Effect of Post-Weld Grinding on the Fatigue Strength of Thin-Walled RHS High-Strength Steel T-Joints Under Different Load Stress Ratios

Abstract

In this work, the focus is laid on the mean stress effect on the fatigue strength of thin-walled rectangular hollow section T-joints made of high-strength steel S960 M x-treme. The specimens are cyclically tested at a stress ratio of R = −1 and R = 0.1 in both as-welded and ground (weld-profiled) conditions. In the context of nominal stress evaluations, the ground specimens demonstrate a distinct advantage in contrast to the as-welded condition, exhibiting an increase of +33% at R = 0.1 and +16% at R = −1. Based on the experimental results, a corresponding Haigh diagram is evaluated, revealing a notable difference in the mean stress sensitivity, with M₁ = 0.58 for the as-welded condition and M₁ = 0.39 for the ground condition. Finally, mean stress factors are presented, enabling feasible application in the fatigue design of welded and post-treated structures. The resulting factors are compared with values from the literature for steel applications, showing an increased mean stress influence using high-strength steel as the base material.

1. Introduction

Cyclically loaded structures commonly experience different types of stress at specific stress ratios. When assessing fatigue behaviour, stresses can be described in terms of compressive, alternating and tensile stresses [1]. The analyses in this paper are based on standardised formulas used in standards like the IIW recommendations [2] and the DVS1612 [3]. In Figure 1, the relationship between the mean stress σ_m and the stress range Δσ = 2σ_a is outlined. Further, the stress ratio R can be calculated in accordance with Equation (1).

The present work focuses on an application-oriented and practical–realistic T-joint-shaped tube sample that is made of high-strength fine-grained structural steel S960, which is tested under specific cyclic loads. Due to increasing requirements, whether in terms of stress or weight reduction (in regard, for instance, to sustainability reasons), high-strength steels are becoming more and more important. This kind of structural detail is applied, for instance, in the fabrication chassis frames and cabin constructions for rail vehicle structures. Therefore, high-strength steel materials are analysed in more detail below.

$$R={\displaystyle \frac{{\sigma }_{min}}{{\sigma }_{max}}}={\displaystyle \frac{{\sigma }_m-{\sigma }_a }{{\sigma }_m+{\sigma }_a}}$$

(1)

The motivation consists of evaluating the differences in the effect of mean stress between an untreated as-welded condition (AW) and a post-treated ground state using weld-profiling with the aid of an angle grinder (GR). Nevertheless, there are several studies that have already focused on the mean stress effect [4,5,6]. This study aims to make a significant contribution to this knowledge, as the analysis of the mean stress effect has so far been based largely on the use of data at 2 × 10⁶ cycles.

It is necessary to use post-treatment processes to exploit the potential of high-strength material in fatigue strength. In fact, there is a variety of treatments, such as high-frequency mechanical impact (HFMI) [7,8,9], shot peening [10,11,12] or TIG dressing [13,14,15]. However, based on previous studies [16,17] and simplified handling in the workshop, with regard to the use of realistic T-shaped tubes, only grinding [18,19,20] is considered here.

In this study, thin-walled rectangular hollow section T-joints with a stress ratio of R = −1 are tested for their fatigue strength. In order to base the comparisons on a broader data set, the test data from an earlier study is used, in which the tests and evaluations were carried out at a stress ratio of R = 0.1. To ensure comparability, the sample geometry, the test setup, and the limited cycle number are identical to those of the preliminary study [21]. These fatigue tests are evaluated in a frequency range of 5–24 Hz.

On the one hand, the aim of this paper is to investigate the scope of application of fatigue behaviour and thereby contribute to existing standards such as the IIW recommendation [2], the FKM guideline [22] and the DVS1612 [3]. The use of higher-strength materials should significantly increase the expected service life compared to the more conservative recommended values in these guidelines. On the other hand, the mean stress effect of the tested specimens, especially from AW to GR specimens, is to be analysed and evaluated. Therefore, a Haigh diagram is evaluated based on the practice-oriented experimental results and the corresponding mean stress sensitivity factors for both conditions, which are calculated using the fatigue strength amplitude σ_{aR = −1} at R = −1 and σ_{aR = 0} at R = 0, as well as the mean stress σ_{mR = 0} at R = 0.

The main scientific contributions of this paper can be summarised as follows:

• Experimental investigation of the mean stress effect for thin-walled rectangular hollow section T-joints as representative specimens for high-strength steel S960 weld structures.
• Development of Haigh diagrams for both as-welded and ground conditions and evaluation of the corresponding mean stress sensitivity values.
• Suggestion of engineering-feasible mean stress factors for each condition, ensuring practicable consideration of the mean stress effect in design.

