Investigation on fatigue strength of cut edges produced by various cutting methods for high-strength steels

Due to the interest in effective light steel constructions, high-strength steels have gained importance. Different thermal cutting processes are frequently used in the metal processing industry. Besides the weld seams, free cutting edges gain technical and economic relevance as locations for potential fatigue cracks. In this investigation, fatigue tests were carried out on 8-mm-thick samples of S355M and S690Q steels at a stress ratio of R = 0.1. The cutting methods used were oxygen, plasma, laser, and waterjet cutting. Quality improvement methods, like shot peening, grinding, and cutting speed reduction, were applied for some series. The surface roughness was measured to classify the specimens into quality groups according to ISO 9013. The cut edge condition was also characterized by hardness and residual stress measurements. The investigation shows that all tested series exceed the FAT100 class and can be classified in FAT125. Specimens ranged in quality group 2 of ISO 9013 according to the roughness achieve FAT140 regardless of cutting technology or material. According to the ISO 9013, most of the specimens are classified in the quality group 2 and group 3. Fatigue strength results are significantly different in one quality group. No prediction can be made. ISO 9013 has a weak connection to fatigue strength. Quality improvement methods have a significant influence on the fatigue strength and can increase it. Due to reduced cutting speeds, the roughness decreases also. It results in an increase of the fatigue strength in all tested series in this study. In order to make a prediction of the fatigue performance, the standard has to be specified and the cutting process as well as the steel strength should be considered.


Introduction
The fatigue strength reducing effects of weld seams is taken into account in the designing phase of welded constructions. Welds are placed in less stressed areas and conventional posttreatment methods, like shot peening, are used. Therefore, free cutting edges gain technical and economic relevance as locations for potential fatigue cracks. Cutting technologies used in the industry produce different execution qualities at the cutting edge. The so-called Kitagawa diagram shows the relation between fatigue strength and defect size based on experimental values [1], as illustrated in Fig. 1. For components with sufficiently large defects, the fatigue strength of the component is characterized by crack propagation phase. With such defect sizes, the fatigue strength does not increase with the tensile strength. The execution quality of cut edges in terms of surface roughness (or "defect size") directly influences the local stress level and therefore the expected lifetime of the specimens. The lifetime is governed by crack growth in the case of large defects and by crack initiation in the case of small defects. Crack initiation is affected by the hardness respectively tensile strength resulting in an increase of fatigue strength with increasing tensile strength in the presence of small defects. A higher material strength therefore leads to a higher fatigue strength of the component due to the suppressed dislocation movement. In general, for thermal cutting edges, the fatigue strength should increase with tensile strength if the defect is smaller than the corresponding limit value.
The influence of thermal cutting edges and their different qualities on fatigue strength has already been analyzed in several studies, for example, by Sperle [1], Remes et al. [2], Stenberg et al. [3], and Lillemäe et al. [4]. Sperle carried out fatigue tests with a comprehensive range of steel grades (yield strengths from 240 to 900 MPa) on 6-mm-thick and 12-mmthick specimens. The specimens were thermally cut by gas, plasma, and laser. He found out that the fatigue strength increases with the steel grade depending on the surface roughness. The investigation compared also the plasma and laser cut edges with machined edges. Sperle determined that the fatigue strength rises with increasing tensile strength depending on the roughness. The increase of the fatigue strength in all cases is comparable, which was explained by the consistently small roughness depth.
Another study was made by Remes et al. [2]. They used 15mm-thick and 17-mm-thick specimens, which were plasma cut, and considered three series: first untreated, second ground, and third ground + sandblasted. The surface roughness, hardness, and residual stresses were determined out for the rolled plate surface and specimen cut edge. Fatigue tests on specimens were conducted, which had yield strengths of 460 MPa or 690 MPa, with different surface treatments. The investigation shows that posttreatments increase the fatigue strength of high-strength steel due to the reduced surface roughness and induced compressive residual stresses. The study was continued by Lillemäe [4], which shows additionally that 690 steel had significantly higher fatigue strength compared with 355 steel. Stenberg et al. [3] investigated the influence of surface roughness on the fatigue strength in high-strength steels (S700 and S960) of 6-mm-thick and 16-mm-thick specimens. They were cut using oxygen, plasma, laser, and waterjet cutting. Surface roughness was measured and classified in the four quality groups according to the ISO 9013 standard [5] for thermal cutting quality tolerances as follows: It was used to assess the produced quality of the thermally cut edges. The testing proved an increased fatigue strength compared with the conservative international guidelines [6]. However, the examinations do not cover all quality ranges. At this point, the question arises, up to which conventional cutting edge quality fatigue strength increases with higher strength of material. The execution quality, for example, according to ISO 9013, of the material or different cutting processes in the guidelines is not considered yet [7,8]. Depending on the process and the cutting parameters used, cutting technologies produce different execution qualities at the cutting edge. The requirements for the quality characteristics of the components are, for instance, defined according to EN 1090 [9] and ISO 9013 respectively. Especially the surface roughness is a factor governing the fatigue strength of thermal cut edges.
In this investigation, fatigue tests were carried out on thermal cutting edges in different execution qualities with a constant force amplitude at a stress ratio R = 0.1. The tested specimens were made from 8-mm-thick plates of S355M and S690Q. The cutting methods used were oxygen, plasma, laser, and waterjet cutting. Quality improvement methods, like shot peening, grinding, and cutting speed reduction, were applied. All samples were characterized in terms of hardness, roughness, and other quality characteristics according to EN 1090 and ISO 9013. Furthermore, the residual stresses were measured. The fatigue strength of all series was analyzed considering the surface quality characteristics.

