2016 Volume 56 Issue 5 Pages 835-843
The relationship between residual stress and fatigue strength has been researched for electron beam welded joints made of sheet metals with a thickness of approx. 10 mm. Residual stresses were analyzed by the Finite Element Method, and fatigue strength was obtained by the three-point bending fatigue test. From the results of examining the effect of heat input, a particular relationship was observed between fatigue strength and residual stress, but the effect of stress concentration on fatigue strength was not significant within the tested range of stress concentration factors with varied bead shapes. As a result of testing the effect of steel type, the correlation between fatigue strength and residual stress was confirmed with exception of S50C steel that resulted in lower fatigue strength deviating from the correlation, the deviation of which can presumably be attributed to localized hardening and narrowed area where compressive residual stress was induced.
Residual stresses in welds significantly affect the fatigue strength and stress corrosion crack resistance of the weld joint, and therefore, numerous studies have been made about the effects. The residual stress induced by electron beam welding (EBW) has also been researched in various approaches, but many of the study reports dealt with thick plates (50 mm or thicker) only.1,2,3,4,5) This was because the EBW process was costly, and thus studies were focused on thick plate materials for such specific applications as observed in the nuclear-power and space-related fields that require extremely high qualities for the weldments.
In these times, however, there have been many cases where the EBW processes are employed in such general applications as autos, electric machinery and apparatuses, and industrial tools. This is owing to reduction in the processing cost through usage expansion of the equipment and corporate efforts by job shops. In particular, EBW is applied for partial penetration welding of parts with a plate thickness of 3–15 mm as observed in car gears and gauge diaphragms. While EBW has lately been expanded in such sheet metal applications, there are little reports of in-depth research on this field. Taking this into consideration, the authors have researched the residual stress characteristics in the partial penetration EBW welds of sheet metals.
For the first study report,6) the effects of beam power and diameter were investigated in an orderly sequence for carbon steel (SS400). As a result, the extent of transverse residual stress at the weld toe, which largely affected the fatigue strength, was clarified to have relation to the weld penetration. In the second study report,7) the transverse residual stress at the weld toe was revealed to become compressive to a larger degree as the yield strength of the testing material was higher. This tendency was observed not only for ferritic and martensitic steel materials but also for other types of metals such as SUS304 and A5052. The present report provides the fatigue strength data obtained by actual testing and discusses how they are affected by the residual stress in relation to the parameters of heat input and steel type.
In the present study, the testing plates are referred to as “sheet metals” in the EBW applications, in contrast to heavy thick metals having a thickness of 50 mm or thicker which have conventionally been welded by EBW.
The testing materials used were steel plates of five different types (SS400, SUS304, S50C, HT60, and SCM440). The dimensions of steel plate were 10 mm for thickness, 100 mm for width, and 100 mm for length — Fig. 1. The partial penetration welding by the EBW process can generally be applied to a plate thickness range from about several millimeters to 10-odd millimeters; in this study, 10-mm thick plates were used because this thickness enabled to obtain partial-penetration welds easily with varied electron-beam output parameters.
Dimensions of testing plate and temperature measuring points.
Bead-on-plate welding was conducted by using the 6-kW EBW equipment shown in Fig. 2. As shown in Table 1, five different combinations of beam power and beam diameter were used in the study of the effect of heat input. The conditions of A-1, A-2, and A-3 were to examine how beam power affect the test result, and A-2, A-4, and A-5 were to investigate how beam diameter influence the test result. The beam diameter was the theoretical value at an intensity of 1/e (about 37% of the peak value). For changing the beam diameter in the conditions of A-4 and A-5, the focal distance was shifted respectively by 20 mm and 40 mm with the focus electric current decreased respectively by 20 mA and 40 mA from that in the condition A-2. In the conditions of A-1–A-5, the amount of heat input was varied. As shown in Table 2, the welding parameters were kept constant for all types of steel in order to study the effect of steel type. Each test plate was restrained with a couple of restraining plates clamped at six points on the right and left sides of the test plate. The restraining plates for the test plate were removed in about three minutes from the start of welding, because it took time to open the chamber to the atmosphere.
