Influence of Repair Welding on the Fatigue Behavior of S355J2 T-Joints

This paper investigated the effect of repair welding on the microstructure, mechanical properties, and high cycle fatigue properties of S355J2 steel T-joints in orthotropic bridge decks. The test results found that the increase in grain size of the coarse, heat-affected zone decreased the hardness of the welded joint by about 30 HV. The tensile strength of the repair-welded joints was reduced by 20 MPa compared to the welded joints. For the high cycle fatigue behavior, the fatigue life of repair-welded joints is lower than that of the welded joints under the same dynamic load. The fracture positions of toe repair-welded joints were all at the weld root, while the fracture positions of the deck repair-welded joints were at the weld toe and weld root, with the same proportion. The fatigue life of toe repair-welded joints is reduced more than that of deck repair-welded joints. The traction structural stress method was used to analyze fatigue data of the welded and repair-welded joints, and the influence of angular misalignment on was considered. The fatigue data with and without AM are all within the ±95% confidence interval of the master S-N curve.


Introduction
Orthotropic steel deck (OSD) is widely used in the construction of various bridges with the advantages of lightweight, low cost, and large bearing capacity [1,2]. However, fatigue cracks often appear on OSD due to welding residual stress and complex traffic load [3,4]. The cracks of the T-joints between the U-rib and deck have the greatest impact on OSD performance. Xiao et al. [5] conducted stress analysis and fatigue evaluation to understand key factors contributing to the cracks of rib-deck T-joints. The test results found that the fatigue cracks were mainly located at the weld toe and weld root. These fatigue cracks can be categorized into four types according to different growth paths, as illustrated in Figure 1, namely toe-deck crack (crack I), root-deck crack (crack II), root-weld crack (crack III) and toe-rib crack (crack IV) [5]. Cracks I and II can penetrate the bridge panel and cause pavement damage [6,7]. Sim et al. [8] conducted a full-scale fatigue test on OSD. The test results found that the fatigue cracking of rib-deck T-joints was more serious at the deck side than at the rib side, and the cracking probability of the weld toe was higher than that of the weld root. Therefore, the research should focus on the fatigue cracks at the deck side.
At present, most of the bridges in service worldwide have used OSD structures [9], which are easily affected by fatigue cracks. In order to prolong the service life of the bridge, it is necessary to repair the fatigue crack in time. Among the many ways to repair cracks, repair welding is the most widely used with the advantages of good economy and high efficiency [10,11]. Numerous studies have been carried out to explore the influence of repair welding on the microstructure and mechanical properties of materials. Aman et al. [12] discovered a decrease in the hardness of the heat-affected zone (HAZ) after repair welding Materials 2023, 16, 3682 2 of 17 on 316 L stainless steel. The decrease in hardness was attributed to the reduction of δ-ferrite in HAZ, and the hardness will decrease with the increase of the number of repair welding. Luo et al. [13] investigated the effect of repair welding heat input on the microstructure of 304 stainless steel. It was found that with the increase of heat input, the δ-ferrite in the weld metal decreased slightly, while the austenite grain size in the HAZ was slightly increased. Katsas et al. [14] conducted tensile and bending tests on both welded joints and repairwelded joints. The results found that a single repair welding had no effect on the ultimate tensile strength of the welded joints. And in the absence of "non-fusion", the repair-welded joints passed the bending test as successfully as the welded joints. Shojaati et al. [15] pointed out that multiple repair welding had no adverse effect on the microstructure and mechanical properties of X20Cr13 stainless steel. It can refine austenite grains in the HAZ and increase impact toughness. Dong [16] proposed that a major detrimental effect of repair welding on structural integrity is the increase of crack driving force caused by the elevation of membrane stress level, and the adjustment of welding size can effectively reduce the membrane stress. At present, most of the bridges in service worldwide have used OSD struct which are easily affected by fatigue cracks. In order to prolong the service lif bridge, it is necessary to repair the fatigue crack in time. Among the many ways t cracks, repair welding is the most widely used with the advantages of good econo high efficiency [10,11]. Numerous studies have been carried out to explore the in of repair welding on the microstructure and mechanical properties of materials. A al. [12] discovered a decrease in the hardness of the heat-affected zone (HAZ) afte welding on 316 L stainless steel. The decrease in hardness was attributed to the re of δ-ferrite in HAZ, and the hardness will decrease with the increase of the nu repair welding. Luo et al. [13] investigated the effect of repair welding heat inpu microstructure of 304 stainless steel. It was found that with the increase of heat in δ-ferrite in the weld metal decreased slightly, while the austenite grain size in th was slightly increased. Katsas et al. [14] conducted tensile and bending tests welded joints and repair-welded joints. The results found that a single repair weld no effect on the ultimate tensile strength of the welded joints. And in the absence o fusion", the repair-welded joints passed the bending test as successfully as the joints. Shojaati et al. [15] pointed out that multiple repair welding had no adver on the microstructure and mechanical properties of X20Cr13 stainless steel. It ca Contrary to the research on microstructure and mechanical properties, there are few studies on the fatigue properties of repair-welded joints. Wang et al. [17] found that the fatigue life of the repair-welded specimen is 7.1 times that of the specimen using the drilling stop-hole method, and 3.3 times that of the specimen reinforced by carbon fiber-reinforced polymer (CFRP). Seo et al. [18] studied the fatigue property of the bogie structure after repair welding. The results found that the fatigue life of repair-welded joints was lower than that of welded joints. When the repair welding width decreased, the fatigue life increased.
Due to the structural complexity, current fatigue design/evaluation guidelines for welded structures are mostly stress range based, likely hot spot stress method [19,20] and nominal stress method [21,22]. A traction-based structural stress definition was introduced by imposing equilibrium conditions [23]. In addition to its demonstrated mesh-insensitivity and convenience when applied to large and complex assemblies, the method has led to the development of the Master S-N curve adopted by ASME Div 2 [24] since 2007.
In this study, the repair welding was carried out on S355J2 steel T-joints, and positions were selected at the weld toe and deck to fit actual working conditions. The microstructure and mechanical properties of the repair-welded joints were studied. In addition, the high cycle fatigue (HCF) properties of the joints are analyzed by the traction structural stress method considering the influence of angular misalignment (AM).

