J Fail. Anal and Preven.(2012)12: 267-272 maximums at 3 us after beginning of the impact, but their residual stress. This mechanism has been verified from locations were, respectively, about 0.3 mm(normal)and Fig. 5 showing that higher(compressive)residual stresses 0.2 mm(oblique)away from the contact points. This result accumulated in the upper part of the metal substrate( the revealed that the crater area under oblique impact was range from 0. 1 to 0.5 mm in x-axis), and then transformed relatively smaller, testifying the phenomena in Fig 3. into the lower(tensile) residual stresses when the depth However, values of these shear stresses, which were grew to the lower part of the substrate(the range from 0.5 responsible for the buckling of the coating, were greater to 1.5 mm in x-axis). This fact revealed that the ductile under oblique impact than those under normal impact. In metal substrate provided adequate structural strength under other words, under oblique impact, higher possibilities impact. Meanwhile, comparatively speaking, the residual would be introduced for separation between coating and stress from normal impact(90%)was a little bit larger than substrate, and also for removal of coating material after that from oblique impact. This phenomenon may be attributed to the fact described as follows as was dis- Actually, for both normal and oblique impacts, great cussed above, the greater the plastic deformation that compressive stresses were engendered upon coatings and occurred in the target under normal impact, the larger the substrates because of erodent particles striking. Through residual stress that was left the shot peening experiment on ductile material ANSI 4340 steel, Torres and Voorwald [36] pointed out that such Comparison with FEM compressive stress only located at limited areas on the target surface, the compressive residual stress increased In this section, results of the normal impact from the below the surface until reaching a maximum depth, and coupled method will be compared to those from the sole then tended to decrease. transforming into the tensile fem and the sole sph models table 2 tabulates the dif ferences between the three models. Indeed. the computation time of sole FEM model was much less than those with SPH particles, and the resulting curve was even smoother. as seen in Fig. 5. However. obvious distortion occurred on the FEM meshes within the impacted area Fig. 6a), while the crater of the SPh particles was free of this problem(Fig. 6b). This could be explained by the fact that the FEM elements were connected by nodes, and thus its result was relatively smooth but inaccurate because of element distortion under such high-velocity impact. Com- prehensively speaking, the coupled method yielded t=9gs」 reasonable results. and also involved cost-effective com- Distance(mm) SPH90° △SPH45° 一FEM90° Distance(mm) Distance beneath the contact point(mm) Fig 4 Interfacial shear stresses between coating and substrate: (a) Fig. 5 Residual stresses beneath the contact points(along the path ofmaximums at 3 ls after beginning of the impact, but their locations were, respectively, about 0.3 mm (normal) and 0.2 mm (oblique) away from the contact points. This result revealed that the crater area under oblique impact was relatively smaller, testifying the phenomena in Fig. 3. However, values of these shear stresses, which were responsible for the buckling of the coating, were greater under oblique impact than those under normal impact. In other words, under oblique impact, higher possibilities would be introduced for separation between coating and substrate, and also for removal of coating material after impact. Actually, for both normal and oblique impacts, great compressive stresses were engendered upon coatings and substrates because of erodent particles striking. Through the shot peening experiment on ductile material ANSI 4340 steel, Torres and Voorwald [36] pointed out that such compressive stress only located at limited areas on the target surface, the compressive residual stress increased below the surface until reaching a maximum depth, and then tended to decrease, transforming into the tensile residual stress. This mechanism has been verified from Fig. 5 showing that higher (compressive) residual stresses accumulated in the upper part of the metal substrate (the range from 0.1 to 0.5 mm in x-axis), and then transformed into the lower (tensile) residual stresses when the depth grew to the lower part of the substrate (the range from 0.5 to 1.5 mm in x-axis). This fact revealed that the ductile metal substrate provided adequate structural strength under impact. Meanwhile, comparatively speaking, the residual stress from normal impact (90) was a little bit larger than that from oblique impact. This phenomenon may be attributed to the fact described as follows. As was dis￾cussed above, the greater the plastic deformation that occurred in the target under normal impact, the larger the residual stress that was left. Comparison with FEM In this section, results of the normal impact from the coupled method will be compared to those from the sole FEM and the sole SPH models. Table 2 tabulates the dif￾ferences between the three models. Indeed, the computation time of sole FEM model was much less than those with SPH particles, and the resulting curve was even smoother, as seen in Fig. 5. However, obvious distortion occurred on the FEM meshes within the impacted area (Fig. 6a), while the crater of the SPH particles was free of this problem (Fig. 6b). This could be explained by the fact that the FEM elements were connected by nodes, and thus its result was relatively smooth but inaccurate because of element distortion under such high-velocity impact. Com￾prehensively speaking, the coupled method yielded reasonable results, and also involved cost-effective com￾putation time. Fig. 4 Interfacial shear stresses between coating and substrate: (a) normal impact; (b) oblique impact Fig. 5 Residual stresses beneath the contact points (along the path of z-axis) 270 J Fail. Anal. and Preven. (2012) 12:267–272 123