6-kW electron beam welding equipment.
Test No. | A-1 | A-2 | A-3 | A-4 | A-5 |
---|---|---|---|---|---|
Steel type | SS400 | ||||
Beam power | 600 W | 1200 W | 1800 W | 1200 W | 1200 W |
Beam diameter | 0.49 mm | 0.49 mm | 0.49 mm | 1.04 mm | 1.53 mm |
Heat input | 72 J/mm | 144 J/mm | 216 J/mm | <144 J/mm | |
Welding speed | 500 mm/min |
Test No. | B-1 | B-2 | B-3 | B-4 | B-5 |
---|---|---|---|---|---|
Steel type | SS400 | SUS304 | S50C | HT60 | SCM440 |
Beam power | 1200 W | ||||
Beam diameter | 0.49 mm | ||||
Heat input | 144 J/mm | ||||
Welding speed | 500 mm/min |
The restraining force was measured by a load sensor, and the resultant pressing force was about 390N for each clamping position. One piece of restraining plate (with a 3000-mm2 contacting area on the test plate) was clamped at three locations, and thus the restraining stress could be calculated as 390N × 3/3000 mm2 = 0.39 MPa. In order to verify the test plate was sufficiently restrained, the gap between the test plate and the fixture was checked with a gap gauge (0.05 mm) before welding after restraining, and before releasing the restraint after welding; as a result, no gap was observed.
2.3. Temperature MeasurementThe surface temperature of the testing steel plate was measured during welding in order to develop the basic data for determining the heat input conditions for the analysis by the Finite Element Method (FEM). The measuring locations were, as shown in Fig. 1, at 3 mm and 8 mm on the top surface as well as 2 mm on the bottom surface respectively from the center. The temperature measurement was carried out for all the test plates, using a MEMORY HiCORDER together with K-thermocouples (glass fiber coated) made of wires with a diameter of 0.32 mm. The tip of the thermocouple was fixed on the temperature measuring point in the test plate by resistance welding, and the other end of the thermocouple was connected to the MEMORY HiCORDER with a compensation wire for K-thermocouple. The measurement frequency was 10 plots per second.
2.4. Analysis by FEMThe FEM analysis was applied for calculation of residual stresses. For the analytical model, a three-dimension model with 20 mm in the longitudinal direction and right-left symmetric 1/2 part in respect of the weld axis, as shown in Fig. 3, was adopted in an attempt to shorten the computation time. The analysis was also carried out with the model having a 100-mm longitudinal length in conformity with that of the actual test plate; as a result, it was verified that there was no significant difference in the analytical results, including temperature and strain as compared with those of the model with a longitudinal size of 20 mm. The restraining condition was set to be consistent with that in the actual welding; i.e., the same position was fully restrained in the Z direction (the direction of angular distortion) for 3 minutes after welding. The welding conditions were also set to be the same as those for the actual welding as shown in Tables 1 and 2. Other analytical conditions were those shown in Table 3. The physical properties applied were those shown in Fig. 4 for examples in consideration of temperature dependency, but density and Poisson’s ratio were kept constant regardless of temperature dependency.
Analytical model for FEM.
Boundary condition | Atmosphere (Vacuum) | Radiant heat transfer at an emissivity of 0.2 |
Restraint | Restrained at the same location for the same time as taken in the welding test | |
Heat source | Internal calorific type | Heat source is input into the penetration area defined with cross-sectional macrograph |
Efficiency of heat input | 70–80% | |
Number of elements | Approx. 15000 |
Typical physical properties (SUS304).
For analysis of residual stresses, the existing data reported in the past6,7) was applied. In relation to these data, the restraining condition set for the analysis on the effect of heat input was obtained by simulating the butt joint. Since the butt joint was completely restrained by using a clamping jig in the previous experiments, the residual stress should have been similar to that in the present case of bead-on-plate welding.