Materials and Test Specimen Preparation
The material of base metal and filler metal used in this study are S355J2 steel and E500T-1 wire, respectively. Table 1 provides their chemical compositions. Metal active gas (MAG) arc welding is adopted in this study with the advantages of convenience on construction. The manufacturing process and dimensions of test specimens are shown in Table 2. The size of the deck plate used in this study is 1800 mm × 600 mm × 18 mm (length × width × thickness), and the thickness of the U-rib is 8 mm. The blunt edge of the welding groove is 0-1 mm, and the groove Angle is 50 • . The length of the weld is equal to the length of the deck plate, and the size of the weld leg is 8-10 mm. The OSD is cut by wire-cutting equipment to fabricate the T-joints. As shown in Table 2, the T-joint has a length of 300 mm and a width of 100 mm. The welded joints, toe repair-welded joints, and root repair-welded joints are prepared in this study. The length of the repair weld is the same as the width of the deck plate. The width and depth of repair welds are 8 mm and 4.5 mm, respectively. Figure 2 shows the schematic diagrams of repair-welded joints. plate. The width and depth of repair welds are 8 mm and 4.5 mm, respectively. Figure 2 shows the schematic diagrams of repair-welded joints.

Microstructure and Mechanical Properties Test
The T-joints were cut into metallographic samples by using a wire-cutting machine. The metallographic samples are corroded with 4% nitrate alcohol for about 10 s after grinding and polishing. The microstructure of the welded and repair-welded joints is observed by the Olympus-DSX510 optical microscope. The tensile specimens were produced in compliance with ISO 6892-1: 2009 [25], and the dimensions are presented in Figure 3. The 100 kN universal material testing machine was employed for the tensile test, and the rate of strain is 0.03/s. The hardness test was carried out on the ARTCAN-300SSI-C microhardness machine. The interval of adjacent test points is 0.5 mm, the load is 0.98 N, and the hold time is 15 s.

High Cycle Fatigue Test
To explore the fatigue property of welded and repair-welded joints, the HCF test was conducted by the MTS high-frequency fatigue testing machine shown in Figure 4a. Tjoints are fixed by the hydraulic wedge, and clamping length is 80 mm, see Figure 4b.

