Residual Stress Analysis in Girth-welded Ferritic and Austenitic Steel Pipes Using Neutron and X-Ray Diffraction

This paper is dedicated to the thorough experimental analysis of the residual stresses in the vicinity of tubular welds and the mechanisms involved in their formation. Pipes made of a ferriticpearlitic structural steel and an austenitic stainless steel are each investigated in this study. The pipes feature a similar geometry and are welded with two passes and comparable parameters. Residual strain mappings are carried out using X-ray and neutron diffraction. The combined use of both techniques permits both near-surface and through-wall analyses of the residual stresses. The findings allow for a consistent interpretation of the mechanisms accounting for the formation of the residual stress fields due to the welding process. Since the results are similar for both materials, it can be concluded that residual stresses induced by phase transformations, which can occur in the structural steel, play a minor role in this regard. Introduction Current fatigue design standards and recommendations, like the ones given by the International Institute of Welding (IIW) [1], are based on the assumption of yield strength magnitude tensile residual stresses if the actual residual stress state is unknown. This postulate reflects uncertainties about the initial residual stress state after welding, which may depend on numerous parameters, as well as about the possible relaxation of residual stresses, which occurs when the static or cyclic yield strength is exceeded locally. Therefore, research concerning both the development and the relaxation of residual stresses is needed to improve the generalized approach given in [1]. Experimental and numerical analyses of welding residual stresses hardly ever show that the conservative assumptions made in the IIW recommendations hold. In girth-welded pipes, the pipe geometry and the heat input have been identified as the governing factors for the residual stress development [2,3], apart from material parameters. Under suitable conditions, pipe wall bending can occur, leading to compressive axial residual stresses at the weld toe of girth welds, but also to tensile residual stresses at the weld root, which was shown by other authors, see e.g. [2,3], and in previous studies on ferritic-pearlitic and austenitic steel pipes using X-ray diffraction [4-6]. In this work, these will be supplemented by neutron diffraction measurements and the results obtained from the two different steels will be compared. The thorough experimental analysis of the residual stress state after welding will serve as a basis for the investigation of residual stress relaxation under loading and the validation of numerical simulations, both of which are subject of future work. Experimental work Sample preparation. Pipes of the ferritic-pearlitic structural steel S355J2H+N and of the austenitic stainless steel X6CrNiTi18-10 were used for the experiments. The yield stress of the base Residual Stresses 2016: ICRS-10 Materials Research Forum LLC Materials Research Proceedings 2 (2016) 229-234 doi: http://dx.doi.org/10.21741/9781945291173-39 230 materials is 355 MPa and 223 MPa, respectively. The tubular specimens of 200 mm length were machined on all surfaces, resulting in outer diameters of 100.5 mm and 114 mm and wall thicknesses of 7.75 mm and 7.5 mm for the structural steel and the austenitic steel, respectively. A v-shaped groove was introduced at half-length as a weld preparation. Before welding, the specimens of the structural steel were stress relieved thermally at 600 °C for 30 minutes and cooled uniformly at about 1 °C/min. Metal active gas (MAG) welding was performed in flat position using a rotary table. The filler metal, ISO 14341-A-G 4Si1 for the structural steel and ISO 14343-A-G 19 9 NbSi for the austenitic steel, was applied in two passes, which were started at the same point and were welded in the same direction. The nominal energy inputs were similar for the structural and the austenitic steel, with 8.6 kJ/cm and 9.1 kJ/cm for the root pass and 11.8 kJ/cm and 11.2 kJ/cm for the second pass, respectively. Each pass was welded at room temperature. Residual Stress Analysis. The residual stresses in the welded samples were determined using Xray diffraction (XRD) and neutron diffraction (ND) in order to analyze the stress state both at the surface and within the pipe wall. The measurements were taken at points along lines perpendicular to the welding direction at φ = 90°, where φ is the circumferential angle marking the welding direction and the start/stop location at φ = 0°. As previous work revealed virtually axisymmetric residual stress states [4-6], φ = 90° is considered to be representative. Due to the symmetry, measurements were only performed on one side of the weld centerline up to a distance of 60 mm. The coordinate x specifies the axial distance of a certain point from the weld centerline. The hoop and axial residual stresses on the surfaces of the pipes were determined by XRD. The inner surfaces were only accessible after sectioning the tubes, the released stresses being monitored by strain gauge measurements. For details concerning the XRD measurements, please


