated.
Investigating all the relevant material aspects is beyond the scope of a
single PhD, and so the Supergen consortium funds a number of projects
working on different material aspects. The purpose of this thesis is to
investigate, and seek a greater understanding of, the behaviour of welds in
P92 steel so that their in-service behaviour may be better understood
particularly the response of the material to post-weld heat treatments
(PWHT), the optimum weld consumable composition and the microstuctural
development during creep-rupture. This understanding has been achieved
through a combination of microstructural characterization, thermodynamic
modelling and mechanical testing.

Specifications for weld metals define a range of compositions;

thermodynamic modelling has enabled a better understanding of how the
composition affects the final microstructure of P92 weld metal (given that
this work is based upon thermodynamic predictions, the understanding
developed here is applicable to both parent and weld metal). Precipitation
strengthening is important to the creep resistance of P92 and the
modelling has revealed how precipitate levels vary based on composition.
Using this knowledge, quality checks on P92 used by industry can better
ensure the fitness for service of a material if an accurate composition is
known; furthermore, this understanding will enable manufacturers to
further tailor compositions to produce the strongest possible material.

Following welding with P92 fillers, post-weld heat treatment is carried out,
and there is a desire to perform this heat treatment close to the Al I
temperature of the materials involved. As such, it is important to
accurately know the Al temperature of the materials being heat treated. A
combination of thermodynamic modelling, experimental thermal analysis
and microstructural characterization was used to investigate the key
transformation of ferrite to austenite. This investigation focused on the

1 AI "" The temperature at which ferrite begins to transform to austenite on heating.

2
effect of composition on the transformation temperature, All and the rate
at which austenite could form during PWHT. An equation to predict the
Ael 2 temperature of P92 is produced and validated. The knowledge of how
composition affects the Al temperature is useful for both welds and parent
material, enabling the design and selection of P92 material that will not
undesirably transform during heat treatments. It is proposed that the
equation for Ael allows the determination of maximum safe heat treatment
temperatures and will reduce the likelihood of poor quality material
entering service. Experimental work has demonstrated that during PWHT
(or parent material tempering), equilibrium conditions are approached,
confirming that Ael should be used to determine maximum heat treatment
temperatures instead of the ACl 3 temperatures which are currently
employed.

Creep testing of three different weld consumables was carried out to

determine which had the best properties for use in service, and to
understand the microstructural features which controlled creep behaviour
of these weld metals. Creep testing of weld metal has indentified that 0-
ferrite causes early failure as the resulting precipitate-free zones (PFZs)
are creep weak. The presence of localized 8-ferrite is caused by an
inhomogeneous distribution of ferrite stabilizers, particularly tungsten
within the weld metal, resulting in greater stability of o-ferrite and its
retention in the weld. Using this knowledge, alloy specifications of weld
consumables and corresponding welding procedures can be improved to
ensure a homogeneous distribution of elements so that localized
weaknesses in a weld can be avoided. There is tentative evidence that
tungsten plays an important role in the creep ductility of P92 and that
variations in tungsten and silicon could lead to an optimization of creep
strength.

The outcomes of this thesis facilitate a better understanding of P92 parent

metal and welds and provide results that are immediately applicable and
useful to the power generation industry.

2 Ael = The theoretical temperature at which ferrite begins to transform to austenite on

heating using only thermodynamic considerations.
3 ACI = The temperature at which ferrite begins to transform to austenite during continuous
heating.

3
Table of Contents
1 Introduction ............................................................................... 6
2 Literature Review ........................................................................ 9
2.1 Steam-Generating Power Plant ............................................... 9
2.1.1 Advances in Boilers .......................................................... 11
2.1.2 Advances in steam temperature and pressure ...................... 11
2.1.3 Advances in materials ....................................................... 12
2.2 Principles of Creep .............................................................. 15
2.3 Development of Power Plant Steels ........................................ 20
2.3.1 Types of Heat Resistant Steels ........................................... 25
2.3.2 The Alloy Design Concept .................................................. 26
2.3.3 Alloying Elements and Microstructure in 9-12%Cr Steels ....... 28
2.3.4 Evolution of microstructure and material properties during creep
32
2.4 P92 Power Plant Steel .......................................................... 36
2.4.1 Compositions of P92 Parent Material ................................... 36
2.4.2 Microstructural Features of P92 Steel .................................. 38
2.4.3 Characteristics of Precipitates ............................................ 39
2.4.4 Mechanisms of transformation ........................................... 42
2.4.5 The a - y Transformation in 9-12%Cr Steels ........................ 46
2.5 Welds in Power Plant Steels .................................................. 53
2.5.1 Microstructure of Welds .................................................... 53
2.5.2 Post-Weld-Heat-Treatments in P92 ..................................... 57
2.5.3 Creep failure of 9-12%Cr steel welds .................................. 59
2.6 Summary .......................................................................... 61
3 Experimental Work .................................................................... 63
3.1 Material and specimen preparation ........................................ 63
3.1.1 Weld pad preparation ....................................................... 63
3.1.2 Specimen preparation ....................................................... 63
3.2 Characterisation Methods ..................................................... 67
3.2.1 Sample Preparation .......................................................... 67
3.2.2 Optical Microscopy ........................................................... 68
3.2.3 Micro-hardness Measurement ............................................ 68
3.2.4 Scanning Electron Microscopy (SEM) ................................... 68
3.2.5 Electron Back-Scatter Diffraction (EBSD) ............................. 69
3.2.6 Dilatometry ..................................................................... 70

4
 3.3 Modelling ........................................................................... 71
4 Results .................................................................................... 73
4.1 Weld Consumable Composition and Modelling ......................... 73
4.1.1 Weld Metal Phase Fraction Diagrams ................................... 74
4.1.2 Precipitate Phase Fraction Diagrams ................................... 76
4.1.3 The Effect of Element Variation on Precipitation .................... 81
4.2 Weld Pad Microstructural Characterization .............................. 83
4.2.1 Optical Microscopy ........................................................... 83
4.2.2 Hardness ........................................................................ 84
4.2.3 SEM Characterization ....................................................... 84
4.3 Austenite Formation During PWHT ......................................... 86
4.3.1 The Effect of Composition on A1 ......................................... 86
4.3.2 PWHT Temperature Dependence ........................................ 95
4.4 Weld Consumable Creep Testing ......................................... 103
4.4.1 Creep Test Results ......................................................... 103
4.4.2 Microstructure Following Creep ........................................ 106
5 Discussion ............................................................................. 114
5.1 Consumable Composition and Modelling ............................... 114
5.2 Weld Pad Microstructure .................................................... 119
5.3 The Effect of Composition on Ai .......................................... 120
5.4 PWHT Temperature Dependence ......................................... 127
5.5 Weld Metal Creep Testing ................................................... 133
6 Conclusions ............................................................................ 137
7 References ............................................................................. 142

5
1 Introduction
The constraints that are currently placed on power generation plant in
terms of environmental impact and economics have focussed attention on
the development of high efficiency, low emission systems. If thermal
efficiencies of generating plants can be increased, fuel can be conserved
(less fuel is required for a given power output) and emissions reduced
(lower fuel consumption means lower emissions of environmentally
damaging gases). An increase in the thermal efficiency of a power plant
can be most effectively achieved by increasing the temperature and, to a
lesser extent, the pressure of the steam entering the turbine. Most modern
steam power stations now in operation reach efficiencies of around 42%
with steam temperatures of up to 600°C and pressures of 25-30 MPa
(Bugge et al. 2006; Chew 2003).

Steam generating fossil fuel fired power stations make up the majority of
electrical power generation across the world; however they are inherently
inefficient. Energy is lost as heat has to be removed to ensure that the
metal, which makes up the majority of the power plant structure, remains
at a safe operating temperature, thereby reducing the thermal efficiency of
the system. Increasing the safe operating temperature of metals within the
power plant reduces the energy that is lost to cooling, thereby increasing
efficiency.

Conventional steam generating power plants currently fall into three

categories depending on their operating conditions;

1. Subcritical plants that make up most of the steam generating power

plants currently operating at temperatures lower than 600°C.
Subcritical means that there is a non-homogeneous mixture of
water and steam in the evaporator part of the boiler. In this case a
drum-type boiler is used because the steam needs to be separated
from water in the drum of the boiler before it is superheated and
fed into the turbine.
2. Supercritical plants operating beyond the critical pressure of steam
(255 bar) and at temperatures approaching 620°e. Supercritical
power stations are also known as "once-through plants" because
boiler water does not circulate multiple times through the steam

6
 drum until it has transformed to steam; rather the water transforms
to steam in the boiler waterwall due to the higher temperatures.
3. Research and development is ongoing to commercialize ultra-
supercritical steam plants. Their "clean coal" technology would push
efficiency rates higher and CO 2 and particulate emissions lower,
compared to supercritical steam plants. Operating parameters have
projected temperatures of 760°C and boiler feed discharge
pressures from 330 bar to 420 bar.

Increasing the efficiency is in part a material issue, as higher operating

efficiencies require materials that are able to operate at a higher
temperature. Metals are the materials currently able to operate at high
temperatures and withstand the stresses created by the steam they
transport. The problem with metals is that when they are held at
temperatures above a third of their melting temperature, they may suffer
from a time dependent deformation process called creep, which eventually
leads to failure.

Steels are the primary metal used for construction of power plants, in
particular the parts which transfer heat in the boiler and transport steam to
the turbines. For decades, efforts have been made to better design steels
that have improved creep resistance, achieved primarily through alloy
design and careful control of the microstructure. The alloying elements
strengthen the steels by solid solution strengthening, precipitate
strengthening and altering/controlling the development of the
microstructure over time. Parent steels generally exhibit good creep
resistance thanks to continuing improvements in material design, the pace
of which has increased rapidly thanks to computer modelling which reduces
the amount of experimental trial and error.

However, power plants are massive constructions made of many

components that must be joined together. The joining of metals in the
power plant is primarily done through welding. While the bulk material of
components (so-called 'parent material') are carefully designed to have a
homogeneous microstructure, welding of the large pipes introduces
localised inhomogeneous microstructures that alter the mechanical
properties of the structure as a whole.

The weld itself introduces a filler material that melts and fuses with the
parent material, but while the composition of the filler may be similar to

7
the parent metal, there will be elements present at different levels, along
with some additional elements, in order to control the microstructural
outcome of the complex multi-pass weld thermal cycle. The thermal cycle
associated with the welding process also affects the parent material
adjacent to the weld metal, often detrimentally altering its microstructure.
The thermal cycle also introduces residual stresses associated with the
expansion and contraction of the filler metal during melting and
solidification. Heat treatments following the welding process seek to reduce
residual stresses and restore the material to a microstructural state which
is acceptable for the application in which the structure is being utilised, but
these heat treatments have to be carefully controlled so as to prevent the
accidental formation of undesirable phases.

Invariably, the weldment becomes creep-weak compared to the parent,

and therefore will be the region most likely to fail first during service. The
steel P92 is currently one of the best performing creep-resistant high-
temperature commercially available steels. It is important to understand
how welds in this material will perform under creep and a full
understanding of this requires an examination of:

• The chemical composition of potential filler metals.

• The microstructure that the welding process produces in the various
zones of the weldment.
• The appropriate post-weld-heat-treatment for the weldment.
• The performance of the welds during creep.

8
2 Literatu re Review

2.1 Steam-Generating Power Plant

The Energy Information Admin istration (EIA) projects that total worldw ide
installed coal-fired generating capacity w ill approach 2000 GW by 2030, up
from 1119 GW in 2003. More than 61% of the projected new generating
capac ity is expected to be in China (546 GW), followed by the U.S .
(16.7%, 147 GW) and India (10.7%, 94 GW) (EIA 2006).The boiler is the
heart of a PC (pulverised-coal) power plant, burning fuel to provide the
steam that drives turbines to generate electricity. Technology
improvements in PC boilers and in other plant components have yielded
significant economies of scale along with improvements in efficiency ,
reliab ility, and environmental performance of the overall power plant . Th is
has contributed to significant cost reductions since the introduction of PC
plant technology (Paul et al. 2005). Figure 2.1 shows the cumulative
installed capacity of pulverized coal-fired plants in the world from 1921 to
2004 (Yeh and Rubin 2007).

- ｾ 900
U
"0"
0
. .,.
_
ｾ 800
700 ----- Worlu
U
..s..ｾ . . ｃ ---
ＭＧ UK

-. .-
ｾ - ,.. 600 -6- Gam:lny

--
ｾ 500
=
c..=:: 400 c:
-4--
Jnpan
Chi na

-=ｾ ........ｾ 300

-ｾ

ｾ Ｎ Ｎ ﾧ 100
,.. 200

-=c.. ｉ ｾ
0 4e, ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ
U ｏ ｾ ｏ ｏ ｎ ｾ ｏ ｾ ｏ ｏ ｎ ｾ ｏ ｾ ｏ ｏ ｎ ｾ ｏ ｾ ｏ ｏ ｎ ｾ ｏ ｾ
ｎ ｎ ｎ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｏ ｏ ｏ ｏ ｏ ｏ ｾ ｾ ｏ ｏ
ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｏ ｏ
- - - - - - - - - - - - - - - - - - - - ('I ('I

Yea.'
Figure 2 . 1 Worldwid e cumul ati v e capacity of PC coal-fired plants (Yeh a nd Rub in 200 7)

While the majority of new PC boilers installed worldw ide since 1990 have
been subcritical units employing 166 bar/538°C/538°C drum boilers (Kitto
1996), supercritical boiler technology, operating at higher temperature and
pressure, continues to be developed in Europe and Asia (primarily Japan).
More recently, several "ultra-supercritical" boilers-with even higher
temperature and pressure-were built in Europe and Japan, where higher
coal prices justified the higher cost of these more efficient plants (Figure
2.2) (Bugge et al. 2006; Kitto 1996). In the 19905, more efficient PC

9
plants using supercrit ical boiler technology ach ieved net plant efficienc ies
of 42-44% in Japan, Ge rmany, Denmark, Netherlands, and most recently
China (Yeh and Rubin 2007). Figure 2.3 shows the recent progress in PC
plant efficiency, ach ieved via higher steam pressure and temperatu re,
double reheat, and other design changes, albe it with an increase in capital
cost. Other studies note that advances in materials and process
components could allow ultra-supercrit ical boilers to ach ieve still higher
efficiency within a decade (Viswanathan et al. 2006) .

c 80
- World - ｾ ｵ ｢ ｣ ｩ ｩ｣ ｉ
CJ
eo:
Q. 70 ｾ World - ｾ ｵ ｰ ･ ｣ ｩ ｬ ｩ ｣ ｉ
eo:
U 60
"C
ｾ

--
ce
Vl
C ｾ
Co-'
-.
50
40
c0e JO
U
"C
ｾ
20
Ｎ ｾ
:..
ｾ
10
ｾ
Q..
:::: o
ｾ Ｇ｢ ""b ｾ h'\.. ｾ ｾ ｾ Ｇ ｢ (\b ｾ ｾＧ｜Ｌ ｾ ｾ
ｾ ｾ Ｍ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ
Year
Fig ure 2 .2 World PC coal - fired plan ts' annu al instal led capacity (i n GW/year) by ty pe of boile r
(Yeh and Rubin 2007)

45 % o A e lorev;lrkel 2 (OK) ｾ ｾ
0
• Bex b;lch II (D)
> ----
:c Ｔ Ｔ ｾ
ｾ ｔ ｡ ｣ ｨ ｩ ｢ ｡ ｮ Ｍ ｷ
ｾ
ｮ EPO OP)
orclj yllancl. va rket 3 (OK ) [] •
:c '-'
D Matsuura EPOC OP)
b. Lu beck (D) b.
n - M atsuura-2 OP) +
c 43%
CJ
X Ro toc k I ( 0 )
Ｎ ｾ • Hern weg 8 (N L )
CJ
• Arner 9 ( L )
E Ｔ ｾ + E bjerg ;lrk et ( OK ) X
ｾ

;Z
-Q,/

41%
+ St:l uclin ger 5 (0)
o Fyns arkel ( OK )
X A edore arket I (OK)
E3 Sludsru 3.4 (OK)
0

:+.
••
X
E3
40q{
1980 1985 1990 1995 2000 2005

Figure 2 .3 Recent prog ress in plant efficien cy of PC coal -f ired power pla nts in European
countries and Japan (Bugge et al. 2006)

10
Early coal-fired boilers typically employed fixed or moving grates on which
chunks of coal were burned to provide the heat needed to generate steam.
The introduction of PC technology, in which coal is pulverized into a fine
powder and injected into the furnace via burners, substantially increased
the surface area of the fuel and improved the speed and efficiency of
combustion. Major subsequent advances in PC boiler design came from
economies of scale together with the increased steam pressure and
temperature that became possible with the development of stronger
metals (such as "superalloy" steels) and other technology improvements
(Bugge et al. 2006). The resulting increase in boiler efficiency allowed
utilities to produce more electricity with less fuel, thereby reducing the
capital and operating costs. The following sections elaborate on the
technology advancements in PC boiler size, steam temperature, steam
pressure, and materials.

2.1.1 Advances in Boilers

The rapid unit cost reduction of new generating plants prior to the 1970s
has been mainly attributed to economies of scale in all power plant
components including the generator, turbine, and boiler. In the early
1900s, a SO MW plant (considered large at that time) housed five 10 MW
steam turbines and typically required 50-60 boilers to power the turbines.
By the 1920s, the introduction of PC, together with improvements in boiler
design that raised steam temperatures and boiler output, reduced
significantly the number of boilers per plant. The subsequent development
of single-boiler, single-turbine systems contributed to more rapid
improvements in thermal efficiency and unit cost reductions (Viswanathan
and Bakker 2001b).

2.1.2 Advances in steam temperature and pressure

Advancements in steam temperature and steam pressure have contributed
greatly to PC plant efficiency improvements. The maximum steam
temperature and pressure increased from about 260°C and 6.9 bar in 1900
to about 593°C and over 276 bar in the 1950s (Yeh and Rubin 2007). In
the early 1960s, the utility industry's move toward larger units was
accompanied by widespread adoption of supercritical boilers operating at
621°C and 310 bar. Of the nearly 11,000 MW in large units committed by
U.S. utilities in 1962 and 1963, 70% were designed for supercritical steam
pressure with either single or double reheat (Yeh and Rubin 2007). Reheat
refers to the process of passing steam that has already been through the

11
turbine back through the boiler to increase the thermal efficiency; this may
be done multiple times depending on the design of the boiler and the
turbine. During the late 1960s and early 1970s, boiler tubes on
supercritical units started to experience metal fatigue and creep, and scale
deposits from boiler walls induced greater corrosion and erosion damage in
the boiler, turbine nozzles, and other parts of the plant. As a result, the
availability of these plants dropped and they became more costly to
operate. The inability at that time to improve the metallurgy of boilers and
turbines led the utility industry to retreat from supercritical units to the
more reliable subcritical units (Bugge et al. 2006; Yeh and Rubin 2007).
Not until roughly 20 years later did utilities in Europe and Japan begin to
adopt improved supercritical units (Figure 2.3).

2.1.3 Advances in materials

In the mid-1930s, metallurgical progress made available superheater
tubing and turbine parts that allowed steam temperatures to be raised to
496°C, thus increasing plant thermal efficiency to 26% (Yeh and Rubin
2007). The subsequent development of superalloys that resisted metal
fatigue and cracking allowed engineers to design boilers for still higher
temperatures and pressures, culminating in the development of
supercritical boilers that began service in 1957 (Viswanathan et al. 2006).
The introduction in the 1950s of the 0.5CrMoV, 1.25CrO.5Mo and
2.25Cr1Mo ferritic steels allowed an increase in steam temperature from
about 450 0 C to 540-568°C with a corresponding increase in thermal
efficiency. For operations at even higher temperatures (i.e. above 568°C),
the use of new high strength austenitic steels (such as the German
17Cr13NiMoN) could have offered a solution but this was not feasible
because these steels have:

• high coefficient of thermal expansion

• low thermal conductivity
• higher cost than ferritics
• poor fatigue behaviour
• they are difficult to inspect by Non-Destructive Examination (NDE)
methods.

At that time, most engineers believed that the extra cost of special alloys
would be compensated for by the fuel savings from more efficient

12
supercritical boilers. However, the sustained material problems noted
above led to lower availability and higher maintenance costs, which ended
the use of supercritical units in the U.S. by the early 1980s. Attention
therefore turned to the 9Cr1Mo and 12Cr steels in the search for ferritic
steels for superheater and steam pipework components that could
withstand operation at up to 600°C and beyond. Originally developed for
the chemical process industries, 9Cr1Mo bainitic steels evolved into the
ASTM P9 that has been used for creep resistant tubing in power plants at
temperatures up to around 540°C. Steel P91 is a martensitic version of
grade P9 which contains micro-alloying additions of vanadium, niobium,
and nitrogen. P91 was originally developed during the 1970s by Oak Ridge
National Laboratory (ORNL) in the USA. This steel was approved by ASTM
Standard A213 as T91 for tubing in 1983. In 1984, ASTM Standard A335
approved Grade 91 material as piping steel and on-site testing began in
the United States, Canada, and Europe. P92 was originally developed by
Nippon Steel as NF616, being approved by ASTM as standard A213 T92 for
tubing in 1994 and by ASTM as standard A335 P92 for piping steel in 1994.
The material is included in EN 10216-2 under the designation
X10CrWMoVNb9-2. P92 was developed to give a material with higher
strength than P91. Exploitation of P92 steel began in the late 1990s with a
small number of installations. Use of the material is increasing now that
further experience has been gained in the fabrication and use of this steel.
Grade 92 steel variations are designated as T92 for tubing, P92 for piping,
and F92 for forgings. Tubing refers to relatively thin-walled components
(typically 6 - 8mm wall thickness) that are used for heat transfer
purposes; piping refers to relatively thick-walled components (typically 25
mm or greater wall thickness) used for the transfer of superheated steam
to the turbine.

The only unit using P92 in the USA is the main-steam pipe in the 530 MW,
Weston 4 power plant of Wisconsin Public Service Corp. which began
operating on June 30, 2008 (steam temperature 585 0 C) although several
such units (for example main-steam pipe in Comanche 3 and John W. Turk
(610°C» are now being planned as a result of the success of units
operating in Europe and Japan since the late 1990s (Yeh and Rubin 2007).
According to Vallourec & Mannesmann Tubes (V&M TUBES), a leading
supplier of P92 steel, between 1990 and 2005, they delivered
approximately 1000 tonnes of T/P92 all over the world. Since 2005 they

13
have delivered a further 79,000 tonnes, mainly to new projects in
Germany and Asia, with 24 large power generating units (400 MW+) using
V&M P92 in Europe, two in the USA, 14 in Asia (excluding China) and 77 in
China. Other manufacturers report similar increases in demand between
2000 and 2012 with China representing the largest uptake for P92 due to
the large number of new fossil-fired plants being constructed.

New materials, such as nickel-based superalloys and further developed

steels, are expected to increase steam temperature beyond 760°C and
pressures up to 345 bar, which is expected to increase plant efficiency
beyond 45% within a decade (Bugge et al. 2006).

14
2.2 Principles of Creep

As has already been mentioned, creep failure of components is a major

barrier to the development of super-critical and USC power plants. Creep
describes the material deformation occurring as a result of long term
exposure to stress at a level that is lower than the yield stress or ultimate
strength of the material and at a temperature greater than 0.4 of the
absolute melting temperature T m. This creep strain is thermally activated
and is a function of stress, temperature, time and material properties and
eventually leads to failure. Creep is a time dependent deformation process
where failure does not occur suddenly upon loading; rather the material
irreversibly strains over a period of time until failure occurs (Evans 1984;
Evans and Wishire 1985).

Tertiary
Creep

--
•ｾ
1
Steady.. ute
Creep
"1-------
-- ......................................... _ .................................. .

Primary
Creep

Initial Elastic Strain

Time (5)

Figure 2.4 Schematic of a high temperature creep curve at constant stress and temperature
(Evans and Wishire 1985)

Figure 2.4 shows a typical creep curve demonstrating the variation of

creep strain with time when a material is subjected to a constant stress Ga.
The three stages during creep are the primary (transient), secondary
(steady state) and tertiary (rupture) stages. During primary creep, after
the initial strain on loading, the creep rate gradually decreases to a
constant value. The strain is a combination of elastic and plastic
deformation, and the increase in dislocation density results in a decreased
creep rate. In secondary creep, the rate of strain hardening is balanced by
the rate of dislocation recovery, so the creep strain rate is constant and
reaches the minimum creep rate, hence the steady-state. Creep here can
be governed by two creep mechanisms, namely dislocation creep and

15
diffusional creep. In the tertiary stage, the strain rate sharply increases
and causes the formation of creep damage voids within the material. This
voiding causes an increase in the local stress within the component which
further accelerates the strain eventually leading to the final rupture. As
creep is also dependent upon material properties, the primary stage may
sometimes be absent or inverted, and the extent of the tertiary stage may
be limited in brittle materials and extensive in ductile materials (Evans
1984).

me:

ＺＺＺｾ
ｾ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ
time

Figure 2.5 Effect of temperature and stress on the strain vs time creep behaviour (Evans
1984)

The time from the first application of stress to final rupture is termed the
creep failure life at a particular stress and temperature, while the total
strain along the stress axis after creep rupture is the total elongation.
Figure 2.5 demonstrates the importance of stress and temperature on the
creep curve. As stress and temperature increase, so does the minimum
creep rate; accordingly, the creep failure life decreases.

The relationship between creep strain rate (t) and temperature (T) is
expressed using the Arrhenius law, as shown in Equation 2.1 (Evans and
Wishire 1993), where A is a material-dependent constant, Qc is the
activation energy for creep which is constant for a given creep mechanism,
R is the universal gas constant and T is the absolute temperature in Kelvin.

Equation 2.1

16
Norton's law, Equation 2.2, describes the relationship between steady state
creep rate Eand stress a, where n is the creep index.

Equation 2.2

The parameter of greatest significance in controlling the creep life is the

steady state creep rate E (see Figure 2.5 where E is the slope of the
secondary stage), due to its extended duration. Life calculations use the
steady state creep rate as E varies with both stress and temperature as
shown in Equation 2.3.

E = u(a) vCT)

Equation 2.3

By substituting veT) from Equation 2.1 and u(a) from Equation 2.2, we get
the power law relationship (Equation 2.4) of E expressed as a function of a
and T (Evans and Wishire 1993).

Equation 2.4

Materials may creep by several mechanisms at elevated temperatures

depending on the stress and temperature. Diffusion creep is controlled by
volume diffusion (Nabarro-Herring creep) or grain boundary diffusion
(Coble creep); dislocation creep is controlled by volume diffusion (high
temperature power law creep) or by pipe diffusion (low temperature power
law creep). The creep rates of all the mechanisms are represented by the
following general equation (Equation 2.5):

Equation 2.5

where Eo is a material constant characteristic of the mechanism and

material, d is the grain size, p is the grain size exponent, G is the shear
modulus and 0 is the diffusion coefficient relevant to the mechanism. The
values of n, p and 0 are typical of each creep mechanism and are listed in
Table 2.1.

17
 Table 2.1 Stress exponent n, grain size exponent p and diffusion mechanism for each creep
mechanism (Abe 2008)

Deformation mechanism n p D
Dislocation Creep
Low temperature power law creep 5-7 0 Op
High temperature power law creep 3-5 0 0 1

Diffusion creep
Coble creep 3 Ogb
Nabarro-Herring creep 2 DI

0" Op and Ogb indicate the diffusion mechanisms of lattice, dislocation pipe
and grain boundary diffusion, respectively. The four creep mechanisms are
independent of each other and the creep strain produced by each
mechanism contributes additively to the total creep strain; therefore, the
mechanism which results in the highest value of E dominates. The different
creep mechanisms can be represented using a deformation mechanism
map as shown in figure 2.6. The diffusion creep mechanisms (Coble and
Nabarro-Herring creep) appear in the lowest stress range while the
dislocation creep mechanisms (high and low temperature power law creep)
appear in the intermediate stress range. A dislocation glide mechanism
without the aid of diffusion takes over the role of plastic deformation above
the athermal yield stress (aa, the stress at which dislocations are able to
move without diffusion). Creep rates of both dislocation mechanisms are
independent of grain size, but the diffusion creep mechanisms see an
increase in creep rate with decreasing grain size. The steel P92 is designed
to operate at high temperatures and stresses where high temperature
power law creep is the dominant mechanism.

18
 '0-1 ------------- \dea\ ｾ ｴ Ｌ Ｎ ･ Ｂ Ｙ ｴ ｨ - - - - - - - - - - - - -

ｾ Dislocation glide
rnVI
...
CI)

ｾ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ ｡ Ｎ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ
1ii
low temperature
"
Ｎ ｾ
CI)
power law creep High temperature
co power law creep
E
...
o
Z

Coble creep Nabarro-Herring

creep

Ｑ Ｐ Ｍ Ｗ ｾ Ｍ Ｍ ｾ Ｍ Ｍ ｾ Ｍ Ｍ ｾ Ｍ Ｍ ｾ Ｍ Ｍ ｾ ｾ ｾ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ ｾ Ｍ Ｍ ｾ Ｍ Ｍ ｾ
o 0.2 0.4 0.6 0.8 1
Normalized temperature, T/Tm

Figure 2.6 Schematic drawing of a deformation mechanism map. Tm is the materials melting
temperature (Abe 2008)

19
2.3 Development of Power Plant Steels
The constraints that are currently placed on power generation plant in
terms of environmental impact and economics have focussed attention on
the development of high efficiency, low emission systems.

The process of steam generation to drive turbines is the same for

supercritical and ultra-supercritical plants with the difference coming from
the operating conditions (and improved thermal efficiency) of ultra-
supercritical plants, while the advantage that both have over subcritical
plants is the absence of a steam drum as the high pressure and
temperature allows water to rapidly turn to steam rather than being re-
circulated as required in a subcritical plant. Ultra-supercritical pressure
refers to steam pressure exceeding conventional supercritical pressure
steam conditions, i.e., 24 MPa with a superheater outlet temperature of
538°C, but can be more accurately defined as representing turbine inlet
steam conditions of at least 24 MPa for main steam pressure and a
temperature of at least 565°C for main steam and reheat steam (see
Figure 2.7 for a process diagram of the supercritical/USC power plant).
Such elevated steam conditions increase plant efficiency and thereby
contribute to improved resource and energy conservation, as we" as
environmental protection. Power plants incorporating these steam
conditions had already been constructed in the US and Europe by the late
1950s. While the improvements in ultra-supercritical pressure power plant
are not obvious merely from steam conditions and plant specifications,
there is in fact a very major technological difference between modern
plants and those of a generation ago. Specifically, the older plants were
designed and constructed as base load plants without the load-adjusting
functions considered essential for the plants being used today. Base-load
refers to plants that are almost continuously generating electricity at or
near their design conditions such as steam pressure and power output of
the turbine and generator. Load-adjusting (or plant cycling) refers to a
mode of operation where power plants are frequently required to either
start-up/shut-down or adjust the output power to meet demand in power
grids that increasingly give priority to energy generated by renewables.
Plant cycling is a problem for older plants as they were typically designed
to experience 1-2 full cycles in a year but are now experiencing hundreds
of cycles per year creating problems with thermal fatigue. Thus, elevated
pressure and temperature requirements formerly met through the use of

20
large quantities of austenitic steels for heavy, thick-walled components
such as headers, piping, and turbine equipment in the older generation of
ultra-supercritical pressure power plants now have potential problems from
thermal stress due to the larger thermal expansion coefficient of the
austenitic steels. For steam power plants to be capable of responding to
changes in electricity demand, or for plants undergoing frequent start and
stop cycles, Sliding pressure operation is a method of controlling the MW
output by a power station by reducing the pressure of steam going to the
turbines. This is done by changing the heat input in the boiler, often in a
short space of time, thus introducing frequent changes in temperature;
this type of operation causes problems if using austenitic steels. It is
preferable to use ferritic steels with their smaller coefficients of thermal
expansion for heavy, thick-walled components in order to reduce thermal
stresses during this mode of operation. Accordingly, the heat resistance
capability of ferritic steels is the major determinant of steam conditions
(Masuyama 2001).