Microstructure and Mechanical Properties Test
The T-joints were cut into metallographic samples by using a wire-cutting machine. The metallographic samples are corroded with 4% nitrate alcohol for about 10 s after grinding and polishing. The microstructure of the welded and repair-welded joints is observed by the Olympus-DSX510 optical microscope. The tensile specimens were produced in compliance with ISO 6892-1: 2009 [25], and the dimensions are presented in Figure 3. The 100 kN universal material testing machine was employed for the tensile test, and the rate of strain is 0.03/s. The hardness test was carried out on the ARTCAN-300SSI-C microhardness machine. The interval of adjacent test points is 0.5 mm, the load is 0.98 N, and the hold time is 15 s. plate. The width and depth of repair welds are 8 mm and 4.5 mm, respectively. Figure 2 shows the schematic diagrams of repair-welded joints.

Microstructure and Mechanical Properties Test
The T-joints were cut into metallographic samples by using a wire-cutting machine. The metallographic samples are corroded with 4% nitrate alcohol for about 10 s after grinding and polishing. The microstructure of the welded and repair-welded joints is observed by the Olympus-DSX510 optical microscope. The tensile specimens were produced in compliance with ISO 6892-1: 2009 [25], and the dimensions are presented in Figure 3. The 100 kN universal material testing machine was employed for the tensile test, and the rate of strain is 0.03/s. The hardness test was carried out on the ARTCAN-300SSI-C microhardness machine. The interval of adjacent test points is 0.5 mm, the load is 0.98 N, and the hold time is 15 s.

High Cycle Fatigue Test
To explore the fatigue property of welded and repair-welded joints, the HCF test was conducted by the MTS high-frequency fatigue testing machine shown in Figure 4a. Tjoints are fixed by the hydraulic wedge, and clamping length is 80 mm, see Figure 4b.

High Cycle Fatigue Test
To explore the fatigue property of welded and repair-welded joints, the HCF test was conducted by the MTS high-frequency fatigue testing machine shown in Figure 4a. T-joints are fixed by the hydraulic wedge, and clamping length is 80 mm, see Figure 4b. plate. The width and depth of repair welds are 8 mm and 4.5 mm, respectively. Figure 2 shows the schematic diagrams of repair-welded joints.

Microstructure and Mechanical Properties Test
The T-joints were cut into metallographic samples by using a wire-cutting machine. The metallographic samples are corroded with 4% nitrate alcohol for about 10 s after grinding and polishing. The microstructure of the welded and repair-welded joints is observed by the Olympus-DSX510 optical microscope. The tensile specimens were produced in compliance with ISO 6892-1: 2009 [25], and the dimensions are presented in Figure 3. The 100 kN universal material testing machine was employed for the tensile test, and the rate of strain is 0.03/s. The hardness test was carried out on the ARTCAN-300SSI-C microhardness machine. The interval of adjacent test points is 0.5 mm, the load is 0.98 N, and the hold time is 15 s.

High Cycle Fatigue Test
To explore the fatigue property of welded and repair-welded joints, the HCF test was conducted by the MTS high-frequency fatigue testing machine shown in Figure 4a. Tjoints are fixed by the hydraulic wedge, and clamping length is 80 mm, see Figure 4b.   Table 3 lists the parameters of the HCF test. The welded joints are denoted and referred to in this study as A, while the toe repair-welded joints and deck repair-welded joints are marked as B-DZ and B-D, respectively. Each group contains four specimens, which are subjected to fatigue loads of 36/360 kN, 40/400 kN, 44/440 kN and 46/460 kN, respectively.

Traction Structural Stress Method
When both ends of the specimen are strained, the stress distribution around the weld is highly nonlinear in the direction of the thickness of the deck and difficult to be calculated directly. Dong [23] proposed that the stress in the direction of the thickness of the deck can be decomposed into two parts based on stress linearization. The first part is the sum of membrane stress σ m and bending stress σ b , which is called traction structural stress σ s , and it is balanced with external load [26,27]. The other part is notch stress σ n generated by local notch effect, which is highly nonlinear and in self-equilibrium state. The stress distribution diagram in the direction of the thickness of deck is shown in Figure 5. At present, the traction structural stress method has been proven to be suitable for describing the four categories of cracks on rib-deck T-joints. This method converts the joint force/bending moment obtained by finite element analysis (FEA) into membrane stress and bending stress to calculate the traction structural stress of the potential failure interface. Table 3 lists the parameters of the HCF test. The welded joints are denoted and referred to in this study as A, while the toe repair-welded joints and deck repair-welded joints are marked as B-DZ and B-D, respectively. Each group contains four specimens, which are subjected to fatigue loads of 36/360 kN, 40/400 kN, 44/440 kN and 46/460 kN, respectively.