Actually, this article starts with the evaluation of materials and components, proceeding to a fracture surface analysis. The following chapter Fatigue Test Results discusses the influence of grinding and shows nominal stress fatigue curves with the influence of mean stress. The curves are evaluated separately for each treatment in order to obtain a detailed overview. In the final section on outcomes, the data is presented in a Haigh diagram, and the mean stress effect on the realistic-oriented T-joints is evaluated. Finally, the findings are summarised in the Conclusion.

2. Experimental Investigations

2.1. Material and Specimen Geometry

The welded material examined is high-strength S960 fine-grained structural steel. The shaped tube has the main dimensions of 60 mm × 60 mm × 3 mm, with a length of 360 mm. The test procedure, including the strain gauge application, is described in detail in an earlier paper [21]. The chemical compositions of the RHS tube are listed in

Table 1. Chemical composition of alform S960 M x-treme tube in wt% according to [23].

C	Si	Mn	P	S	Al	Cr	Ni	Mo	Cu	Ti	V	Nb	Zr	N	B
0.086	0.115	1.62	0.007	0.0009	0.055	0.92	0.480	0.221	0.019	0.024	0.12	0.002	0.002	0.0057	0.0021

. The material exhibits a nominal yield strength of f_y = 975 MPa and a maximum ultimate tensile strength of f_u = 1035 MPa [23].

The test specimen is a single-sided through-welded HV3 robotic seam [24], where the weld layer looks like a backing layer. There are two different welding positions on the specimen. The robotic welding system is a metal active gas (MAG) welding process, which is used with the shielding gas type 273 Corgon^® 18 [25]. On the longitudinal side is a PC layer and on the transversal side a PB layer [26]. The chemical composition of the filler material is listed in

Table 2. Chemical composition of the consumable 960-IG in wt% according to [27].

C	Si	Mn	Cr	Mo	Ni
0.12	0.80	1.90	0.450	0.55	2.350

. The samples are tested for fatigue strength in an untreated state, AW, and a post-treated state, GR. The seam surface was ground by applying weld-profiling using an angle grinder with an A40 grit. Based on preliminary investigations [16,17], this post-treatment method has become particularly established in applications such as forklifts or driving cabin frames in rail vehicle engineering and is therefore investigated further in this paper. The finished welded specimen is presented together with its main dimensions (in mm) in Figure 2. In order to gain more profound insights into the intricacies of robotic welding processes, the welding parameters are illustrated in

Table 3. Robotic MAG welding parameters.

Welding Position in (-)	Wire Diameter in mm	Current Range in A	Voltage Range in V	Wire Feed Speed in m/min	Welding Speed in cm/min	Preheated
PC	1.2	100–160	20–30	5–10	30–45	no
PB	1.2	140–190	20–30	6–12	25–40	no

.

2.2. Experimental Setup

The fatigue tests on the RHS specimens are performed at stress ratios of R = 0.1 and R = −1, whereby details of the tests under R = 0.1 are given in a previous study [21]. The fatigue test termination criterion is either macroscopic crack initiation, resulting in a reduction in the structural stiffness, or run-out, defined as 5 × 10⁶ cycles. A total of 80 specimens is examined in this study, with 20 specimens each being analysed for the stress ratio in both AW and GR. The failure criterion for the fatigue tests is set at a cylinder stroke difference of +/− 0.25 mm. According to the available data, the fatigue test is stopped as soon as a higher value is recorded.

The fatigue test setup is described in Figure 3, where the most important main components are shown. The RHS specimen is screwed firmly on the bottom to the testing machine. A horizontal force acts on the upper section, which is connected to the hydraulic cylinder by a connecting rod. The testing load acts purely horizontally on the upper end of the RHS specimen, which is indicated as “force application”.