21
 Conventional Steam - Supercritical Process Diagram

,....
•...ｾ ｉ .........

-,

"•
J
f

Figure 2 .7 Process diagrams of supercritical steam generating power plant (http ://www.flowserve .com/lndustries/ Power-Generation/ Conventional-Steam) .

22
 tCtO.5Uo

2.25Cr1Mo

ICr1Mo

t2CrtMoV

Pit

Ettt 111111111111111111111111111111111111111111111111111 1111 11111111111111111111111111111111111111111 III

Pl2

I
AlSI Ut

AlSl347

500 525 550 575 825

maximum op....tln" t.mp ....tu ... In ·C, b ..ed on •
100 000 h .v....".
atr1laa ruptu ... atr1lngth of 100
MP.
Figure 2.8 Stress rupture strengths of the currently used and the newly developed power
station steels (Ennis and Czyrska-Filemonowicz 2003)

The next generation of steam power plants should be capable of operating

with steam at 625-650 o C, to enable thermal efficiencies of around 45% to
be achieved. The increasing operating temperatures and pressures impose
increasingly stringent requirements on the materials of construction
(Viswanathan and Bakker 2001a, b; Viswanathan et al. 2006). High
chromium ferritic/martensitic steels such as P91, E911 and P92 (Shibli
2008; Vaillant et al. 2008) have been developed for advanced steam power
plants, in particular those sections requiring thick walled pipe to transport
supercritical steam as shown by the thick red pipe sections in Figure 2.7.
In Figure 2.8, the stress rupture strengths of the currently used and the
new power station steels are compared on the basis of the maximum
service temperature for a 100,000 h stress rupture strength of 100 MPa
(Ennis and Czyrska-Filemonowicz 2003). It can be seen that the maximum
service temperature increases with increasing complexity of the steel
composition and the more highly alloyed steels have sufficient 100,000 h
stress rupture strength to be considered for application at temperatures in
excess of 600°c' Indeed, the new high chromium steels have similar stress
rupture strengths to austenitic stainless steels. There are several reasons
for the reluctance to use the austenitic steels; obviously the increased cost
of the steel with the high chromium and nickel contents is a disadvantage
but there are technical problems because the thermal expansion coefficient
of austenitic materials is at least 50% higher than that of ferritic steels.

23
- - - - - - - - - --

This means that care has to be taken during cooling and heating to avoid
excessive thermal stresses that can lead to fatigue failures. Steels with a
ferritic/martensite microstructure have good thermal conductivity, low
coefficient of thermal expansion and a high resistance to thermal shock
(Cerjak and Letofsky 1998) which make them more attractive for high
temperature applications then austenitic steel.

.-

24
2.3.1 Types of Heat Resistant Steels
Various kinds of heat resistant steels are separately used in power plants
according to their specific purposes. They are generally classified into
ferritic steels and austenitic steels, but are then further sub-divided.
Ferritic steels include carbon steels (C-Mn, etc.), low alloy steels
(O.S%Mo,2.2S%Cr-1 %Mo), intermediate alloy steels (S-10%Cr) and high
alloy steels (12% Cr martensitic steels and 12-18%Cr ferritic steels of the
AISI400 series); steels with 9-12%Cr are also usually considered a class
due to their martensitic microstructure. Austenitic steels include 18%Cr-
8%Ni steels and 25%Cr-20%Ni steels of the AISI300 series, 21 %Cr-
32%Ni steels such as Alloy 800H, and Cr-Mn steels of the AISI200 series
(Masuyama 2001). Figure 2.9 shows the chemical compositions of typical
heat resistant steels used under stresses in the Fe-Cr-Ni ternary phase
diagram (Masuyama 2001). Ferritic steels generally do not contain nickel,
and, because steels with chromium compositions of 2%, 9% and 12% are
particularly high in strength, they are widely used. Among austenitic
steels, materials in commercial use are positioned along the boundary
between the full y phase and the y phase containing a and/or o. The full y
phase steels contain relatively high Ni content, the high cost of which is
typically offset by high creep strength. In contrast, the y phase steels with
a and/or 0, although less costly, require some improvement to elevate the
creep strength. In order to facilitate a better understanding of the different
types of steels, Figure 2.10 shows schematically illustrated microstructures
of ferritic and austenitic heat resistant materials. In both cases, material
upgrades are illustrated from left to right, and the precipitates appearing
therein change according to type.

25
 Cr 0 : Ferritic
6. : Meta-Stable
Austenitic
o : Stable Austenitic

aＫ ｃ ｉ ｟ Ｎ Ｎ ａ ｉ ｲ Ｍ Ｊ Ｍ ｾ Ｍ Ｍ ｴ Ｍ ｾ Ｍ Ｍ Ｚ ｾ Ｍ Ｍ ［ Ｉ ｴ Ｍ ｾ
a+CI+T

l8C
12Cr
Ｙ ｃ ｲ Ｎ Ｎ Ｚ Ｎ Ｌ Ｎ ｶ ｾ ｒ

Ｒ ｃ ｲ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ___ｾ ｾ ｾ ｾ ___ｾ ｾ ｾ ｾ
ｾ M
Figure 2.9 Compositions of heat resistant steels in Fe-Cr-Ni ternary phase diagram at 800°C
(Masuyama 2001)

(Fe. Mn),eNa

18Cr8NI 18Cr8NT1, Nb
CSteei CrMoV (+Cu)(+AO.11) (o.4C25Cr20Ni)(+Nb, 11)
.) Fen1tIc b) Austenitic
Figure 2. 10 Schematic illustration of microstructures of ferritic and austenitic steels
(Masuyama 2001)

2.3.2 The Alloy Design Concept

Heat resistant steels for practical application must be designed by
considering their service conditions and environments, and by examining
their various properties. However, when alloy design is performed based
on modification of existing steels, oxidation and corrosion resistance as
well as their general material properties are expected to be nearly
equivalent to those of the original materials. Hence, chemical compositions
and heat treatment conditions are examined in particular consideration of
creep strength improvement. Figure 2.11 shows the concept of alloy design

26
for heat resistant steels to improve creep strength through the
modification of existing steels. For ferritic heat resistant steels, research on
9-12%Cr system steels is fairly advanced, and approaches for the
improvement of creep strength through solution strengthening,
precipitation strengthening and microstructural stabilization have been
adopted. These techniques are also applicable for the modification of Cr-
Mo low alloy steels as well. On the other hand, chemical compositions of
austenitic steels can be largely classified into the four categories shown in
the figure, and solution strengthening and precipitation strengthening are
designed specifically for each of these categories. 18%Cr-8%Ni steels
based on Type 304 steels include Type 316 steels solution-strengthened
through the addition of Mo, as well as Type 321 steels and Type 347 steels
precipitation-strengthened through the addition of Ti or Nb. However,
these materials were originally developed for chemical equipment, placing
emphasis on corrosion reSistance, but were not designed from the
standpoint of creep strengthening. Accordingly, the further enhancement
of precipitation strengthening by means of "under-stabilizing" carbon
and/or composition design for improved creep strength is used. 15%Cr-
15%Ni or 21 %Cr-30%Ni steels with full y phase structure are capable of
high creep strength in is the as-received condition (with no special need for
alloying for precipitation strengthening of solution strengthening), although
they are costly because of their high nickel content. Steels with chromium
levels of 20% and over are likely to have excellent oxidation and corrosion
resistance, but a costly nickel content of at least 30% is required to
maintain a full V structure. Nevertheless, low-cost, high strength, highly
corrosion-resistant austenitic steel can be designed by adding nitrogen of
about 0.2%, which reduces the nickel content necessary to maintain a full
y structure as nitrogen is also a strong austenite stabiliser, and by
combining the strengthening mechanisms as described above (Masuyama
2001).

27
 Mo. WAdcftion

:
I
........ -....
: :
_....., ........_-,
:._ ....... .
I
_ Co :
leu
V. Nb AdcitIon
Hell StnIn8th 9-12"Cr Steel

ｌ Ｚ Ｚ Ｎ ｾ Ｎ ｾ .. j ｌ ｾ Ｎ ｾ ｪ ｌｾ ｪ (NbVSO.l")
S 0.1"
SolutIon PrecIpjtatIon
StrS'Cthellll. S1reIIC\tlelIiI.
T)08304 r -IType318(Mo) I I Crc.tblde I
r.Ic:ro ｾ ｲ ｷ of
StabIIlzrc T,.".321m). T1C I-- !Ill, n. B
ｾ Ｗ Ｈ ｎ ｢ Ｉ Ｎ NlC ｾ ｾ ｓ ｴ ｮ ｉ ｲ ｗ ｴ ｨ
Under Stllbilzilc 1"'Cr-a»f Steell
'--<T1+Nb)/C ｾ
Alloy ｾ for CnIep
Cu. Mel, W AdcttIon

Ｈ
Ｑ
Ｑ
Ｕ
ｾ
ｾ
ｬ
Ｑ
ｭ
Ｕ
｣
ｾ
ｎ )
T)08 17- I 4CuMo
I SI8II
•
I Creep RaiItInI
NAdIttIon

+ ｾ
Type 310

T)08 AIIoylOOH

Figure 2.11 General concept of alloy design for heat resistant steels (Masuyama 2001)
Cl'wp/ComJIIon
RIIIIatInt SI8II
- I..ow-CoIt
O'eep/CamIIIon
RllllantStMI

Alloying Elements and Microstructure in 9-12%Cr Steels

Although the overall microstructures of the 9-12%Cr steels are quite

similar for most compositions that have been developed, it is the
compositional changes made over the years that have resulted in the
improved properties as shown in Figure 2.12. The effects of the various
alloying additions are outlin&d in the following discussion.

10'h Creeo Rupbre ｳ ｾ at 8OO'C

I 35 MPa i
I
60 MPa I 100 MPa 140MPa 180 MPa
I -C +W !
!I 2.25Crl Mo I :+V .1 2.25Crl MaV I -Me! ttl! ·IUserl.8WVNI I
i(ASMET22)
, ST8A24
!+WeI (ASMET23)
STBA24J I
i .
! I
Ａ｛ ｾ Ｙ ｃ ｾ ｲ ｾ ｉ ｾ ｍ ｯ ｾ ｾ Ｚ Ａ Ｚ Ａ Ａ Ａ Ｎ ｾ Ｙ ｃ ｲ Ｒ ｍ ｯ ｾ ｾ ｖ ｎ ｢ ｾ ｾ
i (ASME TV)
: ｓ ｾ tV
(NFA 48213) iｾ I Ｍ ｍ ･ Ａ
i ｾ Ｋ ｎ ｢ ｾ ｾ ｾ ｾ ｉ ｍ ｯ ｾ ｖ ｎ ｉ ｾ ｾ ｃ Ｎ ｾ ｖ Ｎ ｾ ｎ ｢ Ａ Ｎ Ｎ Ａ Ｚ Ｈ Ｉ ｰ ｴ ｴ ｾ ｮ ｵ ｾ ｾ ｛ Ａ Ｙ ｃ ｲ ｾ ｬ ｍ ｡ ｾ ｖ ｎ ｢ ｾ Ｋ ｗ ｲＺＺＹｑｏＮＵｍｯ］］ＺＭＭＺＭＱＮｉｎｴｖｍｊ］］ｾ
I I I Ｈ ｾ ｩ ｬ ｬ Ｉ Ｈ ｾ Ｚ Ｒ Ｉ
I 12Cr ! +WeI I
I 51410)

!
i
I
Figure 2.12 The development of ferritic steels for power plants (Masuyama 1999)

28
Chromium (Cr)
Chromium is a ferrite stabilising element that is generally added to steels
for oxidation and corrosion resistance and is the main alloying element in
P92 steel, where it also contributes to solid solution strengthening.
However, the main strengthening effect is achieved by the precipitation of
chromium-rich carbides from the solid-solution, which can impede the
movement of dislocations and grain boundaries to increase creep strength
of the steel. Chromium reacts with carbon to form carbides; the chromium
rich carbides usually encountered in P92 steels are M23 C6 which form
during tempering and remain present throughout elevated temperature
exposure. In steels containing nitrogen, chromium rich M2X (Cr2N) can also
form under some conditions (Klueh 2005).

Molybdenum (Mo)
Molybdenum is a ferrite stabiliser and improves the creep properties of P92
steel by solid solution hardening, but can also be detrimental as it
accelerates the growth of M23 C6 carbides (Maruyama et al. 2001). In the
tempered condition, molybdenum is in solid solution and provides solid-
solution strengthening. However, the amount of the molybdenum must be
limited to avoid the formation of is-ferrite and the intermetallic Laves-
phase. Klueh (Klueh 2005) stated that for structural steels being used at
elevated temperatures, the MO eq (where MO eq = Mo + 0.5 W [wt%]) should
not exceed 1%. If the steel has a MO eq greater than 1% when operating in
the 600-650 o C temperature range, the tendency for the intermetallic
Laves-phase to form is increased, resulting in the removal of molybdenum
from solid solution and a reduction in solid solution strengthening (Klueh
2005).

Tungsten (W)
Tungsten is used in P92 steels as a substitute for the higher level of
molybdenum seen in steels from earlier generations such as P91. Like
molybdenum, tungsten is a solid solution strengthener and its addition has
been found to be effective in increasing high temperature creep strength.
The strengthening effect of tungsten can be lowered dramatically during
long time exposure due to the formation of the intermetallic Laves-phase
during long time exposure at high temperatures. The limit of tungsten can
also be defined using the aforementioned MO eq equation (Klueh 2005). It
has been found that according to the tungsten precipitation behaviour, the

29
optimum tungsten content is around l.Bwt%; tungsten contents higher
than 2wt% have been seen to accelerate the intermetallic Laves phase
formation if the chromium content stays at 9 wt% (Hasegawa et al. 2000).

Vanadium and Niobium (V, Nb)

Vanadium and niobium are added to steels in order to promote the
formation of fine, dispersed and thermally stable MX particles within
grains. In P92, the MX preCipitates (V or Nb = M, C or N = X) can be in the
form of carbides MC, nitrides MN, or carbonitrides M(C, N) (Sourmail
2001). Although the density of MX particles is low in the common grades of
high chromium ferritic steel, the particles are necessary in maintaining a
fine subgrain structure in the steels over extended service at high
temperature and to improve creep strength of the steel (Abe et al. 2004;
Abe et al. 2007; Sawada et al. 2001a, 2003; Yin and Jung 2009). A
comparison (Maruyama et al. 2001) was made among stress-rupture
curves of three ferritic steels: 9Cr-1.BW-OAMo-VNb steel (O.16V, 0.06Nb,
O.llC and 0.036N) with tungsten and MX, 9Cr-1Mo-VNb steel (O.22V,
0.09Nb, 0.10C and O.OSlN) with MX but without tungsten, and 9Cr-2W
steel (O.OBC) with tungsten but without MX particles. Although the 9Cr-
lMo-VNb steel contained larger amounts of vanadium, niobium and
nitrogen than the 9Cr-1.BW-OAMo-VNb steel, the former was inferior to
the latter in creep rupture strength because of the absence of tungsten.
However, the rupture life of 9Cr-2W steel was substantially shorter than
those of the 9Cr-1Mo-VNb steel in spite of the tungsten addition, and the
important role of the MX was seen as a major factor in determining these
properties.

Nitrogen and Aluminium (N, AI)

Nitrogen plays an important role in increasing the creep life of steel: it
allows the precipitation of strengthening nitrides in the form of MX, while
nitrogen in solid solution has a greater strengthening effect than carbon
(Sourmail 2001). If the P92 steel contains aluminium, it is also possible for
AIN to form. The AIN removes nitrogen and thus reduces the amount of
vanadium nitride particles in the system leading to a reduction in creep
strength (Gomez et al. 2009). As such, 9-12%Cr steels should have a very
low aluminium content or high N:AI ratio. In this way, aluminium does not
combine with a Significant part of the nitrogen and the precipitated volume
of vanadium nitrides is not significantly diminished.

30
Boron and Phosphorus (B, Pl
Boron segregates to austenite grain boundaries during cooling after
austenitization, During tempering, boron diffuses into prior austenite grains
and is incorporated in growing M23 C6 precipitates. Boron incorporated in
M23 C6 reduces their coarsening rate by reducing the interfacial energy of
this phase (Abe et al. 2004; Czyrska-Filemonowicz et al. 2003;
Golpayegani et al. 2003). Phosphorus can also segregate to the surface of
M23 C6 , and a small amount of phosphorus has been found in the
intermetallic Laves-phase (Ennis et al. 1997).

Carbon
Carbon is a strong austenite stabiliser with a relatively large solubility in
austenite, but a small solubility in ferrite. It is mainly added to P92 for the
formation of the strengthening phases M23 C6 and MX (Ennis et al. 1998).

Nickel and Manganese (Ni, Mn)

Nickel and manganese are austenite stabilisers, lowering the ferrite to
austenite transformation temperature At. The main reason for adding them
is to ensure 100% austenite formation (without (5-ferrite) during the
austenitization treatment and by lowering the temperature for
transformation from austenite to ferrite, ensuring a 100% martensite
structure when cooled (Knezevic et al. 2008). Nickel results in an increase
in toughness of the steel, but at concentrations higher than 0.4 at%, it is
responsible for significant coarsening of M23 C6 (Yin and Faulkner 2003).
Manganese decreases the coarsening rate of M23 C6 at low concentrations;
however, when its concentration reaches about 0.5 at%, it actually
contributes to the coarsening of M23 C6 (Helis et al. 2009). Manganese is
also present in the steel in order to collocate with sulfur, and to increase
the solubility of nitrogen in austenite (Sourmail 2001).

Microstructures of 9-12%Cr steels currently being developed or already

commercially available generally consist of a single phase of tempered
martenSite, with just a few exceptions. High dislocation densities are
generally found in this type of structure, being principally influenced by the
tempering temperature. The dislocation density remains high when the
tempering temperature is low, as in the case of turbine rotor steels. Figure
2.13 shows a representative microstructure observed through optical
microscopy and transmission electron microscopy (TEM) of a typical 9-

31
 12%Cr heat resistant steel. The tempered martensite is composed of
numerous laths, and chromium carbides such as M23 C6 precipitate along
the lath boundaries and along the prior-austenite grain boundaries. Fine
MX carbonitrides of (V, Nb)(C, N) coherently precipitate on the ferrite
matrix in laths, and dislocation networks are formed along the lath
boundaries or the sub-grain boundaries. It is considered that the creep
strength of 9-12%Cr steels is closely associated with the stabilization of
MX carbonitrides and the dislocation structures, and it is inferred that in
tungsten - containing steels, strength rises by suppressing recovery and
recrystallization of martensitic structures during creep.

ｾ Ｚ Ｚ Ｚ ］ Ｚ Ｚ Ｚ Ｚ Ｚ Ｚ Ｚ Ｍ Ｍ ｲ Prior-.,.. Boundary

M23CS and Laves

Precipitation on
Prior-.,.. and lath
Boundaries

/ '
I I •

MX Precipitation Dislocation Network

within Lath on Lath Boundary
a) LOM b) TEM (Illustrated)
Figure 2.13 Typical microstructure of tempered martenstitc 9- 12% Cr steel (Masuyama 2001)

2.3.3 Evolution of microstructure and material properties during

creep

2.3.3.1 Precipitation
The high strength of 9-12%Cr steels is due to a chemical composition
designed to produce 100% austenite during austenitization by balancing
the austenite and ferrite stabilizers, while forming 100% martensite during
the quench (Klueh 2005). The creep strength of the material is also
determined by elements which contribute to solid solution strengthening
and precipitation strengthening; in the latest steels, solid solution
strengthening is provided by tungsten while the main precipitates are
M23 C6 and MX (Anderson et al. 2003). Dislocation motion is an important
mechanism of creep and the preCipitates dispersed in the material can act
as obstacles for dislocation motion. For the dislocations to move around

32
 the precipitates, a higher stress must be applied compared to the stress it
takes to move through the bulk material. Creep deformation proceeds after
the dislocations have overcome the additional stress imposed by the
precipitates, and therefore materials with large numbers of precipitates
normally possess higher creep strength.

The precipitates M 23 C6 and MX are described in detail in section 2.4.3.

M23 C6 is the dominant precipitate in P92 steel and improves creep strength
in the short term, though the strengthening effect is reduced at later
stages due to coarsening of the precipitates. The evolution of particle mean
size d is described by the Ostwald ripening law (Orlova et al. 1998) as
shown in Equation 2.6 where do is the initial particle size at t=O and k is
the particle growth rate.

d3 = ､ ｾ + k. t
Equation 2.6

Creep strength is diminished by precipitate coarsening as it limits the effect

that grain boundary precipitates have on subgrain growth. Large grain
boundary precipitates can also act as the initial locations of creep
cavitation, resulting in the onset of tertiary stage creep.

The minor precipitate MX that forms in the subgrain interior does not
coarsen rapidly during creep. Klueh (Klueh et al. 2005) performed tests on
9-12%Cr steels which showed the presence of elements such as vanadium,
niobium, and nitrogen gave an advantage in creep strength over steels
without these elements where chromium and carbon levels were similar.
Previous authors have also shown that MX precipitates based on
molybdenum, niobium, vanadium, titanium or tungsten are resistant to
coarsening (Nawrocki et al. 2000; Tamura et al. 2003). While M23 C6 at
grain boundaries coarsen during creep with reducing numbers and greater
inter-particle spacing, MX precipitates in the subgrain interior increase
primarily in number with time at a given temperature (Klueh et al. 2005).
In contrast, M23 C6 precipitates within martensite laths decrease in size and
number, indicating that they dissolve and re-precipitate at the grain
boundary (Anderson et al. 2003). Ongoing MX precipitation offsets some of
the loss in creep strength due to M23 C6 coarsening, but eventually the
reduction overwhelms the strengthening effect of MX, leading to creep
cavitation and material deformation.

33
 2.3.3.2 Recovery of martensite and dislocations
Well developed martensite laths (subgrains) with low dislocation density in
their interiors are important characteristic features of long-term creep
exposed materials (Czyrska-Filemonowicz et al. 2003). As shown in Figure
2.14, with the progress of creep deformation, the subgrain width increases
and free dislocation density decreases .

• : Mod.ger-IMo 0: TAF650
e
-:::1

...
.c
:2
ｾ
923K 98.1MPa

-
.c
j

10-1

--
M
I
8
>.
'r;;
c
-8
c
-g
.9

...-
( I)

Q
10120 1 2
Time /10 '5
Figure 2.14 Change of lath width and dislocation density in lath interior as a function of creep
duration (Sawada et al. 1999)

Sawada (Sawada et al. 1999) also looked at the effect of tungsten on lath
recovery and found that the growth of lath width and the annihilation of
dislocations in the lath interiors are slower in tungsten containing steel.
Accumulation of creep strain is suppressed because of the slow recovery of
its lath structure which results in the lower creep rate and the higher creep
rupture strength of the tungsten containing steels.

High dislocation density in tempered martensite structures also contributes

to creep strength. During creep, dislocations move by glide and climb
resulting in the deformation of material. If the stress is high enough,
dislocations can cut or loop obstacles such as precipitates, grain
boundaries and other dislocations. However, coarsening of precipitates and
martensite lath boundaries during creep (such as increasing spacing or
subgrain width) make dislocation movement easier.

34
2.3.3.3 Hardness
Hardness can give an indication of the creep strength of materials. Its well
established relationships with other parameters such as yield stress,
elongation and fatigue make hardness evaluation useful in obtaining plastic
flow and strain hardening parameters associated with creep of materials
(Kohlhofer and Penny 1996).

It has been shown that hardness reduces with increasing temperature and
thermal exposure duration. The effect of stress on hardness degradation
has also been investigated (Watanabe et al. 2004) where it was shown
that a stress-aged specimen has a lower hardness than a solely thermally
aged specimen, indicating that stress has a significant effect on hardness
degradation during creep. Both carbide coarsening and dislocation recovery
have been regarded as the sources of the hardness difference.

35
 Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｎ ｾ Ｍ Ｍ Ｍ _._---

2.4 P92 Power Plant Steel

2.4.1 Compositions of P92 Parent Material

P92 is a ferritic 9Cr-1.75W-O.5Mo steel micro-alloyed with vanadium and
niobium, and with controlled boron and nitrogen content. In modern super-
critical steam power plants, P92 allows higher operating temperatures and
pressures to be employed and therefore higher efficiencies can be
achieved. European experience has shown that P92 is suitable for
components outside the boiler (i.e. pipes and headers) at steam
temperatures up to 625°C (Richardot et al. 2000a).

The chemical composition of grade 92 specified by ASME standards is listed

in Table 2.2, along with various reported compositions of P92 from the
literature. It can be seen from the difference between the literature
average and the ASTM mid-point that some elements are consistently used
at either the upper or lower limits of the ASTM specification. Manganese is
routinely used at an average of 48% above the ASTM mid-point while
nickel is 31 % below. As is described in section 2.3.3, both manganese and
nickel have the same general effect, so manganese may be used in greater
quantities in preference to nickel as it is less expensive. Aluminium is
generally kept very low, if it is present at all, as it forms AIN which
removes the amount of nitrogen available for the precipitation of useful
phases (see 2.3.3). Boron is also generally present in low quantities,
although in some cases, it is not reported to be present even though the
ASTM states a minimum of O.OOlwt%. The greatest standard deviation is
seen in the alloying element which is present in highest concentration
(chromium), possibly affecting the amount of the beneficial phase M23 C6
present and thereby causing a variation in the mechanical properties of
different P92 steels.

36
 Table 2.2 ASTM composition of P92 and reported compositions of P92 from the literature

Name Composition (wt%)

C Mn P S Si Cr W Mo v Nb N B AI Ni Fe
ASTM min 0.070 0.030 8.500 1.500 0.300 0.150 0.040 0.030 0.001 BAL
ASTM max 0.130 0.600 0.020 0.010 0.500 9.500 2.000 0.600 0.250 0.090 0.070 0.006 0.040 0.400 BAL
ASTM average 0.100 0.315 0.010 0.005 0.250 9.000 1.750 0.450 0.200 0.065 0.050 0.004 0.020 0.200 BAL
M. Hattestrand et al. (1998) 0.110 0.460 0.008 0.040 8.960 1.840 0.470 0.200 0.070 0.050 0.001 0.060 BAL
P. J. Ennis et al. (1997) 0.120 9.100 1.800 0.500 0.200 0.060 0.040 BAL
P. J. Ennis et al. (2000) 0.120 0.470 0.011 0.006 0.020 9.070 1.780 0.460 0.190 0.063 0.043 0.003 0.002 0.060 BAL
K. Sawada et al. (2001) 0.110 0.450 0.012 0.003 0.100 8.820 1.870 0.470 0.190 0.060 0.047 0.002 0.010 0.170 BAL
Y.F. Yin et al. (2003) 0.110 0.460 0.008 0.040 8.960 1.840 0.470 0.200 0.070 0.050 0.001 0.060 BAL
K. Sawada et al. (2002) 0.110 0.450 0.012 0.003 0.100 8.820 1.870 0.470 0.190 0.060 0.047 0.002 0.010 0.170 BAL
K. Rodak et al. (2003) 0.094 0.460 0.210 8.840 1.720 0.470 0.210 0.070 0.260 BAL
Wachter et al. (1995) 0.124 0.470 0.011 0.006 0.020 9.070 1.780 0.460 0.190 0.063 0.043 0.003 0.002 0.060 BAL
B S Lim et al. (2005) 0.100 0.480 0.017 0.006 0.220 9.110 1.750 0.470 0.180 0.056 0.041 0.002 0.010 0.180 BAL
C S Jeong et al. (2005) 0.100 0.480 0.017 0.006 0.220 9.110 1.710 0.470 0.180 0.056 0.0405 0.0029 0.010 0.180 BAL
R Sugiura et al. (2007) 0.090 0.470 0.010 0.001 0.160 8.720 1.870 0.450 0.210 0.060 0.050 0.002 0.180 BAL
B Kim et al. (2008) 0.100 0.480 0.017 0.006 0.220 9.110 1.750 0.470 0.180 0.056 0.041 0.002 0.010 BAL
Literature average 0.107 0.466 0.012 0.005 0.123 8.974 1.798 0.469 0.193 0.062 0.045 0.002 0.008 0.138 BAL
Percentage difference between ASTM & 7% 48% 23% -8% -51% 0% 3% 4% -3 %
-5% -10% -40% -61% -31%
Literature average

Standard Deviation (literature) 0.011 0.011 0.004 0.002 0.086 0.141 0.059 0.012 0.011 0.005 0.004 0.001 0.004 0.072

37
 2.4.2 Microstructural Features of P92 Steel
Tempered martensite steels are hardened by dislocations, solutes and
precip itates. In P92, normalization produces a martensitic structure with a
high dislocation density within the martensite laths. The dislocations are
introduced during martensitic transformation which involves severe plastic
deformation. The martensitic transformation and subsequent annealing
fo rms subgrains within the prior austenite grains. The subgrains are
bounded by low- angle boundaries (dislocation networks) as well as high-
angle boundaries (Chilukuru et al. 2009). Precipitates at the subgrain
boundaries are essential in stabilizing the subgrain structure against
coarsening and recrystallization (Abe 2004), while precipitation in the bulk
increases the deformation resistance of the subgrain interiors (Yoshizawa
et al. 2009) . Figure 2.15 shows the martensitic structure of a 9Cr steel
after tempering, with the distribution of the two main precipitates M23 C6
and MX indicated (Abe et al. 2007).

!hI

Figure 2. 15 Illu stra tion of martensitic 9Cr steel after tempering : (a) subgrain structure; (b)
distribution of M23 C6 and MX (Abe et al. 2007)

Dispersion strengthening of the alloy is inversely proportional to the mean

inter- particle distance and proportional to their volume fraction (Abe et al.
2007). Therefore, optimal creep strength of the steel can be obtained by a
dispersion of fine and thermally stable precipitate particles combined with
enough sol id solution strengthening. Zielinska-Lipiec (Zielinska-Lipiec and
Czyrska -Filemonowicz 2007) studied the effect of normalization and
tempering on the microstructure of P92 (see Table 2.2 for composition)
and the results of their work can be seen in Table 2.3. If the optimal creep
strength is given by having fine precipitates, then their work shows that a
temper of 715°C (the lowest temperature that they employed) gave the
finest particles at all normalization temperatures for which they have data.

