Zhao et.al. 36 reported a fatigue strength of 75 MPa for cubic regular specimens with a porosity of ~63% and strut thickness of ~600 µm. In this study, the fatigue strength of cubic regular structure reported is 100 MPa with a porosity of ~76% and as-designed strut thickness of ~450 µm. Considering that the difference in other parameters such as microstructure and heat treatment, the estimated fatigue strength of cubic regular is similar but slightly higher when compared to Zhao et.al36. The fatigue strength of 3.1 MPa obtained from cross shaped structures is very close to the fatigue strength of 2.5MPa reported by Peng et. al. 55 from the fatigue life prediction of BCC (85% porosity) using finite element analysis. Irregularity had a predominant impact which reduced the fatigue strength by almost ten times. Star regular specimens had the highest fatigue strength after cubic specimens. The presence of one vertical strut in star regular specimens increased the fatigue strength by almost seven times compared to cross regular. However, the same extent of increase was not seen when comparing star irregular and cross irregular since the vertical struts of the star topology are inclined with respect to the load. The fatigue strength of trabecular structures is much higher than the cross shaped samples even though they have lesser number of struts per node (4-6 trabecular and 8 cross regular). The number of struts per node for irregular cubic and star structures is also higher than in trabecular structure and yet their fatigue strength is slightly lower. The lower overall porosity of trabecular specimens has for sure an influence, but the main cause for the better behavior is the presence of struts in all the direction and the isotropy of the topology. This is further explained using the frequency variation curve in Fig.8.
The general frequency versus number of cycles curve for regular cross structure showed a linear decrease in the frequency as shown in Fig. 8a. Similar results were obtained for all regular structures and most of the irregular configuration. But some of the specimens, especially in trabecular structures, had a distinctive behavior as shown in Fig. 8b. In this case, the decrease in the frequency was not steady, as the graph indicates multiple peaks and valleys were observed which delayed the failure. The random distribution of the struts eliminates field failure, the failure occurs at the weakest link , but the random struts help in stress redistribution and the failure remains local. This can be seen through the fracture plane analysis in the next section which indicated multiple failure planes in trabecular specimens.