Traction Structural Stress Method
When both ends of the specimen are strained, the stress distribution around the weld is highly nonlinear in the direction of the thickness of the deck and difficult to be calculated directly. Dong [23] proposed that the stress in the direction of the thickness of the deck can be decomposed into two parts based on stress linearization. The first part is the sum of membrane stress and bending stress , which is called traction structural stress , and it is balanced with external load [26,27]. The other part is notch stress generated by local notch effect, which is highly nonlinear and in self-equilibrium state. The stress distribution diagram in the direction of the thickness of deck is shown in Figure 5. At present, the traction structural stress method has been proven to be suitable for describing the four categories of cracks on rib-deck T-joints. This method converts the joint force/bending moment obtained by finite element analysis (FEA) into membrane stress and bending stress to calculate the traction structural stress of the potential failure interface.  In this paper, the FEA based on ABAQUS 6.14 software is used to calculate the traction structural stress of the potential failure surface. To establish the finite element model (FEM) of joints accurately, WiKi-SCAN 2.0 is used to measure the dimensional parameters of joints, see Figure 6. The dimensional parameters of welded and repair-welded joints are listed in Table 4.
In this paper, the FEA based on ABAQUS 6.14 software is used to calculate the traction structural stress of the potential failure surface. To establish the finite element model (FEM) of joints accurately, WiKi-SCAN 2.0 is used to measure the dimensional parameters of joints, see Figure 6. The dimensional parameters of welded and repair-welded joints are listed in Table 4.  The FEM of welded and repair-welded joints are established based on the dimensional parameters, see Figure 7. Because the excess welds on the surface of the deck will be removed after repair welding, the FEM of deck repair-welded joints is the same as that of the welded joints. The boundary condition and loading mode of the welded and repairweld joints model are the same. The degree of freedom in the Y direction of the fixture position of the FEM is constrained, that is, the surface area which is 80 mm away from both ends of the FEM. The movement at the left end of the FEM is restricted in the X and Z directions. At the same time, 1 MPa tensile stress is applied to the right end face of the FEM. It can be seen that the mesh division of the FEM is not uniform. This arrangement can not only improve the calculation efficiency but also ensure the accuracy of the calculation results because of the mesh insensitivity of the traction structural stress method.   The FEM of welded and repair-welded joints are established based on the dimensional parameters, see Figure 7. Because the excess welds on the surface of the deck will be removed after repair welding, the FEM of deck repair-welded joints is the same as that of the welded joints. The boundary condition and loading mode of the welded and repairweld joints model are the same. The degree of freedom in the Y direction of the fixture position of the FEM is constrained, that is, the surface area which is 80 mm away from both ends of the FEM. The movement at the left end of the FEM is restricted in the X and Z directions. At the same time, 1 MPa tensile stress is applied to the right end face of the FEM. It can be seen that the mesh division of the FEM is not uniform. This arrangement can not only improve the calculation efficiency but also ensure the accuracy of the calculation results because of the mesh insensitivity of the traction structural stress method. In this paper, the FEA based on ABAQUS 6.14 software is used to calculate the traction structural stress of the potential failure surface. To establish the finite element model (FEM) of joints accurately, WiKi-SCAN 2.0 is used to measure the dimensional parameters of joints, see Figure 6. The dimensional parameters of welded and repair-welded joints are listed in Table 4.  The FEM of welded and repair-welded joints are established based on the dimensional parameters, see Figure 7. Because the excess welds on the surface of the deck will be removed after repair welding, the FEM of deck repair-welded joints is the same as that of the welded joints. The boundary condition and loading mode of the welded and repairweld joints model are the same. The degree of freedom in the Y direction of the fixture position of the FEM is constrained, that is, the surface area which is 80 mm away from both ends of the FEM. The movement at the left end of the FEM is restricted in the X and Z directions. At the same time, 1 MPa tensile stress is applied to the right end face of the FEM. It can be seen that the mesh division of the FEM is not uniform. This arrangement can not only improve the calculation efficiency but also ensure the accuracy of the calculation results because of the mesh insensitivity of the traction structural stress method.  Nodal forces calculation at the potential failure surface is shown in Figure 8. The surface of interest is defined in FEM, namely toe failure surface (Figure 8a) and root failure surface (Figure 8b). All elements connected to one side of the potential failure surface are identified to extract nodal forces, which are highlighted in Figure 8. Then the traction structural stress σ s can be calculated using Equations (1)- (3).