Introduction
Current fatigue design standards and recommendations, like the ones given by the International Institute of Welding (IIW) [1], are based on the assumption of yield strength magnitude tensile residual stresses if the actual residual stress state is unknown.This postulate reflects uncertainties about the initial residual stress state after welding, which may depend on numerous parameters, as well as about the possible relaxation of residual stresses, which occurs when the static or cyclic yield strength is exceeded locally.Therefore, research concerning both the development and the relaxation of residual stresses is needed to improve the generalized approach given in [1].
Experimental and numerical analyses of welding residual stresses hardly ever show that the conservative assumptions made in the IIW recommendations hold.In girth-welded pipes, the pipe geometry and the heat input have been identified as the governing factors for the residual stress development [2,3], apart from material parameters.Under suitable conditions, pipe wall bending can occur, leading to compressive axial residual stresses at the weld toe of girth welds, but also to tensile residual stresses at the weld root, which was shown by other authors, see e.g.[2,3], and in previous studies on ferritic-pearlitic and austenitic steel pipes using X-ray diffraction [4][5][6].In this work, these will be supplemented by neutron diffraction measurements and the results obtained from the two different steels will be compared.The thorough experimental analysis of the residual stress state after welding will serve as a basis for the investigation of residual stress relaxation under loading and the validation of numerical simulations, both of which are subject of future work.

Experimental work
Sample preparation.Pipes of the ferritic-pearlitic structural steel S355J2H+N and of the austenitic stainless steel X6CrNiTi18-10 were used for the experiments.The yield stress of the base materials is 355 MPa and 223 MPa, respectively.The tubular specimens of 200 mm length were machined on all surfaces, resulting in outer diameters of 100.5 mm and 114 mm and wall thicknesses of 7.75 mm and 7.5 mm for the structural steel and the austenitic steel, respectively.A v-shaped groove was introduced at half-length as a weld preparation.Before welding, the specimens of the structural steel were stress relieved thermally at 600 °C for 30 minutes and cooled uniformly at about 1 °C/min.Metal active gas (MAG) welding was performed in flat position using a rotary table.The filler metal, ISO 14341-A-G 4Si1 for the structural steel and ISO 14343-A-G 19 9 NbSi for the austenitic steel, was applied in two passes, which were started at the same point and were welded in the same direction.The nominal energy inputs were similar for the structural and the austenitic steel, with 8.6 kJ/cm and 9.1 kJ/cm for the root pass and 11.8 kJ/cm and 11.2 kJ/cm for the second pass, respectively.Each pass was welded at room temperature.
Residual Stress Analysis.The residual stresses in the welded samples were determined using Xray diffraction (XRD) and neutron diffraction (ND) in order to analyze the stress state both at the surface and within the pipe wall.The measurements were taken at points along lines perpendicular to the welding direction at φ = 90°, where φ is the circumferential angle marking the welding direction and the start/stop location at φ = 0°.As previous work revealed virtually axisymmetric residual stress states [4][5][6], φ = 90° is considered to be representative.Due to the symmetry, measurements were only performed on one side of the weld centerline up to a distance of 60 mm.The coordinate x specifies the axial distance of a certain point from the weld centerline.
The hoop and axial residual stresses on the surfaces of the pipes were determined by XRD.The inner surfaces were only accessible after sectioning the tubes, the released stresses being monitored by strain gauge measurements.For details concerning the XRD measurements, please refer to [4][5][6].
The ND measurements for the residual stress analysis within the pipe wall were carried out at the 2nd Generation Neutron Residual Stress Mapping Facility (NRSF2) beam line [7] of the High Flux Isotope Reactor (HFIR), Oak Ridge National Laboratory.For the ferritic pipes, the diffraction of neutrons with a wavelength of λ ≈ 1.72 Å on the {211} lattice planes was analyzed, whereas λ ≈ 1.54 Å was used for the {311} lattice planes of austenite, in order to attain a diffraction angle of 2θ ≈ 90°.The planes were chosen due to their similarity to the expected bulk stress/strain response.In addition, these reflections are reported to be the least-sensitive to intergranular strains [8].The stress-free lattice spacing was determined in the base metal (BM) for both materials as well as in the weld metal (WM) and the heat-affected zone (HAZ) of the ferritic-pearlitic samples to take composition gradients into account.At each axial position, measurements were taken at five points that were evenly distributed across the wall thickness.At every point, the strain was determined in hoop, axial and radial direction, assuming that these are the principal directions, in order to compute the multiaxial residual stress state.The cubical measurement volume had an edge length of 2 mm.