3. Investigation of Fracture Surfaces

After testing, the failed specimens exhibiting macroscopic crack initiation are further fully ruptured to enable a fracture surface characterisation. Figure 4 and Figure 5 present an AW and GR specimen, respectively. There is no significant difference in the fracture behaviour. Both specimens demonstrate a fatigue fracture due to the cyclic load at the outer edges, defined here with the index “A”. The areas marked “B” in the illustrations show the subsequent fractures that were generated. In essence, the only discernible distinction is that the initiation of cracking at the outer edges occurs substantially later in the GR specimens. Hence, the specimens cracked where the fracture is expected to occur. This is therefore a consequence of the reduction in notch stress caused by the grinding process after welding. It is essential that the notches are smoothed and the weld seam transitions are ground as notch-free as possible. The finding suggests that the GR specimens demonstrate the capacity to withstand extended load cycles under equivalent loads, thereby exhibiting a level of durability comparable to that of the AW specimens when subjected to higher loads [28,29]. The fracture behaviour exhibited by the specimens is comparable to the simulation results. A large number of research papers has been published that demonstrates analogous fracture behaviour in this area [30,31,32,33].

4. Fatigue Test Results

4.1. Influence of Grinding

The S/N curves are evaluated based on the pearl string method in accordance with the DIN 50100 standard and calculated using Equation (2) to determine the permissible fatigue strength amplitude σ_a for the service lifetime [34]. The stress range, designated as Δσ, is set at 2 × 10⁶ for N, representing the number of load cycles. The definition of both Δσ_i and N_i is shown to be equivalent at a given point, for instance, in the fatigue life curves at 30,000 load cycles. This serves as a starting point for the S/N diagrams.

$$∆\sigma =∆{\sigma }_i{ \left({\displaystyle \frac{N_i}{N}}\right)}^{{\displaystyle \frac{1}{m}}}$$

(2)

As shown in Figure 6 and Figure 7, the data is represented using nominal normal stresses evaluated by analytical computation and is calculated with a survival probability of P_s = 97.5%. It is imperative to note that the calculation of the permissible stress is based exclusively on specimens that have exhibited a maximum of 2 × 10⁶ load cycles. It is evident that all other specimens that have undergone 5 × 10⁶ load cycles or more are to be classified as run outs. This classification is explicitly demonstrated in the following diagrams. The slopes after the knee point are defined for all samples as m = 22, according to the IIW recommendations [2]. As in the previous study, the knee point can be determined numerically using an equation-matching procedure based on Equation (2), the slopes, and the data from the fatigue strength curves. The knee point at R = −1 is calculated using Equation (2). By comparing the two slopes, the curves can be equated, and the required load amplitude and the number of cycles can be determined at the point where the two curves cross. The calculation is constrained by the boundary conditions of N_{_start} = 3 × 10⁵ load cycles at the initiation point and N_{_end} = 5 × 10⁶ load cycles at the endpoint. For the AW specimens, which have slopes m = 4.1 and m = 22, a knee point is determined for Δσ__{AW_kneept_R-1} = 116 MPa at N_{_AW_kneept_R-1} = 3.01 × 10⁶ load cycles. For the GR specimens, using slopes m = 3.5 and m = 22, a knee point is calculated at Δσ_{_GR_kneept_R-1} = 121 MPa and N_{_GR_kneept_R-1} = 4.18 × 10⁶ load cycles. Figure 6 shows a Δσ_{_AW_R0.1} = 78 MPa at a stress ratio of R = 0.1 and 2 × 10⁶ load cycles, with a slope of m = 22 in the AW cycles. The ground specimens show a Δσ_{_GR_R0.1} = 104 MPa with a slope of m = 22. Compared to the fatigue strength, the specimens exhibit Δσ_{_AW_R0.1} = 108 MPa with m = 4 at 5 × 10⁵ load cycles, as well as Δσ_{_GR_R0.1} = 125 MPa with a slope of m = 3.8. In general, therefore, an increase of +33% can be achieved in the GR condition compared to the AW. This gathered data can also be found in

Table 4. Influence of post-treatment at N = 2 × 10⁶ for R = 0.1 and R = −1.

Stress Ratio R in (-)	Number of Load Cycles N in (-)	Δσ for AW in MPa	Δσ for GR in MPa	Benefit GR/AW in %
R = 0.1	2 × 106	78	104	+33
R = −1	2 × 106	128	150	+16

.

The nominal fatigue curves shown in Figure 7 are performed and evaluated with a stress ratio of R = −1; apart from that, the other test parameters are identical to those of the R = 0.1 specimens. It is evident from the analysis of the AW specimens that they demonstrate a Δσ_{_AW_R-1} = 128 MPa in conjunction with a slope of m = 4.1. The data of the GR condition, designated as Δσ_{_GR_R-1} = 150 MPa, has been determined with a slope of m = 3.5. This corresponds to an increase of +16% from the AW to GR condition at N = 2 × 10⁶ load cycles. The fatigue tests results from an earlier study [21] are illustrated in

Table A1. Results from the fatigue tests with R = 0.1.