2.5. Three-point Bending Fatigue TestThe fatigue test specimen with a width of 20 mm was machine cut from the central area of the bead-on-plate weld coupon shown in Fig. 1 and was tested by the three-point bending load. The fatigue test method is schematically shown in Fig. 5. Fatigue test was conducted under the load control conditions with a stress ratio of 0.1 and a load frequency of 15 Hz — Table 4.
Outline of three-point bending fatigue test.
Loading mode | 3-point bending |
Control method | Load control |
Stress ratio | R = 0.1 |
Loading frequency | 15 Hz |
Span | 64 mm |
Vickers hardness was measured at 0.2 mm below the test plate surface at locations ranging from the weld metal center to the base metal. The testing load was 300 gf. The measuring locations are shown in Fig. 6.
Vickers hardness measuring positions.
The analysis results of the residual stress in the transversal direction to the weld axis are shown in Figs. 7(a) and 7(b). Figure 7(a) is a comparison chart on the effect of beam power, and Fig. 7(b) is that on the effect of beam diameter. Every result exhibited such a tendency that the transverse residual stress ranged from zero to compression near the weld toe and changed to tension as the calculation point departed from the weld toe. In the following, noted is the residual stress near the weld toe that significantly affects the fatigue strength of the weld. From the analysis results shown in Figs. 7(a) and 7(b), it is clear that as the beam power increases and as the beam diameter decreases the compression stress at the weld toe becomes stronger. Figure 8 shows cross section macrographs for individual welding conditions. The macrographs suggest that the analysis results shown in Figs. 7(a) and 7(b) can be related to the penetration depth. That is, as the penetration depth increases, the compression residual stress become stronger.
Analysis results of transverse residual stresses (Effect of heat input).
Cross-sectional macrographs (Effect of heat input).
The reason the compression stress at weld toe became larger with an increase in the beam power and a decrease in the beam diameter was detailed in the past reports;6,7) i.e., the existence of the maximum temperature region at the mechanical melting point caused the increase in the maximum tension residual stress zone in the interior of the plate, and the resultant reactive force probably generated strong compression residual stresses in the weld surface area.
3.1.2. Stress ConcentrationWith the macrographs shown in Fig. 8, the shape of weld toe was measured to obtain the flank angle and radius of curvature. In addition, the stress concentration factor of Kt was obtained by the following formulas.8,9) The results are shown in Table 5.
Test No. | Beam power and diameter | Radius of curvature (mm) | Flank angle (degree) | Stress concentration factor (Kt) |
---|---|---|---|---|
A-1 | 600 W, 0.49 mm | 0.97 | 167 | 1.3 |
A-2 | 1200 W, 0.49 mm | 0.27 | 144 | 2.0 |
A-3 | 1800 W, 0.49 mm | 0.19 | 120 | 2.6 |
A-4 | 1200 W, 1.04 mm | 0.73 | 158 | 1.6 |
A-5 | 1200 W, 1.53 mm | 1.21 | 161 | 1.3 |
The stress concentration factor was found to vary from 1.3–2.6 depending on the welding conditions of A-1–A-5. From this finding, the results of the three-point bending fatigue test under the welding conditions of A-1–A-5 were expected to respond to both residual stress and stress concentration.