Specimens
Fatigue strength of different thermal cut edges of two different steel grades with a minimum yield strength of 355 and 690 MPa was analyzed. The uniaxial test specimen used in this study was similar to previous work in [10] and is shown in Fig. 2. The samples were 394 mm long and 38 mm wide in the critical cross section. Besides the two different steel strengths, the specimens were cut from the 8-mm-thick plate using oxyfuel, plasma, laser, and waterjet cutting processes. The conditions for the specimens were untreated and defined "as-cut" and quality improvement treatments (posttreatment processes) such as grinding, sandblasting, and cutting speed reduction were used. The laser cutting process changed parameter from a 10-Hz pulsed laser to continuous laser for both steel grades (S355M, S690Q) using the reduced cutting speed. Fifteen different series were defined. A total amount of round more than 150 fatigue specimens were tested, which were at least 10 samples per test series.
The two different steel grades have a wide range of mechanical properties as follows: -S355M: Thermomechanically rolled weldable fine grain structural steels with minimum yield strength of 355 MPa and a carbon equivalent value (CEV) less than 0.39, EN 10025-4 [11]. Samples were cut from an 8-mm-thick plate. -S690Q: Quenched and tempered high-strength steel with a minimum yield strength of 690 MPa and a carbon equivalent value less than 0.65, EN 10025-5 [12]. Samples were cut from an 8-mm-thick and 20mm-thick plates. Table 1 shows the detailed material properties. Because the specimens were cut from different industrial partners, two different charges of the materials were defined. The specimens using laser cutting processes were cut out of plates from charge 2 (S355M,2 and S690Q,2). The chemical composition including the carbon equivalent value is presented in Table 2.

Cut edge conditions
The cut edge conditions were characterized. The location of measurements of the roughness, hardness, and residual stress is shown in Fig. 3. Hardness measurements were made at the cross section and directly at the cut edge (see below). Table 3 shows the cut surface of steel S355M by using oxygen, plasma, laser, and waterjet cutting processes. The surface topography after using posttreatment methods like shot peening and grinding or due to speed reduction is also  shown. At the top of each figure, there is the upper edge, and at the bottom, there is the lower edge of the cut surface. It can be seen that oxygen cutting implicates a lot of slag and scale, which stick to the edge. It provides a cut surface with varying roughness. Plasma cutting creates the cut surface with the lowest roughness, while laser cutting generates surfaces with the highest roughness value. However, it depends on the cutting speed. The surface after shot peening and grinding looks visually clean. No other defects on the surfaces are visible.

Roughness measurements
Surface roughness measurements were made on the specimens according to ISO 4288 [13]. Measurements were done on the cut edge surface (Fig. 3). The tests were carried out with the MarSurf M 400 surface measuring instrument using the profile method over the length of 17.5 mm. From each series, 3 specimens were measured: 3 lines at the top and 3 lines at the bottom of the cut edge surface. The following parameters were determined: The average of the five highest peaks and lowest valleys R z , the arithmetical average R a , and the maximum peak to valley height R max . No measurements were done at the plate surface.