38
Table 2.3 Influence of a heat treatment on microstructural parameters of P92 steel (Zielinska-
Lipiec and Czyrska-Filemonowicz 2007)

Temp. Temp. Dislocation Mean Mean Mean

Normalization Tempering °C density,10 14 m-2 subgrain size, diameter diameter MX,
°C ｾ ｭ M23 C6 , nm nm

970 715 8.72 ± 1.21

970 775 0.38 ± 0.1

1070 715 8.98 ± 1.15 0.37 ± 0.1 72 ± 16 14 ± 3

1070 775 7.46 ± 0.94 0.42 ± 0.09 81.6 ± 28 16± 3

1070 835 2.26 ± 0.58 0.50 ± 0.1 83.7 ± 22 16 ±4

1145 715 0.37 ± 0.08 53 ± 9 12 ± 3

1145 775 0.42 ± 0.09 68 ± 19 15 ±4

1145 835 2.12 ± 0.52 0.58 ± 0.12 81 ± 27 12 ± 3

2.4.3 Characteristics of Precipitates

The main types of strengthening precipitates in P92 in the as-received
heat-treated conditions are MZ3 C6 and MX, with Laves phase precipitating
at long term high temperature exposure.

2.4.3.1. M 23 C6
MZ3 C6 is the principal precipitate found in P92 in the quenched and
tempered condition. A variety of work has been done to characterise the
M23 C6 produced by normalization and tempering in P92 (Ennis et al. 2000;
Zielinska-Lipiec and Czyrska-Filemonowicz 2007). It has been found that
when normalized at 1070 0 C, any M23 C6 is completely dissolved, but then
forms rapidly during tempering. The carbides precipitate mainly on prior
austenite grain boundaries and subgrain boundaries, as was illustrated in
Figure 2.15. The size of the M 23 C6 precipitates in P92 is typically less than
100 nm (the actual size depending upon the details of the heat treatment,
as shown in Table 2.3). Compared to previous generations of 9-12%Cr
steels such as P91, in which M23 C6 is typically between 200 and 380 nm
(Anderson et al. 2003), the MZ3 C6 in P92 is much finer. M23 C6 contributes to
material strengthening by retarding subgrain growth so the finer size of
the precipitates in P92 increases its strength relative to previous materials.
However, after long term exposure at high temperatures, the precipitates
coarsen and this reduces the creep strength.

39
The M23 C6 has a face-centred-cubic (fcc) structure (Sourmail 2001) and is
typically Cr23C6 with iron, molybdenum and nickel being able to partially
substitute for the chromium. The addition of boron to P92 helps to reduce
the coarsening of M23 C6. During air cooling after normalising, boron
segregates to prior austenite grain boundaries through a non-equilibrium
mechanism. Non-equilibrium segregation means that boron atoms have
been dragged to the boundary by vacancies, not that there exists a binding
energy between boron atoms and the boundary. During cooling from the
normalisation temperature, boron starts to diffuse back into the grain on a
time scale of one or two minutes. Together with the carbon, the boron then
gets incorporated in the growing M23 C6 (which could now be referred to as
M23 (C,B}6, and depending on the distance between the prior austenite
grain boundary and the precipitate, more or less boron is available during
the growth process that takes place during tempering. For precipitates
close to or at the prior austenite grain boundaries, a higher B:C ratio than
in the steel as a whole is possible. As boron diffusion and precipitate
growth occur over the same time scale, the boron does not affect the
number density of M23 C6 as nucleation occurs before any boron has
diffused to the precipitate. The boron has been reported to decrease the
interfacial energy between M23 (C,B}6 and the matrix, thus reducing the
driving force for coarsening (Abe et al. 2004; Czyrska-Filemonowicz et al.
2003; Golpayegani et al. 2003).

2.4.3.2 MX
The MX precipitate is also found in P92 steel. It has a face-centred-cubic
(fcc) structure and precipitates mainly in the subgrain interior (see Figure
2.15). It can take a number of compositions where M = Nb and/or V, X =
C and/or N, and the precipitates can be referred to as carbides, nitrides
and carbonitrides (Sourmail 2001). There are three main types of MX
precipitates that can be found in P92. The first type is NbX which are
spherical and relatively coarse. They remain after austenitsation and are
randomly dispersed in the steel, due to the niobium content being higher
than the limits of the mutual solubility of niobium and carbon and / or
nitrogen in austenite at the austenisation treatment temperatures used for
P92 (Ennis et al. 1998; Maruyama et al. 2001; Zielinska-Lipiec and
Czyrska-Filemonowicz 2007). Secondary MX phases form during tempering
and these take the form of NbX and VX. These MX particles are fine and
are distributed uniformly in the subgrains. The majority of the secondary

40
 MX are vanadium-rich and are between 12-16 nm in size (Zielinska-Lipiec
and Czyrska-Filemonowicz 2007) (see Table 2.3), and compared to the
primary MX in P92 which is typically 100 nm (Anderson et al. 2003), these
secondary MX precipitates are much finer. The third type of MX precipitate
is called a V-wing complex. These occur when the large Nb(C, N) that was
undissolved during austenitization act as nucleation sites for the VN that
are produced during tempering. The VN tend to form at the edges of the
larger Nb(C, N) giving the appearance of a wing hence the term V-wing
complex (Hu et al. 2010).

The fine MX particles produced during tempering improve the creep

strength of P92 by pinning dislocations (Abe et al. 2007; Klueh et al.
2005), while the coarse NbX that remain at austenitization are able to limit
the growth of austenite grains and ensure that the recrystallisation of
martensitic laths during tempering does not result in a loss of creep
strength (Anderson et al. 2003).

2.4.3.3 Laves Phase

Precipitation of intermetallic Laves phase occurs during creep exposure or
isothermal aging at temperatures around 600°C. It typically has a chemical
composition (Fe, Crh(Mo, W). The addition of molybdenum and tungsten
to P92 steel increases the creep strength of the matrix through solid
solution strengthening and, as such, the effect of the Laves phase
precipitation on long-term creep strength has been debated as it promotes
the depletion of molybdenum and tungsten from solid solution (Hosoi et al.
1986a; Hosoi et al. 1986b; Kunimitsu et al. 1991; Miyahara et al. 1995).
Whilst Laves phase has been generally considered detrimental for the
creep strength of 9-12% Cr steels because of the reduction in solid
solution strengthening, it has been argued that they can contribute to the
increase of the creep strength by precipitation hardening under certain
circumstances (Hald 1996). More recently, it has been generally argued
that fine Laves phase may contribute to the creep strength (Oimmler et al.
2003; Hald 1996; Hald and Korcakova 2003; Korcakova et al. 2001). There
are two mechanisms for the nucleation and growth of Laves phase, one of
which is that Laves phase precipitates alone on martensite lath boundaries
and that such precipitates are coherent with one grain but grow into the
adjacent grain which has no rational orientation relationship with them.
The other one is that Laves phases are formed in the regions adjacent to
M23 C6 particles, and then they grow at the expense of the chromium-rich

41
M23 C6 carbides in close vicinity. It has been established that the former of
these is the dominant formation mechanism for Laves phase as Laves
phases are observed to preferentially locate on the prior austenite grain
boundaries and the martensite lath boundaries (Cui et al. 2010).

2.4.4 Mechanisms of transformation

The atomic arrangement in a crystal can be altered (Figure 2.16) either by
breaking all the bonds and rearranging the atoms into an alternative
pattern (reconstructive transformation), or by homogeneously deforming
the original pattern into a new crystal structure (displacive
transformation). In the displacive mechanism, the change in crystal
structure also alters the macroscopic shape of the sample when the latter
is not constrained. The shape deformation during constrained
transformation is accommodated by a combination of elastic and plastic
strains in the surrounding matrix. The product phase grows in the form of
thin plates to minimise the strains. The atoms are displaced into their new
positions in a coordinated motion. Displacive transformations can therefore
occur at temperatures where diffusion is sluggish. Some solutes may be
forced into the product phase, a phenomenon known as solute trapping.
Both the trapping of atoms and the strains make displacive
transformations less favourable from a thermodynamic point of view. It is
the diffusion of atoms that leads to the new crystal structure during a
reconstructive transformation. The flow of matter is sufficient to avoid any
shear components of the shape deformation, leaving only the effects of
volume change. This diffusion is necessary even when transformation
occurs in pure iron (Bhadeshia 1999). In alloys, the diffusion process may
also lead to the redistribution of solutes between the phases in a manner
consistent with a reduction in the overall free energy. Figure 2.17 is a
summary of the main transformations that occur in steels.

42
 · O OO UOU

DisplacivQ

ｲ Ｎ Ｚ Ｍ Ｍ ］ ｏ Ｍ ］ Ｍ Ｚ Ｚｏ Ｚ Ｍ Ｍ ］ ｾ ｏ Ｚ Ｚ Ｍ Ｚ Ｇ Ｂ ｬ ｾ ｦｦｲＭ･Ｍ･Ｍ＼＼Ｚｈｈｾ＾ＭｩＭＭＬ
000 . 0 . 0 o 0 H 0 • 0 Interlace
0 . 00000 ｏ Ｎ ｏ ｏ ｏ
000 0000 00000 . 0
00 . 000 . 00 . 000 0
000 0000 000 0 0 . 0
000000 .
00 ","", 0 . 0
ｏ Ｎ ｏ ｏ ｏ oo o o e o
o oooo e o . o . oo . e
oo e oooo oo e o e oc
ｯ ｯ ｯ ｯ ｯ ･ ｯ ｾ " . 00000
ｾ oO He o
U e o .
• 0 Interface
Recunst ructive .. _I
OO H OOO
ｏ ｏ ｾ ｏ ｏ ｏ
OOCOOOO
0000000
oo ooo e o
Figure 2.16 Main mechanisms of transformation : parent crystal contains two kinds of atoms ;
figures on right represent partially transformed sa mples with parent and product unit cells
outlined in bold; transformations are unconstrained in thi s illustration (Bhadeshia 1999)

RECONSTRUCTIVE DISPLACIVE
Dlflusion 01 all atoms during Invariant plane strain Shape
nucleation and growth. deformation with large shear
Sluggish below about 850 K. component.
No iron or substitutional
solute diffusion .
Thin plate shape .

WI DMANSTATTEN
--1 ALLOTRIOMORPHIC
FERRITE ) FERRITE
Carbon diffusion during -

H )
IDIOMORPHIC para equilibrium nucleation &
growth .
FERRITE

BAINITE & ACICULAR

MASSIVE FERRITE
- No change In bulk
FERRITE
Carbon diffusion during f-
composition . paraequilibrium nucleation. No
diffusion during growth .

PEARLITE MARTENSITE
'---
Cooperative growth of D lffus lonless
-
ferrite & cementite. nucleation & growth .

Figure 2.17 Summary of essential characteristics of solid state transformations in steels

(Bhadeshia 1999)

43
The only strain that cannot then be cancelled by diffusion is the volume
change due to the difference in densities of the parent and product phases
- the strain due to reconstructive transformations in steels is therefore an
isotropic volume change. In contrast, displacive transformations typically
involve an invariant plane strain shape deformation with a large shear
parallel to the invariant plane and a dilatation normal to the plane (the
invariant plane is often referred to as the habit plane). They do not involve
the diffusion of iron atoms or substitutional solutes. Widmanstatten ferrite,
acicular ferrite, bainite and martensite are all products of displacive
transformations (Figure 2.17) (Bhadeshia and Honeycombe 2006). Here
the movement of iron and substitutional solutes occurs in a coordinated
manner, leading to a well defined and reproducible crystallographic
relationship between the parent and product phases. It is emphasised that
while displacive transformations in steels are associated with a volume
change, this change is not isotropic but occurs as a dilatation normal to the
habit plane which remains macroscopically undistorted. Also, the volume
strain is typically 0.03, which is much smaller than the shear strain which
is typically 0.26 (Bhadeshia 2004). The different features of reconstructive
and displacive transformations are illustrated in Figure 2.18.

isotropic expansion general invariant-plane

strain
ｾ sｾ
........•... ｾｽｾＭＭＭＭＬｾＭＮＭＭＭＭＭＭＭＭＺ
I ,
I I
I I
I I
I ,
I ,
, I
, I

I I

1....--.--.............. ..". ..". ..",. ..",...

". . . . ,...
- before transformation
- •--- after transformation
Figure 2.18 Nature of reconstructive transformations (left) and displacive transformations
(right) from austenite on cooling: for bainite and martensite 5 0.22·0.26 and Zi 0.02· = =
0.03 (Bhadeshia 2004)

The volume changes that occur in steels as they are heated and cooled can
be measured by dilatometry, where the change in length of an unloaded
specimen is measured as a function of temperature. Dilatometry is one of
the most powerful techniques for the study of solid - solid phase
transformations in steels, because it permits the real time monitoring of
the evolution of transformations in terms of dimensional changes occurring

44
in the sample by application of a thermal cycle. The applicability of
dilatometry in phase transformation research is due to the change of the
specific volume of a sample during a phase transformation. This technique
is widely used to study the transformation behavior of steels during
continuous heating, cooling, and isothermal holding. By recording the
transformations taking place over a range of conditions, it is possible to
present the results in a graphical form, which shows the formation
temperatures of microstructural constituents that may be obtained for a
given cooling or heating condition. Figure 2.19 shows such an experiment
(Leblond et al. 1986) - the upper straight line represents the expansion of
the body centred cubic phase (ferrite, bainite, martensite) and the lower
line that of austenite (y). Data at locations between the upper and lower
lines correspond to the co-existence of the parent and product phases. The
transformations occurred at different temperatures upon heating and
cooling. The transformation temperature is a function of the cooling rate,
steel composition and austenite grain size. The measured coefficient of
thermal expansion is larger for austenite ("'23 X 10-6 K- 1) than for ferrite
("'15 x 10-6 K- 1 ) (Francis et al. 2007). As a consequence, the volume
change due to transformation is greater during cooling than during
heating. The volume expansion due to the transformation of austenite can
partly compensate for thermal contraction strains arising as a welded joint
cools. Each grain of austenite can in general transform into 24
crystallographic variants of bainite or martensite (Bhadeshia and
Honeycombe 2006). When all of these form, the effect on a macroscopic
scale is that the shear strains average to zero. The volume strain cannot
be cancelled in this way since it is always positive with respect to the
sample frame, but because of the large number of variants that form, the
dilatation observed macroscopically appears isotropic, even though that
associated with an individual plate is not (Figure 2.18) (Francis et al.
2007).

45
 a-y
transformation

12.5 .
,

10

7.5

2.5

200 400 600 800 1000

Temperature (OC)
Figure 2.19 Dilatometric diagram of AS08 class 3 reactor pressure vessel steel heated at 30 K
s-l and cooled at 2 K s-l (Leblond et al. 1986)

2.4.5 The a - V Transformation in 9-12D/oCr Steels

Pure iron exists in two crystal forms at atmospheric pressure; one is a
body-centred cubic (bcc, a-iron) form that remains stable from low
temperatures up to 910°C (the Ae3 temperature); beyond this
temperature, it transforms to a face-centred cubic (fcc, V-iron) form. The a
to V transformation during heating is accompanied by a volume change of
approximately 3% (resulting in a corresponding linear expansion /
contraction of 1%), which is associated with a significant contraction on
the heating dilatometric curve (Figure 2.19). A study of the constitution
and structure of steels must start with the iron-carbon equilibrium
diagram, shown in Figure 2.20. Many of the basic features of this system
influence the behaviour of even the most complex alloy steels. There are
several temperatures or critical points in the iron-carbon equilibrium
diagram that are important, both from the basic and from the practical
point of view. Firstly, there is the Ael temperature at which the eutectoid
reaction occurs, which is 723°C in the binary Fe-C diagram. Secondly,
there is the Ae3 temperature when the final a-iron transforms to V-iron in
heating. For pure iron, this occurs at 910°C, but the temperature of the
transformation start is progressively lowered by the addition of carbon. At
temperatures between Ael and Ae3 the microstructure consists of a mix of

46
a-iron and V-iron. The Ael and Ae3 are easily detected experimentally by
dilatometry during heating or cooling cycles, but some hysteresis is
observed. Consequently, three values for each point can be obtained: (i)
Ac values are those measured on heating through the transformation; (ii)
Ar values are those measured on cooling through the transformation; (iii)
Ae values are theoretical values representing the transformation under
equilibrium conditions. It should be emphasized that the Ac and the Ar
values are sensitive to the rates of heating and cooling, as well as to the
presence of alloying elements.

Carbon (Atomic %)
2 4 6 8 10 12 14 161820
3000
I I'
I
600
A Sｐ ｾ ｳ ･ • UQi td
B Liquid 0 2800
1500
6 Phase"
1 400
ｾ ｊ ｾ I--- ｣ ..･ ｭ LIQuid
･ｮｴ ｾ 2600

ｾ Ｈ ｦ Ｄ
N Delta Phase ....... ｾ ｳ ｴ ･ ｮ ｯ Ｘ Ｂ Ｇ Ｍ Ｍ
• Austenite .. Liquid 2400
300
ｾ I
200

100
Austenite

A"", y
'O'(f

·E
,,
'"
Austenite
.C F
,,
2200

20001iil
000
G ｉ Ｏ ｾ / ,,
.. Cementite
.. Ledeburite +-
, 1Ｘ Ｐ Ｐ ｾ
900 V ,, 1600 ."J
Ｈ ｾ ａＬ Austenite :
.. Cementite '
,,
Curie ｉ ｾＡ ｊ ｴ ｾ -.-.:.:.;;;; . . . . 5/ , , 1400
Point
700
- ［ Ｍ Ｇ Ｍ ｾ ａ
-,, , ,
1200
A" I
'0 ,
600 , ' -;--
,u
uPhss0 .....
PeMite :.t : Pearlite "I 1000
Ｚｾ
ｾ ｾ Ｍ
• Ferrite Pearlite.
,9. Cementite - +.Cementite
200 ' :2
, 0
' • Transformed
' Ledeburite , 0
Hi
:Y : I Ｚｾ
300
100 ,:i ,, 200
0 ,,w ;w 100
0

ｾ Ｚ
3.0 4.0 5.0
Iron : Carbon (Weight ｾ Ｎ Ｉ
..o '.
Hypoeutectold ｾ Hypereutectoid -oo:
Steels 0,;- Cast Irons -----

Figure 2.20 The Iron-Carbon phase diagram (Bhadeshia and Honeycombe 2006)

The great difference in carbon solubility between V- and a-iron (Bhadeshia

and Honeycombe 2006) normally requires the rejection of carbon from the
a- phase on cooling from the v- phase. The transformation of V to a occurs
via a eutectoid reaction that plays a dominant role in heat treatment. The
eutectoid temperature is 723°C and the eutectoid composition is about 0.8
wt%C. On slow cooling of alloys containing less than 0.80 wt%C,
proeutectoid ferrite is formed from austenite in the range 910-723°C with
enrichment of the residual austenite in carbon. At 723°C, the remaining
austenite, now containing 0.8 wt% carbon, transforms to pearlite, a
lamellar mixture of ferrite and iron carbide (cementite) . In austenite with

47
0.80-2.06 wt%C, on slow cooling in the temperature interval from 1147°C
to 723°C, cementite first forms progressively depleting the austenite in
carbon. At 723°C, the austenite, containing 0.8 wt%C, transforms to
pearlite (Bhadeshia and Honeycombe 2006). Ferrite, cementite, and
pearlite are thus the principal constituents of the microstructure of plain
carbon steels as they are subjected to relatively slow cooling rates to avoid
the formation of metastable phases such as martensite and/or bainite.

Rapid quenching of austenite to room temperature often results in the

formation of martensite, a very hard structure in which the carbon,
formerly in solid solution in the austenite, remains in solution in the new
phase. Unlike ferrite or pearlite, martensite forms by a sudden shear
process in the austenite lattice that is not accompanied by atomic diffusion,
as described in the previous section. The martensite reaction in steels
occurs during cooling in a temperature range that can be precisely defined
for particular steels. This transformation begins at a martensite start
temperature, Ms, which can vary over a wide temperature range from as
high as 500°C to well below room temperature, depending on the
concentration of y-stabilizing alloying elements in the steel. Once Ms is
reached, further transformation takes place during cooling until the
reaction ceases at the Mf temperature. At this temperature, all the
austenite should have transformed to martensite, but frequently in
practice, a small portion of the austenite does not transform; this is called
retained austenite. Large volume fractions of austenite are retained in
some highly alloyed steels, where the Mf temperature is well below room
temperature. To favour the martensitic reaction over other competing
reactions, it is usually necessary for the steel to be rapidly cooled so that
the rate of cooling is sufficient to suppress the higher temperature
diffusion-controlled ferrite and pearlite formation reactions, as well as
other intermediate reactions such as the formation of bainite. The fact that
the ferrite and pearlite reactions are essentially high temperature
transformations occurring between 720°C and 550°C and that the
formation of martensite is a low temperature reaction, reveals that there is
a wide range of intermediate temperatures in which neither of these
phases forms. This is the region in which fine aggregates of ferrite plates
(or laths) and cementite particles are formed. The generic term for this
intermediate structure is bainite. Bainite also occurs during continuous
cooling at rates too fast for ferrite and pearlite to form, yet not rapid
enough to produce martensite (Garda de Andres et al. 2002).

48
2.4.5.1 Continuous Heating Transformations
It is the allotropic nature of iron, and its solubility for carbon that makes
possible the large range of transformations, microstructures, and
properties of plain carbon steels. The temperatures at which the
transformation of austenite commences and finishes at different rates of
cooling are of importance for planning and designing many industrial
processes. Moreover, there are other transformations, which cause the
steel to revert to the austenitic condition, in which the knowledge of the
transformation temperatures during heating is vital.

l1L fLo

f ACh
I
,
I
I
I
I
I I h
I
I I
I I
Ilinh:
I
I I I
I I I
I I ¥1CgIOb l I
o<+•• c glOb.,}' : ':
oc+Cglob
,
I
I
I
I
I:
I

I I
､ Ｈ ｦ ｾ ｌ ｊ ｌ ｯ Ｉ Ｏ ､ ｴ

f
-
Figure 2.21 Diagram of the heating dilatometric response of O-+Y transformation and the
carbide dissolution process of martensitic 9-12%Cr steels (Garcia de Andres et al. 2002)

A schematic diagram showing curves of relative change of length as a

function of temperature (6L/Lo =f (T» and its corresponding derivative
(d(6L/Lo)/dt =f (T» are shown in Figure 2.21. This diagram illustrates the
contraction that takes place during the ferrite-to-austenite transformation
by continuous heating of a martensitic stainless steel with an initial
microstructure of ferrite plus globular carbides (M 23 C6 ). These dilatometric
curves, aside from the a to y transformation between ACt and AC3
temperatures, develop a nonlinear length variation associated with the
carbide dissolution process. During heating at temperatures above AC3,

49
there is a continuous increase of slope in the curve ｾ ｬ ｪ ｌ ｯ =f (T) that
corresponds to the progressive dissolution of the carbides in austenite. The
point of inflexion of this section displayed as a maximum in the ､ Ｈ ｾ ｬ ｪ ｌ ｯ Ｉ Ｏ ､ ｴ

=f (T) curve corresponds to the ACe temperature or the temperature at

which the carbide dissolution process ends. Since the presence of carbides
in the as-quenched microstructure of these steels has a decisive effect on
its creep-rupture properties, the study of the carbide dissolution process is
important for the optimization of these materials. In this sense,
dilatometric analysis plays an important role in the detection of the
temperature at which the carbide dissolution process reaches completion
during continuous heating (Garda de Andres et al. 2002). Once the
carbides have been dissolved, significant concentration gradients (in
carbon and other alloying elements) exist throughout the austenite phase.
As the temperature to which the steel has been heated increases, these
concentration gradients are reduced by diffusion, reaching the
homogeneity at the ACh temperature. As shown in Figure 2.21, dilatometric
analysis enables the determination of this temperature at the point where
the ｾ ｬ ｪ ｌ ｯ = f (T) curve starts to display again a linear tendency
characteristic in the thermal expansion of a homogeneous phase.

The key transformation pOints during continuous heating can be

summarised as:

• ACl = The temperature at which the ferrite-to-austenite

transformation begins during continuous heating.
• AC3 = The temperature at which the ferrite-to-austenite
transformation ends during continuous heating.
• ACe = The temperature at which the carbide dissolution process
ends during continuous heating.
• ACh = The temperature at which the microstructure becomes a
single homogeneous phase during continuous heating.

2.4.5.2 Continuous Cooling Transformations

Examples of experimental critical temperatures for the non-isothermal
decomposition of austenite into ferrite and pearlite are displayed on the
dilatometric curve of a carbon manganese steel (Fe-0.2C-1.1Mn-0.34Si)
in Figure 2.22. The Ar3 and Arl critical temperatures correspond,
respectively, to the start and finish temperatures of the nonisothermal

so
 decomposition of austenite into ferrite plus pearlite. The specific austenite-
to-pearlite transformation may not show a clear indication of the
transformation start in many instances. This is not due to lack of
resolution, but to a masking of the individual reactions by the continued
growth of the other phase.

0.010 Austenite
0.009

-=
..c 0.008
ｾ

Go;
.,,;j
0.007

--= 0.006
Go;
ｾ
0.005
=
eo!!
.c
U 0.004

-
Go;
.:: 0.003
.!!
Go;
cz::: 0.002
Ferrite and Pearlite
0.001
0.000
o 200 300 400 500 600 700 800 900
Temperature I DC
Figure 2.22 Cooling dilatometric curve of a carbon manganese steel (Fe- 0.20C-1.1Mn-
0.34Si) obtained at a cooling of 1 K S·1 (Garda de Andres et ai. 2002)

Figure 2.23a shows the dilatometric curve of a low carbon manganese steel
(Fe -0.07C-1.S6Mn-0.41Si) after rapid cooling (234 K S-l) (Garda de
Andres et al. 2002). Dilatometric analysis indicates that two different
transformations take place during cooling in the steel. Metallographic
examination shows that the final microstructure obtained after cooling at a
rate of 234 K S-l is composed of bainite and martensite (Figure 2.23b).
Since bainite is formed at higher temperatures than martenSite, the first
volume expansion displayed on the curve corresponds to the
transformation of austenite into bainite, whereas the volume expansion
shown at the lower temperature corresponds to the transformation of
austenite into martensite. The bainite (Bs) and martensite (Ms)
transformation start temperatures are, respectively, the temperatures at
which the thermal contraction shown by the dilatometric curve deviates
from linearity due to the volume expansion associated to the bainite
transformation first and the martensite transformation later, and when

Sl
transformation is completed (Mf), it returns to linerarity (Bhadeshia and
Honeycombe 2006).

(a)
0.0 10
0.009
0.008
oJ:
OL 0.007
= 0.006
ｾ

= 0.005
tJ
ｾ
c 0.004
'"
oJ:

.....
i..,i
0.003
Ｎ ｾ
0.002
ｾ'"
cz:: 0.00 1 I\larttnsitt and Rainite
0.000
300 4UO 500 600 7UU /SOO 'lOO
-0.00 1 Trmprraturr I"C
-0.002

(b)

Figure 2.23 (a) Cooling dilatometric curve and (b) m icrostructure of a low carbon manganese
steel (Fe - O.07C-1.5 6Mn - 0.41Si) after cooling at a rate of 234 K 5-1 (Garcia de Andres et al.
2002)

The key transformation pOints during continuous cooling can be

summarised as:

• Ar3 = The temperature at which the decomposition of austenite to

ferrite begins during continuous cooling.
• Arl = The temperature at which the decomposition of austenite to
ferrite ends during continuous cooling.
• Bs = The temperature at which Bainite begins to form during
continuous cooling.
• Ms = The temperature at which Martensite starts to form during
continuous cooling.

52
 • Mf = The temperature at which Martensite finishes forming during
continuous cooling.

2.5 Welds in Power Plant Steels

2.5.1 Microstructure of Welds
Fusion welding involves the localised injection of intense heat and its
dissipation by conduction into the parent material. The weld microstructure
at each location is therefore closely related to the thermal history
(Lancaster 1993) . The different zones and their characteristics for a single
pass weld have been well described (Mannan and Laha 1996; Savage et al.
1976) and are shown in Figure 2.24 (Francis et al. 2007). Regardless of
the primary solidification structure, the fusion zone in low alloy steels
transforms to austenite at a temperature not far from the solidification
point, and then undergoes a solid state transformation to a structure that
will depend on both the hardenability of the alloy and the cooling rate.
Adjacent to the fusion zone is a heat affected zone (HAZ), a region that is
not heated sufficiently to cause melting, but nevertheless is
microstructurally altered by the welding thermal cycle. As indicated in
Figure 2.24, the HAZ can be subdivided accord ing to the extent to which
grain growth and austenitization occur as follows; (i) a coarse grained zone
(CGHAZ); (ii) a fine grained zone (FGHAZ); (iii) an intercritical zone
(ICHAZ) and (iv) over tempered base metal.

, . . - - - - - Solidified weld

Ｂ Ｇ Ｍ ｾ Ｍ Ｍ Ｍ Ｂ Ｍ Ｍ Ｍ Ｍ Ｍ Ｎ Ｎ ＭＭ Ｍ ＮＮＭ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ------------' 1500

Grain growth
(CGHAZ)
----------------------- 1100
Recrystallised
(FGHAZ)
---- ａ ｾ

HAZ
Figure 2.24 Metallurg ical zones in single pass weld, categorised accord ing to maximum local
temperature: micrographs correspond to weld in 2.2SCr- l Mo steel (Francis et al. 2007)

53
 2.5.1.1 HAZ and regions of HAZ
The heat affected zone (HAZ) is the area of base metal that has not been
melted by the weld pass but which has had its microstructure and
properties significantly altered by the thermal cycle associated with
welding. The thermal history of the HAZ differs depending on its distance
from the heat source, giving rise to different microstructures that can be
divided into two regions; a high temperature region and a low temperature
region. The high temperature region is mainly the coarse grained HAZ
(CGHAZ) and the low temperature region consists of the fine grained HAZ
(FGHAZ) and the intercritical heat affected zone (ICHAZ), and these
regions are shown in Figure 2.24. During welding, the CGHAZ is adjacent
to the molten weld pool, so it experiences a very high peak temperature.
This peak temperature is much higher than the a to y transformation
temperature AC3, so the microstructure is re-austenised and results in the
coarse grained microstructure. The number of grain boundary precipitates
in the CGHAZ is low as at the high peak temperature, most existing
precipitates are dissolved, and rapid cooling does not allow them to re-
form. The FGHAZ experiences a peak temperature around the AC3 which
allows recrystallisation to take place, resulting in a finer grain structure.
The existing precipitates in the FGHAZ do not dissolve but rather coarsen.
The ICHAZ experiences a peak temperature just above the eutectoid ACl
temperature, so the microstructure is a mixture of fine recrystallised grains
and the unchanged grain structure of the parent metal.