Results
Fig. 1 shows the hoop residual stress on the outer surface for both a ferritic and an austenitic sample as determined by XRD.Except for the weld seam, where tensile residual stresses occur in the ferritic sample of S355J2H+N and compressive residual stresses can be found in the austenitic sample of X6CrNiTi18-10, the results agree well on a qualitative level.Close to the weld seam, a small tensile maximum of about 70 MPa or 200 MPa can be seen, respectively, whereas the residual stresses are in the compressive regime at a distance of 10 to 40 mm from the weld centerline.Here, maximum compressive values of -250 MPa were determined for the ferritic sample and -550 MPa for the austenitic sample.At larger distances, the residual stresses in both samples approach 0 MPa.
The residual stresses on the inner surface after sectioning can be taken from Fig. 2. Again, the distributions are qualitatively similar and mainly differ in the maximum values.In the weld seam, moderate tensile residual stresses were found that are higher in the ferritic sample.At x = -7 mm, a maximum of almost 500 MPa or 800 MPa is reached for the ferritic or austenitic sample, respectively.With increasing distance from the weld, the residual stresses become compressive, reaching maximum values of -200 MPa and -700 MPa, and finally converge to 0 MPa.The residual stress state was affected by the sectioning process.Strain gauge measurements on the inner surface near the weld revealed a decrease of the tensile residual stresses of about 100 MPa for a ferritic pipe and 25 MPa for an austenitic one.A more detailed examination of that effect can be found in [5,6].

Figure 1: Hoop residual stresses on outer surface of ferritic and austenitic samples Figure 2: Hoop residual stresses on inner surface of ferritic and austenitic samples
The results of the neutron diffraction experiments for the hoop stress component are presented as contour plots in Fig. 3 for a ferritic pipe and in Fig. 4 for an austenitic one.Each diagram displays the cross section of the whole pipe wall; the white areas reflect the fact that the center of the fully immersed measurement volume had to be at least 1.5 mm away from the surfaces.The center positions of the measurement volume at each measuring point are also given.In the ferritic sample, the sign of the residual stress mainly depends on the axial distance from the weld.In the weld metal, the heat-affected zone and the adjacent base metal, the residual stresses are tensile for about |x| < 14 mm.For |x| > 14 mm, they are compressive, reaching a maximum at about 25 mm and fading with increasing distance from the weld.Across the pipe wall, the highest absolute values can be found near the inner surface, which holds both in the tensile and the compressive area.In the austenitic sample, a hoop residual stress distribution similar to the one in the ferritic pipe was found, both qualitatively and quantitatively.Distinct differences are the higher tensile residual stresses in the vicinity of the weld, where up to 380 MPa instead of 280 MPa are reached, and the presence of an area of tensile residual stress near the outer surface for |x| > 25 mm.The axial residual stress on the outer surface is given in Fig. 5. Qualitatively, there is a good agreement for both materials within |x| < 30 mm.In the vicinity of the weld, the residual stresses are compressive, taking moderate values of up to -200 MPa in the weld seam and peaking to -400 MPa or -600 MPa near the weld toe, depending on the material.With increasing distance, the residual stresses then reach a relatively small tensile maximum and decrease again, exhibiting a different behavior for the two materials.While the residual stress is almost 0 MPa at x = -60 mm in the ferritic sample, it asymptotically approaches a value of -450 MPa in the austenitic pipe.

Figure 5: Axial residual stresses on outer surface of ferritic and austenitic samples Figure 6: Axial residual stresses on inner surface of ferritic and austenitic samples
The axial residual stress on the inner surface, obtained after sectioning, shows a behavior converse to that on the outer surface, see Fig. 6, featuring small tensile stresses in the weld, a tensile maximum at x = -6…-10 mm, and compressive residual stresses with a maximum at about x = -30 mm.While the latter amounts to -70 MPa in the ferritic sample, it reaches -900 MPa in the austenitic pipe.Moreover, far away from the weld, compressive stresses of -550 MPa are found in the X6CrNiTi18-10 pipe, compared to about 0 MPa in the S355J2H+N sample.Through strain gauge measurements, the residual stress release near the weld due to sectioning was found to be about 125 MPa for the ferritic pipe and 50 MPa for the austenitic sample.

Discussion
Based on XRD results, the governing mechanism responsible for the residual stress formation in the girth-welded pipes investigated here has already been discussed in detail for the ferritic-pearlitic pipes [4,5] and for the austenitic pipes [6].It was shown that in both cases, the circumferential contraction of the weld and the highly heated areas in its vicinity during cooling is constrained by the adjacent material, thus causing tensile hoop residual stress in the former and compressive hoop residual stress in the latter areas, as can be seen in Fig. 1 to 4. Due to the self-constraining tubular geometry, the contraction of the weld also leads to necking of the pipe and thereby to pipe wall bending, which accounts for the axial residual stress state featuring inverse signs on the inner and outer surfaces, see Fig. 5 to 8. In general, the ND results confirm the findings obtained by XRD and thus the interpretation of the formation mechanisms of the residual stress fields.However, there are also slight deviations from the XRD results that will be discussed in the following paragraphs.
Both in hoop and axial directions, the XRD measurements on the inner surface of the ferritic and austenitic pipes show residual stress maxima that are at least 7 mm from the weld centerline, whereas the residual stresses in the weld metal are relatively low.The ND measurements, however, reveal that the highest values occur in the weld metal or at the notch between the weld root and the base metal.This important difference can be ascribed to the sectioning that was necessary to perform the XRD measurements on the inner surface, thereby altering the residual stress state, which was also shown at selected points by strain gauge measurements.It is also possible that the curvature of the weld reinforcement influences the XRD measurements, especially at the weld root.Moreover, the surface of the weld reinforcement is subject to a lower constraint as there is no base material in the immediate vicinity, which might also affect the residual stresses on the surface of the weld seam.The last two points can also be used as an explanation why compressive hoop residual stresses are found in the weld seam on the outer surface of an austenitic pipe, see Fig. 1, whereas tensile residual stresses occur in this region almost across the whole wall thickness, see Fig. 4.
A comparison of the residual stress states in the ferritic and austenitic samples yields a good qualitative agreement.There are, however, significant differences: First, the absolute values of the residual stresses in the austenitic samples are often higher than in the structural steel and by far surpass the yield stress of 223 MPa.This effect is most pronounced in all XRD results, i.e. on the surface.Second, at large distances from the weld, the residual stresses on or near the surfaces of the austenitic steel can approach values significantly different from 0 MPa, which does not occur in the ferritic samples.These stresses can be both tensile, see Fig. 4, and compressive, see Fig. 5 and 6.
Both the higher values and the residual stresses far away from the weld can be explained by the fact that the austenitic pipes, unlike the ferritic samples, have not been heat-treated before welding.Machining leads to significant near-surface plastic deformation, thus causing both a work hardening effect and introducing machining residual stresses.Work hardening is limited to areas 200 -300 µm below the surface [6] and is therefore most distinct in the XRD results, where residual stress values of up to 900 MPa occur.The ND results reveal that the residual stresses in the pipe wall are much lower, see Fig. 4 and 8, which confirms the findings in [6].Since they can still exceed the yield stress to a lesser extent, it can be inferred that work hardening effects not related to the machining process also play a role in the residual stress formation in the austenitic pipes.
Residual stresses resulting from machining, or introduced by the fabrication process in general, extend farther below the surface than the work hardening effect [6].This explains why they can affect the results of ND measurements to a larger extent, as reflected for instance by the relatively high hoop residual stresses near the outer surface at a large distance from the weld, shown in Fig. 4. The fact that tensile stresses can be found here in contrast to the XRD measurements on the surface, see Fig. 1 and 2, can be explained by the considerable variation of the machining residual stresses in different specimens.It was found that they can be tensile, compressive or even negligibly small, presumably depending on the condition of the cutting tool and the machining parameters.
The fact that the results are similar for both materials, apart from machining effects, reveals that phase transformations, which can occur in the ferritic-pearlitic steel but not in the austenitic steel, do not contribute substantially to the formation of the residual stress field.This is due to the relatively long cooling time from 800 °C to 500 °C in the HAZ of about 18 seconds [5], leading to high transformation temperatures and thus to a dominant effect of thermal contraction.

Summary and Conclusions
In this study, the residual stress state in girth-welded pipes of ferritic-pearlitic and austenitic steel has been thoroughly investigated using both X-ray and neutron diffraction.It was shown that the results of neutron diffraction measurements mostly agree well with the X-ray measurements on the surface.Nevertheless, they also contain complementary information, especially regarding the residual stresses near the inner surface of the pipes, which is only accessible for X-ray measurements after sectioning and thus altering the residual stress state.Also, neutron diffraction confirmed that the residual stresses on the surface of the austenitic pipes are significantly affected by the machining process during sample preparation, resulting in very high residual stress values.
From the good qualitative agreement of the residual stress states in pipes of similar geometry but different types of steels it can be inferred that phase transformations play only a minor role in the formation of the residual stress fields in the girth-welded pipes investigated here.

Figure 3 :Figure 4 :
Figure 3: Hoop residual stress distribution in the pipe wall, structural steel S355J2H+N

Figure 7 :Figure 8 :
Figure 7: Axial residual stress distribution in the pipe wall, structural steel S355J2H+N