Influence of alloy type,peak temperature and constraint on residual stress evolution in Satoh test

Abstract

The evolution of residual stress in welds is affected by a variety of factors such as joint design, welding parameters, material properties and the possible presence of phase transformations. The Satoh test can be useful as a method to understand differences in welding residual stresses among various materials. This study compares the evolution of residual stress measured in the Satoh test for ferritic and stainless steels. For weld thermal cycles with a peak temperature above Ac₃, the ferritic alloys exhibit lower residual stresses than the austenitic alloys due to the well known reduction in stress from the martensitic or bainitic transformation. However, the residual stress levels among these alloys are similar for peak temperatures below Ac₃. The application of constraint during both the heating and cooling portions of the thermal cycle affects the final magnitude of residual stress in a way that is not often considered.

Keywords

Welding residual stress Satoh test Phase transformation

Introduction

Residual stresses in welds can be quite significant and have an appreciable influence on weldability and service performance.¹ The evolution of welding residual stress is affected by a number of factors, including weld geometry, joint design, welding parameters and the temperature dependent thermophysical and mechanical properties.^2–4 The large difference in temperature dependent material properties between unique alloys or alloy systems is particularly important and expected to produce wide variations in behaviour. Further complications arise in alloys that experience a solid state phase transformation during cooling, as this transformation also affects the magnitude of accumulated weld residual stress.^6–16 The wide range of complex interactions between welding parameters, weld design and material properties makes it difficult to make straight forward comparisons between expected differences of residual stress among engineering alloys. It is clear that a fundamental assessment of welding residual stress for a given alloy or alloy system is important in assessing its potential performance in a given application or for providing critical feedback such as in a root cause analysis of a component failure. Welding residual stresses can have significant effects in service behaviour such as reheat cracking in ferritic alloys, stress relaxation cracking in stainless steels and stress corrosion cracking in many alloy systems.

The Satoh test has recently been applied as a means to study residual stress evolution in welds in a controlled manner for the purpose of making direct comparisons between engineering alloys.^{5–7, 10, 12–17} As shown schematically in Fig. 1, a sample is locally heated and cooled under conditions that simulate the heating and cooling rates during fusion welding. The peak temperature representative of the coarse grain heat affected zone (CGHAZ) is often used in the weld thermal cycle simulation. In a conventional Satoh test, the sample is permitted to expand freely during the heating stage. When the peak temperature is reached, the sample is cooled while contraction is restricted, and the residual stress that accumulates due to the restricted contraction is then measured with a load cell in line with the sample. This procedure is thought to simulate the restrained contraction that accounts for residual stress accumulation in real welds.⁵

Schematic illustration of Satoh test comparing evolution of temperature and stress as function of time

Work carried out by Dai et al.⁶ and Frances et al.⁷ showed that the transformation temperature in steels affects the final residual stress when using the Satoh test. Jones and Alberry⁸ conclude that residual stresses are best avoided by suppressing the transformation temperature low enough so that the phase change compensates for the thermal contraction. Not only can this reduce residual stresses at ambient temperature, but this can lead to the presence of favourable compressive residual stresses near the weld surface. A lower transformation temperature merits a smaller temperature range after the transformation in which the residual stress can begin to reaccumulate. This effect is well known and is being used to develop welding consumables with low transformation temperatures to help minimise residual tensile stresses.⁹

Shirzadi and Bhadeshia¹⁰ used a modified Satoh test design to help ensure uniform temperature and stress conditions within the test material. The concept involved the bonding of two pieces of non-transforming metal (alloy IN617) to the test material. This was compared with a sample made entirely of the test material. Both samples behaved identically until the start of the phase transformation, with the monolithic sample transforming at a higher temperature than the bimetallic sample. This was attributed to the regions outside of the sample centre that transform before the centre of the sample. The bimetallic sample showed a greater level of residual stress at room temperature. This was attributed to the addition of alloy IN617, which has a higher coefficient of thermal expansion than the test material. Thus, thermal contraction stresses were able to accumulate more during cooling due to this added material. Also described in that work is a design to completely eliminate residual stresses outside of the HAZ by the use of Invar, a material that has zero thermal expansion. With this approach, the material of interest would be the only contributor of residual stress accumulation in the HAZ.