Nodal forces calculation at the potential failure surface is shown in Figure 8. The surface of interest is defined in FEM, namely toe failure surface (Figure 8a) and root failure surface (Figure 8b). All elements connected to one side of the potential failure surface are identified to extract nodal forces, which are highlighted in Figure 8. Then the traction structural stress can be calculated using Equations (1)-(3). Once the traction structural stresses have been determined using the procedures described through the above process, the equivalent traction structural stress commonly used to evaluate the fatigue life of welded joints in engineering can be obtained through Equations (4)-(7) [26]. where: 3.6 (6) Once the traction structural stresses have been determined using the procedures described through the above process, the equivalent traction structural stress commonly used to evaluate the fatigue life of welded joints in engineering can be obtained through Equations (4)-(7) [26]. where: 3. Microstructure and Mechanical Properties 3.1. Microstructure Figure 9 illustrates the variation in the microstructure of the S355J2 steel-welded joint. It can be seen that the welded joint of fusion welding is mainly composed of weld metal (WM), heat-affected zone (HAZ), and base metal. HAZ can be divided into three distinct zones, namely, coarse grain HAZ (CGHAZ), fine grain HAZ (FGHAZ), and inter-critical region (ICHAZ-lying between the base metal and FGHAZ) [28]. The WM (Figure 9a) is mainly composed of the pro-eutectoid ferrite precipitated along the grain boundary of the columnar crystal and the inner acicular ferrite. The CGHAZ (Figure 9b) consists of lath martensite and a little granular bainite. The grains continue to grow after reaching austenitizing temperature, and the overheated structure with coarse grains is obtained. The microstructure of FGHAZ (Figure 9c) contains ferrite, pearlite, and a small amount of granular bainite. Compared with the base metal, the grain size of FGHAZ is small. In ICHAZ (Figure 9d), pearlite is first austenitized and then forms finer ferrite and pearlite after cooling. Ferrite fails to austenitize but continues to grow. Therefore, the microstructure structure of ICHAZ is similar to that of the base metal, and the size of grains varies in this zone. Figure 9 illustrates the variation in the microstructure of the S355J2 steel-welded joint. It can be seen that the welded joint of fusion welding is mainly composed of weld metal (WM), heat-affected zone (HAZ), and base metal. HAZ can be divided into three distinct zones, namely, coarse grain HAZ (CGHAZ), fine grain HAZ (FGHAZ), and inter-critical region (ICHAZ-lying between the base metal and FGHAZ) [28]. The WM (Figure 9a) is mainly composed of the pro-eutectoid ferrite precipitated along the grain boundary of the columnar crystal and the inner acicular ferrite. The CGHAZ (Figure 9b) consists of lath martensite and a little granular bainite. The grains continue to grow after reaching austenitizing temperature, and the overheated structure with coarse grains is obtained. The microstructure of FGHAZ (Figure 9c) contains ferrite, pearlite, and a small amount of granular bainite. Compared with the base metal, the grain size of FGHAZ is small. In ICHAZ (Figure 9d), pearlite is first austenitized and then forms finer ferrite and pearlite after cooling. Ferrite fails to austenitize but continues to grow. Therefore, the microstructure structure of ICHAZ is similar to that of the base metal, and the size of grains varies in this zone.  Figure 10 shows the microstructure of the toe repair-welded joint. The grain distribution and morphology in the repair-welded joint are similar to that in the welded joint, which can be divided into WM, HAZ, and base metal.  Figure 10 shows the microstructure of the toe repair-welded joint. The grain distribution and morphology in the repair-welded joint are similar to that in the welded joint, which can be divided into WM, HAZ, and base metal.  Figure 10 shows the microstructure of the toe repair-welded joint. The grain dist bution and morphology in the repair-welded joint are similar to that in the welded joi which can be divided into WM, HAZ, and base metal. The Image Pro Plus 6.0 software is used to analyze the microstructure of the weld and repair-welded joints. Figure 11 compares the image of the grains in the CGHAZ, a it is found that the austenite grains further grow. The content of lamellar martensite d creases, and the content of lath martensite increases. The average austenite grain area The Image Pro Plus 6.0 software is used to analyze the microstructure of the welded and repair-welded joints. Figure 11 compares the image of the grains in the CGHAZ, and it is found that the austenite grains further grow. The content of lamellar martensite decreases, and the content of lath martensite increases. The average austenite grain area of the weld joint is calculated to be 1645.23 µm 2 , and that of the repair-welded joint is 3368.39 µm 2 , increased by 2.05 times. The coarse grains can decrease the mechanical properties of the welded joints, which may become a weak point in the HCF test. the weld joint is calculated to be 1645.23 µm 2 , and that of the repair-welded joint is 3368.39 µm 2 , increased by 2.05 times. The coarse grains can decrease the mechanical properties of the welded joints, which may become a weak point in the HCF test.