2.5.1.2 Weld Metal

The weld metal (WM) is the region which undergoes melting and re-
solidification during the welding process. Filler metal and parent metal are
melted together and well mixed, with element distribution in this region
generally thought to be uniform (Francis et al. 2007). In a Single-pass
weld, the WM has a coarse columnar structure as indicated schematically
in Figure 2.24. In a multi-pass weld, the WM is made up of a large number
of overlapping weld beads where a repeated heating and cooling cycle has
affected the microstructure of adjacent beads, which includes the coarse
columnar region, a recrystallised coarse grain region and a recrystallised
fine grain region which together are termed the recrystallised region. The
recrystallised region is made up of the coarse columnar zone of previous
beads that have been re-austenitised by subsequent bead deposition
leading to the refined equiaxed structure.

54
 2.5.1..3 Weld interface

The weld interface is defined as the interface between the liquid metal and
the heat affected zone. It can be further defined at a finer scale as shown
in Figure 2.25 (Savage and Aronson 1966; Savage et al. 1976). What is
normally called the fusion line is actually a region with two additional zones
between the weld metal and heat affected zone, namely the unmixed zone
and the partially melted zone.

.-
H.ot Dlrec ted

ｉ ﾷＺＺｾｾ､
" f 1" .. n:,;riCDI (a)
ｾ
-))
Ｍ Ｍ Ｍ
Figure 2 .25 Schematic diagrams of (a) classical weld structure and (b) Savage's definition of
weld interface region (Savage et al. 1976)(Savage et al. 1976)

Unmixed zone
The unmixed zone is the narrow band consisting of melted and solidified
base metal and any composition gradient between the composite region
and the parent metal contained in the unmixed zone, where the base metal
in this region has been completely melted. Savage (Savage et al. 1976)
argues that while molten, the material in this region only experienced
lamellar flow, indicating that the liquid at the fusion boundary does not
experience mechanical mixing but only diffusional mixing.

The significance of the unmixed zone is given more consideration for the
fusion boundary of heterogeneous welds and less for autogenous and
homogeneous welds due to their uniform composition. However, it is
possible that in an autogenous or homogeneous weld, the resulting
composition of the WM may be different from the base metal due to other
factors, such as absorption, outgassing, oxidation, and vaporization of
certain elements. As such, an unmixed zone may be significant even in
these cases (Savage et al. 1976).

Partially melted zone

The partially melted zone is the narrow band region between the unmixed
zone and the HAZ, where the base metal has undergone only partial
melting during welding. Localised melting is caused by the variation of the

55
effective melting temperature from place to place, which is due to the
point-to-point variations in solute content in the base metal. In the
partially melted zone, the grains near the fusion boundary are smaller than
those in the nearby region of the HAZ where no localised melting has taken
place (Savage et al. 1976).

2.5.1..4 Multi-Pass Welds

Fusion welding of thick walled components necessarily involves many weld
passes to fill up the jOint. Weld beads covered by other passes then
experience multiple heat pulses and a further subdivision of metallurgical
zones. Some of the possible combinations of thermal cycles are illustrated
in Figure 2.26 (Francis et al. 2007). With regard to the sequence shaded
dark grey in Figure 2.26, the initial intercritical heating cycle may not have
a significant effect on the final microstructure, since the subsequent cycle
introduces a much higher peak temperature at that location. However,
other sequences can have significant and detrimental effects. For example,
the intercritically reheated CGHAZ is the zone of the CGHAZ which is only
partly reaustenitised during a subsequent thermal cycle and is shaded
black in Figure 2.26. In some steels, a concentration of austenite
stabilisers, such as carbon, can be developed in the regions that are
reaustenitised and, upon cooling, these locations transform into hard
microstructures associated with poor toughness, the so-called local brittle
zones. Given that the austenite in the CGHAZ would have originally been
coarse, the decline in toughness in the intercritically reheated CGHAZ can
be identified via hardness tests (see section 2.3.4.3).

he ,,,"eded zone
CGHAZ

FGHAZ

Inleltrilically
reheated
CGHAZ

Figure 2.26 Examples of thermal cycles in multipass weld (Francis et al. 2007)

56
Filler metals for ferritic steels are generally selected to achieve an
appropriate balance between strength and toughness, or to mitigate
against toughness related problems such as cold cracking (Reddy et al.
2003). Other factors may also be relevant, such as with 9- 12%Cr creep
resistant steels, where it is important to avoid the formation of 5-ferrite at
high temperatures and yet still achieve a complete transformation to
martensite at ambient temperature (Onoro 2006). The design
requirements for weld filler metals generally necessitate a chemical
composition that differs from the parent material . Thus, in a multipass
weld, the composition of one weld bead can vary from the next as a
consequence of changes in dilution levels. A schematic representation of
how dilution might vary in a typical power plant steel weld is given in
Figure 2.27. The extent to which dilution might influence microstructure
will depend on the mismatch between the compositions of the filler metal
and parent material. However, in microalloyed steels, even a relatively
small change in composition due to dilution has been shown to significantly
affect weld toughness (Hunt et al. 1994).

base
metal

D significant dilution
low dilution
Figure 2.27 Variation in dilution of filler metal in power plant weld (Francis et al. 2007)

2.5.2 Post-Weld-Heat-Treatments in P92

As has previously been described, after welding, the resultant weldment
has a complex and inhomogeneous microstructure. This region will have
very different mechanical properties to the parent material and if put into

57
service in the as-received condition, it would be likely that it would suffer
heavy damage. Post-weld-heat-treatment (PWHT) of such weld joints is
necessary to stabilize the microstructure, improve its toughness and also
to relieve residual stresses after welding.
In the 9-12%Cr power plant steels, structures of this type undergo a post
weld heat treatment (PWHT) which tempers the martensite formed in the
HAZ and produces a complex distribution of precipitates which enhances
their resistance to creep deformation (Cool and Bhadeshia 1997). This
PWHT is typically carried out at 765°C for the parent P92 with a tolerance
range of +/- 15°C (Marshall et al. 2002a), with the minimum PWHT being
set in international standards at 730°C to avoid weakening of the materials
mechanical properties such as toughness and creep-rupture strength.
When selecting the PWHT temperature, it is important to consider the A1
temperature of the material; if the PWHT temperature exceeds the Al
temperature (even if transiently), then austenite may re-form in the steel,
resulting in the formation in the final product of either untempered, brittle
martensite (if the cooling rate from the temperature where austenite is
formed is high enough) or soft ferritic regions (associated with slow cooling
rates, or extended thermal soaks at temperatures below the austenite
stability region), both of which are undesirable. Both of these phases are
creep-weak (for different reasons) and will thus promote early component
failure.

The presence of soft ferrite is the more likely consequence of inadvertent

austenite formation during PWHT. A transient period of reaustenitization
(due to temperature overshoot) will result in ferrite being reformed during
the sub-A1 temperature period, be this either the primary soak period, or
the cooling period. The relevant welding standards (BS! 1987b) define a
maximum cooling rate following PWHT; however the actual cooling rate
immediately after the end of the soak period is commonly below the
permitted maximum (Richardot et al. 2000a).

There is very little concern that the heat treatment will result in austenite
formation in the parent P92 itself, since the Al temperatures in the parent
are typically between 830-900 o C (depending upon the specific composition
within the specification) (Richardot et al. 2000a), some 70°C above the
target heat treatment temperature. However, there are concerns that the
alloy composition may result in other undesirable phase transformation
behaviour; Cool and Bhadeshia (Cool and Bhadeshia 1997) conducted work

58
to examine austenite formation during PWHT of 9Cr-1Mo steels, finding
that excessive use of austenite stabiliSing elements such as nickel renders
alloys susceptible to austenite formation. However, it was also found that
the excessive use of ferrite stabilisers prevented the steel from been able
to fully austenitise at high temperatures, leading to the retention of soft 15-
ferrite (Cool and Bhadeshia 1997) in the steel structures.

In addition, it is recognised that microsegregation may exist in weld metals

due to their rapid solidification which may result in different parts of the
microstructure exhibiting different transformation temperatures (Santella
et al. 2003). Nickel and manganese are the primary austenite stabilisers
ensuring full austenite formation (without i5-ferrite) during the
austenitization and thus a 100% martensite structure on cooling (Knezevic
et al. 2008). According to Marshall (Marshall et al. 2002a), the tendency
for undesirable reaustenitization of P92 weld metals during heat treatment
may be understood by considering the sum of the nickel and manganese
levels in the steel; specifically, he proposed that the minimum ACl for P92
weld metal at l.Swt%(Ni + Mn) is 790°C and thus suggested that 775°C is
therefore a safe upper limit for the PWHT temperature assuming that a
margin of error of ± lSoC in the PWHT temperature is required.

2.5.3 Creep failure of 9-12 % Cr steel welds

Weldment structures generally have lower creep strength than normal bulk
steels as a result of their complex and heterogeneous structures formed
during the welding thermal cycle. A typical welded structure consists of
parent metal, heat affected zone and weld metal (see Figure 2.24) which
all have different creep properties. Failure modes of welds during creep
have been well studied (Francis et al. 2004; Issler et al. 2004) and have
been categorised according to the position of the creep fractures within the
weldment;

I. Type I occurs within, and is confined to, the weld metal.

II. Type II occurs within the weld metal but may grow beyond the weld
and into the bulk.
III. Type III occurs in the coarse grained HAZ (CGHAZ, Figure 2.24)
close to the fusion boundary. It is actually reheating cracking that
arises early in life as a result of creep strain absorbed during the
decay of residual welding stress exhausting local ductility.

S9
 a. Type lIla also occurs in the CGHAZ but occurs later in life in
more ductile regions after long-term loading (Brett 2004).
IV. Type IV occurs in the low temperature HAZ, typically the
intercritical HAZ (ICHAZ, Figure 2.24).

Creep cracking has also been found at the inter-bead (weld bead on weld
bead) fusion boundaries which differs from type IlIa cracking which occurs
adjacent to the fusion boundary of weld bead to parent metal (Allen et al.
2007a).

2.5.3.1 Type IV cracking in ICHAZ

Type IV cracking is considered the most important life-limiting failure
mechanism in weldments under creep conditions. It occurs in a
circumferential mode just outside the fully transformed HAZ region
adjacent to the parent metal. During welding, the ICHAZ is heated to a
maximum temperature between the ACl and AC3 resulting in partial
transformation from a to y (Gaffard et al. 2005). Upon cooling, any y
transforms to soft ferrite, leading to a microstructure of tempered
martensite and soft ferrite. As the peak temperature of the heating cycle is
relatively low with a short duration, precipitates do not generally fully
dissolve in the austenite whereas in the CGHAZ, they typically do. The
resulting microstructure of the ICHAZ has a very fine structure consisting
of subgrains with low dislocation density. Hardness measurements reveal
that the ICHAZ has a hardness lower than both the rest of the weldment
and the parent material (Albert et al. 2003; Wu et al. 2004). After creep
exposure, creep cavities form in the ICHAZ resulting in a failure
mechanism involving creep of locally constrained regions (Wu et al. 2004).
Coarsening of M23 C6 and an absence of secondary reprecipitation of fine MX
particles during tempering and creep exposure have been identified as
major mechanisms accelerating the poor creep response of the ICHAZ
(Spigarelli and Quadrini 2002).

60
2.6 Summary

P92 steel is a high-alloy steel that has been specifically designed for
operating at high temperatures (600 0 C - 650°C) and has found wide use in
the power generation industry, particularly since 2005. For the successful
installation and use of this advanced steel all aspects of its behaviour, in
terms of both metallurgy and in-service behaviour, must be understood.

In normal industrial practice, following welding with P92 fillers, post-weld

heat treatment is carried out, and there is a desire to perform this heat
treatment close to the Al temperature of the materials involved to develop
what are considered to be the best mechanical properties in the weld and
heat affected zone. As such, it is important to accurately know the Al
temperature of the materials being heat treated. There are gaps in the
knowledge as to how each of the elements in the material, rather than just
the obvious elements such as manganese and nickel, affect the
transformation temperature, All and the rate at which austenite could form
during PWHT. This gap in the knowledge of how composition affects the Al
temperature is applicable to both welds and parent material.

Specifications for weld metals (and parent metal) define a range of

compositions, but do not offer any guidance on how variations within that
specification may affect the final material properties. As precipitation
strengthening is important to the creep resistance of P92, it is important to
understand how the levels of precipitates may vary even within the
specification.

The creep failure associated with welds has been regarded as a critical
issue in the studies of power plant steels, and there is also little guidance
on how the composition variations often necessary for successful welding
affect the creep rupture performance.

In light of this, the current thesis will seek to examine and seek a greater
understanding of the behaviour of welds in P92 steel so that their in-
service behaviour may be better understood, focussing on the response of
the material to post-weld heat treatments (PWHT), the optimum weld
consumable composition and the microstuctural development during creep-
rupture. From this, alloy specifications of weld consumables may be
improved, quality checks on P92 used by industry could better ensure the
fitness for service of a material if an accurate composition is known and

61
enable manufacturers to further tailor compositions to produce the
strongest possible material. Furthermore, the design and selection of P92
material that will not undesirably transform during heat treatments could
be enhanced.

The outcomes of this thesis should enable a better understanding of P92

parent metal and welds and provide results that are immediately applicable
and useful to the power generation industry through the collaboration that
SuperGen provides.

62
3 Experimental Work

3.1 Material and specimen preparation

3.1.1 Weld pad preparation

(a) (b) (c)

Figu re 3 . 1 Th e three we ld 200mm x 1S0mm we ld pad s ( a) A, (b) B a nd ( c)

As depicted in Figure 3.1, three "'200 x 150 x 30 mm multipass weld pads

were made by Doosan Babcock Energy Ltd. (UK), by depositing 4 mm
diameter flux-coated consumable stick electrodes (A, B or C) onto carbon -
manganese steel base plates using the manual metal arc (MMA) procedure
with DC+ polarity and a current of 160-170 A. The resulting heat input was
around 1.5 kJ mm -1 . Approximately 20 runs were made per layer, and each
pad consisted of at least 10 layers. The deposition was performed in the
200 mm direction only, at speeds between 140-180 mm/min with a weld
bead overlap of around 50% .

A preheat/interpass temperature between 200-300°C was maintained

throughout the welding process, measured using temperature indicating
crayons. Chipping/wire brushing between passes (to remove slag) was
performed where necessary. After welding, a post-weld heat treatment of
760°C was applied for 2 hrs.

3.1.2 Specimen preparation

Standard 50mm gauge-length, ridged, uniaxial (Figure 3.2) specimens
were removed from the weld pads using electric discharge machining
(EDM) so that the test loading axis was transverse to the welding direction .

63
 (a)

I ｾＮ SOmm

(b)
.p1
Figure 3 .2 (a) Orientation of cree p speci men removal (schematic only, not to sca le) and ( b)
schema ti c of the creep speci men

To help ensure consistent test results and to avoid the effect of dilution
from the steel base plates, material close to the weld start and stop edges
of the pad s and material comprising the innermost and outermost beads
was not used . The majority of the specimens were creep tested in-house
and were 130 mm long with a 10 mm diameter gauge section, based on
BS 3500: Part 3. However, a small number of creep tests were performed
by Corus UK Ltd. and these specimens were 88 mm long with a 6.25 mm
d ia meter gauge section, based on ASTM E21 & E8. The results do not seem
to have bee n affected by the different gauge cross section sizes, although
it is known that smaller sections can introduce more scatter in the results
of in homogenous materials such as weld pads (Evans, R Wand B Wilshire.
1985). Where tests have not been carried out in-house, this is indicated in
the results.

Creep tests were performed in air at three different temperatures and

nominal stress conditions, given in Table 3.1. Constant uniaxial tensile
loading was applied in the direction of the specimen axis, transverse to
that of welding. Strain measurement over the gauge length was regularly
logged using output from extensometers attached via the specimen ridges.

64
 Table 3.1 Weld metal creep test parameters

Weld Metal Temp. (0C) Stress (MPa)

A 625 150
A* 625 150
B 625 150
C 625 150
A 625 170
B 625 170
C 625 170
C (rpt) 625 170
A* 670 90
B 670 90

C 670 90
* - tested by Corus UK Ltd., (rpt) - repeat

The uniaxial creep tests were performed on the Denison Creep Testing
Machine. The basic frame of the machine is a 2 m x 2 m 'L' shape, the
vertical section having 4 x ¢50 mm columns and the horizontal section
consisting of a lever system whereby the load is applied.

The graduated load lever is linked to the bottom lever which in turn
connects, through knife edges, with the lower loading bar (within the
afore-mentioned columns).

Since the lengthening of the specimens allows the load lever to gradually
fall to its lower limit, a 'strain take-up' system comes into operation from
time to time to prevent this happening. It is triggered by the end of the
load lever pressing on a micro-switch which in turn causes a DC motor-
drive screw system to draw the upper loading bar upwards. Since the
upper loading bar is connected through the specimen to the lower loading
bar, the load lever is gradually raised to its highest working position and as
this occurs another micro-switch causes the strain take-up motor to be
switched off. The whole of this operation is very slow, taking about half an
hour and hence the load on the specimen remains steady.

All of the specimens have been machined with knife-edge flanges.

Conventional extensometer 'legs' are fitted to these flanges and at the
lower ends of the legs linear voltage displacement transducers produce, via
a conditioning unit, a DC voltage which is proportional to strain. The
voltage is then fed to a computer-based data-logging system.

65
 A 3-zone muff type furnace is lowered to surround the specimen which
previously has had three thermocouples spot welded to it. Each of the
three furnace zones, together with its thermocouples, forms an
independent control loop via a Eurotherm electronic controller and its
associated thyristor.

Both specimen 'caps' have their threads cleaned of debris by a M16 tap in
order to avoid torsional effects when the specimen is fitted to the loading
bars. The lower cap is screwed onto the lower bar and the specimen is
then itself screwed into the M16 hole. Then the upper cap is screwed into
the top of the specimen so that it almost goes beyond the lower end of the
thread. This procedure allows the top cap to have full thread engagement
on both the specimen and the upper loading bar, when fitting is completed.

Three thermocouples are then prepared by using three 1 m lengths of

thermocouple wire. One end of each piece is connected to the controller
input while the other is tightly twisted and spot welded to the appropriate
position on the specimen (two outside the gauge length and one in the
middle). Then the extensometer legs are fitted carefully, ensuring that the
specimen knife edge flanges fully engage with the 'V' shaped grooves at
the top of the assembly.

After a final inspection to ensure that everything is correctly positioned,

the furnace is lowered so that its centre coincides with that of the
specimen.

In order to avoid any damage to the speCimen it is essential that whilst

heating up to the test temperature, the strain take up system (S.T.S.) is
operating. For this reason a minimum over balance is set on the load lever
and the S.T.S. is switched on with the gear train pin engaged (a
modification to the loading machine stops the furnace from being switched
on whilst the S.T.S. is switched off).

In less than 2 hours the specimen should have reached the required
temperature. When this is achieved the loading lever is temporarily
supported by a prop close to its free end. The gear train pin is then
removed and the manual S.T.S. wheel is turned anti-clockwise about 5 to
10 .turns. This will ensure that the knife edges are at minimum
disengagement so that when the load is manually wound on there is no
jolting as contact is re-established. With the loading lever firmly propped,

66
the moveable weight is wound along the lever to the position where the
scale indicates the appropriate force.

The strain transducers are finally positioned so that their plungers are
almost fully retracted and intercept values are keyed into the logger to
ensure that zero strain is displayed. Then the manual S.T.S. wheel is
steadily wound clockwise to bring the loading lever knife edges into contact
and soon after this has occurred the load lever should rise from the prop
and the latter is carefully removed.

After starting the test, it is only necessary to periodically check that

progressive strain increase is displayed and that the load lever has not
'bottomed'. When the load lever reaches the bottom after a period of
testing time, this will show that the specimen has failed.

3.2 Characterisation Methods

3.2.1 Sample Preparation
All creep tested specimens were sectioned using electric discharge
machining (EDM) through the specimens parallel to stress axis so that the
plane of the cross-section lay normal to the welding direction, while
sections of the weld pad were machined using EDM and a cut-off wheel.
Sectioned samples were mounted using cold mounting epoxy resin for
large specimens for examination via optical microscopy (OM) and hardness
tests. Hot mounting with conductive phenolic mounting resin black was
employed for smaller specimens for all forms of microscopy. Cross-sections
of the samples were ground and polished with SiC papers and loose
diamond abrasives down to a Ij.Jm finish for OM and scanning electron
microscopy (SEM), with a further polish to a O.06j.Jm finish with colloidal
silica for electron back-scatter diffraction (EBSD) analysis. Samples for OM
and SEM were etched with alcoholic ferric chloride (5g FeCI 3 , 2ml HCI, 95ml
ethanol) for 20 seconds to reveal the microstructure. Samples that
underwent heat treatment had any surfaces that were to undergo
examination ground back by "'1 mm before polishing in order to avoid
regions that may have undergone decarburisation.

67
 3.2.2 Optical Microscopy
Optical microscopy (OM) was used for the examination of samples up to
magnifications of x40. OM is primarily used to obtain macro-scale
information on the weld structure and montages of the microstructure
were generated using Adobe Photoshop and multiple images. Scale
markers are displayed on individual images.

3.2.3 Micro-hardness Measurement

Micro-hardness of specimens was measured using a Leco M-400 tester with
a Vickers indenter. The indentation time was 15 seconds and hardness
values are obtained by measuring diagonals of hardness indents and using
Equation 3.1 to convert to Vickers hardness readings;

2F sin ｾ 1.854F
Hv = d2 ｾ d2

Equation 3.1

where F is the load in kgf, a = 136°, d is the arithmetic mean of indent

diagonals in mm and therefore Hv is the Vickers hardness in kgf/mm2. The
force used and spacing of indents is stated when results are presented.
The hardness tester was calibrated using a calibration block of known
hardness. For hardness traverses, each reading is taken from a single
point. For bulk hardness the values are an average of five points.

3.2.4 Scanning Electron Microscopy (SEM)

The scanning electron microscope (SEM) uses a high-energy electron
beam, which scans a sample surface, for imaging. The electrons interact
with atoms in the sample to produce signals that provide information about
the sample. Secondary electron (SE) signals provide information
concerning the topography and surface structure of the sample, while
back-scattered electrons (BSE) provide information about element
distribution with contrast being related to the mean atomic number of the
material. Characteristic X-rays were used to identify and quantify
composition using the energy dispersive X-ray (EDX) analysis method. A
Philips XL-30 equipped with a field emission gun (FEG), which facilitates
imaging at higher magnification that that of a traditional tungsten filament
SEM (by producing an electron beam that is smaller in diameter, more
coherent and with a much greater current density/brightness), was used.
Specimens for examination in the SEM must be conductive and so, as

68
stated in 3.1.1, samples were hot mounted in conductive phenolic
mounting resin.

Samples were imaged using a 20 kV accelerating voltage and a working

distance of 10 mm unless otherwise stated. Laves phase precipitates were
quantified according to the method presented by Korcakova et al.
(Korcakova et al. 2001) which relies on the contrast between these
precipitates and the matrix that is evident in BSE images. The method
presented by Korcakova et al. uses the image analysis software Image] to
convert the BSE images into a binary image which, due to the high
contrast, results in the Laves phase been clearly defined from the rest of
the material. The software can then count the number and mean size of
the precipitates. This method could not be used for the MX and M23 C6
phases due to their low contrast with the matrix. To count the size and
number of these precipitates, Image] was again used but the precipitates
had to be manually identified first before the image was turned binary. The
difference between the MX and M23 C6 phases was determined by their
location an relative size based on the description of their morphology in
section 2.4.2. While transmission electron microscopy (TEM) is a more
usually used method for the counting and sizing of precipitates, Korcakova
et al. demonstrated that these methods which use SEM images gave
comparable results and were much less time consuming, allowing a greater
(and therefore, more representative) area of the material to be examined
in the same time. For the sizing and counting of precipitates 40 images per
sample were evaluated.

3.2.5 Electron Back-Scatter Diffraction (EBSD)

Electron back-scatter diffraction (EBSD) is a microstructural-
crystallographic technique, which allows crystallographic information to be
obtained in an SEM fitted with a backscatter diffraction camera. When the
electron beam impinges on a tilted crystalline sample, the diffracted
electrons form a characteristic diffraction pattern on a fluorescent screen
which can be analysed to provide information such as grain morphology,
crystal orientation and grain boundary misorientations. EBSD examination
was carried out at Loughborough University using an EDAX Pegasus
combined EBSD and EDX system attached to a LEO 1530-VP FEGSEM.

Special care must be taken when preparing samples for EBSD

examination; diffraction electrons escape from only a few tens of

69
nanometres from the sample surface, and so it is essential to remove all
mechanical distortion which is generally produced during grinding and
polishing. To this end, following polishing down to a 1 IJm finish with
diamond abrasives, an additional final polishing stage using colloidal silica
was then carried out to remove residual surface damage. The colloidal
silica polish is a chemo-mechanical polish that combines the effect of
mechanical polishing with etching. This polishing stage was conducted for
at least 30 minutes to achieve the desired surface finish. The EBSD
examination used a 20 kV accelerating voltage and an aperture size of 30
IJm. The sample was tilted at 70° with respect to the horizontal.

As producing high resolution images with EBSD can be time consuming, a

series of low quality images were first taken at 5 locations in the samples
to ensure that any final image would be representative of the
microstructure.

3.2.6 Dilatometry
The AC1 temperature of the three weld metals and the P92 steel was
measured by dilatometry, carried out using a thermomechanical analyser
(TA Instruments Q400) with a sample size of tv 5 x 5 x 15 mm and a
heating rate of 100°C/hour unless otherwise stated in order to simulate the
rate used in industrial PWHT; the sample dilatation was measured using a
glass probe resting on top of the sample. The ACl temperature was derived
from the dilatometric traces, being identified as the point at which the
gradient of the line on heating (associated with thermal expansion) began
to decrease (associated with the contraction associated with the phase
transformation from ferrite to austenite). Dilatometry is discussed in detail
in section 2.4.5. The dilatometer was calibrated using an iron-carbon
calibration block of known transformation temperature. For the
determination of the initial AC1 temperature,· three samples per material
were used, and for the examination of the effect of holding temperature
three samples per temperature were used.

70
 3.3 Modelling

Thermodynamic modelling of phases expected in the steels as a function of

composition and temperature was carried out using Thermo-Calc 5 for
Windows with the TCFE6 Steel and Fe-alloys database (Foundation of
Computational Thermodynamics and Thermo-Calc Software AB, Stockholm,
Sweden). Thermo-Calc is a software package used to perform
thermodynamic and phase diagram calculations for multicomponent
systems in materials science and engineering. Calculations are based on
thermodynamic databases produced by expert evaluation of experimental
data using the CALPHAD method. The CALPHAD method is based on the
fact that a phase diagram is a representation of the equilibrium
thermodynamic properties of a system, which are the sum of the
properties of the individual phases. It is thus possible to calculate a phase
diagram by first assessing the thermodynamic properties of all the phases
in a system.

TCFE6 is the steel and Fe-alloy database for Thermo-Calc and can be
applied in steel and Fe-alloy design and engineering. The TCFE6 database
includes data for molar volume calculation of density and lattice parameter
(for cubic structures), coefficient of thermal expansion and/or relative
length change.

Thermo-Calc can be used for all kinds of calculations of thermodynamic

properties, equilibrium and partial/local-equilibrium quantities, chemical
driving forces, and for various types of stable/meta-stable phase diagram
and property diagrams of multi-component systems. Thermo-Calc deals
with systems that are in equilibrium, i.e. in a state stable against internal
fluctuations in a number of variables such as temperature and composition.
If the work that can be exchanged with the surroundings is limited to
pressure-volume work, then the state of equilibrium of a system can be
obtained by assigning values to exactly N+2 state variables where N is the
number of components of the system. The output of phase diagrams can
show the phases present in terms of mole fraction, mass fraction and
volume fraction.

The modelling started with defining the system, which, in this work
amounts to defining the elements contained in the steel with the TCFE6
database. The phases which are considered in the calculation can be
selected from the full list of possible phases. The concentration of each

71
element is defined under a temperature and pressure to obtain the
equilibrium. In all calculations, pressure was kept at atmospheric pressure.

For the production of phase property diagrams, the initial inputs were the
chemical composition (with iron left to make up the balance of the
system), pressure and an initial temperature. To produce the phase
property diagram, a temperature range must also be defined which in this
case was 500·C to 2000·C. The number of steps (iterations) in the
calculation must also be defined, and for all phase property diagrams this
was set at 50. An increased number of steps can increase the accuracy of
the results but also increases calculation time. Initial calculations used
steps of 10, 50, 100, 200 and 500. The 10 step calculations gave different
transition temperatures, while use of all the other numbers of steps gave
the same results. Therefore 50 steps was chosen as the calculation gave
reliable accuracy in the shortest time. The potential phases included in the
calculations must also be defined, and the list of phases included is shown
in section 4.1. For the production of graphs, the resulting data were
exported to Microsoft Excel.

72
4 Results

4.1 Weld Consumable Composition and Modelling

Table 4.1 P92 alloy specification ranges (Richardot et al. 2000b), along with compositions of
the parent P92 and weld metals (Weld Metal A, Weld Metal B and Weld Metal C) studied in
this work.

Element, wt% (balance iron)

C N B Si Mn P 5 Cr Mo Ni AI Co Cu Nb V W

P92 min 0.07 0.03 0.001 0.3 8.5 0.3 0.0' 0.15 1.5

P92 max 0.13 0.07 0.006 0.5 0.6 0.02 0.01 9.5 0.6 0.4 0.04 0.09 0.25 2.0

Pa....tP9Z 0 I 0.47 0.0034 0.45 0.45 0.QI5 0.001 8.62 0.33 0.27 0.019 0.076 0.21 1.86

WMA 0.11 0.047 0.2 0.78 0.011 0.005 8.73 0.45 0.64 0.04 0.043 0.199 1.66

WMB 0.09 0.048 0.42 0.99 0.011 0.006 8.74 0.'9 0.42 0.04 0.08 0.23 1.59

WMC 0.1 om5 0.007 0.23 0.59 O.OIl 0.009 9.82 1.41 0.7 1.09 0.02 0.05 0.25

Table 4.1 shows the measured compositions of the three weld deposit
materials, provided by the Supergen industrial sponsors, that were initially
examined for use with P92. The P92 alloy specification is also shown, as is
the composition of the parent P92 that the weld consumable will be used
with later in the work programme. The weld metals were chosen to
encompass the range of compositions that can be seen in power plant
welds. All are commercially available, with WMA and WMB having been
derived from the P92 specification, while WMC is based on creep-resistant
alloy B2 and has been included due to its potentially greater creep
strength. In order to understand what phases may be expected to make up
the microstructures of the weld metal derived from each of the
consumables (under equilibrium conditions), thermodynamic modelling was
used. The measured composition of each weld metal type was input into
the program ThermoCalc along with the phases (and crystal structures)
that were to be included in the calculation as follows;

• Face Centred Cubic (FCC)

• Body Centred Cubic (BeC)

• Laves Phase

73
 • MnS

• AIN

• BN

• Liquid, Gas (for calculations that are near melting temperatures)

where FCC is the crystal structure of the MX precipitates and austenite;

identification of which phase is being predicted can be achieved by
examining the predicted composition and phase fraction. Ferrite is
represented by the body centered cubic phase. Only phases that have been
observed in the literature for 9Cr steels are included in the calculation.