Satoh test results have been provided to date that show the accumulation of stress for different alloys. However, these tests generally only used one (or two) peak temperature(s). In the case of two peak temperatures,^{13, 15} peak temperature values above and below Ac₃ were used to compare the effect of the phase transformation. However, the HAZ experiences peak temperatures from the solidus temperature to ambient temperature. Thus, use of a single (or two) peak temperature(s) cannot provide information on the range of residual stresses expected in the HAZ. In addition, most tests conducted to date allow free expansion during the heating stage of the thermal cycle. However, this does not represent an actual weld thermal cycle where there is constraint during both heating and cooling. Another style of Satoh tests was conducted where the sample is under constraint during the entire thermal cycle.^{16, 17} In this case, stress accumulates in compression during heating due to the restricted thermal expansion. Depending on peak temperature, compressive stresses may still be present during cooling and will eventually turn to tension due to the applied constraint. The tensile stresses will start to form at lower temperatures since it takes time for the compressive stresses to reduce and turn to tension, thus resulting in a lower final level of residual tensile stress. These studies demonstrate the rather wide range of results that can be observed depending on the test procedure employed. The Satoh tests appear to become a common method for understanding residual stress formation in welds. Thus, it is useful to consider how various procedural parameters affect the results of the tests.

The main objectives of this research are to compare the evolution of residual stress measured in the Satoh test between various alloys and investigate the influence of peak temperature on residual stress evolution. Preliminary tests that apply constraint during both the heating and cooling stages of the thermal cycle are also conducted, and the results are compared to those obtained with the standard Satoh test. The results of this work should be useful for making direct comparisons of residual stress levels in various alloys and should also shed light on procedural issues that should be considered with the Satoh test.

Experimental

In this study, a Gleeble 3500 thermomechanical simulator was used to conduct Satoh tests on the alloys summarised in Table 1. Peak temperatures T_p of 1350, 1100, 1000, 900 and 750°C were used, along with heating and cooling rates of 10 and 30°C s^− 1 respectively. These heating and cooling rates were chosen as they represent a reasonable range of heat inputs for traditional arc welding processes. Cylindrical samples (110 mm long and 10 mm in diameter) with threaded ends were used for all of the tests. The samples had a reduced centre gage section 15 mm long and 6 mm in diameter.

Table 1

Compositions of alloys used in this study/wt-%

Element	T24	P91	304H	347H
C	0.08	0.10	0.04–0.10	0.08
Si	0.28	0.28	0.75	0.75
Mn	0.51	0.56	2.0	2.0
P	0.007	0.017	0.045	0.045
S	0.005	< 0.001	0.03	0.03
Cu		0.04
Ni	0.12	0.17	8–10.5	9–13
Cr	2.43	8.63	18–20	17–19
Mo	0.95	0.89
Al	0.010	0.003
Ti	0.070	< 0.003
V	0.24	0.21
Nb	< 0.01	0.080		0.54
B	0.0042
N	0.007	0.0386
Zr		< 0.001
W		< 0.01

Most tests were conducted in which the samples were heated without constraint to the T_p, locked in place and allowed to cool to room temperature T_r, while stress was measured as a function of temperature. Selected tests were also conducted with constraint applied during both the heating and cooling portions of the thermal cycle. The Gleeble has the added advantage that phase transformations can be monitored in situ during a Satoh test with a dilatometer placed on the centre of the sample (i.e. the same location in which the temperature is monitored and controlled). The dilatometer was used to measure the radial expansion and contraction of the sample during the thermal cycle to detect the Ac₁ and Ac₃ transformation during heating and the martensitic or bainitic transformation during cooling. Figure 2 shows an image of the specimen set up in the Gleeble before the installation of the dilatometer. Copper grips were used for the tests, as well as conical jacks that restrict axial expansion of the specimen in the grips.