Hardness measurements and metallography analysis
Cross sections have been cut from the dog-bone specimens by using a water-cooled abrasive cutter. The surface were ground using different abrasive paper, polished and etched with a Nital solution (2%) to visual the microstructure. The hardness in the heat-affected zone (HAZ) of thermally cut edges was analyzed afterwards. Therefore, Vickers hardness (HV0.1) measurements were made for each series according to ISO 6507 [14]. The investigation was carried out with Wilson Hardness Tukon 1202 instrument. No hardness measurements were made for waterjet cut plates, because no changes to the base material were expected. Additionally the hardness was measured directly at the cut edge according to the instruction from the Appendix in the DIN EN 1090-2:2017 draft. UCI (ultrasonic contact impedance) method was used [15]. Cut edges have to be prepared and ground using a P600 abrasive paper. Valleys of the surface should be visible. Measurements are made as seen in Fig. 4.

Residual stress measurements
The residual stresses were measured on the cut edge of all series using the X-ray diffraction (XRD) technique. Location of the measurements which were in the middle of the cut edge is shown in Fig. 3. For the XRD experiments, a ψdiffractometer with radiation source CrKα was used. The circular collimator diameter was 1.5 mm. The longitudinal residual stresses were calculated from the {211}-Fe diffraction lines determined at 13 ψ-angles with help of the sin 2 ψ-method. Residual stress measurements were done on all 15 different series.

Fatigue tests
All fatigue tests were performed on a 250-kN hydraulic testing machine using cyclic tensile tests with a constant amplitude load (CAL) and a stress ratio R = 0.1 following the recommendations of DIN 50100 [16]. The test frequency was 10-20 Hz, depending on the load level of each test. The main fatigue strength was determined in the life range of 1 × 10 5 to 2 × 10 6 cycles. The run-out point was defined at five million cycles and run-out specimens were excluded from the S-N curve calculation. The tests were stopped after complete failure of the specimen. The fracture surface was analyzed to identify the location of crack initiation (Table 4).

Roughness on the cut edge
The measured roughness parameters on the cut edge are the average of the five highest peaks and lowest valleys R z , the arithmetical average R a , and the maximum peak to valley height R max . The results are summarized in Table 5. The corresponding roughness measurements and allowable ranges, according to ISO 9013, are shown in Fig. 5. It demonstrates the arithmetic mean of surface roughness R a as a function of arithmetical peak-to-peak average roughness R z for different cutting technologies and steel strengths (thickness 8 mm). The different quality groups according to the ISO 9013 for 8-mm-thick plates are also shown on the top of the figure. The cutting technologies have various colors (red-oxygen, blue-plasma, green-laser, black-waterjet) and the cut edge condition is marked as a cycle (as-cut) or as a square (improved quality). All measurements can be classified in range 2 and range 3. Plasma cut edge specimens show the best quality of all cutting technologies, which defined in quality range 2. Oxygen cut edge has a high scatter between ranges 2 and 3. For the laser and waterjet cut edges, the figure shows some differences. Depending on the cutting speed, two different qualities can be reached. For high cutting speed, the roughness is also high and the quality range is 3. A reduction of the cutting speed increases the quality and the roughness is low (R z 27-41, quality range 2).
Shot peening and grinding do not change the roughness values compared with the as-cut conditions at the cut edge surface significantly, even though the shape of the cut edge looks visually smoother after posttreatment methods used (see Table 3). Table 3 Cut surfaces of steel S355M condition as-cut and improved quality using oxygen, plasma, laser, and waterjet cutting processes