It is worth noting at this point that the formation of martensite, the desired
matrix in P92 steels, cannot be predicted in ThermoCalc as it is the product
of a non-equilibrium displasive transformation (see section 2.4.4). Instead
ThermoCalc predicts a-ferrite; the transformation to martensite in actual
P92 material is facilitated by rapid cooling from temperatures above Ai.

4.1.1 Weld Metal Phase Fraction Diagrams

Figure 4.1 shows the phase property diagrams for the three weld metal
types selected over a temperature range from 500-2000°C, thereby
encompassing possible service temperatures, post-weld-heat-treatment
conditions and the high temperature conditions experienced during
welding. The phases predicted to be present are shown in the figure
legends with the primary phases being low temperature a-ferrite, y-
austenite, high temperature ｾ Ｍ ｦ ･ ｲ ｲ ｩ ｴ ･ and liquid above 1500°C. From these
phase diagrams, the Aei temperature and the temperature at which ｾ
ferrite will begin to form on heating can be derived and these are shown
for the three weld metal types in Table 4.2.

74
 Tabl e 4.2 Ae l and a-ferrite start t emp eratures, as ca lculated by ThermoCalc

Weld Meta l 15- ferrite Start T emperature, °C

A 7 58 12 74

B 769 1224

c 77 1 1244

0 .9 - - liqu id

0 .8 - - Ferrite

::: 0 .7 - - Austenite
'"ｾ 0 .6
c: - - (V, Nb)(N, C)
0
B 0 .5
ｾ - - Lave s Phase
u.
& 0 .4
'"
.c
"- 0 .3
M23C6

- - M3P
0 .2

0 .1 MnS

0
500 700 900 1100 1300 1 500 1700 1900
Temperature, °C

(a)

0 .9 - - Liqu id

0 .8 - - Ferrite

0.7 - - Austenite
'"'"
'"
E.O .6 - - (V, Nb )( N, C)
c
0
B 0 .5 - - Nb(N,C)
ｾ
u.
., 0.4 Laves Phase
'"'"
.c
"- 0 .3 - - M23C6

0 .2 M3P

0. 1 - - MnS

0
500 700 900 1100 1300 1500 1700 1900
Temperature, °C

(b)

75
 - - Liquid
0 .9
- - Fern te
08
- - BN
ｾ 0 .7
ttl - - Au stenite
E_ 0.6
c: - - (V, Nb )( N,C)
o
t O.S - - Nb ( N,C)
ｾ
"-
<Ii 0 .4 - - La ves Pha se
'"
ttl
.I: M23 C6
a. 0 .3
- - M23 (C,B)6
0 .2
- - M3P
0 .1
- - MnS

500 700 900 11 00 1300 1500 1700 19 00

Te m perature , °C

(e)

Figure 4 . 1 Predi ct ed phase fra ction diagrams for t he th ree weld metals (a) WMA (b) WMB (c)
WMC at th ermody namic equilibrium

4.1.2 Precipitate Phase Fraction Diagrams

Figure 4,2 shows the phase fraction diagrams for the three weld metals,
with just the precipitate phases being considered. All weld consumables
are predicted to conta in a form of MX, M23 C6 , Laves Phase and the impurity
phases M3 P and MnS . Both WMB and WMC contain two forms of MX, while
WMC also has BN and an additional form of M23 C6 •

0 .02 5
- - ( V, Nb)( N, C)

0 .02
- - Laves Phase
'"
VI
ttl
E
c:- 0.0 15 - - M23C6
o
B
ｾ M3 P
.,
u.
VI
0.0 1
ttl
.I:
a. - - MnS
0 .005

600 700 800 900 10 00 11 00 1200 1300 1400 1500

Temperature, °C

(a)

76
 0.025
- - (V, Nb)(N , C)

- - Nb(N,C)

- - Laves Phase

- - M23C6

M3P

- - MnS

0
500 600 700 800 900 1000 11 00 1200 1300 1400 1500
Temperature, ·C

(b)
0.025
- - BN

- - (V,Nb)(N,C)
0.02
VI
VI - - Nb(N ,CJ
'"
E
c' 0.015 - - Laves Phase
0
B - - M23C6
ｾ
u.
0 .01
5( - - M23(C, B)6
'"
.c
a. M3P
0.005
- - MnS

0
500 700 900 11 00 1300 1500
Temperature , °C

(e)

Figure 4 .2 Predicted phase fraction diagrams for the precipitate phases in the three weld
m etals (a) WMA (b) WMB (c) WMC at thermodynamic equilibrium .

77
4.1.2.1 Predicted composition of MX

The composition of the MX phases predicted in the three weld m etal types
are presented in Figure 4.3, showing that the most prevalent form is
expected to be (V,Nb)(N,C) in all three weld metal types, whil e WMB and
WMC are predicted to conta in a small amount of MX in the form Nb(N ,C).

--c --N - - Nb --v --c --N -- Nb --v

0 .9 0.9
ｾ 0 .8 ｾ 0.8
ｾ 0 .7 ｾ 0 .7
W iii
'0 0.5 '0 0 .5
§ 0.5 § 0.5
ｾ 0.4 ｾ 0 .4
.tIn 0.3 ｾ 0 .3
In
ｾ 0 .2 ｾ 0.2
0.1 0 .1
0 0
600 650 700 500 650 700
Temperature, °c Temperature, °c

( a) (b)

--c --N -- Nb --v --c --N - - Nb --V

1
0.9 0.9
ｾ 0 .8 ｾ 0 .8
ｾ 0 .7 ｾ 0.7
W W
'0 0 .5 '0 0 .6
§ 0 .5 § 0.5
ｾ 0.4 ｾ 0 .4
ｾ 0.3 ｾ 0 ,3
In
ｾ 0 .2 ｾ 0 .2
0.1 0 .1
a 0
600 650 700 600 650 700
Temperature, °C Temperature, °c

(e) (d)

--c --N - - Nb --V

0.9
ｾ 0.8 1----------
ｾ 0 .7
iii
'0 0 .6
§ 0.5
ｾ 0.4
ｾ 0.3
ｾ 0 .2

ｯ Ｎｾ Ａ ﾧ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ
600 650 700
Temperature, °c

(e)

Figure 4.3 Predicted compo si tio n of MX phases in the three weld metals (a) (V,Nb)(N,C) in
WMA, (b) (V,Nb)(N,C) and (c) Nb(N,C) in WMB, (d) (V,Nb)(N,C) and (e) Nb(N,C) in WMC

78
4.1.2.2 Predicted composition of M 23 C6

Figure 4.4 shows the predicted composition of M23 C6 in the three weld
metal types, showing them to be chromium-rich while the additional form
of this phase in WMC is predicted to co ntain predominantly boron as their
non - metallic element, hence it is referred to as M23 (C,Bk

- - C - - Cr - - Fe - - C - - Cr - - Fe
- - Mn - - Mo - - W - - Mn - - Mo - - W

1
ｾ 0 .9 ｾ 0.9
ｾ 08 E 0 .8
!!! 0 .7
w ｾ 0 .7
'0 0 .6
g 0 .5
r --------_ '0 0.6
6 0 .5
r--------_
ｾ 0.4 ｾ 0 .4
u. 0 .3 .t 0.3

ｾ ＮＲ ｅ ｾ
L 0.1 ｾ ｾ ﾧ ｾ ｾ ＮＲ
L 0.1 ｅ ｾ ｾ ﾧ ｾ ﾧ
o o
600 6S0 700 600 650 700
Temperature, °c Temperature, °c

(a ) (b)

- - C - - Cr - - Fe - - v - - Mo --8 - - C - - Cr - - Fe

1
0.9 0 .9
ｾ 0 .8 ｾ O.S
ｾ 0 .7 ｾ 0.7
W 0.6
'0 r-------------------- W
'0 0.6
g 0.5 6 0.5
ｾ 0.4 ｾ 0. 4
ｾ 0.3 ｾ 0 .3
ｾ 0.2+-------------------- ｾ 0.2
ｯ Ｎｾ Ａ ［ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ ｾ
0. 1
oｴ ］ ］ ］ Ｚ Ｚ Ｚ Ｚ ［ Ｚ Ｚ Ｚ ］ ］ ］ ］ ］ ｾ
600 650 700 450 500 550 600
Temperature, °C Temperature, °c

(e) ( d)

Figure 4.4 Composition of M23 C6 phases in the three weld metal types (a) in WM A, (b) in
WMB, (c) and (d) in WMC due to the presence of M23 (C,B)6

79
4.1.2.3 Predicted composition of Laves Phase

Figure 4.5 shows the predicted composition of the intermetallic Laves

Phase in the three weld metal types indicating that in WMA and WMB, it is
tungsten rich while WMC contains a molybdenum rich form.

- - Cr - - Fe Mo -- w - - Cr - - Fe Mo --W

0 .9 ,,0 .9
ｾ 0.8 '"
EO .8
0 .7 ｾ '"
wO .7
W 0 .6 } -_ _ _ _ _ _ _ _ __
ｾ
B05Ｎ Ｖ r-----------
'0
:5 0.5
ｾ 0.4 '" 0.4
.t
ｾ 0 .3 ｾ ＮＳ
VI
;. 0 .2 l: 0 .2

O. ｾ 1======::::;:=::::==:;
600 650 700
ｯ Ｎｾ ｴ Ｚ ］ ］ ］ Ｚ Ｚ Ｚ Ｚ ［ Ｚ ］ ］ Ｍ ］ ］ ｾ ｾ
600 650 700
Temperature, °c Temperature , °C

(al (bl

- - Cr - - Fe - - Mn

- - Mo - - Nb

1
C 0 .9
E 0 .8
ｾ 0.7
w
'0 0 . 6
:5 0.5 j-- - - - - - - - -
ｾ 0 .4
u.. 0.3

l: 0 . 1
ｾ 0 .2

ｏ ｾ Ｍ Ｍ ｾ Ｍ Ｍ
450 4 70
.. Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ ｾ
490
Temperature, °C
5 10 530

(el
Figure 4 .5 Composition of Laves Phase in (a) WMA, (b) WMB and (c) WMC [for WMC the
Laves phase is predicted to be unstable above 530·C by ThermoCalc]

80
4.1.3 The Effect of Element Variation on Precipitation

As the precipitate levels and types have been seen to differ in the two P92 -
similar weld consumables WMA and WMB, the sensitivity of the equilibrium
precipitate fraction and make-up to element variation was investigated by
thermodynamic calculation using ThermoCalc. Using the composition of
WMA as a base individual elements were varied, utilising ranges for each of
the elements as defined by the P92 alloy specification minimum values and
maximum values (see Table 4 .1). Each element was varied individually,
with iron making up the balance. Figure 4.6 shows the effect of niobium
content on the predicted fractions of the MX precipitates at 625°C. Figure
4.7 shows the elements that affect the precipitate fractions of the M23 C6
and Laves Phases, with again, the predictions being made for a
temperature of 625°C.

- - Nb( N,C) - (V,Nb)(N ,C)

3 .00E -0 3

2.S0E -0 3

:::
'"
E
2.00E-0 3
c-
o
B 1. 50E -03
...ｾ
Q)
VI
l. OOE -03
'"
.I:
Cl.

5 .00 E-04

O.OOE+OO
0 .04
L- ____-_--___--.. . .ｾ
0 .045 0 .05 0 .055 0.06 0 .06S 0 .07
ｾ ｾ Ｚ Ｚ Ｚ Ｚ Ｚ ］ ］ Ｚ ］ ］
0 .07 5 O.OB 0 .OB5 0.09
Nb, wt%

Figure 4 .6 Phase fraction of the two types of MX as Nb content varies within the limits for P92
(all other elements at WMA levels). at a constant temperature of 62SoC

81
 Laves Phase - M23C6 Laves Phase - - M23 C6

0 .03 0.03

0.025 0.025
:::
'"
E 0 .02
'"'"'"
E 0.02
c' c'
0 0
B
- B
0 .0 15 0 .015
ｾ ｾ
u. u.

''"" 0.01
'"'" 0.01
'"
.<:
"- '"
.<:
0-
0.005 0.005

0 0
0 .07 0 .09 0 . 11 0.13 0.3 0.4 0 .5 0 .6
C, wtO/o Mo, wtO/o

(a) (b)

Laves Phase - M23C6 Laves Pha se - M23C6

0 .03 0.03

0.025 0.025
'" '"'"
''""
E 0 .02 '"
E 0.02
c' c'
Q 0
t> 0.015 B 0.Ql5
ｾ ｾ
"- u.

''"" 0.01
'"'"'"
0.0 1
'"
.<:
"-
.<:
0-
0.005 0.005

0 0
1.5 1. 6 1.7 1. 8 1.9 2 0 0. 1 0.2 0.3 0.4 0.5
W, wtO/o 51, wt%

(c) (d)

Figure 4 .7 The effect of element varia tion on the precipitate fractions of Laves Phase and
M23 C6 at 625°C. Elements were varied within the P92 specification li mits with the composition
of WMA as a base. Only those elements that significantly affected precipitate fractions are
shown

82
4.2 Weld Pad Microstructural Characterization

4.2.1 Optical Microscopy

Figure 4.8 shows OM images of the three weld metals in the as-received
state (deposited and heat treated) sectioned perpendicular to the welding
direction, which all have the expected martensite microstructure.

(a)

(b)

(el
Figure 4.8 Optical micrographs of (a) WMA, (b) WMB, (c) WMC in the as-received (deposited
and heat-treated) state

83
 4.2.2 Hardness

Figure 4.9 gives a hardness profile for the three weld metals section ed
pe r pendicular to the welding direction over a distance of 15 mm which
encompassed three weld beads, with intervals of lmm between ind ents
and a load of 200 gf. The average hardness was 294 .2 kgf mm -2 for WMA,
2
280.8 kgf mm - for WMB and 297.4 kgf mm -2 for WMC. Notably, as well as
being the material with the lowest hardness, WMB also had the greatest
variation in hardness, with some very low hardness values be ing measured
in this material.

330
320
• I.

•
310
300
I.

•
•
290 t .t. • I.
- .!
N
g 280
- •• •
•
I
270 • •
260
250 • •
240
230
o 2 3 4 5 6 7 8 9 10
Spacing of hardness indents, mm

Fi gure 4 .9 Hardn ess profiles of th e three weld metals across three weld bead s, with Imm
spacing betw ee n ind ents

4.2.3 SEM Characterization

Figure 4.10 shows SEM images (both BSE and SE images of the same
regions) that were taken in order to characterize the precipitates present in
the three weld metals in the as-received condition. BSE imaging shows
that precipitates have a similar contrast to the matrix, especially the larger
precipitates. EDX analysis revealed the larger precipitates to be chromium
rich, indicating that these are M23 C6 . The other precipitates were too small
to be analysed by EDX, but based on thermodynamic predictions and
comparison with the literature (Abe et al. 2007; Sourmail 2001), it can be
surmised that they are MX particles. No Laves phase was observed.

84
 (a) (b)

(c ) ( d)

(e) (f)

Figure 4.10 Pairs of SE and BSE images showing precipitation, large precipitates are M23 C6
and smaller precipitates are MX. (a) and (b) WMA, (c) and (d) WMB, (e) and (f) WMC. Dark
regions in the BSE images are a result of porosity

85
4.3 Austenite Formation During PWHT

Following modelling and initial characterisation of the three weld metals,

the post-weld-heat-treatment conditions were investigated. The
importance of PWHT was described in section 2.5.2, in particular the PWHT
conditions which are currently recommended for welds in P92 steels.
Initially, dilatometry was used to measure the ACI of the three weld metals
for comparison with the measured Ael (4.3.1.1). Following this,
ThermoCalc thermodynamic modelling' was used to investigate the effect of
individual elements on the Al temperature resulting in an equation to
calculate the Ael for weld consumables for use with P92 (4.3.1.4). The
effect of PWHT temperatures on the a to y transformation was investigated
using a combination of ThermoCalc thermodynamic modelling (4.3.2.1),
experimental simulation of PWHT and measurement of phase changes
using dilatometry (4.3.2.2) and microstructural investigations of the
resulting microstructures (4.3.2.3).

4.3.1 The Effect of Composition on Ai

4.3.1.1 Weld Metal ACJ and AeJ

Figure 4.11 shows the measured dilatometric curves of the three weld
metals, with their ACI temperatures indicated (ACI values where
determined using the method shown in Figure 2.21). The results of
dilatometric measurement of ACI and equilibrium calculations of Ael using
ThermoCalc are shown in Table 4.3 (the Ael values being taken from Table
4.2), along with the data for the parent P92, the composition of which can
be found in Table 4.1. These data have been plotted (see Figure 4.12) as a
function of the (Ni + Mn) content in order to investigate whether the
recommendation that (Mn + Ni) < 1.5wt% (see section 2.5.2) is a
sufficient condition to avoid the possibility of reaustenitization in these
steels during PWHT (Marshall et al. 2002a) as it will ensure that the Al
temperature will be above the selected PWHT.

86
 ｾ Ｌ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ

E
-=
CD
0>
C
III
.c 150
U
c
0
in
c:
CD
E
0

Wel d Metale
Wad Mf!tal A
Weld Metal B

Ｑ Ｐ Ｐ Ｔ Ｍ ｾ ｾ ｾ Ｍ ｲ Ｍ ｲ ｾ ｾ Ｍ Ｇ Ｍ Ｇ Ｍ Ｍ ｲ Ｍ ｾ ｾ ｾ ｾ Ｇ Ｍ ｾ ｾ Ｍ ｲ Ｍ ｲ Ｍ ｲ ｾ Ｍ Ｇ Ｍ Ｇ ____ｾ ｾ ｾ ｾ ｾ
650 700 750 800 B50 900 950
Temperature (" C)

Figure 4 . 11 Dilatometric measurement of thermal expansion for the three weld m etals
exam ined in this work, with Ac, temperatures highlighted

Table 4.3 Ac, temperatures (measured by dilatometry) and Ae , temp era tures for the parent
P92 and th ree weld metals employed in this work. Ae , te m peratures are determined from a
si ngle calculation in ThermoCalc where the phase fraction of austenite is set to O.

Ae, I 'c Ac, I ·c

Parent P92 B19 B30

WMA 75B 792

WM B 769 BOO

WM C 774 797

87
 840
, ____ _. . - - - Paren t P92 Weld Metals
830 I Ｎ Ｍ Ｍ ｾＬ -----------------------/

---.
I
I
820
ｾ - - - __ I

810 ------------------------...;..: - ----- -'-

' -

i 800
:.
I

B
ｾ
'" 790
0.
E
ｾ
780 ------------------------+--------- ｾ

•
I

770

I
Ｗ Ｖ Ｋ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ ＱＭ
I

750
o 0.2 0. 4 0.6 0 .8 1. 2 1.4 1. 6
NI + Mn, wt%

Figure 4 . 12 AC I te m peratures ( m eas ured by dila t om etry ) and Ael temperatures (pre dicted
with ThermoCa lc) as a function of th e ( Ni + Mn) co ntent of th e stee l for th e parent P9 2 and
the three weld m et al t ypes exa m ined in th is wo rk

4.3.1.2 The Effect of Material Starting Condition and Heating Rate

on the ACI
Th e parent P92 stee l used in t he invest igations had already undergone a
full normalizin g and t e mpering hea t t re atment befo re the ACl shown in
Tabl e 4 .3 wa s measured. In order t o investigate any poss ible effects tha t
the prev ious heat treatme nts may hav e on the m easured ACl , a heat
treatme nt cycle was perform ed in whi ch the steel was auste nit ized (Ac !
being measured during the heati ng st age), coo led to form an untempered
stru cture, then reaustenitized (w ith th e AC t be ing measured again on
heating) . The results of th is experime nt are shown in Figure 4 .13 .

The sam e ex periment was repeate d on a sa m ple of WMA and the re sults
are shown in Figure 4 . 14 . The tren dlin es of t he data were produced us ing
an 8 point moving average.

88
 1200 ｾ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ

1000 ｴ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ ｾ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ
u
o
ｃＱｉｾ 800
...::l
-:0... 600 Ｋ Ｍ Ｍ Ｍ ｉ
C1I
Co
E 4 00 Ｋ Ｍ ｾ ｦ Ｍ Ｍ Ｍ
C1I
I-
200

o Ｋ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｎ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｎ Ｍ Ｍ Ｍ Ｍ ｾ ｟ Ｎ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｟ Ｎ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ ｾ Ｍ Ｍ ｾ
o 500 1000 1500 2000 2500 3000
Time, min

(a)

1. 00 15 Ｎ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｲ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｌ Ｍ Ｍ Ｍ Ｍ Ｍ ｴ Ｍ Ｍ ｲ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｌ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｲ Ｍ Ｍ Ｍ Ｍ Ｌ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ

1.001

...
1:1 1. 0005
......
,.....
o
ｾ 1
...I
<l
:0 0 .9995
0
AC 1 = 830 (

0.999

O.9985 Ｋ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｋ Ｍ Ｍ Ｍ Ｍ Ｍ Ｋ Ｍ Ｍ Ｍ Ｍ Ｋ Ｍ Ｍ Ｋ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｉ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｋ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｉ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ
780 800 820 840 860 880 900 920
Temperature, °c

(b)

1.0015 '1

I I
1.001 I

I
ｾ 1.0005 --- r--
ｾ
.......
,..... .,.
Ｎｾ
. Ie
o rr :'1" IT_
t- h:i.
"...,

"'
ｾ 1 "I'
...I
<l
:c 0.9995 I--
ｾ I

AC 1 = 826°( I Ii
0 .999 II

0.9985
780 800 820 840 860 880 900 920
Temperature, °c

(c)

Figure 4 .13 Thermal cycle (a) and AC I measurement of parent P92 during (b) the first heating
cycle and (c) the second heating cycle

89
 t-
1. 00 15 ｾ ---,----.,...--

1.001 I
... 1.0005
1:1
-.....
"'""
C 1
...J
-.....
...J 0 .9995

720 740 760 780 800 820 84 0

Temperature, °C

(a )

1.0015 I
1. 00 1 ---fo--
... 1.0005
1:1
-.....
"'C"" 1
...J

ｾ 0.9995
'-'
1:1 0.999
AC 1 =792 °C
0 .9985

0.998
700
J 720
--+--
740
Ｍ Ｍ Ｍ Ｍ Ｋ Ｍ ｾ Ｍ Ｋ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｋ Ｍ Ｍ Ｍ ｉ Ｍ Ｍ
760 780 800 820 840
Temperature, °C

(b)

Fig ure 4 . 14 ACt meas ure ment of WMA during (a) the first heating cycle and (b) the second
hea t ing cycle

The effect of the heating rate on the measured AC l value was also
investigated for the parent metal and WMA. Heating rates of 1000Cjhr,
50 0Cjhr and 20 0Cjhr were used. The results are shown in Figure 4.15.

90
 • Parent _ WMA

840

830

u
820
•
0
810
...
QJ

::l
ｾ 800 -t-
...

t
QJ
Q. 790
E
QJ
•
•
780
....f-
<
770

760 -- -

750 L - - - ----,-----.,-------,------,--
o 20 40 60 80 100 120
Heating Rate °C/hr

Figure 4 . 15 The effect of hea t ing ra t e on meas ure d AC I t emperatu re for pa rent P92 stee l and
weld fille r metal (WMA) chosen for later creep test s. Th e valu es with a hea t ing rate of zero
represent t he calculated Ael valu es.

4.3.1.3 The Effect of Element Variation on Ael

Th ermoCa lc was used to investigate wh ich elements have the greatest

effect on t he predicted Al temperature. In itial ly, a base composition for P92
steels was re quired for which a ba seli ne Ae l temperature could be defin ed .
Table 4.4 shows the allowe d m inimum and maximum compositions of each
eleme nt according to th e pa re nt P92 specification with increased limits for
manganese an d nicke l, as wel l as add it ional elements such as cobalt and
copper that are prese nt in we ld metals (Richardot et al. 2000a). The base
composition fro m which t he variations of elemental compositions were to
be made was defined as being t he m id -po int of each elements range and is
also identified in Table 4.4. Using t his composit ion, a base value for the
Ael was established as bei ng 800.6°C. The compositions of individual
elements were then varied in tu rn to thei r max imum and min imum
allowable compositi ons according t o the specification limits indicated in
Ta ble 4.4 while all ot her elements remained at the base value with the iron
content making up the balance, and the Ael temperature of these
compositions predicted using Th ermoCalc. The influences of the ind ividua l
ele m ents (across the ra ng e of t he specification composition for the
element) are sh own in Fig ure 4.16.

91
 Table 4.4 Minimum and maximum compositions of each element according to the parent P92 specification with increased limits for Mn, Ni and the addition of Co and Cu.
The selected composition was used as a base from which to investigate the effect of individual elements on Ael when varied between the maximum and minimum.

Name C N B Si Mn P S Cr Mo Ni Co Cu Nb V W
Weld
0.07 0.02 0.001 0.001 0.3 0.001 0.001 8.5 0.3 0.001 0.001 0.001 0.04 0.15 1.5
minimum
Weld
0.13 0.07 0.007 0.5 1.0 0.02 0.01 10.0 1.5 0.7 1.1 0.04 0.09 0.25 2.0
maximum
Selected
0.1 0.045 0.004 0.25 0.65 0.011 0.006 9.25 0.9 0.351 0.551 0.021 0.065 0.2 1.75
composition

92
 830

820

ｾ - 8 10
ｾ
a
'"c.ｾ 800
.
E
f-
--
0; 790
<{

780

770 +---__,..__- Ｍ ｾ Ｍ Ｍ ｲ Ｍ Ｍ Ｎ ｟ ｟ Ｍ ｾ Ｍ Ｌ Ｎ ｟ ｟ _ ｟ ｟ Ｌ Ｎ Ｎ ｟ ｟ ｟ ｟ ｟ ｟ Ｎ Ｍ ｾ Ｍ ｾ Ｍ ｾ Ｍ ｟ Ｌ

c N B Sl Mn P s Cr Mo Ni Co Cu Nb V w
Element

Figure 4 . 16 Effect of in dividu al ele ment variation from th ei r base levels t o the P92
specification limits on the Ae , temperature . Dark sha ding within the bars indicates the
element at its maximum allowa ble co mposi tion whil st th e light shading within the ba r
indicates its m inimum co mpositio n. Red bars are austenite stabilizers, bl ue are ferrite
stabi lizers .

It is shown that in all cases, the values of the Ael temperature predicted
with ThermoCalc varied approximately linearly with the composition of
ind ividual element between the min imum and maximum values defined for
that element in the speci fica t ion (the method used was the same as that
published by Santella (Santella 2010). In addition, simultaneous variations
of multiple elemen ts were also conducted; it was shown that the variation
in Ae l with each elem ent was essentially independent of variations made in
other elements . Accordingly, the Ael temperature sensitivity for each
element within the compos itional range could be calculated independently,
and values for these are shown as the element coefficients in Equation 4.l.
The elements that have a negative Ae l temperature sensitivity result in
re du cti on in the Ael temperature with increasing element composition in
the alloy and vice versa .

93
4.3.1.4 An Equation for Calculating AeJ in P92
Equation 4.1 predicts the Ael temperature for any alloy within the
compositional range shown in Table 4.4, as follows:

Ae1(°C) = S09.7 - 99.5C -196.4N - 67.4Mn -78.SNi - 9.85Co - 47.2Cu + 20B

+ 10.9Si + 35.S? + 95.65 + 3.7Cr + 19.7Mo + 31.4Nb + 53.8V
+ 10.6W

Equation 4.1

where the symbols indicate the composition (wt%) of that alloying element
in the alloy in question (within the range for P92 steels as defined in Table
4.4). The equation was derived by first using ThermoCalc and the selected
composition in Table 4.4 to calculate a value of Ael (corresponding to
809.7°C). Following this, each element was individually varied to its
maximum allowable value and minimum allowable value and the
corresponding Ael calculated. In this way, elements that increased or
decreased the Ael temperature were identified. The element coefficients
correspond to the increase/decrease in Ael temperature that a 1 wt%
increase in that element would give. Table 4.5 shows a comparison of the
Ael temperatures directly calculated using ThermoCalc compared with
those calculated with Equation 4.1 (itself generated from ThermoCalc
outputs) for the compositions of the materials of interest in this work
(shown in Table 4.1) and for two parent P92 steels where a detailed
composition was available in the literature. It is again notable that across
the three weld consumables employed in this work, a difference in the Ael
values of 13°C is predicted by Equation 4.1, which is slightly higher than
the 8°C difference in measured ACl as shown in Figure 4.12.

Table 4.5 Comparison of Ael temperatures of seven compositions of P92 steel both from use
of Equation 4.1 and by direct calculation with ThermoCalc

Ael eq. Ael TCW

Name (oq
4.1(oq
Selected
800 800
Compo
Parent P92 816 819
Sawada
(Sawada et 821 817
al.2001b)
P.l. Ennis
(Ennis et al. 828 821
2000)
WMA 758 758
WMB 769 769
WMC 771 774

94

I
4.3.2 PWHT Temperature Dependence
Due to the difference between the calculated values of Ae1 and measured
va lue of AC l fo r the range of materials examined (see Figu re 4.12), an
investigation into the use of ca lculated Ae l and measured AC1 temperatures
as a means to identify suitable conditions for PWHT was conducted.
Samp les from WMA were machined into specimens for dilatometry (see
section 3.1.6) and a cycle simulating a PWHT was conducted wh ich
involved heating at a rate of 100°C/hour, holding at a set temperature for
180 m in and then cooling at 50°C/hour (these rates were determined using
BSI " 2633: 1987 Specification for Class I arc welding of ferritic steel
pipework for carry ing flu ids" and discussions with Supergen industrial
partners ). The ho ld temperatures selected were 760 °C (typical of PWHT for
th ese weld typ es) and then a range of higher hold temperatures (namely
770°C, 780 °C, 800°C, and 820 °C) chosen to examine the role of hold
temperature on the austenitization of the welds and the subsequent
m icrostructural characteristics following cooling to room temperature. The
use of these temperatures was designed to provide ins ight into the effects
of PWHT at erroneously high temperatures on the weld metal properties.

4.3.2.1 Thermodynamic Modelling of the a y Transformation

Figure 4.17 shows the transformation of ferrite to austenite as calculated

by ThermoCalc for WMA, showing that under equilibrium conditions these
test temperatures lie between Ae 1 and Ae 3 (volume fraction is presented to
enable comparison with dilatometer results).