Image of Satoh test specimen in Gleeble before addition of dilatometer

Results from some of the preliminary tests are shown in Fig. 3 and demonstrate abnormal behaviours in the stress at low temperatures, shown by the red circle. At temperatures close to T_r, the stress deviated from linear accumulation and in some cases began to decrease, suggesting a source of slippage in the test apparatus. As shown in Fig. 2, the specimen has threaded ends with nuts that apply pressure on the outer surface of the copper grips. During thermal contraction, the grips were observed to slip inwards during the test due to the force from the nuts, thus causing a reduction in the stress. This problem was avoided by pulling the sample to 90% of the yield strength before the test and tightening the outermost nuts that constricted expansion of the sample in the grips. This induced a compressive stress on the sample, thus securing the copper grips into the pocket jaw assembly. The sample was again pulled in tension to 90% yield strength and relieved by tightening the conical nuts. This process was repeated a couple times to secure the copper grips in the Gleeble jaws. Any stress imposed during the securing of the sample was zeroed out by controlling the stroke of the Gleeble. Any presence of stress (tensile or compressive) is also eliminated once the test starts since, during the heating cycle of the test, the sample is not constricted and allowed to achieve a stress free state. This procedure was conducted before each test to eliminate sample slippage and provide accurate residual stress measurements. Although the exact cause of the sample slippage has not been considered in detail, evidence for similar types of stress reduction at the lower temperatures has been observed on other Satoh tests results.¹⁵

Preliminary Satoh test results demonstrating influence of slippage during test that causes reduction in final stress, shown by red circle

The temperature distribution was also measured along the length of the sample during the thermal cycle for various peak temperatures. Thermocouples were spot welded to different locations of the sample in order to measure thermal cycles for locations further from the sample centre. Figure 4a and b shows the temperature distribution for alloy P91 and 304H respectively. These measurements show the temperature gradient through the sample at the moment the peak temperature of interest is reached. Figure 4c demonstrates the difference in temperature gradients between materials. This can be attributed to the lower thermal conductivity in 304H (14.5 W m^− 1 K^− 1) than P91 (26.7 W m^− 1 K^− 1).

Qualitative temperature distribution in Satoh test specimen showing temperatures at different locations for a P91 and b 304H and c difference between two materials; location of thermocouples in a and b are indicated by symbols at axial locations 27, 33, 41, 49 and 55 mm, with 41 mm being specimen centre

To summarise, the procedure for conducting a Satoh test in the Gleeble involves the following critical aspects:

placing the sample in the copper grips and into the Gleeble jaws

pulling the sample in tension to 90% of the yield strength

tightening both conical nuts (this will relieve the tensile force and create a compressive force)

pulling the sample in tension again and securing the copper grips in the jaws

tightening both conical nuts to relieve the tensile force

repeating steps 2–4 as necessary.

As described above, any force on the sample from this procedure will be relieved when test starts.

Results and discussion

Ferritic alloys

Figure 5a shows the accumulation of residual stress as a function of temperature for alloy P91 exposed to various peak temperatures. These results were obtained with no restraint applied during heating. The dilatometry results acquired for a T_p of 1350°C are shown in Fig. 5b, where the dilation is shown as a function of temperature. There is a rather large deviation between the dilation on heating and cooling that is associated with additional dilation from restricted contraction of the sample. The Ac₁ and Ac₃ represent the start and finish of the austenite phase transformation during heating respectively, while the M_s and M_f represent the martensite start and finish temperatures of the alloy during cooling. There are several interesting observations from the results in Fig. 5. The alloy exhibits a reversal of residual stress at temperatures near M_s temperature. However, for a T_p of 1350°C, the stress reversal in the Satoh test occurs ∼500°C, whereas the dilatometry results show the martensite transformation does not begin until 375°C. The results also show that decreasing T_p produces higher final residual stress values for peak temperatures that are above the Ac₃. In contrast, the residual stress is among the highest when the T_p is below the Ac₃ (peak temperature of 750°C). This observation matches the reported behaviour of single pass laser welds, such as in Ref. 18 where the highest welding residual stresses are in a location in the HAZ, removed from the fusion line and consistent with a T_p < Ac₁.

