In fossil and nuclear power plants, circumferential butt weld is the most common type of joint in piping system. In circumferential butt weldments, welding residual stresses are always formed due to localized heating induced by the welding process and subsequent rapid cooling. As we all know, when assessing the risk for growth of minor defects and static fracture in weldments, the welding residual stresses may be the main contributor. Furthermore, the magnitude and distribution of residual stresses always play a fundamental role in the failure of welded structure. Among the residual stresses, tensile residual stress is a main factor enhancing the occurrence of stress corrosion cracking, fatigue damage and brittle fracture [1–4]. Consequently, predicting the residual stresses distribution accurately and evaluation of their effects are crucial to assess the integrity and reliability of a welded structure. ASME-grade P92 ferritic steel, also known as NF616 in Japan, is widely used in pipelines serviced in modern ultra-supercritical (USC) power plant and also under active consideration as candidate materials for high-temperature structural components in Generation IV nuclear power plants . Because of the potentiality of P92 steel for the application of all modern power plants and advanced reactor systems, the level of interest in its weld characteristics is increasing.
In recent years, finite element method (FEM) has become a very popular tool for the prediction of residual stresses. A substantial amount of the simulation and experimental work focusing on circumferential welding with emphasis on pipe welding is available in literatures. Many studies [6–12] have indicated that welding residual stresses depended on several factors including the wall thickness, the structure restraint, the type of weld joint, the number of welding passes, welding heat input, material properties and many other welding parameters. Lee and Chang  investigated the effects of pipe diameter on residual stresses in circumferential welding, and the effects of wall thickness were discussed by Teng and Chang . Sattari-Far and Farahani  studied the effects of weld groove shape and weld pass number on welding residual stresses in butt-welded pipes and the numerical models were verified by using hole drilling strain-gauge method. Jiang and his co-workers  estimated the residual stresses and deformation in the repair weld of stainless steel clad plate by FEM at different welding layer numbers and heat inputs, and results showed that using multiple-layer welding and high heat input it was useful to decrease the residual stresses in the repair weld of clad plate. Heinze et al.  compared the residual stresses of test plates welded under free shrinkage conditions with the plates under transverse shrinkage restraint and results showed an increase of transverse residual stresses at the restraint conditions. Deng and his co-workers  clarified the influence of yield strength of weld metal on the welding residual stresses by means of numerical simulation in SUS304 stainless pipe. Furthermore, the influence of preheat and interpass temperature on welding-induced residual stresses was described in the investigation performed by Heinze et al. . More comprehensive review of these factors affecting the welding residual stresses in welded structures was presented in Leggatt’s  study. Among the influenced factors listed above, it could be noted that the effect of arc travelling speed during welding process was rarely discussed in the published studies and the studies on the effect of weld pass sequences were also very limited.
In this paper, 3D FEM is developed to study the effects of the arc travelling speed and welding sequences on the magnitude and distribution of residual stresses for P92 pipes. In each case, other impacting parameters are kept fixed. This paper aims to provide a reference for optimizing the welding technology and improving the fabrication process of ferritic P92 welded pipes.
ASME-grade P92 steel base material, and CROMOCORD 92 weld filler metal are used in the present investigation and their chemical compositions are shown in Table 1. The pipe with an outer diameter of 105 mm and a wall thickness of 24 mm was supplied in the normalized and tempered conditions; geometry of the multi-pass butt-welded P92 pipe is shown in Figure 1. In this analysis, the base metal and weld metal are assumed to be the same material. The thermal physical properties and mechanical properties of P92 steel, which were also used in our previous study , are shown in Figure 2. The material properties are specific heat capacity cp, coefficient of thermal expansion α, elastic modulus E, thermal conductivity k, yield strength σy, Poisson’s ration μ; and the density of the material is assumed to be constant at a value of 7,850 kg/m3.
Finite element modelling
To capture the temperature fields and the residual stresses in the welded P92 pipes accurately, 3D finite element model has been developed using ABAQUS software. Thermo-mechanical behaviour of the weldments during welding has been simulated by using sequential coupling analysis. Firstly, the temperature distribution and its history during welding are computed by a heat conduction analysis, and then the subsequent mechanical analysis is conducted based on the thermal analysis results. Figure 3 shows the finite element meshes used in the finite element analysis which includes the thermal and mechanical analysis. Element type of DC3D8 is used in the thermal analysis and C3D8 for the mechanical analysis. For the multi-pass welding, an element add and remove technique is used to simulate the deposition of weld metal.