- - Ferrite - - Austen ite - - MX - - MnS - - M23C6

0 .9
0.8
OJ
§ 0.7
o
> 0.6

e
c·
0 .5
ｾ 0.4

ｾ " 0 .3
VI

Il.
0.2
0.1

ｯ ｾ ］ ］ ］ Ｍ ｾ ］ ］ ｾ ］ ］ ］ ］ ］ ］ ］ ］ ｾ ］ ］ ］ ］ ］ ］ ］ ］ ｾ ］ ］ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ
750 770 790 8 10 830 850
Temperature, ·C

Figure 4 .17 The transfo rmation of ferrite to austen ite under equil ibrium conditions, as
calculated by ThermoCalc for WMA

95
4.3.2.2 Measurement of Austenite Formation During PWHT
In order to experimentally quantify the phase fractions of ferrite and
austenite from the dilatometric curves, the lever rule was used . This
involves first determining the thermal ex pansion of ferrite and austenite . A
sample of WMA was heated at a rate of 10QoCjhr from room temperature
to lOOQoC in order to ensure that complete transformation had taken
place. The length change was measured using the dilatometer and the
result of this is shown in Figure 4 .18. Temperature is shown on the x- axi s
while strain, derived from the change in length ｾ ｌ divided by the orig inal
length La, is on the y - axis. The thermal expansion of the individual phases
is re presented by the extended li near sections .

- - fernte - - Transformation - - Austenite - - Linear (ferrite) - - Linear (Austenite)

0 .0 16

0.0 14

0 .0 12

0.0 1
.:i
ｾ 0 .008

0.006
y; 3.0 I E-OSx - 1. 64E -02
0 .004

0 .002

ｏ Ｋ ｾ Ｍ Ｍ Ｍ Ｍ Ｍ Ｇ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ
o 200 400 600 800 1000 1200
Temperatu re, °C

Figure 4 . 18 Th erm al ex pan sio n during continuous heati ng at 100' C/ hr of ferrite and au stenite
for WMA sa mples

The following figures show t he dil ato m eter curves at th e tested hol d
temperatures.

96
 - - 760 PWHT - - Linear (Ferrite) - - linear (Austen it e)

0.016

0.014

0.012

0 .01
.:J
::r 0.008
<l

0.006 y 3.0 1E-OSx - 1.64E -02

0.004

0.002

0
0 200 400 600 800 1000 1200
Temperature, °c

Fig ure 4. 19 Dilato m eter curve of t he 760°C PWHT of WMA . Th e dilatomet ry data for an
eq uivale nt samp le hea ted t hrough to 10000C ( From Fig . 4. 18) is also plotte d for co mparison .

- - 770 PWHT - - linear (Ferrite) - - Linear (Au stenite)

0.0 16

0.0 14

0.0 12

0 .01
.:J
::r 0.008
<l

0.006
y = 3.01 E-O Sx - 1.64E-02
0.004

0.002

oｾ ｾ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ Ｍ Ｎ Ｍ Ｍ Ｍ Ｍ Ｎ Ｍ Ｍ Ｍ Ｍ ｾ
o 200 400 600 800 1000 1200
Temperature, °c

Figure 4 .20 Dilatometer curv e of the 770°C PWHT of WMA . The dilatometry data for an
equival ent sample hea ted through to 10000C (Figure 4.18) is also plotted for comparison .

- - 780 PWHT - - Linear (Femte) - - Linear (Austen ite)

0 .016

0 .014

0.012

0.0 1
.:J
::r
<l
0 .008

0 .006
= 3 .01E -OSx - 1.64E-02
0 .004

0.002

0
200 400 600 800 1000 1200
-0.002
Temperature, °c

Figure 4 .21 Dilatometer curve of the 780°C PWHT of WMA . The di latometry data for an
equivalent sample heated through to 10000C (Figure 4 .18) is also plotted for compariso n.

97
 - - 800 PWHT - - linear (Ferrite) - - linear (Austen ite)

0.016

0.014

0.012

0.0 1
.J'
::J 0.008
<l

0.006
y = 3.0 1E-05x - J.64E -02
0.004

0.002

a
200 400 600 800 1000 1200
-0.002
Temperature, °c

Figure 4.22 Dilatometer curve of th e 800 °C PWHT of WMA . Th e dil atom etry data for an
equiva len t sa mpl e heat ed t hrough to 10000C (Figure 4.18) is al so plotted for compa ris on .

0 .016 - - 820 PWHT - - Linear ( Ferrite) - - Linear ( Austenite)

0.014

0 .012

0.01

0.008
.J'
Ｐ ｾ Ｐ Ｖ
0 .004 y = 3.0 1 E-05x - l.64 E-02

0.002

oｾ ｾ ｾ Ｍ Ｍ Ｍ Ｌ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ Ｍ Ｍ Ｍ Ｍ Ｍ Ｍ ｾ
200 400 600 800 1000 1200
-0 .002
Temperature , °c

Fig ure 4.23 Dilatom eter curve of the 82 0°C PWHT of WMA. The d ilatomet ry data for an
equ iva lent sa m ple heated t hrough to lOOO °C (Figure 4 . 18) is also plott ed for compari son.

The va lue of stra in for pure ferrite and pure austenite can be calculated
from t he equat ions of linear expansion. A line can then be drawn between
t hese two li nes or th ei r extrapolations (the ferrite-austenite tie-line at a
given t empera t ure ) as shown in the schematic diagram below (Figure
4 .24). Th e va lue of strain for the PWHT sample after the 180 min hold can
be extracted from the dilatometer data. This value is subtracted from the
strain of pure ferrite. The resulting value represents the portion of the tie -
lin e t hat is austen ite, which is converted to a percentage of the line as a
whole giv ing the amount of austenite formed in the weld metal during the
high te mpera tu re hold of the PWHT. Table 4.6 shows the results and Figure
4 .25 shows the rate of austenite formation.

98
 a = 1.58 x 1 o-sr - 1.65 x 10- 3 =
strain of vure f errit e
b = 3.01 x 10 - s r - 1.64 x 10- 2 =
strain o f pure aust enite

c = tran s formation strain

a- c
% austenite = --b
a-
x 100

Figure 4.24 Schematics showing how th e value of strai n for pure ferrite and pure aust enite
ca n be ca lcula t ed from the equ at io ns of linear ex pansion and th en used to calculate the
percen tage of austenite at that t emperatu re

Table 4 .6 Am ou nt of auste nite after 180min at th e hold temperature, before coo ling

010 Austenite predicted

PWHT Conditions, °c 0/0 Austenite
by ThermoCalc
760 1.5 1.5
770 8 .3 9 .9
78 0 29 .5 31.6
800 61.3 63 .7
820 84.4 88.4

- - 760 - - 770 - - 780 - - 800 - - 820

100
90
80
70
2! 60
'c
2!VI SO
::>
«
ｾ 40

30
20
10
0
0 20 40 60 80 100 120 140 160 180
T im e f ro m reaching PWHT tem perature, m in

Figure 4 .25 The rate of austenite formation upon reaching the specified PWHT temperature

99
 4.3.2.3 Microstructure Following PWHT

Hardness tests were carrie d ou t fo llowi ng the simu late d PWHT and th e
resu lts of this are shown in Figure 4 .26. Ten indents w ere ta ken at a
spacing of 1 mm alon g t he vert ical axi s of th e samp les. Th e averag e
hardness was 263. 5 kg f mm -2 for 760 0
e, 256.1 kgf mm -2 for 770 0
e, 237 .8
2
kg f mm - for 780
0
e, 213.8 kgf mm-2 for 800 0
e and 208 .3 kgf mm -2 for
82 0°(, Figure 4.2 7 shows optical images of the microstructu res of the weld
metals. I n order to identify ma te rial t hat transformed to austen it e, EBSD
imag ing of the microstructures was carrie d out and the re sults of this are
presented in Figure 4.28 show ing th e mostly martensitic microstructure at
760
0
e (a) devel oping to an equ iaxe d m icrostructure at 820 0
e (d) .

• 760 . 770 . 780 x 800 e 820

300 ...
290
280
270
• , • • • •
;;: 250
260
...• •
•
- .., ••
s: 240 A A

-
A

230
• ...
ｾ
&

220
:I(
:I:

... • .. :I:
•
..
:I(
210
200
:I:
; • ...
190
•
o 2 3 4 5 6 7 8 9 10
Distance from the top (where the probe rested during testing) of the sectioned dllatometry
sample, mm

Temperature, °c
760 770 780 800 820
256. 1 237 .8
I Mean Hardness 263.5 213.8 208.3
Standard DeViation 16.3 13.2 12.6
I 12.2 9 .4

Figure 4 .26 Hardness profile of the we ld meta l after th e si mulat ed PW HT . Th e numbers In t he

legend in dicate th e hold t em peratu re in ·C

100
Figure 4.27 Optical images of the PWHT weld metal A following hold at temperatures of (a)
760 D e, (b) 770 o e, (c) 780 D e, (d) 800 D e and (e) 820 D e . Each image shows a clearly visible
martensite microstructure but limited visible evidence of other transformation products

101
 (a) (b)

(c) (d)

111

00 I 101

(e)

Figure 4.2 8 200IJm x 200IJm inverse pole figure EBSD images of the wel d metals after PWHT
at (a) 760 0 C, (b) 770°C, (c) 780 0 C, (d) 820 ° C; (e) shows the inverse pole figure key . Using
EBSD clearly shows a more equ iaxed grain structure in (c) and (d) res ulting from
transformation to austenite during PWHT and recrystallisation during cooli ng

102
4.4 Weld Consumable Creep Testing
4.4.1 Creep Test Results
Creep tests were performed in air at three different temperatures and
nominal stress conditions, given in Table 4.7. Constant uniaxial tensile
loading was applied in the direction of the specimen axis, transverse to
that of welding. Strain measurement over the gauge length was regularly
logged using output from extensometers. The recorded rupture times and
failure strains are presented in Table 4.7, and the full set of strain-time
curves can be seen in Figure 4.29.

Table 4.7 Weld metal creep test results

Weld Metal Temp. (OC) Stress (MPa) Rupture Time (h) failure Strain (%)

A 625 150 2240 10

A* 625 150 1669 9

B 625 150 1301 4

C 625 150 1609 21

A 625 170 1090 15

B 625 170 656 7

C 625 170 696 16

C (rpt) 625 170 574 15

A* 670 90 983 14

B 670 90 668 3

C 670 90 865 14
* - tested by Corus UK Ltd., (rpt) - repeat test

103
 A 625°C 150MPa - - A 625°C 150MPa*
25
- 8 625°C 150MPa =C 625°C 150MPa

20

ｾ
ｾ
15
c:

C/)
......
til

10

5
ｾ
ｾ ｾｉ
0
--
0 500 1000
Duration (h)
1500 2000 2500

( a)

A 625°C 170MPa - 8 625°C 170MPa

25
=C 625°C 170MPa - - C 625°C 170MPa

20

ｾ
ｾ 15
c:
.j!
...
C/)
10

1500 2000 2500

Duration (h)
(b)

25 A 670' C 9OMPa* - 8 670' C 90MPa c=oC 670· C 90MPa

20

-...
ｾ
0

c:
l!
15

(/)
10

0
0 500 1000 1500 2000 2500
Duration (h)
( e)

Figure 4.29 Strain-time curves for all creep test conditions; (a) 625°C lSOMPa, (b) 625°C
170MPa and (c) 670°C 90MPa

104
It was found that specimens removed from the weld pad made using
consumable electrode A (i.e. WMA) gave consistently longer rupture times
for all test conditions. WMB had the shortest time to rupture, exhibiting
only 60% of the creep life of weld metal A at the higher stress (150 and
170MPa) tests, with little tertiary creep and comparatively poor ductility.
WMC exhibited a decrease in the time to rupture at the highest stress
(170MPa), and an unusually high amount of primary creep strain (Figure
4.29b). A repeat test confirmed the consistency of the markedly different
primary creep behaviour of WMC at 625°C/170 MPa. This repeat test also
gives an indication of the amount of scatter present in the results.

As to be expected, the time to rupture of all three weld metals decreased

with increasing temperature, and the rupture times at 625°C/170 MPa and
670 o C/90 MPa are broadly the same. At 670°C under a more realistic
power plant operating stress of 90 MPa (but at a much higher temperature
than would be experienced in a plant environment), the rupture times of
the three weld metals are relatively closer together; however, the failure
strain of WMB remained significantly lower than that of WMA and WMC.

Minimum creep strain rate, Emin and time to rupture, t r , are often found to
be related through the Monkman-Grant equation (Equation 4.2);

Equation 4.2

where kMG is the Monkman-Grant constant for the material and a is

material constant, typically about 1.0. If a material conforms to this
relationship, this may be regarded as evidence that the controlling
mechanisms for deformation and fracture are essentially the same over the
range of test conditions examined (Anderson et al. 2003).

105
 • A 625·C
• 8625· C • C 625·C o 8670· C .l C 670·C
ｾ A 625·C • ｾ A 670·C • A - 8 --- C

3.50

-
:S
3.30 •
....
GI
E
310
A --
--
A

-
GI
:::J
Q.
:::J
a:: 2.90
Q/I
.9 --A
270

2.50
-5.20 -5.10 -5.00 -4.90 -4 .80 -4 .70 -4.60 -4.50 -4 .40 -4 .30
log [Min. strain rate (l/hr))

Figure 4.30 Creep rupture times against minimum strai n ra tes (log-log) fo r weld meta ls A, 8
and C

Figure 4.30 shows the data plotted in th e form of Equation 4.2. It can be
seen that over the limited ra nge of test conditions examined, all three weld
metals exhibit the same cree p mechanisms and predictions of rupture time
could be made using the mini mum stra in rate and the Monkman -Grant
relation. It Is also evident from Figu re 4.30 that WMA exhibited a greater
rupture time for the same minim um strai n rate compared to WMB and
WMC.

4.4.2 Microstructure Following Creep

FollowIng creep rupture, microstructura l examination of the failed creep
specimens was performed parallel to the applied stress. Figure 4.31 shows
the microstructures of the three weld metals after testing at 625°C and
150 MPa. The weld metals under all test cond it ions conta ined MX, M2J C6
and Laves phase. In Figure 4.31 pairs of secondary electron (SE) and back
scattered electron Images (BSE) are shown. BSE images help differentiate
precipitates; Laves phase has a high contrast to the matrix, showing high
brightness (due to its high mea n atom ic number), while M2J C6 has a
contrast level similar to that of the matrix. Th e smaller precipitates are MX.
Of particular Interest was any difference in the size and precipitation of
Laves phase between the three weld metal s. A method described by
Korcakova (Korcakova et al. 2001 ) enables th e quantification of Laves
phase using SEM Images. Due to t he high mean atomic number of Laves
phase (due to high concentrations of tungsten and molybdenum), it
appears much brighter on BSE images compared to the matrix and other

106
 precipitates. Using the software ImageJ, the bright Laves phase can be
isolated, counted and the average size of the particles deduced. The Laves
phase in the 625°C and 150 MPa samples was quantified in terms of
average size with the results being shown in Table 4.8. Energy-dispersive
X-ray spectroscopy (EDX) analysis was also carried out on the Laves phase
present. While the precipitate size was too small for accurate compositional
measurements, it did reveal that the Laves phase in weld metals A and B
was tungsten-rich while Laves phase in weld metal C was molybdenum-
rich.

(a)

' .
.4
.
Ｎ ｾ
'\ ｾ

:-
.. ,
, .
ｾ
" ,
•

I
. •. «
,tI'

11 . .-

.- ｾ

ｾ
-
4, '. '1,,' 10' 1)1. I,' ,,, I ... '" "
:11; ｾ Ｎ
"' . ' '< t ".'j !'.,

(e) (d)

(e) (f)

Figure 4.31 SE and BSE images of precipitation in the three weld metals A (a) and (b), B (c)
and (d), C (e) and (f) after creep rupture at 625°C and 150MPa. Spherical precipitates with
low contrast to the matrix in BSE images are Mn C6, precipitates with a high contrast to the
matrix In BSE images are Laves Phase, small precipitates with low contrast to the matrix fn
BSE images are MX.

107
 Table 4.8 Approximate quantification of Laves phase in the three weld metals following creep
testing at 625°C and 150MPa

Approximate Mean diameter

Weld Metal Number of Particles Measured
(nm)
A 1051 852
B 998 750
C 888 845

It should be noted for Table 4.8 that using SEM to quantify and measure
precipitates of this size introduces uncertainty; the signals produced by the
electron beam have a minimum resolution of approximately llJm so
anything smaller then this cannot be accurately measured. Therefore the
mean diameter and number of particles measured are indications only.

Creep rupture tests showed that weld metal B had a low rupture strength
compared to the other weld metals, despite close compositional similarities
with weld metal A. Microstructural investigations revealed that weld metal
B contained dense regions of precipitate-free zones near the fracture in all
tests, with these precipitate-free zones exhibiting heavy creep damage
(Figure 4.33). Figure 4.32 shows a precipitate free zone (PFZ) in the
sample tested at 625°C and 150 MPa; creep damage is clearly visible
within the PFZ and is highlighted well using back scattered electron images
where damage shows up as black regions. For the specimen as a whole,
cracks of the type observed in Figure 4.32 are localised to these PFZs,
indicating a causal link between these two features. The initial examination
of the weld pad did not reveal any PFZs, although the initial examination
took only a small section of the weld pad.

Hardness tests were carried out on the PFZ and the bulk specimen in weld
metal B following creep testing at 625°C and 150 MPa. Figure 4.34 shows
that the PFZs are harder than the bulk of the creep specimen.

108
 (b)

Figure 4 .32 SE (a) and BSE (b) images of creep damage in weld metal B concentrated in
precipitate free zones

109
 Direction of stress
Fracture surface

10 mm

Direction of stress

(b)

Figure 4.33 Optica l images of precipitate free zone (PFZ) in weld metal B tested at 625°C and
150 MPa showing (a) PFZ concentrated near the fracture surface and (b) damage
concentrated in PFZ near to the fracture surface as indicated in (a)

110
 ( a) (b)

• Bulk Material _ PFZ

320 --
310
--
300
290
• •
ｾ 280 •
ｾ 270 ....
260 .... ....
250
• ｾ

• .......
• •
ｾ

240 r--
230 •
o 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Position of hardness indent with standa rd error, see micrographs

(c )

Figure 4.34 Hardn ess ind ents in (a) precipitate free zone (PF2) and (b) the bulk specim en in
we ld m etal B tested at 625°C and 150 MPa showing that (c) the PFZ are harder than th e bulk
of th e cree p speci m en (only th e first indent at point 15 is shown , the repeated ind ent is
ignored)

Selected area EDX analysis was carried out on the PFZs and the
surrounding bulk material at a number of locations (Figure 4.35, Table
4.9) . The EDX analysis was carried out on 10 pairs of PFZ and bulk
material regions and the average compositions taken in order to identify
any patterns. Areas with minimal creep damage were chosen in order to
minimise any possible effect of creep voiding on the results. Elements such
as sulfur and phosphorus were excluded from the analysis due to their
small wt% and the difficulty in accurate measurement of the compositions
of these elements. It should be noted that elements such as tungsten are
also difficult to measure in EDX due to the intensity of the peaks being low
and overlapping with other elements therefore the mean compositions of
the two regions are only indicators of the composition difference.

111
 , , l
Ｂ Ｂ ｾ ｾ ｚ ｨ ｟ ｯ Ｂ .. 1110(22\.1'1
Spectrum Si v Cr Mn Fe Ni Nb Mo w
Spectrum 1 0 .27 0 .33 8.52 1.12 86 .27 0.58 0.34 0 .73 1.95
Spectrum 2 0 .59 0.07 9.06 1.03 86 .22 0.69 0 .11 0 .59 1. 61
Figure 4 .35 Method used for EDX analysis of PFZ and surrounding bulk m ateria l in order to
id entify any composi t ional differences (elements are given in wt%)

112
 Table 4.9 EDX analysis of PFZ and bulk pairs in weld metal B following creep testing at 625°C
and 150 MPa

Bulk Spectrum 51 v Cr Mn Fe NI Nb Mo w
1 0.37 0.20 8.92 1.03 86.64 0.49 0.06 0,55 1.75
2 0.36 0.22 8.79 0.97 86.54 0.47 0.06 0,57 2,13
3 0.38 0.18 8.67 1.00 86.63 0.52 0.08 0,65 1.89
4 0.41 0.16 9.00 1.03 86.40 0,47 0.10 0.56 1.87
5 0.28 0.20 8.92 0.91 86.45 0,38 0.05 0.58 2,24
6 0.34 0.16 8.91 1.08 86.38 0,36 0.05 0,42 2,38
7 0.29 0.27 9.00 0.98 86.44 0.49 0.04 0,54 1.95
8 0.44 0.17 8.81 1.02 86.63 0.36 0.03 0,56 2,06
9 0.35 0.18 9.03 1.10 86.41 0,39 0.04 0,57 2,02
10 0.33 0.24 8.85 1.07 86.27 0.40 0.04 0,55 2,26
Mean 0.36 0.20 8.89 1.02 86.48 0.43 0.06 0.56 2.06
Standard Deviation 0.05 0.04 0.11 0.06 0.13 0.06 0.02 0.06 0.20

PFZ Spectrum 51 v Cr Mn Fe NI Nb Mo w
1 0,39 0,27 8,85 1.08 86,16 0,45 0,02 0,72 2.08
2 0,50 0,32 8,95 1.09 85,67 0,51 0.11 0,81 2,05
3 0.41 0,19 8,94 1.17 86,13 0,42 0,02 0,44 2,31
4 0,33 0,22 9,19 1.02 85.18 0,51 0,16 0,68 2,71
5 0.43 0.27 9.19 1.07 85.21 0,55 0,11 0.56 2.61
6 0.35 0,31 9,00 1.06 85.43 0.43 0.09 0,69 2.63
7 0.45 0,16 8.86 1.15 85.94 0.36 0,07 0,77 2,37
8 0.38 0.19 9.02 0,99 86,24 0,51 0.22 0,44 2.45
9 0.41 0.24 9,18 1.12 85.40 0,54 0,06 0,70 2.48
10 0,34 0.24 8,91 1.15 85,59 0,55 0,05 0,60 2.57

Mean 0.40 0.24 9.01 1.09 85.70 0.48 0.09 0.64 2.43

Standard Deviation 0.05 0.05 0.13 0.06 0.40 0.06 0.06 0.13 0.23

113
 5 Discussion

S.l Consumable Composition and Modelling

Figure 4.1 indicates that all three weld metals are predicted to go through
the expected transformation process during the solidification stage of the
welding process as follows; Iiquid/ ｾ Ｍ ｦ ･ ｲ ｲ ｩ ｴ ･ -> ｾ Ｍ ｦ ･ ｲ ｲ ｩ ｴ ･ /austenite ->
austenite -> austenite/a-ferrite -> a-ferrite (this sequence excludes other
phases such as carbides, nitrides etc). Table 4.2 indicates that the
transformation temperatures for all three compositions are different, the
reasons for which will be discussed later. Figure 4.2 shows the
thermodynamic predictions of the precipitation behaviour and of particular
interest is the region between 500°C and 700°C as the operating
temperatures for P92 steels in power plants lie in this range. All three
compositions are predicted to contain minor amounts «O.lwt%) of the
impurity phases M3 P and MnS. WMA and WMB have P92-similar
compositions so are predicted to form the precipitates phases MZ3 C6 , MX
and Laves phase that are found in the parent P92. WMC has a composition
based on steel B2 rather than P92 although the major precipitate phases in
this steel are similar to P92.

It should be noted that the compositions ahown in Table 4.1 have been
determined from the weld pads. The composition of the weld consumable
before use would have varied as during welding some elements will have
oxidised, therefore the amount measured in the completed weld pad may
be less then would be measured if the weld consumable was analysed.

All three contain the phase MX; however WMA and WMB are predicted to
contain more MX than WMC, while WMB and WMC are predicted to contain
two different types of MX. Both types are predicted to have an FCC crystal
structure which is in accord with the literature (Sourmail 2001). The
composition of the two types of MX was investigated using ThermoCalc and
the results of this are presented in Figure 4.3. The MX in WMA is predicted
to contain primarily vanadium with smaller amounts of niobium as the
metal, and primarily nitrogen with small amounts of carbon as the non-
metal. Figures 4.3 (a), (b) and (d) reveal that all three weld compositions
are predicted to contain this form of MX and also indicate that it is also the
most prevalent. The second form of MX predicted to be present in WMB
and WMC is rich in niobium with small amounts of vanadium as shown in

114
 Figure 4.3 (c) and (e). It is also only predicted to present in small amounts
as shown in Figure 4.3. These predictions are in line with similar
observations in the literature; Sourmail (Sourmail 2001) showed that MX
can take a number of compositions where M = Nb and/or V, X = C and/or
N, and can be referred to as carbides, nitrides and carbo-nitrides. Sourmail
described three types of MX precipitates that can be found in P92. The first
type is NbX that remains after austenitsation due to the niobium content
being higher than the limits of the mutual solubility of niobium andcarbon
and / or nitrogen in austenite at the austenisation treatment temperatures
used for P92 (Ennis et al. 1998; Maruyama et al. 2001; Zielinska-Lipiec
and Czyrska-Filemonowicz 2007). This is predicted as being present in
WMB and WMC and can be linked to niobium levels in these two weld
metals being higher than those in WMA (see Table 4.1). Sourmail also
described secondary MX phases forming during tempering and these have
been observed to take the form of NbX and VX, the majority of which are
vanadium-rich (Zielinska-Lipiec and Czyrska-Filemonowicz 2007). Whilst all
three weld metals are predicted to contain this form of MX (Figure 4.2),
the amount present in WMC is lower. This can be explained by the lower
amount of nitrogen in this composition, which is at approximately half the
level present in the weld metals WMA and WMB. While Sourmail describes
these MX precipitates as being either NbX or VX, the results of
thermodynamic modelling shown in Figure 4.3 reveal compositions that are
more complex, with either niobium or vanadium taking a dominant role in
the two forms but never being totally alone.

A mention should be made of the third type of MX precipitate that Sourmail

described, namely a V-wing complex. These occur when the large NbX
precipitates that remain undissolved during austenitization act as
nucleation sites for the VX that is produced during tempering, forming the
V-wing complex. The presence of this type of MX cannot be predicted by
ThermoCalc but it is possible that it could be present in both WMB and
WMC as both of these are predicted to contain the residual NbX at the
austenitising temperatures that could act as a nucleating site for a V-wing
complex (Figure 4.3).

The level of niobium needed to induce the formation of NbX was also
investigated using ThermoCalc, the result of which is shown in Figure 4.6.
Here the niobium content was varied within the limits of the P92
specification (see Table 4.1) while the other elements were held at the

115
 levels present in WMA. Figure 4.6 reveals that if the niobium levels were to
rise above 0.067wt%, the formation of NbC could be expected. The other
MX forming elements, namely carbon, nitrogen and vanadium are at
similar levels in WMA and WMB, but the niobium level in WMA is
0.043wt%, '" half of that in WMB ( 0.08wt%); this suggests that niobium
levels are likely to control the formation of NbX.

Both forms of MX are desirable in the steel. The fine MX precipitates

produced during tempering improve the creep strength of P92 by pinning
dislocations (Abe et al. 2007; Klueh et al. 2005), while the coarse NbX
particles that remain at austenisation are able to limit the growth of
austenite grains and ensure that the recrystallisation of martensitic laths
during tempering does not result in a loss of creep strength (Anderson et
al. 2003). The action of NbX could be of particular importance to a multi-
pass weld since this material undergoes numerous cycles of
recrystallisation during the welding process itself, meaning that higher
levels of niobium that encourage the formation of NbX could result in a
weld that has higher creep strength. A note of caution must be added to
this as Figure 4.6 also shows that the formation of NbX reduces the
amount of VX; as such, a balance would need to be struck between pinning
dislocations and reducing grain growth.

The primary carbide, M23 C6 , is predicted (see Figure 4.2) in all three
compositions with an FCC crystal structure, although the amount present is
predicted to vary; moreover, WMC is predicted to contain two forms of
M23 C6. The predicted compositions of M23 C6 are shown in Figure 4.4. In
WMA and WMB, the metal content is primarily chromium with some iron,
tungsten, molybdenum and manganese. The literature states that M23 C6
has a face-centre-cubic (FCC) structure (Sourmail 2001) and is typically
Cr23C6 with iron, molybdenum and nickel being able to partially substitute
for the chromium. However, ThermoCalc predicts that manganese rather
than nickel will act as the primary substituting element. The composition of
the M23C6 carbides is also predicted to vary with temperature, with the iron
and tungsten levels increasing with temperature while the chromium and
molybdenum levels are predicted to decrease. The compositions of M23 C6 in
WMC differ due to the lack of tungsten (which results in a higher
molybdenum content) and the addition of boron. The literature states that
together with the carbon, the boron gets incorporated in the growing M23 C6
(which could now be referred to as M23 (C,B)6) (Abe et al. 2004; Czyrska-

116
 Filemonowicz et al. 2003; Golpayegani et al. 2003); in addition,
ThermoCalc also predicts the presence of a small amount of M;n(C,B)6 with
the metal being comprised of chromium and iron. However, ThermoCalc
also predicts that at temperatures above 610°C, the M23 (C,B)6 will dissolve
and the boron will form BN. The BN is an undesirable phase as it reduces
the amount of MX present and removes boron from the M23 C6, which thus
limits its effect of reducing coarsening of this precipitate type (Czyrska-
Filemonowicz et al. 2003; Golpayegani et al. 2003).

Laves phase is predicted to be present in all three weld compositions but

only at lower temperatures in WMC. In contrast to M23 C6 which remains
relatively stable in the potential operating range of 600°C to 700°C, the
amount of Laves phase predicted to be stable in WMA and WMB reduces
rapidly with increasing temperature until fully dissolving at temperatures
around 700°C. In WMC, the Laves phase is predicted to dissolve at much
lower temperatures; however microstructural observations show it to be
present at higher temperatures and this will be discussed in more detail in
following sections. Despite this rapid reduction in the stable fraction of
Laves phase with increasing temperature, the composition of the phase
itself remains a stable mix of chromium, iron, molybdenum and tungsten
as shown in Figure 4.5. The primary element in this Laves phase is
tungsten, which is not present in WMC. The predicted form of Laves phase
in WMC consists of molybdenum and iron. The effect of the Laves phase
precipitation on long-term creep strength has been debated widely in the
literature, since it is known that the precipitation of Laves phase promotes
the depletion of molybdenum and tungsten from solid solution (Hosoi et al.
1986a; Hosoi et al. 1986b; Kunimitsu et al. 1991; Miyahara et al. 1995).
As such, Laves phase has been generally considered detrimental for the
creep strength of 9-12% Cr steels because of the associated reduction in
solid solution strengthening. On the other hand, the presence of Laves
phase can contribute to the increase of the creep strength by precipitation
hardening under certain circumstances (Hald 1996). Recently, it has been
generally agreed that fine Laves phase may contribute to the creep
strength (Dimmler et al. 2003; Hald 1996; Hald and Korcakova 2003;
Korcakova et al. 2001). As such, the lack of Laves phase predicted in WMC
will favour solid solution strengthening, while in WMA and WMB,
precipitation strengthening will be provided by Laves phase at the expense
of solid solution strengthening.