a Satoh test results and b dilatometry test results for alloy P91 with heating rate of 10°C s^− 1 and cooling rate of 30°C s^− 1

The reversal of stress associated with the martensite start temperature is well recognised and is associated with the volume expansion from the austenite to martensite transformation that counteracts the tensile strain during cooling.^6–16 However, the difference between the actual M_s temperature and the temperature at which the stress begins to reverse in a Satoh test is not always recognised. The cause for this difference becomes apparent when examining the Satoh test results in which the stress reversal begins at ∼500°C (Fig. 5a), the dilatometry results in which the martensite start temperature is 375°C (Fig. 5b), and the temperature gradient through the sample (Fig. 4a). The temperature is measured at the centre of the sample, and this is the temperature that is conventionally plotted in the Satoh test results. The dilatometer also detects the phase change at the centre of the sample. However, there are parts of the sample away from the centre that are heated above the Ac₃ temperature but are at a lower temperature than the sample centre. This region of the sample will undergo the phase transformation before the centre of the sample. The phase transformation in this region counteracts the thermal contraction, so there is a reduction in stress before the phase transformation measured with the dilatometer at the sample centre. This effect accounts for the apparent differences between the dilation reading and the Satoh test stress reversal. Thus, care must be taken not to use the temperature at the start of the stress reversal on the Satoh test as an indication of the start of the phase transformation when a temperature gradient exists through the sample.

The difference between transformation temperature and temperature at which the stress reversal begins has been noted in other work on SA508 steel.¹⁵ In this case, the stress reversal occurred at ∼600°C, whereas the reported bainite start temperature was 550°C. In previous work, the apparent increase in the bainite start temperature was attributed to the presence of stress, but results presented in this work demonstrate that the temperature gradient in the sample is responsible for this difference. Figure 6 shows dilatometry results on alloy P91 comparing dilation results from a Satoh test and dilation results for athermal cycle under zero stress. The results show that both tests produce the same martensite start temperature, indicating no significant influence of stress on the transformation start temperature.

Dilatometry test, conducted on alloy P91, shows applied stress had no significant effect on transformation temperatures during heating and cooling stages

The influence of T_p on the magnitude of the residual stress in the steels used here can also be understood with reference to Fig. 5a. For peak temperatures above Ac₃, a lower T_p generally accumulates greater stress before the transformation (at a temperature just before the stress reversal), as well as when cooled to T_r. This is also attributed to the temperature distribution throughout the sample. Regions away from the centre of the sample exhibit a lower temperature, but are still above Ac₁ (shown in Fig. 4a) and therefore exhibit the phase transformation during cooling before the sample centre. At higher peak temperatures, larger portions of the sample reach a temperature above Ac₁. Thus, for a higher T_p, a larger portion of the sample undergoes the volume expansion associated with the phase transformation during cooling, leading to a larger reduction in stress accumulation. Furthermore, since there is less stress accumulated before the transformation, there is less final residual stress at T_r. Note that the peak temperature of 750°C is below Ac₁ and therefore will not experience a phase transformation during cooling that can counteract the residual tensile stress. Also note this results in the highest residual stress value at T_r. The final residual stress for the test conducted with a peak temperature of 900°C lies between the results for the samples heated above and below the Ac₃, since this test represents an intercritical peak temperature that is between Ac₁ and Ac₃. Thus, this intermediate residual stress level is attributed to partial transformation that occurred during heating.