Hardness measurements
The various cutting technologies affect the material close to the cut edge, as shown schematically in Fig. 3. Cutting speeds related with the cooling rates have an influence on the size of the heataffected zone (HAZ). The CEVof the steel grades determine the possible maximum hardness. Figure 6 shows hardness measurements on the cross section for S355M, 8 mm and S690Q, 8 mm. The curves are plotted just for thermal cutting technology oxygen, plasma, and laser condition as-cut. All measurements were done approx. 2 mm from both edges into the unaffected base material. Concerning the hardness measurements, the HAZ is less than 500 μm wide. Higher strength steels coming up with higher CEV implicate higher hardness values at the thermally cut edges. Comparing the cutting processes, plasma cutting achieves the highest hardness value. No hardness increase is observed at the oxygen cut S335M. The rapid change in the hardness profile of 8-mm-thick plates indicates a very small heat-affected zone. Laser cut specimens have a small HAZ with only one hardness measurement point. Regardless of the steel grade, the maximum hardness value is between 200 HV0.1 and 510 HV0.1. It implicates a relatively increased hardening compared with the base material from 10 to 170%. The results shown are comparable to previous work [2,4,10].
Additionally the local hardness of the cut edges of all series tested is shown in Fig. 7. The hardness was measured directly at the cut edge. Due to the higher test load (HV10), the local hardness at the cut edge is smaller than the maximum on the cross section. Because of the thin hardened layer, the preparation of the cut edge may remove the HAZ a little bit. Shot peening and grinding do not change the hardness value. The cutting speed reduction increases the cooling rate and the local hardness at the cut edge decreases (laser cut edges).

Residual stress
The residual stresses on the cut edge for the different cutting technologies and materials used are given in Fig. 8. It was measured parallel to the loading direction. Measurement results are very different. Waterjet cut edges show compressive residual stresses over the cut edge about − 200 MPa. The shot peened surface shows compressive residual stresses also, approx. − 200 MPa. Oxygen cut high-strength steels, all plasma and laser cut edges show tensile residual stress. Compressive stress on oxygen cut edges and on waterjet cut edges was also seen in [3]. They measured tensile residual stress just on plasma and not on laser cut edges. Remes et al. as well as Lillemäe show also only compressive stress in their investigation. No stress relaxation was observed due to surface grinding.

Fatigue test results
All fatigue test results are summarized in Table 6 including results from previous studies. It shows the number of specimens including run outs, the slope of the S-N curve, Log C  11, and 12. Each figure presents one cutting technology with the different materials and quality improvement methods used. Two design curves from the IIW recommendations (FAT 125, FAT140) are also added. Figure 9 shows the test data from oxygen cut specimens made from S355M and S690Q.  Test data from plasma cut specimens is shown in Fig. 10. Results of S355M specimens (as-cut) show fatigue strength at Δσ = 239 MPa with a slope of k = 5.2. A slight improvement of grinding the cut edge surface was observed. The fatigue strength was calculated to Δσ = 255 MPa at 2 million load cycles with k = 5.2. The benefit of using high-strength steels is most prominent at higher number of load cycles. Run outs have a significant higher stress range. But the fatigue strength was also calculated to a comparable level because of the steep slope k = 3.9 (Δσ = 232 MPa).
Fatigue test results of laser cut edges of 8-mm-thick specimens are illustrated in Fig. 11. For both materials, an increase of fatigue strength at 2 million load cycles was determined when the cutting speed is reduced. S355M Δσ = 300 MPa (as-cut)➔ Δσ = 338 MPa (reduced speed) and S690Q Δσ = 179 MPa ➔ Δσ = 209 MPa. The slopes of S335M S-N curves are significantly higher (k between 9.8 and 12.3) than the slopes of S690Q S-N curves (k between 3.0 and 3.4). It is also observed that the fatigue strength decreases for the highstrength steel. The cutting speed reduction leads to an improvement of the fatigue limits. One reason could be the geometrical improvement of the surface, despite the tensile residual stress on the cut edges.
The highest fatigue strength results were calculated for waterjet cut specimens. Figure 12 shows the test data from specimens made from S355M and S690Q with normal (ascut) and reduced cutting speed. The regression lines of all four S-N curves are very flat and the slopes are comparable (k = 10.2-18.3). As well as the laser cut edges, waterjet cut edges show an increase of fatigue strength at 2 million load cycles using a reduced cutting speed. S355M Δσ = 351 MPa (ascut) ➔ Δσ = 373 MPa (reduced speed) and S690Q Δσ = 449 MPa ➔ Δσ = 504 MPa. The effect of an increased fatigue strength using high-strength steels is seen.
A comparison of all various cutting technologies of S355M specimens with a thickness of 8 mm is shown in Fig. 13. The plasma cut S-N curve shows the lowest fatigue strength Δσ = 239 MPa with k = 5.2. Oxygen cut edges have a slightly higher fatigue strength level at Δσ = 264 MPa with k = 5.8. The fracture surface analysis points out three different locations of crack initiation: on the cut edge, at the corner between cut edge and rolled plate surface, and at the plate surface. Figure 14 presents an example of each location of the crack initiation. In Table 6, the percentage distribution of all the test specimens is summarized. The crack initiated mostly from the corner or even from the rolled plate surface (cf. [4]).

