Influence of the repair length on the residual stress in P92 steel repair welds

0.   Introduction

P92 steel is widely used in supercritical steam pipes to reduce the weight of the pipes and boiler and improve the pipe system’s flexibility to achieve excellent creep performance and high temperature strength^[1-2]. However, some defects may be exposed due to improper welding processes or damage incurred during the installation of P92 steel, which pose a significant threat to the safe operation of the equipment. After the occurrence of relevant defects, it is usually necessary to repair them by repair welding, which is economical and helps to maintain the structural integrity of the system. As a ferritic heat-resistant steel, the heat-affected zone (HAZ) of P92 steel can contain type-Ⅳ cracks, which may eventually cause structural failure. Some research has shown that residual stress has an important influence on type-Ⅳ cracks^[3]. Therefore, it is of great theoretical significance and engineering application value to research the distribution of residual stress arising in P92 steel during the repair-welding process.

With the booming development of computer technology, the finite element method (FEM) is widely used for the prediction of residual stress. Based on plates of different shapes, Qiao et al.^[4] conducted a butt joint welding deformation and residual stress analysis using FEM and indicated that the residual stress mainly distributes near the weld and the edges that are clamped. According to Yong’s ^[5] research, under multiaxial stress, the plastic deformation at the root of the notch can accelerate material damage and shorten the creep life. Zhang et al.^[6] believed that the crack propagation characteristics of HR3C /T91 joints can be accurately characterized by stress triaxiality. Liu et al.^[7] analyzed the generation and distribution mechanisms of residual stress generated during the welding process of P92 by considering the influence of solid state phase transitions on the residual stress. Jiang et al.^[8-9] simulated and analyzed the repair-welding process of stainless steel, and explained the influence of factors such as the repair welding length and width, on the residual stresses that manifest during repair welding.

Based on the finite element simulation software SYSWELD, as well as existing models and experimental data presented in the literature, this paper verifies the feasibility of SYSWELD in simulating the residual stress distribution of P92 steel welding, simulating the repair welding process of P92 steel plates, and analyzing the residual stress field obtained by different repair welding lengths, to determine the influence of the weld length on the residual stress generated during repair welding.

1.   Finite element simulation method

1.1   Analysis model and grid division

The length, width and height of the welded sheet are 200 mm, 200 mm and 10 mm, respectively, as shown in Fig 1. As shown in Fig 2, when constructing 3D mesh, to improve the calculation efficiency and obtain an accurate solution, the mesh in the weld area and HAZ was more densely distributed, while the mesh was distributed more sparsely in positions far from the weld.

Figure  1.  Schematic the simulated specimen ( mm) (a) Top view of the simulated specimen (b) Main view of the simulated specimen

DownLoad: Full-Size Img PowerPoint

Figure  2.  Schematic of the 2D and 3D grids (a) 2D grid (b) 3D mesh grid

DownLoad: Full-Size Img PowerPoint

1.2   Welding parameters and material performance parameters

The simulation process was divided into butt welding, repair welding and heat treatment. The simulated butt welding consisted of multi-layer and multi-pass welding, while the repair welding was conducted during a single pass. The welding method consisted of argon tungsten arc welding, where the welding parameters are shown in Table 1. Post weld heat treatment, the parameters of which are shown in Fig 3, was conducted after the butt-welding and repair-welding steps. Under the assumption that the welding material and the base material have the same material properties, the specific material performance parameters are shown in Fig 4.

Table  1.  Welding parameters

Layer	Welding current I /A	Arc voltage U/V	Welding speed v /(mm.s−1)
1	100	10	1
2–3	125	12	1
4	125	15	1
repair weld	130	20	1

 | Show Table

DownLoad: CSV

Figure  3.  Heat-treatment parameters

DownLoad: Full-Size Img PowerPoint

Figure  4.  Thermal parameters and yield strength of the P92 steel used in this study (a) Thermophysical parameters of the P92 steel (b) Yield strength of the P92 steel

DownLoad: Full-Size Img PowerPoint

1.3   Boundary conditions and initial conditions

In the simulations, the initial temperature and room temperature were both set to 20 ℃. As shown in Fig 2, to prevent the rigid displacement of the specimen during the welding process, the grid model had rigid constraints along the x and y directions. The 3D, x, y and z directions were respectively set at the three endpoints of the specimen.

1.4   Simulation verification of welding residual stress in P92 steel

To ensure the reliability of the material parameters and simulated results, the test data and model, which are shown in Fig 5, consisted of a length of 200 mm, a width of 200 mm, a height of 6 mm, a weld width of 10 mm and a penetration depth of 2 mm; these parameters were chosen to coincide with those used in Deng’s study^[10], and which were used to verify the P92 simulated processes considered in this paper. Fig. 6 shows the comparison between the simulation results of residual stress in this study and the simulation results in Deng’s study and the measured residual stress values using the blind hole method. It can be seen from the figure that the distribution of the residual stress calculated by the P92 model established in this study is basically consistent with the simulated distribution seen in the literature study, and the measured distribution in the blind hole method(BHD). These results indicate that the material parameters and simulation results adopted here can be used to successfully simulate P92 welding and its repair welding process.