117
 Figure 4.2 (a) and (b) highlight a significant difference in the amount of
M23 C6 that is predicted to form in WMA and WMB, as well as a slight
difference in the amount of Laves phase. ThermoCalc was used to
investigate the effect of composition on the amount of precipitation of
M23 C6 and Laves phase, the results of which are shown in Figure 4.7. The
individual elements were varied within the limits of the P92 specification
while other elements remained at the levels present in WMA and the
temperature was kept constant at 625°C. Increasing the level of carbon
was shown to greatly increase the amount of M23 C6 present while varying
other elements has relatively little effect. From Figure 4.7 (a), the amount
of M23 C6 at a given carbon level can be observed and corresponds well to
the phase property diagrams of Figure 4.2. This explains why WMB, with
0.09wt% C, has significantly less M23 C6 then WMA which has 0.11 wt% C.
The amount of Laves phase present is affected by the levels of carbon,
molybdenum, tungsten and silicon. That Laves phase precipitation would
increase with increasing molybdenum and tungsten is not surprising as
these are its major elements. Figure 4.7 (a) indicates that Laves phase
decreases with increasing carbon content as the M23 C6 fraction increases;
this is because M23 C6 contains chromium, iron, molybdenum and tungsten,
all of which also make up Laves phase so that the former forms at the
expense of the latter. The effect of silicon on Laves phase precipitation is
interesting as silicon is predicted to remain in solid solution and does not
directly precipitate, yet an increased level of silicon results in an increased
amount of Laves phase. However, an increase in Laves phase due to
increased silicon has also been observed in the literature and described as
a kinetic effect, where higher levels of silicon have been shown to promote
nucleation of Laves phase and additionally to reduce its coarsening rate
(Hosoi et al. 1986b).

118
 5.2 Weld Pad Microstructure

Characterisation of the experimental weld pads was carried out using OM,
SEM and micro-hardness testing. Figure 4.8 shows OM images of the three
welds in PWHT state, which all have the expected martensite
microstructure and showed no weld defects such as porosity. No residual
i5-ferrite was observed in the welds, although the complex martensite
microstructure makes it difficult to identify other ferrite phases. Figure 4.9
gives a hardness profile over a length of 10 mm in each sample
encompassing approximately three weld beads. The average hardness was
294.2 kgf mm- 2 for WMA, 280.8 kgf mm- 2 for WMB and 297.4 kgf mm- 2 for
WMC. The hardness profiles show variation over the 10 mm profile,
particularly in WMB which had a maximum of 317 kgf mm- 2 and minimum
of 255 kgf mm- 2 • Variations in hardness were expected, due to the complex
multi-pass weld cycle, since the various HAZ features such as the FGHAZ
and CGHAZ will have differing hardness (see Figure 2.26). Figure 4.10
shows SEM images that were taken in order to characterize the precipitates
present in the weld metal. BSE imaging shows that the precipitates have a
similar contrast to the matrix, especially the larger precipitates. EDX
analysis revealed the larger precipitates to be chromium rich, indicating
that these are M23 C6 • The other precipitates were too small to be analysed
by EDX but, based on thermodynamic predictions and comparison with the
literature, it can reasonably be concluded that these are MX particles. No
Laves phase was observed in any of the weld pads, in contrast to the
thermodynamic equilibrium predictions. This is due to the widely reported
slow precipitation kinetics of Laves phase (Aghajani et al. 2009; Cui et al.
2010; Hald and Korcakova 2003); it is expected that the system will
eventually reach equilibrium and the Laves phase will be present in this
condition.

119
 5.3 The Effect of Composition on A1

It can be seen from Table 4.1 that the compositions of the weld
consumables have some key differences that set them apart from the
parent P92. The elements manganese and nickel are added in amounts
that exceed that which is allowed for parent P92. Nickel and manganese
are austenite stabilisers, lowering the ferrite to austenite transformation
temperature ACl. The main reason for adding these elements is to ensure
100% austenite formation (without the danger of ｾ Ｍ ｦ ･ ｲ ｲ ｩ ｴ ･ formation)
during initial cooling from the melt (and subsequent high-heat input during
multi-pass welds) thereby increasing the stability of austenite at high
temperature, and thus ensuring a 100% martensite structure when cooled
(Knezevic et al. 2008). However, their levels must be restricted to ensure
that during PWHT, there is no danger of reaustenitization. As can be seen
from Figure 4.1 and Table 4.2, the thermodynamic modelling has enabled
the prediction of the Ael temperature and the ｾ Ｍ ｦ ･ ｲ ｲ ｩ ｴ ･ start temperature
for all three weld consumables. If nickel and manganese are the most
important austenite stabilizers and, as such, are the only ones usually
considered (as suggested by Marshall (Marshall et al. 2002a», then it
would be reasonable to assume that the material with the highest
combined content would have the highest ｾ Ｍ ｦ ･ ｲ ｲ ｩ ｴ ･ start temperature and
the lowest Ael' The values in Table 4.1 give a (Ni+Mn) content of 1.42wt%
for WMA, 1.41wt% for WMB and 1.29wt% for WMC. Based on this, it would
be reasonable to expect WMA and WMB to have Aet and ｾ Ｍ ｦ ･ ｲ ｲ ｩ ｴ ･ start
temperatures which are very close to each other, with WMC having a
higher Aet temperature and a lower ｾ Ｍ ｦ ･ ｲ ｲ ｩ ｴ ･ start temperature. However,
Table 4.2 shows that this is not the case as WMA has the lowest Aet and
highest ｾ Ｍ ｦ ･ ｲ ｲ ｩ ｴ ･ start temperature (respectively, significantly lower and
higher than those of WMB) which means that its composition must confer
the strongest austenite stabilization. The composition of WMC gives a ｾ
ferrite start temperature that lies between the others and an Ael that is
the highest. This indicates that the elements nickel and manganese should
not be the only elements considered when designing the weld consumable
composition. Other elements within the three weld compositions are clearly
affecting both the ferrite and austenite stabilization to a significant degree.
The results shown in Table 4.3 and Figure 4.12 show that there is also a
difference between the measured ACt and the calculated Ael of typically
20-30 o C which may be understood as follows. Firstly, ACt is measured
under continuous heating conditions, and Yang and Bhadeshia (Yang and

120
 Bhadeshia 1989) have demonstrated that the higher the heating rate, the
higher the measured ACl value. To approach Ael experimentally would
require an infinitely slow heating rate. Secondly, Ael is produced from a
thermodynamic calculation that does not take into account reaction
kinetics. The transformation from o-+Y is diffusional (Bhadeshia 1999),
with austenite first forming at the prior-austenite grain boundaries, with
fine austenite then precipitating in the grains as the reaction progresses
(Shirane et al. 2009). The formation of austenite during the o-+Y
transformation is controlled by the rate of carbon diffusion as the
thermodynamic driving force for the formation of austenite increases with
increasing available carbon (Cool and Bhadeshia 1997); however in P92
steels, most carbon is held in the form of Cr23 C6 at the start of the o-+Y
transformation which means that these precipitates must dissolve to
release the carbon, leading to sluggish kinetics and thereby increasing the
measured Al temperature.

Recently, Santella (Santella 2010) has used computational

thermodynamics to predict the Ael temperature in 9wt% Cr steels, and has
presented the following equation for the prediction of Ael in parent P92:

Ae 1 (OC) = 778 - 80.6C - lSO.7N - SS.lMn - 68Ni + 22.9Si + 4.9Cr + 22.6Mo

+ 20.2Nb + 43.6V + 10.8W

Equation 5.1

The P92 composition range used by Santella was tailored towards the
parent material which means that the range of composition of certain
elements considered was narrower than that commonly observed in weld
metals used for P92, whilst other elements commonly present in the welds
(which are not present in bulk P92) were absent from the calculations.
Indeed, the compositions of P92 weld consumables are often very different
from that of the parent material, generally resulting in much lower Al
temperatures than the parent; in some cases, there is concern that the Al
temperature of these consumables may be low enough to result in
undesirable reaustenitzation during heat treatment. According to Marshall
(Marshall et al. 2002a), the tendency for undesirable reaustenitization of
P92 weld metals during heat treatment may be understood by considering
the sum of the nickel and manganese levels in the steel; specifically, he
proposed that the minimum ACl for P92 weld metal at l.Swt%(Ni+Mn) is
7900C and thus suggested that 775°C is therefore a safe upper limit for

121
 the PWHT temperature assuming that a margin of error of ± 15°C is
required. As was argued in this thesis previously, this (Ni+Mn) rule is not
deemed sufficient to describe the different Ael temperatures that the three
weld compositions exhibit.

Figure 4.16 indicates that the most significant elements (in terms of
fractional variation in the allowed compositional range) in controlling the Al
temperature of steels of this type (be they weld consumables or parent)
are as identified by Marshall (Marshall et al. 2002a), namely nickel and
manganese. However, other elements are observed to also have a
significant effect, the largest of these being for molybdenum. Marshall's
(Marshall et al. 2002a) proposed formula by which the potential for
reaustenitization of these steels was described by the sum of the nickel
and manganese contents implied that the effects of nickel and manganese
on this reaustenitization potential were equal. This simplification is
generally supported by Equation 4.1, where the influence of nickel and
manganese are shown to be very similar (with the effect of nickel actually
being 1.17 times that of manganese on a per wt% basis). To quantify the
effect of such a difference, it can be shown that for a composition defined
by the selected composition in Table 4.4 (but allowing nickel and
manganese to vary but their sum to be 1.5 wt%), Equation 4.1 predicts an
Ael temperature of 773°C if the alloy contained 0 wt% Ni and 1.5wt%Mn
whereas an Ael temperature of 762°C is predicted if the alloy contained
1.5 wt% Ni and 0 wt% Mn; this difference of over 10°C indicates the value
of the refined predictions of the Ael temperature which would be provided
by use of Equation 4.1 compared to use of Marshall's rule.

The recommendation that Mn + Ni < 1.5wt% (Marshall et al. 2002a) in 9-

12wt% Cr steels has previously been used as a primary factor in weld
consumable design for these steel types to ensure that the Al temperature
is not reduced so far that it may begin to impinge upon the standard heat
treatment temperatures for tempering of martensite. In addition, for weld
metal design, other elements typically remain in the range specified by the
parent alloy. However, Figure 4.12 shows that this recommendation does
not sufficiently describe the effect that the alloy composition has on the Al
temperature. Weld metals WMA and WMB both have similar (Ni + Mn)
content, yet the predicted Ael temperature and measured ACt temperature
show that WMA has a transformation temperature 1QoC lower than WMB.
Comparing the differences in composition with the aid of Figure 4.16 and

122
Equation 4.1 can help explain this difference. Weld metal WMB has silicon
and niobium levels approximately twice those of weld metal WMA and
Equation 4.1 shows that an increase in both of these elements increases
the Al temperature. Figure 4.12 and Table 4.3 also indicate that while
WMC has a much higher Ael than WMA or WMB, their ACl values are
relatively close. Recent work indicates that this can be attributed to the
reaction kinetics of the reaustenitization reaction since it has been argued
that increases in the levels of manganese, silicon and chromium retard the
kinetics of the a-y transformation (Miyamoto et al. 2010). Table 4.1
shows that weld metals WMA and WMB have a higher (Mn + Si) content
(compared to weld metal WMC) and this is likely to slow the
reaustenitization kinetics and thereby increase the measured ACl
temperature above the Ael temperature more than that for weld metal
WMC.

The P92 composition range used by Santella in the work on developing a

formula for the Ael temperature was tailored towards the parent material,
thereby introducing key differences between their equation (reproduced in
this thesis as Equation 5.1) and the equation derived as part of this current
work (Equation 4.1). In Equation 4.1, the elements manganese, chromium,
molybdenum and nickel have been given a greater range as these
elements are found in greater concentrations in weld metals (Table 4.1).
The austenite stabilisers cobalt and copper, present in the commercially
available weld consumables in this study, were not included in Santella's
study, but are included here. In addition, the elements boron, phosphorus
and sulfur were ignored by Santella as their effect on the Ai temperature
was thought to be small. Table 5.1 shows the differences between the
element coefficients in Equation 4.1 and Equation 5.1, and the effect that
these differences would have on the change in the predicted Al
temperature at both the minimum and maximum composition of that
element in the weld metals (the values of maximum and minimum
composition being taken from Table 4.4).

123
 Table 5.1 Comparison of element coefficients in Equation 5.1 (Santella 2010) and Equation 4.1, showing the difference in Ael that Santella's equation would produce at
composition minimum and maximum

V Nb S P B W Cu C N Mo Si Ni Co Cr Mn
Equation 5.1 coefficients
43.6 20.2 10.8 -80.6 -150.7 22.6 22.9 -68 4.9 -55.1
°C/wt%
Equation 4.1 coefficients
53.8 31.4 85.6 35.8 20 10.6 -47.2 -99.5 -196.4 19.7 10.9 -78.8 -9.85 3.7 -67.4
°C/wt%
Weld comp min
0.15 0.04 0.001 0.001 0.001 1.50 0.001 0.07 0.02 0.30 0.001 0.001 0.001 8.50 0.30
(wt%)
Weld comp max
0.25 0.09 0.01 0.02 0.007 2.00 0.04 0.13 0.07 1.50 0.50 0.70 1.10 10.00 1.00
(wt%)
Difference in Ael @ comp min
-1.5 -0.4 -0.1 0.0 0.0 0.3 0.0 1.3 0.9 0.9 0.0 0.0 0.0 10.2 3.7
(0C)
Difference in Ael @ comp max
-2.6 -1.0 -0.9 -0.7 -0.1 0.4 1.9 2.5 3.2 4.4 6.0 7.6 10.8 12.0 12.3
(0C)

124
 A number of elements which were not included in Santella's work have
been included in the current work such as boron, copper and cobalt. Of
these, the effect of including these (even at the highest composition of the
relevant element) is less than 1°C for the elements sulfur, phosphorus and
boron, 1.9°C for copper and 10.S0C for cobalt, indicating the importance of
cobalt in the calculation of Ael temperature. In addition, the differences in
the coefficients attached to the elements in the two equations results in
some significant differences, with both manganese and chromium leading
to differences of greater than 12°C at the highest composition of the
relevant elements, with somewhat smaller differences for other elements.
Moreover, whilst these differences are of course highest at the maximum
composition of each element in the weld metal, the difference remains high
even at the lowest composition of chromium. As such, although the
differences in Ael temperature predicted for the three weld metals
considered is less than 12°C (Table 4.4), the potential for much greater
differences cannot be ignored, and makes the use of Equation 4.1 more
appropriate for design of such weld metals.

The investigation into the dependence of the Ael temperature on

composition using ThermoCalc has quantified the relative influence of the
austenite and ferrite stabilisers and gives a clear indication that, within the
specified composition range, large amounts of nickel and manganese will
decrease the Ael significantly as has been previously observed. However,
the element coefficients in Equation 4.1 show that nickel actually
contributes more to the lowering of Ael on a per wt% basis, whilst Figure
4.16 indicates that the limits of the allowed compositional variations of
these two elements in the P92 specification will confer a slightly larger
range of Ael temperatures associated with nickel variations. The element
coefficients in Equation 4.1 highlight another factor that is not immediately
obvious from Figure 4.16. Elements that are present in minor wt% ranges
such as carbon, nitrogen, and sulfur have coefficients that are much
greater than those of elements present in large wt% ranges. The wt% of
these elements in the alloy is typically difficult to accurately control and
these coefficients demonstrate that even a minor increase or decrease in
the amount of these elements present in the alloys can alter the Ael
temperature significantly. The benefit of Equation 4.1 is that it will
facilitate the balancing of elements to produce an Ael at a desirable level.
However, as demonstrated by Cool (Cool and Bhadeshia 1997), elements
such as nickel and silicon do not just affect the transformation of C-+y but

125
also affect the formation and retention of ｾ Ｍ ｦ ･ ｲ ｲ ｩ ｴ ･ which must be taken
into consideration when designing weld consumables as the presence of ｾ

ferrite would be detrimental to mechanical properties.

126
 5.4 PWHT Temperature Dependence

The investigation into the Ael temperature of the weld metals has
highlighted and quantified the importance of alloy composition on the
transformation of o-y, but it has also highlighted the difference between a
calculated value of Ael and measured value of AC1. The investigation into
which is more applicable to PWHT focused on WMA as at this stage in the
SUPERGEN project, it had been shown that this was the weld consumable
that was expected to perform best under creep conditions. In order to
assess the formation of austenite during the PWHT, a dilatometer was used
as the literature had demonstrated that this was an effective means of
observing the transformation via changes in volume (see section 2.4.5).
The standard 2633:1987 (BSI 1987a) is currently used to determine the
PWHT parameters and from this, the recommended heating rate of
100°C/hr was obtained. The cooling rate of 50°C/hr was also recommended
in the specification, although this cooling rate is at the lower end of
allowable cooling rates (BSI 1987a). This low cooling rate was chosen so
as to investigate whether the decomposition of any austenite formed
during the heat treatments would result in un-tempered martensite or soft
ferrite. The selected hold temperatures of 760°C, 770°C, 780°C, 800°C and
820°C were chosen in order to investigate temperatures close to a
standard PWHT of 760°C ± 15°C while also investigating temperatures that
were within the observed transformation range during continuous heating
(Figure 4.18). ThermoCalc was also used to calculate the equilibrium
transformation of ferrite to austenite (Figure 4.17), presented in volume
fraction terms so it can be directly compared to the results of dilatometry
(Table 4.6).

Figure 4.16 shows the 760°C PWHT cycle and this closely follows the
expansion experienced in the original continuous heating diagram. During
the three hour hold, there is a dimension change that indicates'" 1.5 vol%
austenite has formed, which is in good agreement with the results of the
equilibrium (ThermoCalc) predictions for this temperature (Table 4.6).
During cooling, the variation of strain with temperature also closely
followed the original continuous heating diagram. In Figure 4.17, the
dilatation for the 770°C PWHT cycle is presented; in this case, there is a
clearly visible contraction during the hold, indicating 8.3 vol% austenite,
again in good agreement with the equilibrium (ThermoCalc) prediction
(Table 4.6). The 780°C PWHT exhibits an even higher contraction during

127
 the ho ld period, this occurring at a temperature well before the measured
ACl is reached, with 29 .5 vol% transformation to austenite (Table 4.6).
These three PWHT temperatures (760°C, 770 °C and 780°C) lie between the
calcu lated Ae l of 758°C and the measured Ac) of 792°C and all of them
ex hibit transformation to austenite during extended hold .

The 800 °C and 820 °C heat treatment cycles both exhibit significant
t ransfo rmation of 61.3 vol% and 84.4 vol% respectively (Table 4.6) . The
significant amount of transformation also allows the decomposition of the
austen ite to be observed, especially so in the case of the 820°C heat
treatment cycle wh ich has been reproduced below in Figure 5.1 with
annotations on the phases expected to be present at each stage.

- - 820 PWHT - - Linear (Ferrite) - - Linear (Austenite)

0.016

0 .014

0 .012

0.01
..J Tempered
::J 0 .008 Martensite + MX+ Martensite + MX +
<J
Mn C6 Ml)C 6 + austen ite
0.006

0 .004
Tempered
Iempered Martensj te + MX +
0 .002 Martensite + MX + M2)C. + austenite +
M2)C. + fernte ferrite
0
600 800 1000 1200
-0 .002 Tempered Martensrte +
MX + M2)C. + ferrite Temperature. ·C
+d lsplasive product

Fig ure 5 .1 Th e ph ase tran si tions in WMA during t he thermal cycle of a simulated 820°C PWHT
with a heating rate of 100° C/hour and a cooling rate of 50°C/ hour

The initial microstructure of the material in all samples is as described in

section 4.5.2, namely a martensite matrix with precipitates of M23 C6 and
MX. During the hold, the transformation to austenite has been
demonstrated but at the end of the hold, as the material never completely
transforms to austenite, in all cases the microstructure will be a mix of
austenite, heavily tempered martensite and the precipitates. As the
material cools, a large expansion is observed corresponding to the
transformation of austenite to ferrite, leaving a mixture of tempered
martensite, ferrite and the precipitates. Around 400°C there is a further
small dimension change, corresponding to the marteniste transformation
temperature (M s) for P92 steels (Richardot et al. 2000a), although the
volume change is small. This leaves the resulting microstructure following
a PWHT where there has been transformation to austenite consisting of a

128
 mixture of tempered martensite, un-tempered martensite, ferrite and
precipitates. This microstructure is undesirable as the ferrite has weaker
creep properties compared to the tempered martensite, whilst the
untempered marteniste will be initially brittle, but will temper to some
degree in service (remembering that service temperatures are much lower
than the standard PWHT temperatures). Faster cooling rates will result in a
greater proportion of the austenite transforming to un-tempered
martensite.

As has been previously highlighted, there is a difference between the

measured ACl and the calculated Ael' Table 4.3 and Figure 4.12 highlight
these differences. As the parent P92 steel used in the investigation had
already undergone a full normailizing and tempering heat treatment before
the ACl shown in Table 4.3 was measured, it was therefore prudent to
investigate what effect the material starting condition would have on the
measured ACl and this was conducted on both the parent P92 and WMA.
Figure 4.13 shows the results for the parent P92, demonstrating a 4°C drop
in the measured ACl from 830·C to 826°C after repeating the thermal cycle.
In comparison, Figure 4.14 shows the effect on the weld metal WMA,
where no difference in the measured ACl was observed. This can be
explained as follows. The parent P92 has previously undergone normalising
and tempering heat treatments; normalising creates a microstructure of
untempered martensite and when this is tempered, precipitation of the
carbides M23 C6 and MX occurs. This is the microstructure at the start of the
first thermal cycle, giving an ACl value of 830°C (measured using a heating
rate of 100·C/hr). This first thermal cycle is essentially a normalising
treatment which dissolves all the carbides and results (on cooling) in a
microstructure of untempered martensite. In this structure, the carbon is
in solution rather than having been bound up in precipitates. In the second
thermal cycle, this free carbon increases the driving force for the ferrite to
austenite transformation, thereby reducing the measured ACt value.

The situation is different for the weld metal (WMA). When the weld metal is
deposited, it is molten, cooling rapidly and solidifying to form a
microstructure that ends up as untempered martensite with an
inhomogeneous distribution of elements. As this is a multipass weld (in
these samples approximately 3 weld beads per sample), the weld will
undergo multiple additional thermal cycles. This results in a mixed
microstructure consisting of fine grain and coarse grain martensite with

129
little precipitation. In the first thermal cycle, the weld metal is heated and
the ACl temperature measured. It was then held at high temperature and
normalised, and then cooled, resulting in a microstructure with little
precipitation. As such, the microstructure following one thermal cycle was
very similar to that before the thermal cycle. As such, in the second
thermal cycle, and the measured ACl temperature was the same as in the
first thermal cycle.

Further investigations were conducted into the effect of heating rate on the
measured ACl value. Samples of both parent P92 and WMA were
continuously heated at rates of 20°C/hr, 50°C/hr and 100°C/hr and their
ACl values measured. The results of this experiment are shown in Figure
4.15, showing that for both the parent material and the weld metal, as the
heating rate decreases below the industry standard of 100°C/hr so does
the measured value of ACl which approaches the calculated value of
Ael.The question remains as to why the difference between the calculated
Ael value and measured ACl is greater in the weld metal than the parent
metal. As both materials exhibit a decrease in measured ACl towards Ael
with decreasing heating rate, this difference must be due to kinetics of
transformation. Miyamoto (Miyamoto et al. 2010) investigated the effects
of manganese, silicon and chromium additions to a iron-carbon alloy on the
reverse transformation to austenite. They first calculated the
transformation values using ThermoCalc (with database TCFE5) and then
measured the transformation temperatures with increasing alloying
additions. They found that increasing additions of the elements retarded
the transformation kinetics in the same way as observed in this work. A
comparison of the compositions of the parent metal and WMA (Table 4.1)
shows that managase is the element (of those studied by Miyatomo) with
the most significant difference (0.45wt% in the parent, 0.78wt% in WMA).
Oi (Oi et al. 2000) conducted a similar study that looked at both
manganese and nickel additions to an iron-carbon steel. They found that
the effect of nickel was similar to the effect of manganese. Comparing the
nickel content of the parent metal and WMA reveals that this is another
element with significant difference (O.27wt% in the parent, 0.64% in
WMA). It is therefore proposed that it is the high concentrations of these
elements in WMA which is responsible for the greater difference between
the calculated Ael and measured ACl' The experiments conducted in this
work (both on the effect of heating rate, Figure 4.15, and the temperature
of PWHT, Figure 4.19 - 4.23) demonstrate that the calculated Ael

130
 temperature is more relevant to design of PWHT schedules than the ACl
value obtained via continuous heating.

Figure 4.25 shows the rate of austenite formation during the PWHT. At
760°C, the rate of transformation is slow with measurable transformation
occurring after a hold time of 2 hours. At 770°C there is a faster rate of
transformation, with measurable amounts of austenite formation from the
start of the PWHT. The same is true of 780°C which again shows an
increase in the rate of austenite formation. For 800°C and 820°C there is
already a significant amount of austenite present by the time the hold
temperature is reached, while the rate of austenite formation is initially
rapid before stabilizing. The PWHT at 800·C shows that the rate of
austenite formation at this temperature falls to zero after about an hour; in
contrast, in the PWHT at 820·C, austenite continues to form throughout the
hold. This difference is likely to be due to the M23 C6 not dissolving at a
significant rate at the lower of these two temperatures (see Figure 4.2)
which means that the carbon required to force the transformation is still
locked up in the precipitates. At the higher temperature of 820·C, the
M23 C6 will be dissolving rapidly and releasing the carbon into solution to
drive the formation of austenite (Cool and Bhadeshia 1997). The reason for
the ever increasing initial rate of austenite formation with increasing PWHT
temperature is that both the diffusion rates and the driving force for the
transformation are increasing with temperature, resulting in the continuous
increase in the rate of transformation (Figure 5.2).

The average hardness of the samples (seen in Figure 4.26) shows a

softening of the microstructure following the PWHT. Some softening is
expected as the martensite tempers, but after PWHT at 800°C and 820°C,
the microstructures are particularly soft due to the presence of large
amounts of soft ferrite. There is also large scatter in the hardness values of
the lower PWHT temperatures which is generally expected in multi-pass
welds due to the differences in thermal profile that have been experienced
during the welding process. In the low temperature PWHTs (760·C, 770·C),
little re-transformation will have taken place, and so the varied hardness
profile remains. However, during higher temperature PWHTs, significant
transformation to austenite will have taken place; as such, it is likely that
some variation in hardness will be due to differences between the soft
ferrite or hard un-tempered martensite (the products of austenite
transformation on cooling) and the original martensite that is tempered

131
during the PWHT. Figure 4.27 shows optical images of the five PWHT
samples where it can be seen that it is difficult to identify the different
forms of ferrite that may be present in the microstructure. To this end,
EBSD was used due to its ability to determine crystal orientation and grain
boundaries. It can be seen from the EBSD images presented in Figure 4.28
that there is an identifiable difference in the microstructures of the samples
heat treated with the different hold temperatures. For the sample PWHT at
760°C, there is a clear martensite structure and the same can be said of
the sample PWHT at 770°C, although the structure at 770°C appears to be
finer (this is most likely due to the fact that the region imaged was an area
of FGHAZ, rather than being an effect of the increased PWHT
temperature). At 780°C and 820°C, there has been a clear change in the
microstructure, with equiaxed grains mixed with areas of martensite.
These equiaxed grains are likely to be the areas of soft ferrite, although
their growth has been limited due to the high levels of precipitates present
at a" the PWHT temperatures (see Figure 4.8).

Ｑ Ｔ ｇ ｙ ｾ ｯ ｬ
Ae
1
0
ｉａＨｉｇｾＮＬｉ

Time
Figure 5.2 The TTT curves for the Y-+o transformation and the reverse O-+y transformation
(Bhadeshia and Honeycombe 2006)

132
 5.5 Weld Metal Creep Testing

The earlier failure and low failure strain of WMB under creep has been
shown to be due to the presence of precipitate free zones which are
associated with the development of cracks during creep as shown in Figure
4.32. These zones experienced heavy creep damage as the precipitation
strengthening important to the creep strength of P92 steels was absent. It
is proposed that the PFZ has formed as the result of retained C5-ferrite in
the weld. The retention of C5-ferrite in weld metal for these types of steels
has been observed before (Cai et al. 1997; Falat et al.). They tend not to
be observed in initial examinations of the weld pads as C5-ferrite occurs in
either regions where the complex thermal cycle of multi-pass welding has
created the right conditions (a peak temperature above the C5-ferrite start
temperature, Table 4.2), or where weld solidification has created a
localised region high in ferrite stabilizers which lowers the C5-ferrite to
austenite transformation temperature. As these regions will be very
localised, they can easily be missed in initial examinations of a large weld,
only becoming apparent during creep testing as they are creep weak
(OnorO 2006). In Onoro's study (Onoro 2006), the morphology of the
retained C5-ferrite was found to vary depending upon the location within the
weld pad, with C5-ferrite in the centre of the multi-pass weld having an
elongated spherical morphology and C5 -ferrite near the fusion boundary
having a jagged and blocky morphology. All samples in the current study
were machined from the centre of the weld pad, where the C5-ferrite
morphology is expected to consist of isolated islands due to additional heat
treatments produced by further weld passes. This morphology is
undesirable because it has been shown to significantly reduce creep and
toughness properties of the weld metal due to the C5-ferrite remaining
precipitate free (Cai et al. 1997). The retained C5-ferrite remains
precipitate-free since, when the high temperature C5-ferrite first forms, it
contains little carbon due to the low solubility of carbon in ferrite (the
carbon will have partitioned to the austenite in which it has high solubility).
The rapid cooling experienced by the weld traps carbon in the newly
formed martenSite, from where it is consumed in the rapid precipitation of
the M23 C6 and MX precipitates. The reason why C5-ferrite, and the resulting
PFZ, has formed only in WMB is due to compositional differences in the
weld consumable. The first of these compositional differences to note is the

133
 (Mn + Ni) content. For weld consumables designed for the steel P91, it has
been proposed that (Mn + Ni) < 1.5wt% in order to keep the ACl
temperature in the range suitable for PWHT while still suppressing the
formation of l)-ferrite (Marshall et al. 2002b). As previously stated, in the
design of weld consumables for use with P92, this rule has been followed in
the design of this consumable. Based on these values, it would be
expected that WMA and WMB would have similar l)-ferrite formation
temperatures while WMC would have a lower one. Table 4.2 shows that
this is not the case as it is WMB that has the lowest l)-ferrite formation
temperature. This is because other elements also play a role in the
formation of c3-ferrite. High chromium or high silicon levels promote the
formation of c3-ferrite during austenitization at around 1100 0 C (Bhadeshia
and Honeycombe 2006). Through experimental investigation of chromium
and nickel equivalents, Onoro determined that in a 9Cr steel containing
tungsten and molybdenum, the silicon and vanadium content should each
be below 0.25wt%, while niobium should be below 0.006wt%, in order to
avoid the formation of c3-ferrite during welding (Onoro 2006). It can be
seen from Table 4.1 that WMB has twice the amount of the ferrite
stabilizers silicon and niobium compared to WMA, which has resulted in the
formation of c3-ferrite during welding. The small size of the Laves phase in
WMB, as shown in Table 4.7, is probably due to the higher levels of silicon
which have been shown to act as nucleation sites for Laves phase and
reduce their coarsening rate (Hosoi et al. 1986b). In addition to this, EDX
analysis of the PFZ and surrounding bulk material (Figure 4.35 and Table
4.9) has revealed compositional differences. The PFZ has a higher amount
of ferrite stabilizers, in particular tungsten. This has led to c3-ferrite being
stable at lower temperatures in these regions. It is worth noting that there
is difficulty in measuring the levels of elements like tungsten due to the
peak positions in the EDX spectrum (Figure 4.35), which have low intensity
and overlap with those of other elements. However, the general trend for
higher tungsten levels in the PFZ can be identified. It is likely that this
localised high-tungsten content has come about due to inhomogeneities in
the welding consumable. The weld consumables are modified 9CrMo basic
coated low hydrogen electrodes for high temperature creep applications,
which consist of a mild steel core rod coated by a flux containing the
necessary alloying elements. Discussions with industry have indicated that
elements can become segregated in the flux of the weld consumable,
particularly dense materials such as tungsten. When the welding is

134
 conducted, certain areas will have higher levels of tungsten introduced due
to the inhomogeneous nature of the weld consumable. It is likely that this
was the source of the tungsten imbalance and the cause of the PFZ which
lead to an early creep failure.