Discussion
This document investigated experimentally the influence of the quality characteristics according to ISO 9013 on the fatigue strength of thermal cut edges. Fatigue tests were conducted on thermal cut edges in different execution qualities with constant amplitude loads at a stress ratio R = 0.1. Two different materials S355M and S690Q with a plate thickness of 8 mm were considered. The cutting methods used were oxy-fuel cutting, plasma cutting, laser cutting, and waterjet cutting. Quality improvement methods, like shot peening, grinding, and cutting speed reduction, were used. All samples were characterized in terms of hardness, roughness, and residual stresses.
Different fatigue strengths were determined for the cutting technologies and materials. Most of the specimens were tested "as-cut" without using any posttreatment method, like grinding or shot peening. Corners were not removed. For this cut edge surface quality, the IIW recommends FAT100. The investigation shows that all series exceed the FAT class and can be classified in FAT125. Specimens ranged in quality group 2 of ISO 9013 according to the roughness achieve FAT140 regardless of cutting technology or material.
The fatigue strength can be increased by quality improvement treatments. Shot peening as well as grinding on plasma cut edges shows higher fatigue strength. The treatment removed the defects on the surface and the corners. This is also shown in previous work (see, e.g., [2,4]).
The cutting speed affects the roughness on the cut edge. Due to reduced cutting speeds, the roughness decreases also. This results in an increase of the fatigue strength in all tested series in this study (laser and waterjet cut edges). It also can be explained with in reduction of the defect on the cut edges.
Using waterjet cutting technologies, higher strength steels increase the fatigue strength. The roughness on the cut edge correlates with the fatigue strength and can be defined as the critical defect of the specimens. Waterjet cut edges show also the highest fatigue strength compared with the other cutting technologies. The defects are sufficiently small and the lifetime is governed by crack initiation.  Cut edge/corner/plate surface According to the roughness measurements on the cut edge surface, the plasma cut specimens were predicted to have the highest fatigue strength also shown in [3]. Fatigue test results show that they have the lowest fatigue strength level for S355M steels. The residual stress varied over all series. Waterjet cut specimens, shot peened specimens, and S355M oxygen cut edges show compressive stress or approx. zero. The positive influence of compressive stress for the fatigue strength was analyzed in previous studies already, e.g., [17]. Tensile residual stress was observed at S690Q oxygen cut edges as well as on all plasma and laser cut edges. Due to the stress ratio R = 0.1 and the resulting high load, the influence of the tensile residual stress on the fatigue strength is low. However, the residual stress state, hardness, and the surface roughness should be considered. The surface condition in terms of surface roughness, hardness, and residual stress  According to these quality groups, the ISO 9013 has a weak connection to fatigue strength, also seen in [3,10]. In order to make a prediction of the fatigue performance, the standard has to be specified and the cutting process as well as the steel strength should be considered. The fracture analysis shows that the crack initiates mostly from the corner. The location of crack initiation weakens the surface roughness as a critical quality characteristic and has to be considered. To modify the IIW recommendation for fatigue design, the ISO 9013 should be connected to it and waterjet cutting could be added in the standard. Different cutting methods reach different fatigue limits. Standards should cover that and should diversify cutting methods. A bonus factor could be considered using quality-enhancing methods.

Summary
In this investigation, fatigue tests were carried out on 8-mmthick samples with a minimum yield strength ranging from 355 to 690 MPa at a stress ratio of R = 0.1. The cutting methods used were oxygen, plasma, laser, and waterjet cutting. Quality improvement methods, like shot peening, grinding, and cutting speed reduction, were used. The surface roughness was measured to classify the specimens into quality groups according to ISO 9013. The cut edge condition was also characterized by hardness and residual stress measurements. The following conclusions can be made: