Fatigue strength of shot-peened as-welded joints and post-weld-treated and subsequently clean-blasted fillet weld joints

Surface treatment methods, such as shot peening and clean blasting, can improve the fatigue strength of welded joints. The aim of the current work is to experimentally evaluate the effects of clean blasting on the fatigue performance of post-weld-treated joints made of normal, high-strength, and ultra-high-strength steels, as well as obtain the fatigue strength improvement gained by the shot peening processes in welded connections made of different steel grades using literature data. A literature review was carried out to extract fatigue test data of welded joints, considering both clean blasting and shot peening processes. This data was statistically re-evaluated and used for a comparison to the results obtained in the current experimental work which focuses on evaluating the improvement gained by the clean blasting after post-weld treatments. Experimental fatigue tests were carried out on non-load-carrying transverse attachment joints prepared with gas metal arc welding, and made of S355, S700, and S1100 structural steels. TIG dressing and HFMI treatment post-weld treatments were implemented and, subsequently, the specimens were blast-cleaned using two different abrasives: corundum and sand. The re-analysis of existing fatigue test data indicated higher improvement by shot peening, i.e., average improvements of 1.71 and 1.52 in the fatigue strength for butt weld and fillet weld joints, respectively, than by clean blasting process for which an average improvement of 1.19 in fatigue strength was obtained. The enhancement factors, however, highly varied among different data sets indicating a clear impact of processing parameters on the improvement level. The statistical re-analyses considering all data sets of shot-peened specimens (butt weld and fillet weld joints) showed that one fatigue (FAT) class higher fatigue strength could be recommended for shot-peened joints compared to the as-welded condition with weld toe failures. The experimental work on the post-weld-treated joints indicated that the fatigue strength of clean-blasted joints was similar to that of non-blasted joints, and thus showing no major advantages or disadvantages by the clean blasting post-weld-treated joints with corundum and sand.


Introduction
In the green and digital transition, both academia and industry are looking for efficient ways to enhance the performance of steel structures, enabling a production of more environmentally friendly and resource efficient steel structures. In this context, new emerging technologies for producing new fossil-free steels are put into action [1]. Simultaneously, compared to the conventional manufacturing processes, the electricity-intensive and hydrogen-based steel production also evokes needs to minimize steel material production volume. The utilization of high-strength and ultrahigh-strength steel (HSS/UHSS) materials is an attractive means to reduce weight and thus enable lightweight design and efficient material usage. As the principal aim in the utilization of HSSs and UHSSs is to increase stress levels, considering both static and cyclic loads, the fatigue strength of weldments becomes a substantial design criterion as an increase in the material strength does not provide additional 1 3 benefit to the fatigue performance of normal workshop quality joints in the as-welded (AW) condition. To overcome the issues related to the fatigue performance, an introduction of post-weld treatments is of paramount importance in welded structures made of HSS and UHSS grades. It has been even identified that particularly with the high-frequency mechanical impact (HFMI) treatment, higher benefit can be claimed with HSSs and UHSSs than mild steels [2] as the higher compressive residual stresses can be obtained with higher material strengths [3].
In the past (particularly in the 1980s with many experimental studies), shot peening methods have also been well-recognized as efficient techniques to improve fatigue strength of weldments through the reduction of high tensile residual stresses caused by the welding. The mechanism of shot peening, changing residual stresses, is based on the plastic strain at the surface due to the particle indentation, followed by the relaxation of elastic stresses resulting in compressive residual stresses [4] -similar to other mechanical residual stress modification techniques [5]. An overview on the published literature of shot peening, together with other residual stress-related PWT techniques was undertaken by Farajian et al. [6] within the Working Group 6 (Residual Stress Effects in Fatigue) activities. Shot peening has also been successfully introduced to enhance and retrofit the fatigue strength of existing steel bridges [7,8].
Blast cleaning, which is usually regarded as a surface treatment prior to corrosion protections, is a shot peeninglike process that can also enhance the fatigue strength of weldments [9]. From the engineering viewpoint, this is also an important aspect since many structures undergo clean blasting before painting and other surface treatments. However, far too little attention has been paid to the fatigue performance of combined or subsequent post-weld treatments. To account for the combined effect of geometrical improvement and residual stress modification in fatigue assessments, the use of 4R method was introduced for the ground and mechanically peened welded joints in [10]. In addition, the fatigue properties of clean-blasted or shot-peened HSS and UHSS grades, due to their recent development, have been less investigated. In a recent work, Hensel et al. [11] evaluated the fatigue strength of butt weld joints made of S355N and S960QL grades under uniaxial constant amplitude (CA) loading with the applied stress ratio of R = 0.1. In their work, a shallower slope parameter of S-N curve (m) was obtained for normal strength steel but, however, the fatigue strength of the S960QL specimens was still much higher than that of the S355N specimens. Werner et al. [12] investigated TIG dressing and clean blasting with steel shots (diam. 0.3-0.6 mm) for improving CPW800 and S960QL UHSS grades, respectively, and obtained nominal characteristic fatigue strength up to 400 MPa (m = 12.5) at two million cycles for the shot-peened S960QL butt welds. These studies thus indicate that clean blasting and shot peening processes are viable options to enhance the fatigue strength of UHSS weldments. Section 2 presents a literature review and reanalysis of published fatigue test data on the clean-blasted and shot-peened joints.
Subsequently, the aim of the current work is to evaluate the fatigue strength of HSS and UHSS fillet weld joints post-weld-treated with the HFMI treatment and TIG dressing after the clean blasting process. S700 HSS and S1100 UHSS grades were selected as steel materials under investigation in the experiments. Experimental fatigue tests with the supporting measurements, i.e., surface roughness, residual stress and geometry characterization, are carried out on fillet-welded non-load-carrying transverse (NLCT) joints. The reference data for the clean-blasted joints is extracted for similar joints and material grades tested in the previous tests by the authors and supplemented with additional tests in this work.

Description of the extracted data and statistical analysis
A literature review was carried on the experimental studies to extract fatigue test data to evaluate the improvement gained by the shot peening and clean blasting methods.
Referring to the topic of this work, only the experimental studies, including shot peening or clean blasting as PWT techniques, were considered in the data extraction. For a reference, joints in the AW condition and other PWT methods were also considered to evaluate improvement compared to the joints in the AW condition and, on the other hand, to compare the shot peening and clean blasting method to other PWT techniques. The aim of the re-analysis was to obtain knowledge about the improvement level gained by the shot peening and clean blasting processes and compare these results to the experiments conducted for the post-weldtreated joints (section 3). Each extracted data set was statistically evaluated to determine the mean fatigue strength Δσ c,50% , corresponding to the survival probability of P s = 50% with both fixed (m = 3) and free slope parameter. For obvious reasons, and as also shown in Fig. 2, PWTs provide an increase in the slope parameter and S-N curve represents shallower slopes than in the as-welded condition. However, to determine an enhancement in fatigue strength obtained by PWTs, the S-N curves for post-weld-treated joints were also determined using the slope parameter of the corresponding joints in the AW condition. The statistical evaluations were conducted based on the standard statistical approach [13]: where N f is the cycles to failure (dependent, unknown variable), Δσ nom is the nominal stress range (independent, known variable), and C is the fatigue capacity. It is also worth mentioning that the nominal stresses were either graphically obtained from S-N curves or from the tabulated stress-life values, without considering potential misalignment factors. However, it can be assumed that there are no major changes in the misalignments due to the shot peening or clean blasting since they only modify the surface conditions. As shown in Table 1 and Fig. 1, the majority of the extracted prior investigations on the shot peening and clean blasting have focused on either butt-welded joints or fillet weld joints (failing from the weld toe). Consequently, for these two joint types treated with shot peening, statistical analyses were carried out considering different data sets in the same analysis. For clean blasting processes, only few data sets were found from literature and, consequently, joint type-specific S-N curves were not obtained, and enhancement factors were only obtained for individual data sets (see section 2.2 and Fig. 3).

S-N curves and improvement by clean blasting and shot peening
The extracted data was categorized as butt-welded details (Fig. 2a) and fillet-welded details (Fig. 2b) and mean (P s = 50%) and characteristic (P s = 97.7%) fatigue strengths were obtained with both free and fixed (m = 3) slope parameters of S-N curve. As it can be seen from Table 2, the fatigue data was highly scattered in the shot-peened specimens, and thus, the enhancement factors gained by shot peening are different depending on whether the mean fatigue strength or characteristic fatigue strengths are used in the evaluation. In terms of the characteristic fatigue strengths, enhancement of around 20% in fatigue strength was obtained while for the mean fatigue strength, the enhancement factors were up to 1.6. Table 2 summarizes the obtained fatigue strengths and corresponding slope parameters shown in Fig. 2. Figure 3 presents the enhancement factors, f SP and f CB , obtained by the improvement in fatigue strength by shot peening and clean blasting, respectively, compared to the joints in the AW condition in different data sets. To obtain a fair comparison in this evaluation, an identical slope parameter (m free,AW for joints in the AW condition) was used for both joints in the AW and post-weld-treated conditions. These factors were obtained based on the free slope parameters obtained for joints in the AW condition although typically a change in the slope parameter due to the PWTs occurs. The average values for the enhancement factors, considering all data sets, were f SP = 1.72 and f SP = 1.51 for butt-welded and fillet-welded data sets, respectively (see (1) log N f = −mlogΔ nom + log C, Fig. 2a). For the clean-blasted joints, the available data was less representative, particularly in butt welds, but for the fillet-welded joints, an average of f CB = 1.19 was obtained. When using free slope parameters for the PWT data sets, relative slope parameters (m PWT /m AW ) were >1.25, corresponding to a change from m = 3 to m = 4. However, it is worth mentioning that in many data sets, the number of specimens was limited in the data sets and/or the tests were carried out on almost identical stress level (see Table 1) which hinders a comprehensive evaluation of the changes in the slope parameters.

Materials
Three different steel grades were chosen for the study to examine the effect of clean blasting on the fatigue strength of post-weld-treated welded joints. A thermo-mechanically manufactured normal strength steel S355MC (Domex® 355MC), HSS S700MC (Strenx® 700MC Plus), and UHSS S1100MC (Strenx® 1100 Plus) with the plate thickness of t = 8 mm. Of these, the S1100 steel is quenched and tempered (QT) grade. The specimens (see section 3.2) were gas metal arc-welded using strength-matching filler solid wires, and ESAB OK Autrod 12.51, ESAB OK Aristorod 69, and Böhler Union X96 (all with a wire diameter of ⌀1.0 mm) were used as welding consumables for the S355, S700, and S1100 grades, respectively. For the S1100 grade, the X96 wire is strength undermatching by the nominal values but it has been experimentally verified to provide equal strength with the S1100 base metal [24]. Tables 3 and 4 present the mechanical properties and chemical compositions of the studied materials.

Specimen geometries and welding preparation
The NLCT specimens were manufactured from the sheet metals (t = 8 mm) by laser cutting, and the welding preparation was carried out using a robotized GMAW with the fillet weld size of a = 4 mm at the horizontal-vertical (PB) welding position with a torch angle of 45° and travel angle of 0-5° pushing. The applied welding parameters are presented in Table 5. The weld run-on and run-off parts were machined and ground to flush after welding and post-weld treatments.
The shape and dimensions of the test specimens, as well as the welding preparation work, are shown in Fig. 4. A total number of 34 specimens (22 specimens in total from both S355 and S1100 steel grades, and 12 specimens from the S700 steel grade, see also Appendix 1) were manufactured and fatigue-tested. Table 1 Extracted data sets for the fatigue analysis. t is the plate thickness, f y and f u are the yield/proof strength and ultimate tensile strength of material, respectively, n is the number of tests in the series, R is the applied stress ratio, and Acr. is the acronym referring to the first author of publication in extracted data sets (see also Fig

Post-weld treatments
As the main objective of the study was to examine the fatigue strength of HSS and UHSS grades with usually rather thin plate thicknesses (t = 8 mm in this case), TIG dressing and HFMI treatment were selected as PWT techniques. The PWTs were only applied in the fatigue-critical weld toes next to the base plate, as shown in the macrographs in Fig. 5. TIG dressing was manually prepared in the laboratory condition with the following parameters: U = 13 V, I = 180 A,  Table 2). The scatter bands are for the survival probabilities between P s = 10% and P s = 90% v travel = 2.4-3.4 mm/s. The HFMI treatment was manually applied using a pneumatic HFMI device, namely a commercial Dynatec HiFIT "Basic" device [25], with the pin diameter of 4 mm. The quality of the HFMI-treated areas was visually inspected, and a depth of the indentation around 0.15-0.2 mm was achieved, as recommended in [2].

Clean blasting process
Followed by the PWTs, clean blasting was conducted on the specimens. Two different abrasives were chosen for this study, sand and corundum, as they both are usually readily accessible in the workshops. Clean blasting with corundum was conducted in a blasting cabinet in the laboratory conditions (at LUT University) with a particle size of 0.4-0.6    mm and line pressure of 6.0-6.5 bars, while the grit (sand) blasting was performed by a collaborative industrial member (outside LUT University) using a centrifugal wheel blasting device and the treatment was performed with Sa 3 grade [26], i.e., blast cleaning to visually clean steel. In some specimens, the clean blasting was not performed to supplement the reference data (joints in the post-weld-treated condition without blasting), see details in Appendix A. The visual surface conditions are presented in Fig. 6. Visual inspections reveal that the grit blasting with sand resulted in a rougher surface condition compared to the clean blasting with corundum. Section 3.3.1 quantitatively evaluates these differences in the surface conditions.

Surface roughness
Surface roughness measurements were carried out using a Keyence VR-3200 3D-profilometer by measuring the postweld-treated joints before and after the clean blasting. Two different magnifications were used: a high-magnification Weld run-on and run-off parts machined and ground to flush after welding Mat. S355, S700, or S1100 Visual surface conditions in the S1100 specimens of the study: a a joint in the as-welded condition, b TIG-dressed, and c HFMI-treated joints before clean blasting (and with a strain gage for fatigue testing), and TIG-dressed joints clean-blasted with d sand and e corundum to valley height R z , and averaged ten maximum peak to valley height R z,10 over the measurement length (corresponding to the Japanese Industrial Standard JIS B 0601-2001 [27]). Figures 8 and 9 present the measured surface roughness at the plate surface and at the post-weld-treated weld toes in the as-received (hot-rolled) and blast-cleaned conditions.

Residual stresses
Residual stresses were measured before and after the clean blasting using an X-ray diffractometer (XRD), namely the Stresstech X3000 G3 device (collimator diameter of 1 mm). The applied XRD parameters are presented in  Fig. 9 Measured R a , R z , and R z,10 surface roughness at the treated areas: a HFMI-treated joints, b TIG-dressed joint Table 6. The XRD measurements were conducted at the treated area along the plate surface at the center line of the specimen, and the measured stresses were parallel to the loading direction. In the study, only the transverse residual stress measurements were conducted due to the expected fatigue failures at the weld toes, transverse to the loading direction. Consequently, it can be justified that the transverse residual stress conditions have a higher impact on the fatigue performance of the studied joint type. The residual stresses were determined with the material constants of the modulus of elasticity E = 210 GPa and Poisson's ratio v = 0.3. Figure 10 presents the residual stress measurement results for the S700 and S1100 specimens in the HFMI-treated and TIG-dressed conditions, and before and after the clean blasting treatment (including both blasting with corundum and sand).

Fatigue testing
The fatigue tests were carried out using two different fatigue testing machines: 750 kN servo hydraulic (MOOG, test frequency f = 1-2 Hz) and 700 kN full-resonance (RUMUL, f ≈ 80 Hz) fatigue testing machine under sinusoidal uniaxial CA tension loading with the applied stress ratio of R = 0.1. The servo-hydraulic testing device was used in the tests resulting in short fatigue life around 100 thousand cycles, and high-frequency RUMUL device for testing at the intermediate life (more than 100 thousand cycles). The testing equipment and force transducers were validated by the strain gage measurements before starting the fatigue testing. In the test conducted in the servo hydraulic machine, the total rupture of test specimen was the failure criterion. In the full-resonance machine (due to the different types of testing system), the frequency drop was monitored, and test was stopped before a total rupture of specimen. However, in these specimens, the fatigue cracks were substantially large and, thus, the results obtained using the different fatigue testing equipment are comparable. Each specimen was equipped with a strain gage, positioned at the 0.4t distance from the weld toe (see Fig. 6c), to monitor the secondary bending stresses due to the angular distortion caused by the welding.

Fatigue test results
All fatigue test specimens failed from the treated weld toe area with exceptions given in Appendix 1 for individual test specimens. The angular misalignments were significantly lower in the HFMI-treated joints (k m ≈ 1.25) than in the TIG-dressed joints (k m ≈ 1.55), as also found in [28]. This is caused by the TIG dressing with additional heat pass deposited to the asymmetric T-joint geometry. To enable a comparison in the nominal stress system, an effective misalignment factor k m,eff was considered in the analysis:

Fig. 10
Residual stress measurement results: a S700 specimens and b S1100 specimens before and after the blast cleaning processing. The distance (horizaontal axis) refers to the deepest of point of the dent in the post-weld-treated area Mat. S700 Mat. S700 Mat. S1100 Mat. S1100 After PWT (before cleaning) Blast-cleaned with corundum Blast-cleaned with sand (a) ( b) and the corrected nominal stress: where k m, already covered assigns the misalignment factor covered by the S-N curve in the nominal stress system. For the studied NLCT joint type, k m, already covered = 1.25 is recommended in [13]. Table 8 (Appendix 1) provides the corrected stress ranges for each test result. Due to the limited number of specimens, there is uncertainty related to the characteristic fatigue strengths and the characteristic value k for determining characteristic fatigue strength was obtained as follows [13]: (2) k m,ef f = k m k m, already covered , (3) Δ nom,corr. = k m,eff Δ nom , In the case of clean-blasted joints and joints without blasting merged into the same scatter band -the data points were included in the same statistical analysis (see Table 7). For the geometrically improved joints, a slope parameter of m = 4 has been found to be suitable [29][30][31], and for the HFMI-treated joints, m = 5 is recommended [2,32]. Consequently, these were applied in the statistical analyses. As a further note, according to the obtained results, no major or distinct difference in the fatigue strength between the studied clean blasting processes (corundum and sand) was found, and thus, these results were not differentiated in the S-N plots. Figure 11a and b present the fatigue test results for the S700 and S1100 specimens, respectively. Of these, the S1100 specimens were tested first, and based on those findings (no difference between the clean-blasted and non-blasted HFMI specimens), TIG dressing was chosen for the reference case in the S700 specimens (Fig. 11a). The results of the Mat. S1100  Fig. 11 Fatigue test results for the a S700 specimens and b S1100 specimens. The standardized S-N curves (solid lines) extracted for the NLCT joint type from [2,13], and the reference data for the S1100 specimens (AW, HFMI, and TIG without clean blasting) from [28] S355 specimens are presented in Fig. 12. Table 7 presents the obtained fatigue strengths shown in Figs. 11 and 12.
The reference data, shown in Fig. 11, includes test results conducted for similar specimens to the ones tested in this study (plate thickness and weld size) and thus provides comparable results.

Discussion
In the current work, the fatigue strength of the NLCT joints post-weld-treated with TIG dressing and HFMI, and subsequently processed with clean blasting, was experimentally studied. To support the experimental work, a literature review was carried out to obtain improvement gained by clean blasting and shot peening that can be regarded as rather similar processing technique but specifically intended for fatigue strength improvement. The statistical analyses revealed (see section 2.2) that the fatigue strength improved by factors of 1.59 and 1.56 (mean fatigue strengths), and factors of 1.18 and 1.23 (characteristic fatigue strengths) in the shot-peened butt-welded and fillet-welded joints ( Fig. 2 and Table 2), respectively. Statistical analyses were also carried out for individual data sets (post-weld-treated and AW conditions), and enhancement factors were obtained for each data set. Using such an approach, an average improvement factors 1.72 and 1.51 were determined for shot-peened butt welds and fillet weld joints. An overview of the extracted data sets, however, a high variation in the fatigue strength improvement between the different data sets (Fig. 3). The applied techniques and abrasive materials and their particle sizes in the extracted data sets varied, which potentially explains the high scatter of shot-peened data. The extracted data sets also included studies in which all processing parameters were not comprehensively reported. In addition, it should be noted that a change in the slope parameter is usually present due to the shot peening process (compared to the AW data), and potential for the fatigue strength improvement at the high-cycle regime is increased due to an increased crack initiation period. However, as the extracted data was highly scattered, particularly in the case of shot-peened data, comparisons and evaluation between different data sets were complicated. The best practice for both design methodologies and practical implementation of shot peening techniques similar to the HFMI treatment [2], TIG dressing [30] and grinding methods [29], is required to have a robust and trustworthy post-weld treatment technique that can be applied beyond the academia. This study contributed to know the improvement gained the shot peening, and by these results, one or maximum of two FAT class improvement could be suggested compared to the AW condition, i.e., FAT100 for butt welds and FAT90 for transverse fillet weld joints. Based on the conducted literature review, only few studies have been specifically focused on the clean blasting, albeit a differentiation between the clean blasting (as a surface condition treatment) and shot peening processes (as a fatigue strength improvement technique) can be regarded as an open question is some extend. However, those few studies (see Fig. 3a) focusing on the clean blasting showed that the fatigue strength improvement (an average of 1.19) is lower than that of gained by shot peening. The clean blasting treatment processes the metal surface in similar way to the shot peening but most likely the different (softer) abrasive materials (usually grit or corundum) do not necessarily majorly change the residual stress in the in-depth direction and thus are not that efficient to improve the fatigue strength. Due to these reasons, further analysis on the effects of abrasive materials and processing techniques should be carried out. The extracted data, considering both data sets of shot-peened and clean-blasted specimens, did not include high variety of steel grades with different yield strength. Consequently, a comprehensive analysis on the material strength effects cannot be conducted but by a visual observation (Fig. 3a), a tendency for having a higher improvement level for high-strength steel can be observed, similar to the HFMI-treated joints [2]. In addition, the higher slope parameters were obtained for high-strength steels (Fig. 3b). The experimental work on NLCT specimens made of S355, S700, and S1100 steel grades revealed that the clean blasting, as intended, increases the surface roughness at surfaces (Fig. 9). Meanwhile, from the fatigue viewpoint, compressive residual stresses are introduced at the surface that expectedly compensate the detrimental effects of higher surface roughness on the fatigue strength capacity. The measurements showed low or compressive residual stresses in the joints in the TIG-dressed condition before blasting, which is related to the formation of residual stresses in small-scale specimens (Fig. 10). Consequently, it would be interesting to extend these findings to large-scale specimens in which high tensile residual stresses are likely present before clean blasting process. After the clean blasting processing, compressive residual stresses were found in all joints. The study applied two different abrasive materials, i.e., sand (grit blasting) and corundum, and it was found that grit blasting caused higher surface roughness than corundum, while lower compressive residual stresses were measured in the specimens clean-blasted with corundum compared to the grit-blasted specimens.
By observing the fatigue test results (Figs. 11 and 12), it can be found that the fatigue strength of the cleanblasted specimens expectedly exceeded the recommended S-N curves for post-weld-treated joints and, consequently, clean blasting processes can be trustworthy applied for structures without any concerns related to the fatigue performance of welded joints in the post-weld-treated conditions. However, the results suggested that the additional improvement gained by the clean blasting is not always achieved: in the clean-blasted TIG-dressed S1100 specimens, a slight improvement was found compared to the reference data without the clean blasting treatment (Fig. 11b) but, on the other hand, in similar joints made of the S700 steel grade, the fatigue test results fitted to the same scatter band (Fig. 11a). In the S355 specimens, albeit the very limited number of non-blasted specimens were tested, an indication of improvement could be observed ( Fig. 12) but still further studies would be needed to confirm these findings. From the engineering and workshop point-of-view, these results suggest that in the case of clean blasting process is applied for a welded steel structure, there is no major advantages or disadvantages to perform clean blasting before or after PWTs. Considering different PWT techniques, these results confirm that with the low stress ratios (R < 0.1), welded HSS and UHSS grades benefit most from the HFMI treatment, while the geometrical improvement techniques, such as TIG dressing, are the most efficient for mild steels [2,33].