These results carry important implications for future tests. Satoh test results reported to date typically utilise a single T_p (typically associated with the CGHAZ). This is occasionally carried out to evaluate the effectiveness of an alloy to counteract residual stresses via the martensitic transformation.^{5–7, 10, 12, 14, 16, 17} However, the HAZ will experience the full range of peak temperatures during welding (from ambient to the solidus temperature), and the residual stress behaviour can thus not be captured with results from a single T_p. In fact, Onsoien et al. suggested that it is most relevant to compare residual stress levels in the Satoh test with the maximum axial stress in girth welds.¹⁶ They measured residual stress levels in X70 pipeline steel using Satoh test and compared that to values measured in real welds.¹⁹ The good correlation between maximum residual stresses found in the Satoh test and the maximum values measured in real welds led to this conclusion. In this case, the maximum residual stress acquired in the Satoh test should then also be used for a proper comparison. Use of a single T_p well above Ac₃ in the Satoh test would produce a relatively low residual stress level that may not be representative of the maximum residual stress measured in the HAZ. The results in Fig. 5a suggest that Satoh tests should be conducted over a range of peak temperatures to adequately capture this behaviour.

The results shown for P91 in Fig. 5a were duplicated for alloy T24, and the results are summarised in Fig. 7 in which the final residual stress is shown as a function of peak temperature. The martensite or bainite start temperature was measured for each alloy and is also noted in Fig. 7. Note that the residual stress level is reduced as the transformation temperature is reduced. With a lower transformation temperature, there is a smaller temperature range after the transformation in which the residual stress can begin to reaccumulate. However, as shown in Fig. 5a, it is important to note that the difference in residual stress level depends on T_p. Also note that the highest residual stress levels are expected to occur for regions of the HAZ in which the peak temperature is below the Ac₁. These effects may not always be recognised and should be considered in future Satoh tests.

Effect of peak temperature on residual stress measured at room temperature during Satoh testing of T24 and P91

Austenitic alloys

Figure 8a and b shows typical Satoh test results for two austenitic alloys 304H and 347H respectively. The austenitic alloys do not experience a solid state phase transformation (other than possible carbide formation at longer times/lower cooling rates) and thus accumulate greater residual stress levels than the ferritic alloys when cooled to T_r. In contrast to the ferritic alloys, use of a lower T_p leads to the accumulation of less residual stress due to the smaller temperature range associated with the thermal contraction.

Satoh test results for a alloy 304H and b 347H at various peak temperatures with heating rate of 10°C s^− 1 and cooling rate of 30°C s^− 1

A summary of the Satoh test conducted on all alloys of interest can be seen in Fig. 9. Use of peak temperatures above Ac₃ produces significantly less stress in ferritic alloys, especially at temperatures in the CGHAZ. This can be attributed to the transformation that counteracts the thermal contraction. However, for a T_p of 750°C, where the ferritic alloys do not transform, the residual stress can be comparable or even higher than the austenitic alloys, depending on the alloy.

Satoh test results for alloys P91, T24, 347H and 304H at peak temperatures of 750, 900,1000, 1100 and 1350°C

Effect of applied constraint

The Satoh test is typically conducted by heating the sample without constraint. However, in actual welds, there is significant restraint during both the heating and cooling portions of the weld thermal cycle. Thus, tests were conducted to investigate the possible effect of restraint applied during both the heating and cooling stages. Ferritic alloy P91 and alloy 347H stainless steel were heated to three peak temperatures, 1350, 1100 and 750°C to compare the difference in residual stress of the conventional Satoh test with a test conducted under full constraint. The results can be seen in Figs. 10 and 11. A T_p of 1350°C shows a compression stress accumulation upon heating for both alloys due to the restricted expansion from the constraint. The stress decreases at elevated temperatures due to the lack of strength and associated plastic deformation. At 1350°C, the stress is almost zero, so the accumulation during cooling follows the same path as the conventional Satoh test. Somewhat similar results are observed for both alloys when a peak temperature of 1100°C is used, regardless of the presence of a phase transformation. These results suggest that the use of constraint does not have a significant effect during heating when peak temperatures associated with the CGHAZ are used.