Microhardness
To eliminate the influence of point position on hardness, it must be ensured that the test position is the same in welded and the repair-welded joints, see Figure 12a. The hard-

Microhardness
To eliminate the influence of point position on hardness, it must be ensured that the test position is the same in welded and the repair-welded joints, see Figure 12a. The hardness test results are demonstrated in Figure 12b. The zone 1 to the left of the black line in Figure 12b is the WM of welded joints. Due to the grain growth, the average hardness of this zone decreases by about 30 HV after repair welding. The zone 2 to the right of the black line in Figure 12b is the WM of repair-welded joints, and it can be found that the hardness of this zone increases by about 60 HV. This is because repair welding remelts this zone and reduces the grain size, which results in the enhancement of hardness.

Microhardness
To eliminate the influence of point position on hardness, it must be ensured that t test position is the same in welded and the repair-welded joints, see Figure 12a. The har ness test results are demonstrated in Figure 12b. The zone ① to the left of the black li in Figure 12b is the WM of welded joints. Due to the grain growth, the average hardne of this zone decreases by about 30 HV after repair welding. The zone ② to the right of t black line in Figure 12b is the WM of repair-welded joints, and it can be found that t hardness of this zone increases by about 60 HV. This is because repair welding reme this zone and reduces the grain size, which results in the enhancement of hardness.

Tensile
The tensile properties of welded and repair-welded joints are presented in Figure  and

Tensile
The tensile properties of welded and repair-welded joints are presented in Figure 13 and Table 5

Fatigue Test Results
The HCF test results of the welded and repair-welded joints are presented in Table 6. By calculation, the fatigue properties of toe and deck repair-welded joints are reduced by about 60% and 30% under the same fatigue load compared with the welded joints. The fracture position of each welded joint is divided into two places: weld toe and weld root. Figure 14 show pictures of A1 (fracture at weld root) and A2 (fracture at weld toe), respectively. The A-2,3,4 fractures at the weld toe and A-1 fractures at the weld root. Among four deck repair-welded joints (B-D-1,2,3,4), two breaks at the weld toe and two at the weld root. This indicates that the fatigue property of the weld toe and weld root is comparable. While toe repair-welded joints (B-DZ-1,2,3,4) are all broken at the weld root. The data in Table 4 show that the angle between the weld and deck (toe angle2) increases from 120 • to 155 • , reducing the stress concentration at the weld toe, which results in the failure positions of toe repair-welded joints all located at the weld root. More fracture picture of joints are show in Appendix A. The fracture position of each welded joint is divided into two places: weld toe a weld root. Figure 14 show pictures of A1 (fracture at weld root) and A2 (fracture at w toe), respectively. The A-2,3,4 fractures at the weld toe and A-1 fractures at the weld ro Among four deck repair-welded joints (B-D-1,2,3,4), two breaks at the weld toe and t at the weld root. This indicates that the fatigue property of the weld toe and weld roo comparable. While toe repair-welded joints (B-DZ-1,2,3,4) are all broken at the weld ro The data in Table 4 show that the angle between the weld and deck (toe angle2) increa from 120° to 155°, reducing the stress concentration at the weld toe, which results in failure positions of toe repair-welded joints all located at the weld root. More fracture p ture of joints are show in Appendix A.