Specimen ID	Condition	Stress Ratio R in (-)	Testing Frequency in (Hz)	Load-Cycles Ni in (-)	Nominal Stress Δσ in (MPa)
10	AW	0.1	5	69,881	191
9	AW	0.1	7	125,394	167
78	AW	0.1	15	38,152	215
56	AW	0.1	15	373,191	119
60	AW	0.1	15	1,319,807	95
61	AW	0.1	15	5,000,000	72
62	AW	0.1	15	33,275	238
63	AW	0.1	15	164,790	143
64	AW	0.1	15	774,986	107
65	AW	0.1	15	276,327	131
69	AW	0.1	15	131,321	155
67	AW	0.1	15	94,143	179
79	AW	0.1	17.5	1,595,113	83
77	AW	0.1	20	5,000,000	77
80	AW	0.1	24	5,000,000	83
81	AW	0.1	24	5,000,000	89
47	AW	0.1	24	1,810,795	95
48	AW	0.1	24	5,000,000	89
52	AW	0.1	24	1,112,695	95
8	GR	0.1	7	162,786	226
6	GR	0.1	7.5	457,768	198
1	GR	0.1	7	84,507	226
20	GR	0.1	15	169,778	169
14	GR	0.1	15	907,124	141
25	GR	0.1	20	1,051,872	113
23	GR	0.1	20	95,729	212
26	GR	0.1	20	230,918	183
15	GR	0.1	24	283,684	155
45	GR	0.1	24	1,268,253	127
27	GR	0.1	24	5,000,000	99
40	GR	0.1	24	5,000,000	106
35	GR	0.1	24	5,000,000	113
37	GR	0.1	24	5,000,000	120
39	GR	0.1	24	1,301,614	127
34	GR	0.1	24	486,681	141
31	GR	0.1	24	1,321,460	120
42	GR	0.1	24	307,537	148
30	GR	0.1	24	193,464	176
18	GR	0.1	24	149,480	205

in the Appendix A.

For both S/N curves in Figure 6 and Figure 7, the scatters are listed in the upper right corner. The slopes are almost identical in the AW, which suggests a similar dispersion of TN_{_AW_R0.1} = 1.7 and TN_{_AW_R-1} = 1.8. This phenomenon is attributable to a highly consistent robotic welding process. The deviations for the GR condition of the two scatter plots are greater at TN_{_GR_R0.1} = 2.5 and TN_{_GR_R-1} = 1.7. This can be attributed to the manual grinding process. Due to the dexterity and skill required to handle the angle disc grinder, there are greater differences in fatigue strength, especially at higher load cycles. As demonstrated in Figure 12 based of another study [35], differences between series due to factors such as a tested frequency range from 5 to 24 Hz and the applied load have no negative influence on the test results. Influences such as the residual stresses have not been considered in this study. As demonstrated in

Table A2. Results from the fatigue tests with R = −1.

Specimen ID	Condition	Stress Ratio R in (-)	Testing Frequency in (Hz)	Load-Cycles Ni in (-)	Nominal Stress Δσ in (MPa)
7	AW	−1	7.5	920,059	191
2	AW	−1	7	354,175	215
59	AW	−1	7	265,144	212
68	AW	−1	7	211,354	239
74	AW	−1	10	1,142,725	159
58	AW	−1	5	159,663	266
55	AW	−1	12	5,000,000	133
54	AW	−1	17.4	1,676,754	146
72	AW	−1	16.4	892,569	173
73	AW	−1	16.4	408,326	199
75	AW	−1	16.4	2,950,451	139
76	AW	−1	16.4	208,131	226
70	AW	−1	16.4	154,627	252
57	AW	−1	16.4	550,371	186
71	AW	−1	16.4	1,413,393	153
66	AW	−1	16.4	2,108,936	139
50	AW	−1	16.4	2,491,612	133
49	AW	−1	16.4	5,000,000	120
51	AW	−1	18	5,000,000	126
53	AW	−1	18	2,330,815	133
5	GR	−1	7.5	480,584	227
4	GR	−1	7	424,960	255
3	GR	−1	7	351,937	283
17	GR	−1	18	1,152,812	189
13	GR	−1	16	2,026,833	157
32	GR	−1	16	5,000,000	126
46	GR	−1	16	267,701	267
16	GR	−1	16	475,231	236
29	GR	−1	16	1,089,388	204
21	GR	−1	16	1,309,309	173
19	GR	−1	16	4,005,717	142
24	GR	−1	16	5,000,000	134
12	GR	−1	16	3,925,322	142
22	GR	−1	12	2,852,268	149
36	GR	−1	12	1,168,657	181
28	GR	−1	16	719,686	212
41	GR	−1	8	317,761	244
44	GR	−1	16	301,466	280
11	GR	−1	8	4,631,213	134
38	GR	−1	12	5,000,000	126

, which is cited in the Appendix A, the outcomes of the above-mentioned fatigue tests are illustrated.