For thermal analysis, a moving heat source presented by Goldak et al.  is travelled along the circumference of the weld to generate the weld thermal cycles. The heating period is controlled by the torch travelling speed and the length of the welded pass currently deposited. The mechanical calculation is achieved by using the temperature fields computed previously by the thermal analysis. The elasto-plastic mechanical analysis is based on the equilibrium and constitutive equations. Same finite model used in the thermal and mechanical analysis is employed here, except for the element type and boundary conditions. Mechanical boundary conditions are applied to reflect the actual welding restraint conditions and prevent rigid body motion. Details about the finite element modelling have been described in our previous study .
To validate the finite element modelling results, P92 pipes were welded manually in factory. Weldments were performed by multi-pass welding under the condition shown in Table 2. The welds were divided into two parts, firstly, Gas Tungsten Arc Welding (GTAW) for two-layer root pass welding, followed by Shield Metal Arc Welding (SMAW) for the remaining passes; the welding sequences and the dimensional details of the groove are given in Figure 4. Residual stresses on the outside surface of the pipe were measured through X-ray diffraction technique. Figure 5 compares the simulated axial stress and the corresponding measured data on the outside surface along path 3 as shown in Figure 3. It’s obvious that the simulated results are in good agreement with measured data.
Results and discussion
Residual stresses distribution
Residual stresses distributions of the model used in the work have been discussed carefully in our previous work  and the following results have been achieved:
For residual stresses on the pipe outer surface, the highest hoop, axial and von Mises stresses are found in the last weld bead. And the peaks of tensile residual stress for full welded section take place at the middle thickness of the pipe in heat-affected zone (HAZ).
The through-wall hoop stress at the weld centre line and the HAZ line is mostly tensile. Hoop stress and axial stress are in the same shape and both of them are in very small compressive stresses on the inner surface.
Along the pipe circumference at the outside surface, larger and unevenly distributed residual stresses can be found near the welding start point.
Effect of arc travelling speed
Keeping the rest welding parameters constant, three cases of different arc travelling speed (130 mm/min, 150 mm/min, 170 mm/min) are studied here to discuss the influence of arc travelling speed. Distributions of the residual stresses on the inner surface, HAZ zone and outer surface of butt-welded pipes are shown in Figures 6–8, respectively.
Figure 6 shows the distribution of axial and hoop residual stresses for the three cases along path 1 shown in Figure 3. We can see from Figure 6(a) that the peaks of axial stress in the three cases do not display significant differences. Both the axial stresses in compressive state at the right side and left side near the welded zone decrease with decreasing welding speed. Near the welded root, slower welding speed gives smaller axial residual stress. Figure 6(b) shows the distributions of hoop residual stress; it can be seen that they have the same trend at different welding speeds. Peaks of the tensile hoop stress increase from 45 MPa to 130 MPa when the arc travelling speed increases from 150 mm/min to 170 mm/min.
Figure 7 shows the effect of arc travelling speed on the residual stresses along path 2 shown in Figure 3, in which path 2 is located at the right side HAZ of the weldment. Axial and hoop stresses are shown in Figure 7(a) and 7(b), respectively, from which, significant change of the residual stresses can be seen due to the different welding speeds. Figure 7(a) shows that peaks of the axial residual stress decrease from 357 MPa (in the case of 170 mm/min) to 311 MPa (in the case of 150 mm/min) and 233 MPa (in the case of 130 mm/min). And it can also be seen that the welding speeds do not change the location of the peak axial stress. Nevertheless, as for the hoop stress, the location of peak hoop stress slightly moves towards the inner surface with the decrease in welding speed, which can be found in Figure 7(b). The peaks of the hoop residual stress decline from 582 MPa (in the case of 170 mm/min) to 526 MPa (in the case of 150 mm/min) and 424 MPa (in the case of 130m m/min). In other words, welding speed decreasing from 170 mm/min to 130 mm/min results in the peaks of hoop stress decrease about 27% and axial stresses decrease about 35%. Obviously, the decrease in welding speed affects the residual stresses significantly and causes the reduction of both axial and hoop stresses.
Figure 8 shows the influence of arc travelling speed on the residual stresses along path 3 on the outer surface of the pipe, which is presented in Figure 3. It is observed that the residual stress distribution is not obviously affected by the different welding speeds. Only slight difference can be seen near the welded zone, and the slower the welding speed, the smaller the residual stresses.
From the above analysis, it can be concluded that the arc travelling speed affects the residual stresses at HAZ greatly, while it has little effect on the residual stresses on the pipe outer surface. When the rest welding parameters are constant, the slower arc travelling speed results in the larger heat input which causes the welding temperature to increase. As the yield stress is decreased with the increasing temperature, higher temperature makes the deformation of the pipe materials to increase, which can release more residual stresses. Therefore, the residual stresses are decreased as the arc travelling speed decreases.