Satoh test results comparing general Satoh test with fully constrained specimen for alloy P91 at peak temperatures of a 1350°C, b 1100°C and c 750°C with heating rate of 10°C s^− 1 and cooling rate of 30°C s^− 1; red and blue arrows show heating and cooling stages respectively

Satoh test results comparing general Satoh test with fully constrained specimen for alloy 347H at peak temperatures of a 1350°C, b 1100°C and c 750°C with heating rate of 10°C s^− 1 and cooling rate of 30°C s^− 1; red and blue arrows show heating and cooling stages respectively

However, at 750°C, there is a significant difference between the two methods of measuring stress. During heating, the sample reaches a state of compressive residual stress. This stress is still present when cooling begins because the peak temperature was too low to permit stress relaxation. During cooling, the compressive stress needs to be relieved before tensile stresses accumulate. The constraint during heating causing the compressive stress has a major effect on the magnitude of residual stress when cooled to T_r. Tensile stresses do not start to form until ∼500°C, so there is a smaller temperature range for the stress to accumulate. This is in contrast to the conventional Satoh test, where the tensile residual stresses start to form at their respected T_p.

Conclusions

Satoh tests were conducted on ferritic and austenitic alloys. The influence of peak temperature and the application of constraint applied during heating were investigated. The following conclusions can be drawn from this work.

For the ferritic alloys investigated in this study, the magnitude of residual stress decreases with decreasing martensite/bainite start temperature for tests conducted with peak temperatures above the Ac₃. This is attributed to the volume expansion associated with the austenitic–martensitic transformation. Lower transformation temperatures permit less of a temperature range for the residual stress to reaccumulate with continued cooling.

For ferritic steels tested without constraint during heating, lower peak temperatures above the Ac₃ accumulate greater residual stress when cooled to room temperature because less of the sample experiences the austenite–martensite/bainite transformation that counteracts the thermal contraction.

Ferritic alloys tested above the Ac₃ generally showed lower residual stresses relative to austenitic alloys tested at comparable peak temperatures because of the stress reduction associated with the austenite–martensite/bainite transformation during cooling. However, this large difference among the residual stress levels in the ferritic alloys does not occur when the peak temperature is below Ac₁.

The application of constraint during both the heating and cooling portions of the thermal cycle effects the final magnitude of residual stress relative to the cycle in which constraint is only applied during cooling. The difference is negligible at high peak temperatures due to stress relaxation. However, at lower peak temperatures, the compressive stresses cannot completely relax during the heating cycle, and the corresponding final tensile stresses are therefore reduced when the sample is constrained during the heating stage.

Footnotes

Acknowledgements

The authors gratefully acknowledge the financial support from the Electric Power Research Institute and the NSF Industry/University Collaborative Research Center for Integrative Materials Joining Science for Energy Applications (CIMJSEA) under contract no. IIP-1034703.

References

Kudryavtsev

and Kleiman

: ‘Fatigue of welded elements: residual stresses and improvement treatments’. Proc. IIW Int. Conf. on ‘Welding and materials’, Dubrovnik, Croatia, July 2007, IIW, 255–264.

Deng

and Murakawa

: ‘Finite element analysis of temperature field, microstructure, and residual stress in multi-pass butt-welded 2.25Cr–1Mo steel pipes’, Comput. Mater. Sci., 2008, 43, (3), 681–695. doi: 10.1016/j.commatsci.2008.01.025

Deng

and Murakawa

: ‘Prediction of welding residual stress in multi-pass welded modified 9Cr–1Mo steel pipe considering phase transformation effects’, Comput. Mater. Sci., 2006, 37, (3), 209–219. doi: 10.1016/j.commatsci.2005.06.010

Deng

and Murakawa

: ‘Numerical simulation of temperature field and residual stress in multi-pass welds in stainless steel pipe and comparison with experimental measurements’, Comput. Mater. Sci., 2006, 37, (3), 269–277. doi: 10.1016/j.commatsci.2005.07.007

Kou

: ‘Welding metallurgy’, 2003, Hoboken, NJ, Wiley-Interscience.

Dai

, Francis

J. A.

, Stone

H. J.

, Bhadeshia

H. K. D. H.

, Karlsson

and Withers

P. J.

: ‘Characterizing phase transformation and their effects on ferritic weld residual stresses with z-rays and neutrons’, Met. Mater. Trans. A, 2008, 39A, 3070–3078. doi: 10.1007/s11661-008-9616-0

Francis

J. A.

, Kundu

, Bhadeshia

H. K. D. H.

, Stone

H. J.

, Rogge

R. B.

and Withers

P. J.

: ‘Transformation temperatures and welding residual stresses in ferritic steels’, Proc. ASME Pressure Vessels and Piping Division Conf., San Antonio, TX, USA, July 2007, ASME, 949–956.