Fracture Characteristics
The macro-fracture characteristics of B-DZ-1 (fracture at weld root) and B-D-1 (fr ture at weld toe) are presented in Figure 15. It can be seen that the HCF fracture surfa are divided into fatigue crack initiation, propagation, and fast fracture zone.
The fatigue crack of B-DZ-1 initiates on the surface of the specimen and propaga along the direction of the thickness of the deck. The color of the crack initiation is da After the fatigue crack propagation, the effective bearing area continues to decrease un the instant break. The fatigue crack of B-D-1 originates at the middle of the toe and verges to both sides. Due to obvious welding defects such as pits and undercuts at

Fracture Characteristics
The macro-fracture characteristics of B-DZ-1 (fracture at weld root) and B-D-1 (fracture at weld toe) are presented in Figure 15. It can be seen that the HCF fracture surfaces are divided into fatigue crack initiation, propagation, and fast fracture zone. ture at weld toe) are presented in Figure 15. It can be seen that the HCF fracture surface are divided into fatigue crack initiation, propagation, and fast fracture zone.
The fatigue crack of B-DZ-1 initiates on the surface of the specimen and propagate along the direction of the thickness of the deck. The color of the crack initiation is dark After the fatigue crack propagation, the effective bearing area continues to decrease unt the instant break. The fatigue crack of B-D-1 originates at the middle of the toe and d verges to both sides. Due to obvious welding defects such as pits and undercuts at th weld toe, it is easy to produce stress concentration and cause fatigue crack initiation.

Fatigue Life Assessment Based on Equivalent Traction Structural Stress
Through calculation of the procedures in Section 2.3, the equivalent traction struc tural stress of the toe surface and root surface of the welded joint is 1.599 MPa and 1.68 MPa, respectively. The equivalent traction structural stress of the toe surface and root su face of the toe repair-welded joint is 1.608 MPa and 1.679 MPa, respectively. The equiva lent traction structural stress under other fatigue loads can be calculated based on thes data, as demonstrated in Table 7. The fatigue crack of B-DZ-1 initiates on the surface of the specimen and propagates along the direction of the thickness of the deck. The color of the crack initiation is dark. After the fatigue crack propagation, the effective bearing area continues to decrease until the instant break. The fatigue crack of B-D-1 originates at the middle of the toe and diverges to both sides. Due to obvious welding defects such as pits and undercuts at the weld toe, it is easy to produce stress concentration and cause fatigue crack initiation.