3.1.3. Three-point Bending Fatigue TestThe results of the three-point bending fatigue test are shown in Figs. 9(a) and 9(b). These figures show charts to compare the effects of beam power and beam diameter, respectively. The stress shown on the vertical axis of the individual figure is not nominal one but the local one associated with bead shape. The local stress was obtained with the nominal stress multiplied by the stress concentration factor of the weld toe, which was further multiplied by 4/5 according to the JSSC guideline10) for the effect of the bending stress. The rupture locations in all the ruptured test specimens were the weld toe. Firstly, the effect of stress concentration will be discussed. If fatigue life is significantly affected by stress concentration, it ought to become longer with a smaller stress concentration factor. In the present experimental results associated with the parameters of beam power and beam diameter, no such tendency was found, rather observed was a reverse tendency. Next, the effect of residual stress will be considered. When the residual stress at the weld toe is compressive at a greater degree, the fatigue life ought to be longer. As shown in Fig. 7, the residual stress became compressive to a greater extent with a larger beam power and a smaller beam diameter. As observed in Fig. 9, the compressive residual stress significantly affected fatigue strength as presumed above. From these experimental results, it has been clarified that the fatigue strength of the partial penetration weld made by EBW is influenced predominantly by residual stress, not by stress concentration.
Results of three-point bending fatigue test (Effect of heat input).
As shown in Fig. 10, the surface residual stresses (σT) in the transverse direction are compression stresses in the area near the weld toe and tend to shift toward the tension stress side reaching zero as the testing point departs from the weld toe for all the steel types. This tendency corresponds to that of the test results on the effect of the heat input. Figure 11 shows cross-sectional views of the analytical model, which exhibits residual stress distributions for each type of steel. There can clearly be observed the distributions of compression stress in the steel surface area and of maximal tension stress near the plate thickness-wise center. This also has a good matching to such findings, which were reported for some past researches on the complete penetration weld of thick plate, that compression stress generated in the top and bottom surfaces of the plate and tension stress at the midpoint of the plate thickness.1,2,11)
Analysis results of transverse residual stresses (Effect of steel type).
Cross-sectional distributions of transverse residual stress.
Since the welding condition was kept constant, penetration depth and bead width resulted in the similar dimensions for all the steel types as shown in Fig. 12; therefore, the difference in residual stress can be deemed to be the effect of steel type. As shown in Fig. 13, it has been found that the residual stress at the weld toe correlates with the room-temperature yield strength of each type of steel. In the transverse direction, the residual stress became compression to a larger extent with the steel type having a higher yield strength. To explain the dependency of yield strength on the transverse residual stress, the effect of addition of the compression stress induced by the transformation expansion of the weld was first taken into consideration. Figure 14 shows the analysis results with or without consideration of transformation expansion. From this figure, it can be said that almost no difference is observed in the residual stresses with or without consideration of transformation expansion; i.e., the effect of transformation expansion is not significant.
Cross-sectional macrographs (Effect of steel type).
Room-temp. yield strength vs. residual stress at weld toe.
Analysis results with or without consideration of transformation expansion (Transverse direction, Residual stress at weld toe).
In the study of the effect of heat input discussed above, it was clarified that the residual stress at the weld toe became compression as the result of the reaction of internal tensile stress at a greater extent in proportion to penetration depth. By contrast, in the experiment of the effect of steel type, penetration depths were almost the same for all the steel types. Taking into account these results and the fact that the longitudinal residual stress becomes tensile to a greater extent with an increase in the yield strength of steel, the mechanism of the behavior of the transverse residual stress at the weld toe can be assumed as follows; i.e., in EBW with a narrow bead width, the transverse residual stress near the weld toe that associates with the fatigue strength is affected remarkably by the compression stress corresponding to the Poisson’s ratio related to the longitudinal tensile residual stress.
3.2.2. Stress ConcentrationWith the macrographs shown in Fig. 12, the shape of weld toe was measured to obtain the flank angle and radius of curvature. Also, the stress concentration factor of Kt was obtained by the formulas shown in Section 3.1.2. The results of measurement and calculation are shown in Table 6.