The high failure strain exhibited by WMC can also be attributed to

compositional differences. While it has the same precipitate phases as
WMA and WMB, the absence of the solid solution strengthening effect of
tungsten has resulted in a higher ductility and therefore a higher failure
strain. The addition of cobalt has not provided the same level of
strengthening, and the formation of Laves phase has removed
molybdenum from the matrix and thereby also reduced its solid solution
strengthening effect. Klueh (Klueh 2005) stated that at equilibrium, the
MOeq (where MOeq = Mo + 0.5W [wt%]) should not exceed 1%. If the steel
has a MO eq greater than 1 wt% when operating in the 600-650 o C range,
the tendency for the intermetallic Laves-phase to form is increased,
resulting in the removal of molybdenum from solid solution and a reduction
in solid solution strengthening (Klueh 2005). The MO eq of WMC is well
above this value at 1.41wt%. The high ductility exhibited by WMC can also
be attributed to compositional differences, since this material was based on
steel B2 rather than P92. This weld metal was expected to perform well
since the alloy on which it is based is judged to be significantly stronger
than P92 (Gianfrancesco et al. 2009). The current work and that published
by others (e.g. (Allen et al. 2007b» suggests that the compositional
factors which make a parent material creep-strong do not necessarily carry
across on a one-to-one basis to weld metals.

At this point it is important to refer back to section 4.1 and in particular

Figure 4.2 which showed the equilibrium phase fraction diagrams. After
being held at high temperature during the creep tests for long periods of
time, it is expected that equilibrium conditions will have been neared if not
reached. This means that precipitation of the Laves phase, (previously
described as having sluggish precipitation kinetics resulting in it not being
observed in the weld pads) should be complete. Indeed, Laves phase is
present in all three of the welds following the creep tests. However,
ThermoCalc predicted that there would be no Laves phase present in WMC
at these temperatures (see Figure 4.2c), but Figure 4.31 and Table 4.8
clearly demonstrate that it is present with EDX analysis also revealing that
it is molybdenum-rich. The Laves phase predicted by ThermoCalc is also

135
molybdenum-rich (see Figure 4.Sc) but was not predicted to be stable at
the temperature the creep test was conducted at. To understand the
differences between the microstructural observations and the ThermoCalc
predictions regarding the presence of Laves phase at these temperatures
for WMC requires an examination of the mechanisms of Laves phase
formation. There are two mechanisms for the nucleation and growth for
Laves phase, one of which is that Laves phase precipitates alone on
martensite lath boundaries, with the precipitate being coherent with one
grain but growing into an adjacent grain with which it has no rational
orientation relationship. The other mechanism is that Laves phase
precipitates are formed in the regions adjacent to M23 C6 particles, and they
then grow at the expense of the chromium-rich M23 C6 carbides in close
vicinity. It has been established that the former mechanism is the
dominant formation mechanism for Laves phase as it has been argued that
Laves phase precipitates prefer to locate on the prior austenite grain
boundaries and the martensite lath boundaries (Cui et al. 2010).
ThermoCalc predictions are based upon the principle of thermodynamic
equilibrium, but these weld metals have a martensitic matrix, which is
produced via a displasive process and cannot be predicted through
thermodynamic equilibrium calculations. Laves phase is present as the
martensite matrix in WMC provides numerous nucleation sites.

Overall, WMA was found to have the highest creep strength for the
conditions tested and is therefore most likely to be able to enable power
generation plant-realistic weld HAZ failures in cross-weld specimens within
reasonable test time.

The consumable that differed the most in terms of performance was WMB.
This material had significantly lower creep ductility and the highest
minimum creep strain rates.

136
6 Conclusions
The purpose of this thesis was to understand the behaviour of welds in P92
steel so that their in-service behaviour may be better understood. This
understanding has been achieved through a combination of
thermodynamic modelling, mechanical testing and microstructural
characterization.

It has been industry's experience that different heats of P92 material, all
within the P92 specification, can have very different mechanical behaviour
Thermodynamic modelling has enabled a better understanding of how the
composition affects the equilibrium phases of P92 and the knowledge
learnt here is applicable to both parent and weld material. Precipitation
strengthening is important to the creep resistance of P92 and the
modelling has revealed how precipitate levels vary based on composition.
The key findings are summarized below;

• High levels of carbon increase the amount of M23 C6 precipitated

while other elements have little effect (Figure 4.7). The P92
specification gives limits of 0.07-0.13 wt%C so there can be
significant variation between P92 heats, e.g. from the weld metals
studied here, weld metal A (O.l1wt%C) has 25% more M23C6 then
weld metal B (O.09wt%C).
• The amount of Laves phase is decreased with increasing carbon, but
is increased with tungsten, molybdenum and silicon.
• The stability of Laves phase rapidly decreases over the temperature
range of typically 600°C-70QoC which has implications for the
validity of accelerated creep testing used to determine the long
term creep strength of P92 at its operating temperature which is
typically 500°C-620°C.
• High levels of niobium will result in the formation of two different
kinds of MX precipitate, observed in the literature to have different
morphologies.
• The addition of boron results in two different forms of M23C6. The
literature indicates that boron improves creep resistance as it
reduces the coarsening rate of M23 C6.

Using this knowledge, quality checks on P92 used by industry can better
ensure a materials fitness for service if an accurate composition is known

137
and manufacturers can further tailor compositions to produce the strongest
possible material.

Some of the biggest problems that industry have had when using P92 (and
its predecessor P91) have been caused by incorrect heat treatments that
result in a non-martensitic material entering service (that fails early in
service life), or the cracking of material due to the presence of untempered
martensite (resulting in delays to construction, additional cost and unfit
material entering service). A combination of thermodynamic modelling and
mechanical testing was used to investigate the key transformation of
ferrite to austenite. This investigation focused on the effect of composition
on the transformation temperature, Al , and the rate at which austenite
could form during PWHT. The key findings are summarized as follows;

• The Al value produced by thermodynamic modelling (Ael) is

typically 30°C lower than the value arrived at via thermo-
mechanical testing (Aclt which is currently the industry standard for
determining maximum PWHT temperatures).
• A typical heating rate used by steel manufacturers to measure the
AC1 temperature is 100·C / hr. Experiments conducted in this work
have shown that as the heating rate decreases, the measured AC1
value approaches the calculated Ae1'
• Compositional modelling has revealed the degree to which
individual element variation can affect the A1 temperature. It is
demonstrated that the rule which defines that (Ni + Mn) ｾ 1.5wt%
is insufficient when designing weld consumables that are to be
PWHT at the industry standard of 760°C ± 15°C.
• An equation has been developed that predicts the Ae1 temperature
for any weld consumable for use with P92 using compositional data;

Ae 1 (°C) = S09.7 - 99.SC - 196.4N - 67.4Mn - 7S.SNi - 9.85Co - 47.2Cu

+ 208 + 10.95i + 35.SP + 95.65 + 3.7Cr + 19.7Mo + 31.4Nb
+ 53.8V + 10.6W
• Using thermo-mechanical testing, the transformation of ferrite to
austenite has been investigated for a material with an Ae1 = 758°C
and an ACl = 792°C. It has been demonstrated that, for a range of
PWHT temperatures between Ae1 and AC1, significant transformation
to austenite takes place.

138
 • Microstructural characterisation reveals that a slow cooling rate will
result in retained austenite forming soft ferrite while a fast cooling
rate results in the formation of hard untempered martensite.
• Based on the results of experiments conducted as part of this work,
it is suggested that calculations of Ael would be a more effective
way of determining key transformations for the development of
industry standards for high-alloy steels such as P92.

Using this knowledge, better heat treatments can be devised. The

knowledge of how composition affects the Al temperature is useful for both
welds and parent material, enabling the design and selection of P92
material that will not transform during heat treatment. The equation for
Ael allows the determination of maximum safe heat treatment
temperatures, and should be used in preference to currently measured ACl
values; this change will reduce the likelihood of poor quality material
entering service. The mechanical testing has demonstrated that during
PWHT (or parent material tempering), equilibrium conditions are
approached, confirming that Ael should be used to determine maximum
heat treatment temperatures instead of AC1'

Creep testing of three different weld consumables was carried out to find
out which had the best properties for use in service and the key findings
are summarized as follows;

• Creep testing of weld metal has indentified that 8-ferrite causes

early failure, as the resulting PFZs are creep weak.
• Localized 8-ferrite is caused by inhomogeneous distribution of ferrite
stabilizers such as tungsten, resulting in greater stability of 8-ferrite
and its retention in the weld.
• The source of this tungsten imbalance is discussed and attributed to
a weld consumable with poor tungsten distribution.
• Thermodynamic modelling has been used to calculate the
temperature at which 8-ferrite forms for a number of compositions.
• Due to the localized nature of the 8-ferrite, it is difficult to locate in
initial microstructural examinations, only becoming apparent upon
failure as it is creep weak. Identification of significant retained 8-
ferrite in a P92 weld entering service is almost impossible as
destructive examination is not permitted.

139
 • The absence of tungsten in the weld metal composition (WMC)
results in a material with greater ductility, but lower creep strength.
It has been noted in the literature that P92 can have low creep
ductility and this indicates that tungsten may have a role in that.
• It has been reported in the literature that silicon acts as a
nucleation site for Laves phase and can reduce coarsening. It is
observed that the weld metal with the highest silicon content
(0.42wt%, WMB) has the smallest average diameter Laves phase,
but the same volume fraction of Laves phase as a weld metal with
less silicon (0.2wt%, WMA).

Using this knowledge, quality checks of weld consumables can be initiated

to ensure a homogeneous distribution of elements so that localized
weaknesses in a weld can be avoided. There is tentative evidence that
tungsten plays an important role in the creep ductility of P92 and that
variations in tungsten and silicon could lead to an optimization of creep
strength.

The outcomes of this thesis are very relevant to the power generation
industry and will enable the more successful application of P92 steel and
welds.

140
141
 7 References

Abe, F. 2004. "Coarsening behavior of lath and its effect on creep rates in tempered
martensitic 9Cr-W steels." Materials Science and Engineering A 387-389:565-569.

Abe, F., T. Horiuchi, M. Taneike and K. Sawada. 2004. "Stabilization of martensitic

microstructure in advanced 9Cr steel during creep at high temperature." Materials
Science and Engineering A 378(1-2):299-303.

Abe, F., M. Taneike and K. Sawada. 2007. "Alloy design of creep resistant 9Cr steel
using a dispersion of nano-sized carbonitrides." International Journal of Pressure
Vessels and Piping 84(1-2):3-12.

Abe, Fujio. 2008. Creep-resistant steels. Cambridge: Woodhead Publishing limited.

Aghajani, A., F. Richter, c. Somsen, S. G. Fries, I. Steinbach and G. Eggeler. 2009. "On
the formation and growth of Mo-rich Laves phase particles during long-term creep of
a 12% chromium tempered martensite ferritic steel." Scripta Materialia 61(11):1068-
1071.

Albert, S. K., V. Ramasubbu, N. Parvathavarthini and T. P. S. Gill. 2003. "Influence of

alloying on hydrogen-assisted cracking and diffusible hydrogen content in Cr-Mo
steel welds." Sadhana-Academy Proceedings in Engineering Sciences 28:383-393.

Allen, D. J., B. Harvey and S. J. Brett. 2007a. ""FOURCRACK" - An investigation of the

creep performance of advanced high alloy steel welds." International Journal of
Pressure Vessels and Piping 84(1-2):104-113.

Anderson, P., T. Bellgardt and F. L. Jones. 2003. "Creep deformation in a modified

9Cr-1Mo steel." Materials Science and Technology 19(2):207-213.

Bhadeshia, HKDH. 2004. "Developments in martensitic and bainitic steels: role of the
shape deformation." Materials Science and Engineering A 378(1-2):34-39.

Bhadeshia, HKDH. 1999. "Some phase transformations in steel." Materials Science

and Technology 15:22 - 29.

Bhadeshia, HKDH. and RWK. Honeycombe. 2006. Steels: Microstructure and

Properties. 3rd Edition. Oxford: Elsevier.

Brett, S. J. 2004. "Type lila cracking in Ij2CrMoV steam pipework systems." Science
and Technology of Welding and Joining 9(1):41-45.

BS!. 1987a. "2633:1987 Specification for Class I arc welding of ferritic steel pipework
for carrying fluids." London: BS!.

BSI. 1987b. "Specification for Class I arc welding of ferritic steel pipework for carrying
fluids." In 2633:1987. London: BSI.

142
 Bugge, J., S. Kjaer and R. Blum. 2006. "High-efficiency coal-fired power plants
development and perspectives." Energy 31(10-11):1437-1445.

Cai, G. J., H. O. Andren and L. E. Svensson. 1997. "Effect of cooling after welding on
microstructure and mechanical properties of 12 pct Cr steel weld metals."
Metallurgical and Materials Transactions A-Physical Metallurgy and Materials Science
28(7):1417-1428.

Cerjak, H. and E. Letofsky. 1998. "The effect of welding on the microstructural

development of advanced 9-12 Cr steels." Microstructural Stability of Creep Resistant
Alloys for High Temperature Plant Applications(2):323-337.

Chew, P. E. Ｒ Ｐ Ｐ Ｓ ｾ "PF-fired supercritical power plant." Proceedings of the Institution

of Mechanical Engineers Part a-Journal of Power and Energy 217(A1):35-43.

Chilukuru, H., K. Durst, S. Wadekar, M. Schwienheer, A. Scholz, C. Berger, K. H. Mayer

and W. Blum. 2009. "Coarsening of precipitates and degradation of creep resistance
in tempered martensite steels." Materials Science and Engineering: A 510-511:81-87.

Cool, Tracey. and HKDH. Bhadeshia. 1997. "Austenite formation in 9Cr-1Mo type
power plant steels." Science and Technology of Welding and Joining 2(1):36 - 42.

Cui, H. R., F. Sun, K. Chen, L. T. Zhang, R. C. Wan, A. D. Shan and J. S. Wu. 2010.
"Precipitation behavior of Laves phase in 10%Cr steel X12CrMoWVNbN10-1-1 during
short-term creep exposure." Materials Science and Engineering A 527(29-30):7505-
7509.

Czyrska-Filemonowicz, A, K Bryla, K Spiradek-Hahn, H Firganek, A Zielinska-Lipiec and

PJ Ennis. 2003. "The Role of Boron in the 9% Chromium Steels for Steam Power
Plants." In Sixth International Charles Parsons Turbine Conference. Dublin:365-377

Dimmler, G., P. Weinert, E. Kozeschnik and H. Cerjak. 2003. "Quantification of the

Laves phase in advanced 9-12% Cr steels using a standard SEM." Materials
Characterization 51(5):341-352.

EIA. 2006. "International Energy Outlook 2006.". Washington, DC:

Energy Information Administration, U.S. Department of Energy.

Ennis, P. J. and A. Czyrska-Filemonowicz. 2003. "Recent advances in creep-resistant

steels for power plant applications." Sadhana-Academy Proceedings in Engineering
Sciences 28:709-730.

Ennis, P. J., A. Zielinska-Lipiec and A. Czyrska-Filemonowicz. 1998. "Quantitative

comparison of the microstructures of high chromium steels for advanced power
stations." Microstructural Stability of Creep Resistant Alloys for High Temperature
Plant Applications(2):13S-143.

Ennis, P. J., A. Zielinska-Lipiec and A. Czyrska-Filemonowicz. 2000. "Influence of heat

treatments on microstructural parameters and mechanical properties of P92 steel."
Materials Science and Technology 16(10):1226-1232.

143
 Ennis, P. J., A. ZielinskaLipiec, O. Wachter and A. CzyrskaFilemonowicz. 1997.
"Microstructural stability and creep rupture strength of the martensitic steel P92 for
advanced power plant." Acta Materialia 45(12):4901-4907.

Evans, H. E. 1984. Mechanisms of Creep Fracture. London and New York: Elsevier
Applied Science Publisher.

Evans, R Wand B Wilshire. 1993. Introduction to Creep. London: The Institute of

Materials.

Evans, R. W. and B. Wilshire. 1985. Creep of Metals and Alloys. London: The Institute
of Metals.

Falat, L., A. Vyrostkova, V. Homolova and M. Svoboda. 2009. "Creep deformation and
failure of E911/E911 and P92/P92 similar weld-joints." Engineering Failure Analysis
16(7): 2114-2120

Francis, J. A., HKDH Bhadeshia and P. J. Withers. 2007. "Welding residual stresses in
ferritic power plant steels." Materials Science and Technology 23(9):1009-1020.

Francis, J. A., W. Mazur and HKDH Bhadeshia. 2004. "Estimation of type IV cracking
tendency in power plant steels." ISIJ International 44(11):1966-1968.

Gaffard, V., A. F. Gourgues-Lorenzon and J. Besson. 2005. "High temperature creep

flow and damage properties of 9Cr1MoNbV steels: Base metal and weldment."
Nuclear Engineering and Design 235(24):2547-2562.

Garda de Andres, c., F. G. Caballero, C. Capdevila and L. F. Alvarez. 2002.

"Application of dilatometric analysis to the study of solid-solid phase transformations
in steels." Materials Characterization 48(1):101-111.

Gianfrancesco, A. Di, L. Cipolla, D. Venditti, S. Neri and M. Calderini. 2009. "Creep

Behaviour and Microstructural Stability of FB2 (CrMoCoB) Steel Trial Rotor." In 2nd
ECCC Creep Conference, eds. I. A. Shibli and S. R. Holdsworth. Zurich, Switzerland:
DEStech Publications, Inc.

Golpayegani, Ardeshir, Mats Hattestrand and Hans-Olof Andren. 2003. "Effect of

Boron and Copper on Precipitate Growth and Coarsening in Martensitic Chromium
Steels." In Sixth International Charles Parsons Turbine Conference. Dublin:347-363

Gomez, M., L. Rancel and S. F. Medina. 2009. "Effects of aluminium and nitrogen on
static recrystallisation in V-microalloyed steels." Materials Science and Engineering:
A- 506(1-2):165-173.

Hald, J. 1996. "Metallurgy and creep properties of new 9-12%Cr steels." Steel
Research 67(9):369-374.

Hald, J. and L. Korcakova. 2003. "Precipitate stability in creep resistant ferritic steels-
Experimental investigations and modelling." ISIJ International 43(3):420-427.

144
 Hasegawa, Y., T. Muraki, M. Ohgami and H. Mimura. 2000. "Optimum tungsten
content in high strength 9 to 12% chromium containing creep resistant steels." Creep
and Fracture of Engineering Materials and Structures 171-1:427-435.

Helis, L., Y. Toda, T. Hara, H. Miyazaki and F. Abe. 2009. "Effect of cobalt on the
microstructure of tempered martensitic 9Cr steel for ultra-supercritical power
plants." Materials Science and Engineering: A 510-511:88-94.

Hosoi, Y., N. Wade, S. Kunimitsu and T. Urita. 1986a. "Precipitation behavior of laves
phase and its effect on toughness of 9Cr-2Mo ferritic martensitic steel." Journal of
Nuclear Materials 141:461-467.

Hosoi, Y., N. Wade and T. Urita. 1986b. "Effect of Si and Mn on precipitation behavior
of laves phase and toughness of 9Cr-2Mo steel." Transactions of the Iron and Steel
Institute of Japan 26(1):B30-B30.

Hu, Xiaoqiang, Namin Xiao, Xinghong Luo and Dianzhong Li. 2010. "Transformation
Behavior of Precipitates in a W-alloyed 10 wt pct Cr Steel for Ultra-supercritical
Power Plants." Journal of Materials Science & Technology 26(9}:817-822.

Hunt, A. c., A. O. Kluken and G. R. Edwards. 1994. "Heat input and dilution effects in
microalloyed steel weld metals." Welding Journal 73(1):S9-S15.

Issler, 5., A. Klenk, A. A. Shibli and J. A. Williams. 2004. "Weld repair of ferritic welded
materials for high temperature application." International Materials Reviews
49(5}:299-324.

Kitto, J B. 1996. "Technology development for advanced pulverized coal-fired boilers

." In Power-Gen International. Orlando, FL

Klueh, R. L. 2005. "Elevated temperature ferritic and martensitic steels and their
application to future nuclear reactors." International Materials Reviews 50{5}:287-
310.

Klueh, R. L., N. Hashimoto and P. J. Maziasz. 2005. "Development of new nano-

particle-strengthened martensitic steels." Scripta Materialia 53(3):275-280.

Knezevic, V., J. Balun, G. Sauthoff, G. Inden and A. Schneider. 2008. "Design of

martensitic/ferritic heat-resistant steels for application at 650 degrees C with
supporting thermodynamic modelling." Materials Science and Engineering A 477(1-
2):334-343.

Kohlhofer, W. and R. K. Penny. 1996. "Hardness testing as a means for creep

assessment." International Journal of Pressure Vessels and Piping 66{1-3}:333-339.

Korcakova, L., J. Hald and M. A. J. Somers. 2001. "Quantification of Laves phase

particle size in 9CrW steeL" Materials Characterization 47(2):111-117.

Kunimitsu, 5., Y. You, N. Kasuya, Y. Sasaki and Y. Hosoi. 1991. "Effect of

thermomechanical treatment on toughness of 9Cr-W ferritic-martensitic steels
during aging." Journal of Nuclear Materials 179:689-692.

145
 Lancaster, J. F. 1993. Metallurgy of Welding. 5th Edition: Chapman & Hall.

Leblond, J. B., G. Mottet and J. C. Devaux. 1986. "A theoretical and numerical
approach to the plastic behavior of steels during phase-transformations .1.
Derivation of general relations." Journal of the Mechanics and Physics of Solids
34(4):395-409.

Mannan, S. l. and K. Laha. 1996. "Creep behaviour of Cr-Mo steel weldments."

Transactions of the Indian Institute of Metals 49(4):303-320.

Marshall, A. W., Z. Zhang and G. B. Holloway. 2002. "Welding consumables for P92
and T23 creep resisting steels." In 5th International Conference Welding and Repair
Technology for Power Plants. Alabama. P6-1

Maruyama, K., K. Sawada and J. Koike. 2001. "Strengthening mechanisms of creep

resistant tempered martensitic steel." ISIJ International 41(6):641-653.

Masuyama, F. 1999. "New developments in steels for power generation boilers."

Conference on Advanced Heat Resistant Steels for Power Generation, San Sebastian,
Spain:33-48.

Masuyama, F. 2001. "History of power plants and progress in heat resistant steels."
ISIJ International 41(6):612-625.

Miyahara, K., J. H. Hwang and Y. Shimoide. 1995. "Aging phenomena before the
precipitation of the bulky laves phase in Fe-10-percent-Cr ferritic alloyS." Scripta
Metallurgica Et Materialia 32(12):1917-1921.

Miyamoto, G., H. Usuki, Z. D. Li and T. Furuhara. 2010. "Effects of Mn, Si and Cr

addition on reverse transformation at 1073 K from spheroidized cementite structure
in Fe-0.6 mass% C alloy." Acta Materialia 58(13):4492-4502.

Nawrocki, J. G., J. N. DuPont and A. R. Marder. 2000. "A study on the carbide
precipitation in a ferritic steeL" In Microbeam Analysis 2000, Proceedings, eds. D. B.
Williams and R. Shimizu. Bristol: lOP Publishing Ltd.:17S-176

ai, K., C. Lux and G. R. Purdy. 2000. "A study of the influence of Mn and Ni on the
kinetics of the proeutectoid ferrite reaction in steels." Acta Materialia 48(9):2147-
2155.

Onoro, J. 2006. "Weld metal microstructure analysis of 9-12% Cr steels."

International Journal of Pressure Vessels and Piping 83(7):540-545.

Orlova, A., J. Bursik, K. Kucharova and V. Sklenicka. 1998. "Microstructural

development during high temperature creep of 9% Cr steel." Materials Science and
Engineering a-Structural Materials Properties Microstructure and Processing
245(1):39-48.

Onoro, J. 2006. "Martensite microstructure of 9-12%Cr steels weld metals." Journal

of Materials Processing Technology 180(1-3):137-142.

146
Paul, I., R. Taud and D. O'Leary. 2005. "Tailor-made off the shelf: reducing the cost
and construction time of thermal power plants .. " The World Bank Group

Reddy, G. M., T. Mohandas and D. S. Sarma. 2003. "Cold cracking studies on low alloy
steel weldments: effect of filler metal composition." Science and Technology of
Welding and Joining 8(6}:407-414.

Richardot, D, JC Vaillant, A Arbab and W Bendick. 2000. The T92 P92 Book: Vallourec
& Mannesmann Tubes.

Santella, M. L., S. S. Babu, R. W. Swindeman and E. D. Specht. 2003. "In-situ

characterization of austenite to martensite decomposition in 9Cr-IMo-v steel welds."
Symposium on Austenite Formation and Decomposition, Chicago IL:247-256.

Santella, Michael. 2010. "Influence of chemical compositions on lower ferrite-

austenite transformation temperatures in 9Cr steels." J. Pressure Vessel
Technol. 134(2}

Savage, W. F. and A. H. Aronson. 1966. "Preferred orientation in the weld fusion

zone." Welding JournaI4S(2}:8S-89.

Savage, W. F., E. F. Nippes and E. S. Szekeres. 1976. "Study of weld interface

phenomena in a low-alloy steeL" Welding Journal 55(9}:S260-S268.

Sawada, K., K. Kubo and F. Abe. 2001a. "Creep behavior and stability of MX
precipitates at high temperature in 9Cr-0.5Mo-1.8W-VNb steeL" Materials Science
and Engineering A 319:784-787.

Sawada, K., K. Kubo and F. Abe. 2001b. "Creep behavior and stability of MX
precipitates at high temperature in 9Cr-0.5Mo-1.8W-VNb steel." Materials Science
and Engineering A 319-321:784-787.

Sawada, K., K. Kubo and F. Abe. 2003. "Contribution of coarsening of MX

carbonitrides to creep strength degradation in high chromium ferritic steeL"
Materials Science and Technology 19(6):732-738.

Sawada, K., M. Takeda, K. Maruyama, R. Ishii, M. Yamada, Y. Nagae and R. Komine.

1999. "Effect of W on recovery of lath structure during creep of high chromium
martensitic steels." Materials Science and Engineering A 267(1):19-25.

Shibli, A. 2008. "Performance of modern high strength steels (P91, P92) in high
temperature plant." Proceedings of the ASME Pressure Vessels and Piping
Conference 2007, Vol 9:161-170.

Shirane, T., S. Tsukamoto, K. Tsuzaki, Y. Adachi, T. Hanamura, M. Shimizu and F. Abe.

2009. "Ferrite to austenite reverse transformation process in B containing 9%Cr heat
resistant steel HAZ." Science and Technology of Welding and Joining 14{8}:698-707.

Sourmail, T. 2001. "Precipitation in creep resistant austenitic stainless steels."

Materials Science and Technology 17(1):1-14.

147
Spigarelli, S. and E. Quadrini. 2002. "Analysis of the creep behaviour of modified P91
(9Cr-1Mo-NbV) welds." Materials & Design 23(6):547-552.

Tamura, M., H. Sakasegawa, A. Kohyama, H. Esaka and K. Shinozuka. 2003. "Effect of

MX type particles on creep strength of ferritic steel." Journal of Nuclear Materials
321{2-3) :288-293.

Vaillant, J. c., B. Vandenberghe, B. Hahn, H. Heuser and C. Jochum. 2008. "T/P23, 24,
911 and 92: New grades for advanced coal-fired power plants-Properties and
experience." International Journal of Pressure Vessels and Piping 85{1-2):38-46.

Viswanathan, R. and W. Bakker. 2001a. "Materials for ultrasupercritical coal power

plants - Turbine materials: Part II." Journal of Materials Engineering and Performance
10(1):96-101.

Viswanathan, R. and W. Bakker. 2001b. "Materials for ultrasupercritical coal power

plants-boiler materials: Part 1." Journal of Materials Engineering and Performance
10(1):81-95.

Viswanathan, R., K. Coleman and U. Rao. 2006. "Materials for ultra-supercritical coal-
fired power plant boilers." International Journal of Pressure Vessels and Piping
83{11-12):778-783.

Watanabe, T., M. Yamazaki, H. Hongo, M. Tabuchi and T. Tanabe. 2004. "Effect of

stress on microstructural change due to aging at 823 K in multi-layer welded joint of
2.25Cr-1Mo steel." International Journal of Pressure Vessels and Piping 81(3):279-
284.

Wu, R., R. Sandstrom and F. Seitisleam. 2004. "Influence of extra coarse grains on the
creep properties of 9 percent CrMoV (P91) steel weldment." Journal of Engineering
Materials and Technology-Transactions ofthe ASME 126(1):87-94.

Yang, J R. and H K D H. Bhadeshia. 1989. "Reaustenitization Experiments on some

High-strength Steel Weld Deposits." Materials Science and Engineering A 118:155 -
170.

Yeh, Sonia and Edward S. Rubin. 2007. "A centurial history of technological change
and learning curves for pulverized coal-fired utility boilers" Energy 32:1996-2005.

Yin, F. S. and W. S. Jung. 2009. "Nanosized MX Precipitates in Ultra-low-Carbon

Ferritic/Martensitic Heat-Resistant Steels." Metallurgical and Materials Transactions
a-Physical Metallurgy and Materials Science 40A(2):302-309.

Yin, Y. F. and R. G. Faulkner. 2003. "Simulations of precipitation in ferritic steels."

Materials Science and Technology 19(1):91-98.

Yoshizawa, M., M. Igarashi, K. Moriguchi, A. Iseda, Hassan Ghassemi Armaki and K.

Maruyama. 2009. "Effect of precipitates on long-term creep deformation properties
of P92 and P122 type advanced ferritic steels for USC power plants." Materials
Science and Engineering A 510-511:162-168.

148
Zielinska-Lipiec, A. and A. Czyrska-Filemonowicz. 2007. "Characterisation of the
micro- and nanoscale structure of new creep-resistant steels for use in advanced USC
steam power plants." Materials Transactions 48(5):931-935.

149