Fatigue Life Assessment Based on Equivalent Traction Structural Stress
Through calculation of the procedures in Section 2.3, the equivalent traction structural stress of the toe surface and root surface of the welded joint is 1.599 MPa and 1.687 MPa, respectively. The equivalent traction structural stress of the toe surface and root surface of the toe repair-welded joint is 1.608 MPa and 1.679 MPa, respectively. The equivalent traction structural stress under other fatigue loads can be calculated based on these data, as demonstrated in Table 7. In engineering, Equation (8) is used to evaluate the fatigue life of welded structures through equivalent traction structural stress.
The C d and h are constant parameters obtained by fitting a large number of fatigue test data. Table 8 lists the constant parameters of the master S-N curve provided by ASME standard [29].
In the actual welding process, the AM generated by nonuniform temperature field distribution, assembly clamping, and other factors will affect the welding joint's fatigue performance [30]. However, the equivalent traction structural stress calculated above does not consider the influence of AM, so the introduction of a correction formula can further improve the accuracy of fatigue life evaluation of the welded joints. Figure 16 is the schematic diagram of the AM of the welded joint. The welding root is set as the critical position, and t (mm) is the path length of fatigue crack growth, namely, the thickness of the deck plate. The δ (mm) is the distance between the lower edge of the misalignment side and the clamp block. The α ( • ) is the degree of AM of the specimen. The L (mm) is the distance between the test machine blocks. The distance between the critical position and the two blocks is defined as L 1 (mm) and L c (mm), respectively.  The position of the toe repair welding is on the same side of the deck a weld, so the AM is increased relative to the welded joint, see Figure 17. Whil of deck repair welding is on the different sides of the deck from the origina AM is reduced relative to the welded joint. The AM correction coefficient can be calculated by Equation (9)  The position of the toe repair welding is on the same side of the deck as the original weld, so the AM is increased relative to the welded joint, see Figure 17. While the position of deck repair welding is on the different sides of the deck from the original weld, so the AM is reduced relative to the welded joint. The position of the toe repair welding is on the same side of the deck as the origin weld, so the AM is increased relative to the welded joint, see Figure 17. While the positio of deck repair welding is on the different sides of the deck from the original weld, so th AM is reduced relative to the welded joint. The AM correction coefficient can be calculated by Equation (9)  The AM α of each specimen can be obtained by using three-dimensional scannin technology and computer-aided design (CAD). The detailed parameters of the AM size each specimen and the calculation of the correction coefficient are shown in Table 9. The AM correction coefficient k α can be calculated by Equation (9) [31]. Then, the traction structural stress σ s is modified with AM by Equation (10).
The AM α of each specimen can be obtained by using three-dimensional scanning technology and computer-aided design (CAD). The detailed parameters of the AM size of each specimen and the calculation of the correction coefficient k α are shown in Table 9. The fatigue data with and without AM are plotted in the master S-N curve, as shown in Figure 18. The AM means that the AM is considered. It is observed that all fatigue data are within the ±95% confidence interval of the main curve. It can be observed that the data considering AM are more aggregated to the main curve, meaning that the traction structural stress method with AM is applicable to predict the fatigue life of repair-welded joints. In addition, the fatigue properties of deck repair-welded joints are higher than that of the toe repair-welded joints. Considering that the AM of toe repair-welded joints is larger than that of deck repair-welded joints, it can be determined that the larger AM will reduce the fatigue life of the repair-welded joints. The fatigue data with and without AM are plotted in the master S-N curve, as shown in Figure 18. The AM means that the AM is considered. It is observed that all fatigue data are within the ±95% confidence interval of the main curve. It can be observed that the data considering AM are more aggregated to the main curve, meaning that the traction structural stress method with AM is applicable to predict the fatigue life of repair-welded joints. In addition, the fatigue properties of deck repair-welded joints are higher than that of the toe repair-welded joints. Considering that the AM of toe repair-welded joints is larger than that of deck repair-welded joints, it can be determined that the larger AM will reduce the fatigue life of the repair-welded joints.

Conclusions
In this paper, the microstructure and mechanical properties of the S355J2 steel welded and repair-welded joints are studied. The HCF properties of the T-joints are evaluated by testing and FEA. Moreover, the influence of AM on HCF performance is emphatically considered. The main conclusions are as follows: 1. Repair welding increases the grain size of WM and HAZ, which reduces the hardness of WM by about 30 HV. The tensile strength of the welded joint is the same as that of the base metal, while the tensile strength of the repair-welded joint is reduced by Figure 18. The master S-N curve of the welded and repair-welded joints.

Conclusions
In this paper, the microstructure and mechanical properties of the S355J2 steel welded and repair-welded joints are studied. The HCF properties of the T-joints are evaluated by testing and FEA. Moreover, the influence of AM on HCF performance is emphatically considered. The main conclusions are as follows:

1.
Repair welding increases the grain size of WM and HAZ, which reduces the hardness of WM by about 30 HV. The tensile strength of the welded joint is the same as that of the base metal, while the tensile strength of the repair-welded joint is reduced by about 20 MPa. Compared with the base metal, the elongation of the welded and repair-welded joint decreases by about 25%.

2.
The fracture positions of the welded joints are both weld toe and root, and the ratio is even, which is the same for deck repair-welded joints cracking. However, all of toe repair-welded joints fracture at the weld root. The macro-fracture characteristics of the root surface show that the fatigue cracks initiate on the surface of the specimen and propagate along the thickness of the deck, while the cracks of the toe surface initiate at the middle of the weld toe and propagate to both sides. 3.
The fatigue data with and without AM are all within the ±95% confidence interval of the master S-N curve, while the data considering AM is less discrete to the main curve. Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A
For the readability of the article, the fracture pictures of welded and repaired joints corresponding to Table 6 are shown in Figures A1-A3. Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.

Conflicts of Interest:
The authors declare no conflict of interest..

Appendix A
For the readability of the article, the fracture pictures of welded and repaired joi corresponding to Table 6 are shown in Figures A1-A3.