Figure  5.  Schematic of the simulated FEM ( mm)

DownLoad: Full-Size Img PowerPoint

Figure  6.  Comparison of residual stress by FEM and BHD (a) Transverse residual stress contrast (b) Longitudinal residual stress contrast

DownLoad: Full-Size Img PowerPoint

2.   Numerical simulation results and analysis

2.1   Welding temperature field distribution

In the temperature distribution field of the welded section shown in Fig 7, it can be seen that the gray area, in which the temperature is higher than the melting point of P92 steel (1350 ℃), is larger than the area of the cut groove. Therefore, it is considered that the weld had reached the melting state during the repair-welding process.

Figure  7.  Temperature distribution of the welded section (a) Peak temperature reached during the butt welding (b) Peak soldering temperature

DownLoad: Full-Size Img PowerPoint

2.2   Distribution of the welding residual stress field

Fig. 8 shows a distribution diagram of transverse and longitudinal residual stresses after repair welding, and it can be seen that tensile stress appears in the weld and in the vicinity of the HAZ. The transverse and longitudinal residual stresses are below 330 MPa and 480 MPa, respectively, while the maximum residual stresses occur in the HAZ.

Figure  8.  Residual stress distribution during welding (a) Transverse residual stress distribution cloud map (b) Longitudinal residual stress distribution cloud map

DownLoad: Full-Size Img PowerPoint

3.   Effect of the repair welding length on the residual stress

The residual stresses of 60 mm, 100 mm, and 140 mm long repair welds were simulated using the welding parameters of a 100 mm long repair weld. To analyze the residual stress distribution in more detail, three important paths (Fig. 9) were selected in this paper, among which P1 occurs along the symmetry axis of the structure, P2 is located in the HAZ, and P3 is located in the center of the repair weld. According to the related literature research^[11], type-Ⅳ cracks generally appear in a fine-grained HAZ in which the temperature is higher than Ac3 (900–920 ℃), and the critical HAZ has a temperature range between Ac1 (800–835 ℃) and Ac3. Therefore, it is very important to choose the appropriate path (P2) in the HAZ. As shown in Fig. 10, this paper extracted thermal cycle curves at three points (a, b, and c), which are marked in Fig. 7b. During the repair welding, it can be seen that the highest temperatures of points a, b, and c are 958 ℃, 910 ℃, and 863 ℃, respectively. Point b shows the presence of type-Ⅳ cracks, hence, path P2 choose the line where point b is.

Figure  9.  Locations of the different reference paths

DownLoad: Full-Size Img PowerPoint

Figure  10.  Thermal cycling curve of different positions in the welding process

DownLoad: Full-Size Img PowerPoint

3.1   Effect of the repair welding length on the surface residual stress (P1)

Fig. 11 shows the influence of the repair welding length on the surface residual stress (P1). The transverse residual stress augments gradually from the center of the weld to the edge of the weld, decreases dramatically in the HAZ, increases gradually after passing through the HAZ, and then decreases gradually after reaching the peak. The longitudinal residual stress gradually increases from the center to the edge of the weld, and then gradually declines. By comparing the influence of different repair welding lengths on the residual stress along the P1 path, it is found that with an increase of the repair welding length, the horizontal stress level shows a gradually declining trend, while changes in the longitudinal stress are not obvious.

Figure  11.  Influence of the repair welding length on the residual stress on the surface (a) Transverse residual stress (b) Longitudinal residual stress

DownLoad: Full-Size Img PowerPoint

3.2   Effect of the repair welding length on the residual stress in the HAZ (P2)

In our study of repair welds of different lengths, the normalized distance and the ratio of the starting point length of repair welding to the respective repair welding length were considered. Fig. 12 shows the influence of the repair welding length on the residual stress in the HAZ. For repair welds of different lengths, as the repair welding length increases, the transverse residual stress in the HAZ decreases gradually, while the effect on the longitudinal residual stress is less.

Figure  12.  Influence of the repair welding length on the residual stress in the HAZ (a) Transverse residual stress (b) Longitudinal residual stress

DownLoad: Full-Size Img PowerPoint

3.3   Influence of the repair welding length on the residual stress of the weld (P3)

Fig. 13 shows the influence of the repair welding length on the weld’s residual stress. With an increase of repair welding length, both the transverse residual stress and the longitudinal residual stress show a decreasing trend.

Figure  13.  Influence of the repair welding length on the residual stress in the weld joint (a) Transverse residual stress (b) Longitudinal residual stress

DownLoad: Full-Size Img PowerPoint

4.   Conclusions

（1） After the repair welding process, the tensile stresses of the repair welding seam and in the nearby HAZ were relatively large, while the stress of the base material was relatively small.

（2） In terms of transverse residual stress, as the repair welding length increased, the stress at the center of the structure surface, the stress in the HAZ and the stress at the weld line, all tended to decrease.

（3） In terms of longitudinal residual stress, the stress in the weld center tended to decrease with an increase of the weld length, but the length had no effect on the structure’s center and the HAZ.