Conclusions
In the present work, the fatigue strength improvement of welded joints, considering butt-welded and filletwelded components, by means of the shot peening and clean blasting was evaluated using the available literature data on experimental fatigue test results. Subsequently, experimental work was carried out on TIG-dressed and HFMI-treated S355, S700, and S1100 fillet-welded specimens by fatigue testing joints in the post-weld-treated conditions followed by the clean blasting process. Based on the literature data evaluation and experimental study, the following conclusions can be drawn: • Evaluation of literature data revealed that the average improvement in the fatigue strength (compared to the AW condition) by clean blasting was 1.19 which is clearly lower than that of obtained for shot-peened data sets. • The fatigue strength of clean-blasted post-weldtreated joints was similar to that of post-weld-treated joints that have not been clean-blasted. In some specimens, S355 and S1100, some improvement was found but more comprehensive experimental work should be carried out to also consider the statistical variation of test results. Considering different available PWTs, TIG dressing was found more beneficial in the S355 normal strength steel and HFMI treatment more beneficial in the S700 HSS and S1100 UHSS steel grades. • Between the two experimentally investigated abrasive materials, sand and corundum, no difference in fatigue strength was found even though the grit blasting resulted in higher surface roughness and the clean-blasting with corundum provided slightly higher compressive residual stresses than the grit blasting.  S1100  X96  HFMI  C  488  362600  PVS11TL_2  S1100  X96  TIG  C  460  124792  PVS11HK_3  S1100  X96  HFMI  S  488  234101  PVS11TK_4  S1100  X96  TIG  S  482  86624  PVS11HL_5  S1100  X96  HFMI  C  702  66883  PVS11TL_6  S1100  X96  TIG  C  445  162176  PVS11HK_7  S1100  X96  HFMI  S  717  87463  PVS11TK_8  S1100  X96  TIG  S  491  116566  PVS11_10  S1100  X96  AW  n/a  543  23000  PVS11T_11  S1100  X96  TIG  n/a  403  97548  PVS11H_9  S1100  X96  HFMI  n/a  540  147181