Jones

W. K. C.

and Alberry

P. J.

: ‘The role of phase transformations in the development of residual stress during the welding of some fast reactor steels’, in Proc. Int. Conf. on ‘Ferritic steels for fast reactor steam generators’, (ed. Pugh

S. F.

and Little

E. A.

., 771–476; 1978, London, British Nuclear Energy Society.

Thomas

S. H.

and Liu

: ‘Analysis of low transformation temperature welding (LTTW) consumables – distortion control and evolution of stresses’, Sci. Technol. Weld. Join., 2014, 19, (5), 392–401. doi: 10.1179/1362171814Y.0000000199

10.

Shirzadi

A. A.

and Bhadeshia

H. K. D. H.

: ‘Accumulation of stress in constrained assemblies: novel Satoh test configuration’, Sci. Technol. Weld. Join., 2010, 15, (6), 497–499. doi: 10.1179/136217110X12731414739998

11.

Dai

: ‘Modelling of residual stress minimization through martensitic transformation in stainless steel welds’. Proc, 9th Int. Seminar on ‘Numerical analysis of weldability’, Graz, Austria, September 2010, TechnisheUniversitat Graz, 239–252.

12.

Moat

R. J.

, Stone

H. J.

, Shirzadi

A. A.

, Francis

J. A.

, Kundu

, Mark

A. F.

, Bhadeshia

H. K. D. H.

, Karlsson

and Withers

P. J.

: ‘Design of weld fillers for mitigation of residual stresses in ferritic and austenitic steel welds’, Sci. Technol. Weld. Join., 2011, 16, (3), 279–284. doi: 10.1179/1362171811Y.0000000003

13.

Paddea

, Francis

J. A.

, Paradowska

A. M.

, Bouchard

P. J.

and Shibli

I. A.

: ‘Residual stress distributions in a P91 steel-pipe girth weld before and after post weld heat treatment’, Mater. Sci. Eng. A, 2012, A534, 663–672. doi: 10.1016/j.msea.2011.12.024

14.

Shirzadi

A. A.

, Bhadeshia

H. K. D. H.

, Karlsson

and Withers

P. J.

: ‘Stainless steel weld metal designed to mitigate residual stresses’, Sci. Technol. Weld. Join., 2009, 14, (6), 559–565. doi: 10.1179/136217109X437178

15.

Francis

J. A.

, Turski

and Withers

P. J.

: ‘Measured residual stress distributions for low and high heat input single weld beads deposited on to SA508 steel’, Mater. Sci. Technol., 2009, 25, (3), 325–334. doi: 10.1179/174328408X372074

16.

Onsoien

M. I.

, M'Hamdi

and Akselsen

O. M.

: ‘Residual stresses in weld thermal cycle simulated specimens of X70 pipeline steel’, Weld. J., 2010, 89, 127-s–132-s.

17.

Zhang

Z. L.

, Gundersen

, Aune

, Odegard

, Grong

and method

: ‘A method for retrieving temperature and microstructure dependent apparent yield strength for aluminum alloys’, Comput. Mater. Sci., 2006, 34, 35–45. doi: 10.1016/j.commatsci.2004.10.004

18.

, Chen

, Jin

, Li

and Lu

: ‘Effects of weld metal strength and transformation temperature on welding residual stress’, Sci. Technol. Weld. Join., 2015, 20, (3), 208–215. doi: 10.1179/1362171814Y.0000000274

19.

Law

, Prask

, Luzin

and Gnaepel-Herold

: ‘Residual stress measurements in coil, linepipe and girth welded pipe’, Mater. Sci. Eng. A, 2006, A437, 60–63. doi: 10.1016/j.msea.2006.04.062