4.2. Influence of Mean Stress

To illustrate the obtained data more clearly in terms of the influence of the mean stress state, all results are presented in the AW in Figure 8 and in the GR condition in Figure 9. In these both diagrams, regardless of the post-treatment, the S/N curves are reduced from the stress ratio R = −1 to R = 0.1. There are further research papers that present a similar nominal stress range for the determined stress ratios [36,37].

Table 5. Influence of mean stress condition at N = 2 × 10⁶ for AW and GR condition.

Condition	Number of Load Cycles N in (-)	Δσ_R = −1 in MPa	Δσ_R = 0.1 in MPa	Reduction from R = −1 to R = 0.1 in %
AW	2 × 106	128	78	−39
GR	2 × 106	150	104	−31

shows the influences of mean stress conditions at 2 × 10⁶ load cycles. The differences between R = 0.1 and R = −1 are presented and evaluated in both treatments. Taking an even deeper look at the results, the reduction is lower, from 39% in the AW to 31% in the GR samples.

5. Evaluation of Haigh Diagram

Based on the FKM guideline [22], a linear decrease in the fatigue strength amplitude σ_a between the value of σ_{aR = −1} at R = −1 and σ_{aR = 0} at R = 0 by the mean stress sensitivity M is defined according to Equation (3). In the region between σ_{aR = 0} at R = 0 and σ_{aR = 0.5} at R = 0.5, a shallower decrease by M/3 is defined. The available data of the FKM interpolations allows general data for welded standard steel grades to be obtained. These equations are derived from the experimental data points in the Haigh diagram using a straight-line equation. A detailed derivation of these equations is described in the following literature [38]. As the value of σ_{aR = 0.1} at R = 0.1 is evaluated based on the conducted fatigue tests, Equation (4) is applied to evaluate the Haigh diagram in this range. Based on the statistically evaluated values σ_{aR = −1} and σ_{aR = 0.1} of the tests carried out, the values σ_{aR = 0} and σ_{aR = 0.5} are calculated using the given equations. The resulting Haigh diagrams for the AW and the GR condition are depicted in Figure 10 and the corresponding values are summarised in

Table 6. Fatigue strength amplitudes and corresponding mean stress values at 2 × 10⁶ load cycles and Ps = 97.5%, as the basis for the Haigh diagram for both conditions.

Stress Ratio R in (-)	Model Interpretation in (-)	Stress Amplitude σa_2 × 106_AW in MPa	Stress Amplitude σa_2 × 106_GR in MPa	Mean Stress σm_AW in MPa	Mean Stress σm_GR in MPa
−1	test	64.24	74.75	0	0
0	model	40.65	53.35	40.65	53.35
0.1	test	39.09	52.13	47.78	63.71
0.5	model	30.70	43.17	92.10	129.50

.

$$M_1={\displaystyle \frac{{\sigma }_{aR=-1}-{\sigma }_{aR=0}}{{\sigma }_{mR=0}}}={\displaystyle \frac{{\sigma }_{aR=-1}}{{\sigma }_{aR=0}}}-1$$

(3)

$$M_2={\displaystyle \frac{M_1}{3}}={\displaystyle \frac{{\sigma }_{aR=0}-{\sigma }_{aR=0.1}}{{\sigma }_{mR=0.1}-{\sigma }_{mR=0}}}\Rightarrow {\displaystyle \frac{{\sigma }_{aR=0}-{\sigma }_{aR=0.1}}{{\sigma }_{mR=0.1}-{\sigma }_{aR=0}}}$$

(4)

The Haigh diagram shows that the AW specimens have a notable steeper decline of the slope than the ones in the GR condition. This indicates that the GR specimens have lower mean stress sensitivity, which suggests the influence of the post-weld treatment process. The allowable stress amplitude decreases if the means stress increases. This, in turn, has a negative impact on the fatigue strength of the components subjected to cyclic loading in both conditions.