Effect of welding sequences
It should be noted that in the published studies different types of welding sequences were adopted to weld the thick-wall pipes, and the types could be divided into case A, case B and case C, as shown in Figure 9. Case A is the type we used in Section “Finite element modelling” to verify the finite element model. As is well known, for multi-pass weldments, choosing the proper welding sequences is a more efficient and simpler way to reduce welding residual stresses than changing other welding parameters. Hence developing an available welding sequence is necessary. According to the studies performed by Yaghi et al.  and Brickstad et al. , residual stresses of P91 pipes and stainless steel pipes were investigated in the type of case A, while case B was used by Yaghi et al. [19–21] and Paddea et al.  in many studies. In the following section, all of the three types are performed to investigate how the welding sequences affect residual stresses in multi-pass butt-welded pipes. The diameter and length of the pipes in all models are the same, and the same material properties in all analyses are used.
Figure 10 represents the distributions of the residual stresses along path 1. As this figure reveals, the stress distributions are symmetrical at the weld centre line in case B. Maximum of the compressive axial stress occurs in case A. Both of case A and case C have the same residual stresses distribution regardless of axial or hoop stress.
Figure 11 depicts the effect of welding sequences on residual stresses along path 2. It is clearly seen that welding sequences of case A and case C have no influence on the residual stresses in HAZ. However, in case B, both axial and hoop stresses decrease significantly, peaks of axial stress decrease from 310 MPa to 170 MPa and peaks of hoop stress decrease about 16%.
Figure 12 shows a comparison among residual stresses of the three cases along path 3. It is also shown that case A and case C have the same stress distribution and nearly no difference of the magnitude can be seen between them. When the welding sequences is the type of case B, the maximum compressive residual stresses move to the place where the last weld bead is to be deposited. It can also be seen that case B displays the distributions of residual stresses more symmetrical at the weld centre line.
As a result of comparison among the residual stresses of these three cases at the different places, it is observed that, in general, case A and case C have no significant difference on the distribution and magnitude of the residual stresses. This is due to the fact that inter-pass temperature (200°C–300°C) is controlled strictly during the welding simulation. In the HAZ, the residual stresses of case B are smaller than the other two cases. The reason is that HAZ suffers from the proper post-weld heat treatment and pre-heating effect during welding. Compared with axial stress, welding sequences have less impact on the distribution and magnitude of the hoop stress.
The FEM is employed herein to investigate the effect of arc travelling speed and welding sequences on residual stresses in welded sections of new ferritic P92 pipes. Based on the results in this study, the main conclusions can be drawn as follows:
Both arc travelling speed and welding sequences have influence on the residual stresses. In the HAZ of the weldment, the most significant influence can be found.
The residual stresses increase with the increasing arc travelling speed. Welding sequences affect the distribution and magnitude of axial stresses more significantly than hoop stresses.
Using the slow arc travelling speed and the case B welding sequences can be useful to decrease the residual stresses, which provides a reference for optimizing the welding technology for new ferritic P92 welded pipes with small diameter and thick wall.
The authors gratefully acknowledge the support provided by innovation program for graduate students in JiangSu Province of China (No. CXZZ13_0430) and the Ningbo Natural Science Foundation (No. 2009A610023).
Putanunda SK, Sengupta A, Schaefer J. High Temp Mater Process 1994;13:113–26.Google Scholar
Ren XB, Zhang ZL, Nyhus B. Eng Fract Mech 2010;77:1325–37. Google Scholar
Giroux PF, Dalle F, Sauzay M, Malaplate J, Fournier B, Gourgues-Lorenzon AF. Mater Sci Eng A Struct 2010;527:3984–93. Google Scholar
Teng TL, Lin CC. Int J Pres Ves Pip 1998;75:857–64.Google Scholar
Lee CH, Chang KH. Mater Sci Eng A Struct 2008;487:210–18. Google Scholar
Heinze C, Schwenk C, Rethmeier M. J Constr Steel Res 2012;72:12–19. Google Scholar
Yaghi AH, Hyde TH, Becker AA, Williams JA, Sun W. J Mater Process Technol 2005;167:480–7. Google Scholar
Yaghi AH, Hyde TH, Becker AA, Sun W. Int J Pres Ves Pip 2013;111:173–86. Google Scholar
Paddea S, Francis JA, Paradowska AM, Bouchard PJ, Shibli IA. Mater Sci Eng A Struct 2012;534:663–72. Google Scholar
About the article
Published Online: 2015-02-21
Published in Print: 2016-02-01