Test No. | Steel type | Radius of curvature (mm) | Flank angle (degree) | Stress concentration factor (Kt) |
---|---|---|---|---|
A-1 | SS400 | 0.24 | 145 | 2.2 |
A-2 | SUS304 | 0.23 | 137 | 2.5 |
A-3 | HT60 | 0.24 | 143 | 2.3 |
A-4 | S50C | 0.27 | 145 | 2.1 |
A-5 | SCM440 | 0.22 | 142 | 2.3 |
The stress concentration factors were found to fall in a range of 2.1–2.5 for all the welding conditions of B-1–B-5. This range was smaller than that of the stress concentration factors in the conditions of A1–A5 for testing the effect of heat input (Section 3.1). The stress concentration, therefore, was assumed not to be the main factor that affected the fatigue strength; this matter will be discussed in the following section.
3.2.3. Three-point Bending Fatigue TestThe results of the three-point bending fatigue test are shown in Fig. 15. The stress value shown on the vertical axis of the figure is not nominal one but the local one associated with bead shape, like the case of testing the effect of heat input. The rupture locations in all the test specimens were the weld toe. The residual stress at weld toe tended to become compression to a greater extent for the steel type with a larger yield stress (see Fig. 10). In the fatigue test results, such a tendency could be found that higher fatigue strength was obtained with the steel type having a larger compressive residual stress at weld toe; i.e., residual stress affected the fatigue strength. However, S50C steel was revealed to deviate from this tendency. As regards the effect of stress concentration, no correlation was observed between stress concentration factor and fatigue strength (see Fig. 15 and Table 6). Figure 16 was developed for all steel types with a stress concentration factor of 2.1 unified to that of S50C, which shows the similar relationship to that of Fig. 15 between steel type and fatigue life. In this case, S50C resulted in outstandingly low fatigue life similarly to the result observed in Fig. 15 where the stress concentration factor was varied from 2.1 to 2.5 according to steel type. That is, even in the test parameter range where the effect of steel type was researched, it can be assumed that fatigue strength was affected greatly by residual stress, not by stress concentration. Figure 17 shows the results of Vickers hardness test. In the case of S50C steel with high quench hardenability, the weld toe was verified to be hardened excessively by the rapid heating and cooling caused by welding. In addition to S50C, SCM440 exhibited high hardness, too. This is why, the low fatigue life of S50C is unlikely to be caused by high hardness alone. As regards the residual stress at the weld toe shown in Fig. 10, S50C exhibited a large compression stress but its compression range was narrower than SCM440. From these data, the low fatigue life of S50C can probably be attributed to the following multiple reasons: (1) high quench hardenable S50C exhibited the highest hardness near the weld toe due to the rapid heating and cooling during welding, resulting in the significant hardness difference between the weld and base metal which increased the degree of stress concentration; and (2) the compression stress range from the weld toe was narrower.
Results of three-point bending fatigue test (Effect of steel type).
Results of three-point bending fatigue test (with a stress concentration factor of 2.1 unified for all steel types).
Results of Vickers hardness test.
Furthermore, according to a past example of fatigue test results of the medium-carbon steel weld made by EBW,12) there were segregated inclusions and microcracks in the final solidification zone, which were assumed to have become the initiation site of premature rupture. By contrast the microstructure observation of the present test specimen verified that the weld contained no such inclusion or crack.
The relationship between residual stress and fatigue strength has been researched with the partial penetration welds made by EBW of sheet metals, and the following conclusions have been obtained.
(1) With a steel type of SS400, the three-point bending fatigue test was conducted in the five conditions assorted for comparing the effect of heat input; as a result, a particular relationship was observed between the residual stress at weld toe and the fatigue life. However, the effect of stress concentration on fatigue life was not significant, although bead shape was varied (with stress concentration factors of 1.3–2.6) in association with the welding condition.
(2) With five steel types (SS400, SUS304, HT60, S50C, and SCM440), the weld coupons were prepared with the constant heat input to conduct the three-point fatigue tests; consequently, the correlation between the residual stress at the weld toe and fatigue strength was verified. However, S50C steel resulted in lower fatigue strength deviating from the correlation. This deviation can presumably be attributed to localized hardening and narrowed area where compressive residual stress was induced.