1 Introduction

The demand for high-strength fine-grained structural steel in modern steel constructions increased in recent years in a growing number of applications in civil engineering and mobile crane manufacturing [1]. This led to a development of numerous high-strength base and filler materials by the steel producers with yield strength from 960 MPa [2]. The economical application of these steels is determined by high safety requirements of the welds among achieving adequate mechanical properties. More challenges in welding are also a result of closer technical boundaries and a higher elastic ratio. In particular, high tensile residual stresses in the welds are able to diminish a component’s safety [3]. At highly restrained weld conditions, such as repair welds or in thick-walled support structures, residual stresses may reach crack critical levels [4] (see Fig. 1). Codes and standards regarding manufacturing and design of welded components assume residual stresses on the amount of the yield strength [6, 7]. The bases of these assumptions are studies, which had been conducted predominantly at free shrinking laboratory samples from mild steels. Component relevant restraints and heat conduction conditions as well as the phase transformation behaviour of high-strength structural steels had rather not been considered. As a rule, the consequence is a highly conservative design of welded constructions. Hence, the economic benefit using high-strength steels is limited. However, recent studies revealed that residual stresses in those welds are lower than the yield strength in general [8,9,10,11].

Fig. 1
figure 1

Examples of typical weld seams with high restraints in mobile crane fabrication [5]

Full size image

The residual stress state of a welded component is subjected to a complex interaction of the applied material, the restraint condition based on the design and the process conditions as well as the heat control [12]. Using an adequate heat control during welding, a minimisation of the local restraint stresses in the weld seam and of the global reaction forces and stresses due to shoring effects from an external restraint is attainable [8, 9, 13, 14]. Within the scope of an AiF-research project (IGF.-Nr. 17978 N), various kinds of experimental load analyses were conducted [5, 15]. It was aspired to examine the welding process regarding the produced welding loads and stresses. For this purpose, the transfer of restraint conditions of real welded structures to welding experiments in bench-scale models is mandatory [16, 17]. Therefore, massive slot specimens of high-strength steel S960QL were used (see Fig. 2).

Fig. 2
figure 2

Experimental determination of restraint intensity of the slot specimen: a test setup (1 stereo camera system, 2 DIC-system (Aramis), 3 hydraulic pump, 4 spreader, 5 specimen); b displacement in the analysed area due to loading during restraint intensity determination as per [5]

Full size image

2 Experimental

2.1 Restraint condition

These slot specimens as compact self-restraining structures exhibit realistic conditions in terms of heat conduction and shoring effects. To compare restraint conditions of different weld constructions, the restraint intensity concept was established by Satoh [18]. The restraint intensity transverse to the weld RFy is the spring rigidity of the surrounding structure towards the weld seam based on the seam length LS. The specimen was experimentally and numerically analysed to determine the restraint intensity transverse to the weld direction.

For the experimental analyses, the slot of the weld seam was expanded using a hydraulic spreader. A defined load of Fy = 115 kN was applied at three various positions of the slot (see Fig. 2a). In the course of this, a digital image correlation (DIC) system was used to measure the displacement in the surrounding area of the slot (see Fig. 2b).

The numerical determination of the restraint intensity was performed by means of FE calculation. Therefore, a solid model of the slot specimen was created and meshed. In the FE calculation, a distributed load of p = 100 N mm−2 was applied on the fusion face transverse to the weld seam and the displacement was calculated (see Fig. 3a). This analysis was conducted for a square weld according to experimental determination and for a V-type weld (see Fig. 3b).

Fig. 3
figure 3

Numerical determination of restraint intensity of the slot specimen: a displacement in the analysed area due to loading; b analysed seam geometries as per [5]

Full size image

Both methods the experimental and the numerical analysis revealed qualitatively and quantitatively similar strain fields in the vicinity of the weld with increasing restraint at both ends of the weld seam (compare Figs. 2b and 3). Hence, the results of restraint intensity determination are in very good accordance (see Table 1). Moreover, these values are in the scope of typical values in high-strength steel fabrication according to [19].

Table 1 Comparison of restraint intensity determination
Full size table

2.2 Welding experiments

The slot specimens were used to perform multilayer GMA weld tests in comparison to welding experiment at free shrinking specimens (see Figs. 4 and 5). Plates of high-strength quenched and tempered fine-grained structural steel S960QL (EN 10025-6 [20]) with a thickness of 20 mm were used. Welding was performed with a similar high-strength solid wire according to ISO 16834-A [21]. The chemical compositions and mechanical properties are shown in Table 2. In both test setups, the groove angle was varied. For the narrow seam geometry, a modified spray arc welding process was applied in all weld runs [14]. The 45° V-type seam geometry was welded with a conventional GMAW process. The welding parameters are given in Table 3. The measured cooling times ∆t8/5 are between 6 and 8 s, which is within the recommended working range for this material. Hence, the mechanical-technological properties tested in the free shrinking specimens were in accordance to the standard requirements.

Fig. 4
figure 4

In situ analyses of the temperatures and stress-strain fields adjacent to the weld, while preheating, welding and cooling, local residual stress analyses via XRD [5]

Full size image
Fig. 5
figure 5

Geometry of the free shrinking and slot specimen, test setup of the weld tests [5]

Full size image
Table 2 Chemical compositions (FES, Fe balanced) and mechanical properties of the test materials
Full size table
Table 3 Welding parameters and variation of seam geometry and restraint condition
Full size table

Displacements and strains which occur in the vicinity of the restrained weld seam due to local heat input and while cooling were detected with the DIC system (see Fig. 5). These measurements were performed at certain time and temperature steps while welding and cooling of the slot specimens. A stochastic greyscale value pattern was applied at one side of the weld vicinity (see Figs. 4 and 5). The stereo camera optic of the ARAMIS system allowed the analysis of the spatial displacement of each point on the stochastic pattern.

Temperature measurements were conducted using thermocouples (type-K) attached 10 mm adjacent to the weld. The working temperature was measured on the weld seam. Δt8/5-cooling times were determined at the weld surface using a two-colour pyrometer (measuring range, 350 to 1300 °C). With a thermographic camera, the temperature field was observed.

Local residual stresses were analysed by means of X-ray diffraction (XRD) using the sin2ψ method as in [9,10,11] with 5 to 7 ψ steps between 0° and ± 45° (10 to 20 s each step; CrKα, power, 30 kV at 6.7 mA). The measurements were taken out with a 2-mm colimator along the specimen centre line (x = 0 mm) on the surface of the base material, HAZ and weld metal. The measuring point density was from 1 to 10 mm, depending on the residual stress gradient. The analysis was primarily focussed on transverse residual stresses to investigate the superposition of local and global stresses and since restraint was varied in transverse direction (see Fig. 4).

In recent studies at BAM, also welding experiments with a unique 2-MN testing facility were performed [5, 8, 9, 12,13,14, 22]. Details of this testing facility and weld tests are given in [14]. In the 2-MN testing facility, specimens are hydraulically clamped into a three-dimensional operating test frame. The test frame consists of a hydraulic cylinder system with three piston rods. Hence, global lateral and angular distortions are restrained and the reaction forces, moments and stress could be measured synchronously while preheating, welding and cooling of the specimen. After cooling, at room temperature analyses of the superimposing local restraint stresses and the global stresses on the top surface of the welded specimen by means of XRD directly within the testing facility are enabled. This allowed a comparison with the results of the aforementioned analysis using free shrinking and slot specimens and a further discussion.

3 Results and discussion

3.1 Global strains and stresses

In Fig. 6, contour plots of the global displacement adjacent to the weld in transverse direction −∆y(x,y) for the whole welding procedure of weld test no. 3 are shown. While local preheating of the weld seam, the thermal expansion causes a major displacement off the seam. After root welding, shrinkage of the weld produces a local movement of the weld vicinity towards the weld seam. With each weld sequence, the displacement off the seam in farer regions of the weld increases. In the weld vicinity, shrinkage and displacement towards the weld seam intensify. Subsequent cooling after cap weld pass causes a shifting of the complete measured pattern towards the weld due to the transverse shrinkage, which reaches a maximum at room temperature. The greatest displacement of approx. −∆y = 0.22 mm is revealed in the area near the specimen centre line (x = 0) adjacent to the weld seam. Towards the ends of the weld seam, displacement decreases due to comparable higher restraints in these areas (see Figs. 2 and 3).

Fig. 6
figure 6

Contour plots of transverse displacement −∆y(x,y) at certain phases of welding procedure (test no. 3) [5]

Full size image

−∆ym(t) and εy,m(t) graphs as mean values, measured with the DIC system 10 mm adjacent to the weld flank (y = 20 mm), and temperature T(t) are presented in Fig. 7 for test no. 3. In accordance with Fig. 6, preheating, cooling to interpass and to room temperature cause transverse displacements towards the weld seam. Shrinkage is interrupted by phases, in which the heat input of the next weld run leads to a transient thermal expansion. Therefore, strain increases while heat input occurs and decreases while weld cools to interpass temperature. The maximum strain of εy,m,max = 0.14% is observable after cap bead welding. Due to shrinkage while cooling, the strain decreases continuously and reaches a remaining strain at the weld vicinity (y = 20 mm) of about εy,m,end = 0.04% at room temperature.

Fig. 7
figure 7

−∆ym(t)− and εy,m(t) graphs and temperature T(t) measured in test no. 3; mean values measured at y = 20 mm adjacent to the weld seam [22]

Full size image

By comparison of average strains measured next to weld εy,m(t) of the conventional weld (test no. 3, compare Table 3) and the modified weld with a narrow weld groove (weld test no. 4), two qualitatively equal graphs are exposed (see Fig. 8). Quantitative differences reveal for the total welding time. The lower amount of weld runs for the narrow groove weld leads to shorter welding time. Although the strain level while welding is more or less equal for both processes, the lower total heat input and volume of inserted weld metal for the modified weld lead to significant lower strains εy,m,end remaining after cooling to ambient temperature.

Fig. 8
figure 8

εy,m(t) graphs for two weld tests with different weld geometries (weld test nos. 3 and 4) [22]

Full size image

This is the consequence of the higher number of weld runs needed for the conventional weld (conv. weld: 7 passes, mod. weld: 4 passes, compare Table 3) and a higher eccentricity of the weld in connection with a higher amount of weld runs which had to be inserted asymmetrically into the weld seam. All this causes an increased thermal expansion of about ∆εy = 0.02% at the top of the specimen, which cannot be compensated during temperature equalisation.

Hence, the global strain field εy(x,y) after subsequent cooling to ambient temperature of the modified weld (Fig. 9b) shows significantly reduced values compared to the conventional weld (Fig. 9a), especially in the area around the weld centre line (x = 0 mm). In the areas at the weld seam ends (x = ± (100 to 125) mm), higher negative strains occur in the modified weld. In accordance with the strain fields, in Fig. 9c, d, the stress fields of both welds are shown. The stresses were calculated with the DIC system based on the strain measurements, the plate thickness and elastic constants (Young’s modulus EM = 215 GPa, Poisson’s ratio ν = 0.3). The weld seam ends of the narrow weld reveal higher transverse compressive stresses. Particularly in the areas near the weld (y = (20 to 50) mm), the tensile stresses due to welding are approx. 100 MPa higher in the conventional welds compared to those in the narrow groove welds due to the higher restrained transverse shrinkage (compare Fig. 8).

Fig. 9
figure 9

εy,m(t) graphs (a and b) and ∆y(x,y) graphs (c and d) for two weld tests with various weld geometries (weld test nos. 3 and 4) after cooling to room temperature [22]

Full size image

3.2 Local stress analysis

The XRD measurements transverse to the weld direction along the centre line of the specimens (x = 0 mm, compare Fig. 9c and d) are shown in Fig. 10 as residual stress distributions σrsy(y) of all four weld tests according to Table 3. Both XRD (see Fig. 10) and DIC analyses (Fig. 9) revealed comparable results regarding welding stress level in compliance with the restraint intensity determinations [5].

Fig. 10
figure 10

σrsy(y) distributions for a conventional welds (test nos. 1 and 3) and b modified welds (test nos. 2 and 4), welded and measured under different restraint conditions [5]

Full size image

All graphs exhibit qualitative equal, typical residual stress profiles for steel welds with an undergoing phase transformation according to common concepts [11]. The microstructure of the welds as well as the influence of solid state phase transformation while cooling on the local and global welding stresses and loads is widely discussed in [5, 9,10,11, 15]. Along with material properties, heat control and cooling conditions while welding may have a strong effect on the residual stress evolution. In this study, the same materials and constant heat control parameters were used to mask these effects.

By comparison of the residual stresses in Fig. 10, a broader stress profile reveals for the wider weld groove (α = 45°). Maximum residual stresses are typically on the weld metal surface. The weld interface is affected by high residual stress gradients. Secondary maxima of the residual stresses occur in the HAZ. In the free shrinking specimens, which were welded and measured without restraint, the stresses decrease to the level of the initial state towards the unaffected base material. Note that the base material surface was blast cleaned and reveals residual stresses of about −160 MPa. Hence, a superposition of the reaction stresses and the local restraint stresses in the weld area is revealed for the specimen welded and measured under restraint. In the base material, the stress level is elevated by approx. 50 to 150 MPa in the restrained slot specimens. According to the DIC analyses, the reaction stresses in the conventional weld are significantly higher, which is evident in the areas nearby the weld seam flank or HAZ, respectively. At the weld metal of the modified welds, higher tensile residual stresses of the amount of 60% of Rp0.2 of the weld metal are obviously almost independent of the restraint intensity, which is due to a deeper weld penetration [5, 14]. The lowest residual stresses in weld metal can be found in the conventional weld of the free shrinking specimen (weld test no. 1). However, the maximum tensile residual stresses in the conventional welds (σrsy,max) grow with increasing restraint intensity up to 70% of Rp0.2.

In general, especially if occurring in the HAZ, high residual stresses are detrimental since a critical microstructure and, therefore, significantly degraded mechanical properties may be present. From the DIC and XRD measurements, it could be stated that the narrow weld groove in context with the selected process conditions (welding process, heat control) has a positive effect on the loads and stresses especially in the vicinity of the weld. For a safe welding manufacturing and for a minimal risk of premature failure, such findings should be taken into account particularly with regard to severe restraint conditions.

3.3 Comparison with recent studies

However, residual stress analyses of the slot specimens (see Fig. 10) revealed some divergent results compared to earlier studies that were carried out with special testing facilities [5, 8, 9, 12,13,14, 22]. Hence, in Fig. 11, residual stress distributions of conventional welds with the same materials and welding parameters as weld test nos. 1 and 3 (see Table 3) are presented. These welds were conducted at the 2-MN testing facility under two various restraint conditions [5, 14]. The restraint conditions were varied by means of different weld seam lengths (Lw = 100 and 200 mm).

Fig. 11
figure 11

σrsy(y) distributions for conventional welds in the 2-MN testing facility for two various restraint conditions [5, 14]

Full size image

In comparison with Fig. 10a, considerably lower residual stresses are revealed at the surface of the weld metal. This is due to the lower restraint intensity transverse to the weld direction, which is 30 and 40% of the slot specimen, respectively. In contrast, the residual stress levels in the HAZ of both specimens welded in the testing facility are significantly higher. According to [14, 22, 23], these high residual stress levels in the HAZ are due to increased bending moments and, therefore, bending stresses occurring in the weld seams of the specimens that were welded conventionally using a larger groove angle of α = 45°. Since the bending stresses are a result of a restrained angular distortion, the higher residual stresses in the HAZ should be attributed to a higher bending restraint intensity [18].

Hence, Fig. 12 shows the determination of the bending restraint intensity RMx of the different test setups of the 2-MN testing facility and the slot welds. The comparably longer lever arms of the testing facility involve a bending restraint that revealed to be more than eight times higher than that in the slot weld. The reduction of the weld length from LS = 200 to 100 mm of the specimen in the 2-MN testing facility would additionally double the bending restraint to RMx = 16 kNm (°·mm)−1 whereas the restraint intensity transverse to the weld increases merely from RFy = 3 to 4 kN (mm mm)−1.

Fig. 12
figure 12

Analytical determination of bending restraint intensities for the 2-MN testing facility and the slot weld test setup (LS weld length, bS weld width, ß angular distortion, C spring rigidity, H plate thickness, Wx section modulus) [5]

Full size image

Finally, Fig. 13a shows that the transverse residual stress level in the weld metal under equal material and welding parameters is a linear function of the restraint intensity transverse to the weld seam. In contrast, the occurrence of bending stresses due to hindered angular distortion induces a linear correlation between the residual stress levels in the HAZ and the bending restraint intensity (see Fig. 13b). Thus, it should be noted that especially high bending restraints might lead to crack critical tensile stresses in the vicinity of high-strength steel welds if inadequate welding or heat control parameters are applied.

Fig. 13
figure 13

a Residual stress level in the weld metal σrsy,wm/Rp0.2 of the conventional welds vs. restraint intensity transverse to the weld RFy, σ. b Residual stress level in the HAZ ∆σrsy,HAZ/Rp0.2 of the conventional welds vs. bending restraint intensities RMx [5]

Full size image

4 Conclusions

Derived from real high-strength structural steel components in the mobile crane industry, a specimen geometry was selected to undertake weld tests under defined restraint conditions. The restraint intensity of the test setup was analysed and multilayer GMA welds were performed with two different weld geometries and welding parameter sets at these component-related specimens and at free shrinking specimens from S960QL. With in situ DIC, temperature and XRD measurements, the resulting welding stresses were analysed and compared with results from recent studies. From this work, the following conclusions can be draw:

  1. 1.

    The numerical and experimental analysed restraint intensities of the slot specimens are in very good accordance and in the scope of typical high-strength steel weldments.

  2. 2.

    In the self-restrained slot specimens using a DIC system, in situ measurements of the induced strain were carried out and the strain and stress fields were analysed. As shown in recent analyses, the weld seam with the smaller groove angle causes significantly decreased strains and stresses in the vicinity of the weld.

  3. 3.

    A local analysis of stress superposition using XRD in the weld seam area revealed a high effect of the restraint condition on the transverse residual stresses, especially in the weld metal and the HAZ of the welds, which were prepared with a larger groove angle. Using same material and welding conditions, the residual stress level in the weld metal correlates with the transverse shrinkage restraint.

  4. 4.

    Results of recent studies of welds in a unique testing machine of the BAM [5, 14, 22] revealed that local transverse residual stresses in the HAZ are higher in the specimens despite a lower restraint intensity in the transverse direction. An analytical estimation of bending restraint intensities in each test setup showed values up to 16 times higher in the 2-MN testing facility. Hence, it could be clarified that welding induced bending stresses increase with the bending restraint and cause higher local residual stresses in the HAZ. This was shown by a correlation of the stress level in the HAZ and the bending restraint intensity.

  5. 5.

    Slender seam geometries and adequate process conditions may lead to considerably lower bending stresses and tensile residual stresses in the HAZ, despite severe restraint conditions. This is important to consider while welding high-strength steel structures, since detrimental microstructures and degraded mechanical properties may be evident in the HAZ, involving a higher susceptibility for premature failures if tensile stresses occur.

These results should help in deriving optimal heat control parameters for certain weld tasks in high-strength steel fabrication in order to reduce welding loads, which beneficially affects the economic application and exploitation of the potential of high-strength structural steel.