Based on the evaluated Haigh diagram and the underlying fatigue strength amplitude values, the mean stress sensitivity M₁, applicable between R = −1 and R = 0, and M₂, applicable between R = 0 and R = 0.5 and equalling M₁/3, is defined and given for both conditions in

Table 7. Mean stress sensitivity for P_S = 97.5%.

Condition	M1 in (-)	M2 in (-)
AW	0.580	0.193
GR	0.401	0.134

. The presented values show that due to the post-treatment process, the mean stress sensitivity decreased from M₁ = 0.580 for the AW down to M₁ = 0.401 for the GR condition. A comparison of these values with the data from other studies in the literature [36,39,40] related to steel grades lower than those investigated in this study is warranted. Such a comparison would reveal an increase in the mean stress sensitivity when high-strength steel is used as the base material.

In order to more feasibly represent the effect of the mean stress state, the corresponding mean stress factors are evaluated, as seen in Figure 11. Due to the use of a higher-strength material, the data reveals increased factors compared to a previous study [39], generally valid for steel applications.

Table 8. Mean stress factors of different steels and treatments.

Mean Stress Factor	AW Steel [39] in (-)	AW S960 in (-)		GR Condition S960 in (-)
-	PS = 97.7%	PS = 97.5%	PS = 50%	PS = 97.5%	PS = 50%
f1	1.20	1.58	1.61	1.40	1.32
f2	1.10	1.32	1.34	1.24	1.19
f3	1.32	2.09	2.16	1.73	1.57

shows that the mean stress factor f_{3_Steel} = 1.32 (P_S = 97.7%) given in [39] is lower compared to the results of the AW S960 material, which is evaluated as f_{3_S960_AW} = 2.09 (P_S = 97.5%) based on the results of this study, and can be mainly explained due to the increased mean stress sensitivity of high-strength materials corresponding to [1,5]. Further, Table 8 demonstrates that the factor of the GR samples f_{3_S960_GR} = 1.73 (P_S = 97.5%) is reduced compared to the AW, revealing a decreased mean stress influence if applying grinding as a post-treatment process. The resulting mean stress factors of this study are additionally evaluated for a mean probability of survival of P_S = 50%. Comparing the results of S960 at P_S = 97.5% and P_S = 50%, it can be observed that the factors are quite similar in the case of the AW. Related to the GR condition, a slight decrease in the mean stress factors at P_S = 50% is evaluated, which can be explained by the influence of different scattering in the fatigue tests results at R = −1 and R = 0.1. However, as is also shown in the case of P_S = 50%, the GR condition leads to a decreased mean stress effect in comparison to the AW.

The stated mean stress sensitivities M₁ and M₂, as well as the mean stress factors f₁, f₂ and f₃, are evaluated based on the test results of this study. These values do not represent a general recommendation, and additional test results, especially for high-strength steel welded joints, are necessary to establish statistically consolidated correction factors for design applications.

6. Conclusions

In this paper, the fatigue strength of thin-walled S960 M high-strength steel rectangular hollow section T-joints is analysed considering as-welded and ground (by weld profiling) conditions. Evaluations for the complete creation of the Haigh diagram are derived from the measurement results of the fatigue test, with additional results calculated numerically using the mean stress influence factor equations. The cyclic tests are conducted at load stress ratios of R = −1 and R = 0.1, leading to the following conclusions of this study:

Fatigue cracks appeared on both sample series at the outer corners. It was only evident that the GR specimens withstood higher loads due to the post-weld treatment process.

The nominal stress results show a benefit of the GR condition compared to the AW by +33% at R = 0.1 and by +16% at R = −1. Therefore, the great potential of the post-treatment process is observed.

Based on the evaluated Haigh diagram, a mean stress sensitivity M₁ applicable between R = −1 and R = 0 is evaluated as 0.580 for the AW and 0.401 for GR condition. The results show that due to the post-treatment process, the mean stress effect is reduced.

Finally, corresponding mean stress factors are evaluated, which represent the decrease in the fatigue strength at higher load stress ratios compared to R = −1. The results are compared to values for AW steel applications with comparable joint geometries, which were the result of an earlier study [39]. It reveals an increase in the mean stress effect using a higher-strength steel. However, due to the post-weld treatment, the factors again show a decrease in the mean stress effect for the higher-strength steel as the base material.

The reduction in mean stress sensitivity after grinding is probably the result of two factors. Firstly, the notch effect of the weld profile is reduced. Secondly, tensile residual stresses were the result of grinding. More in-depth research on that topic could be interesting.