**3.5 Tension-tension fatigue testing**

In analyzing the feasibility of fused deposition modeling to fully evolve into a rapid manufacturing tool, it is important to assess the fatigue properties of FDM specimens and their dependence upon raster orientation. A comprehensive fatigue study is warranted and is currently underway by the authors. At this time, however, a pilot fatigue study utilizing a tension-tension loading configuration has been completed. The fatigue tests utilized specimens originally with natural undamaged surfaces. The maximum load of the fatigue cycles was set to 70% of the mean failure load as determined in the static tensile tests for each specific raster orientation. The minimum cycle load was set as 1/10 of the maximum load, and the resulting stress ratio R for the fatigue tests was 0.1. The value of 70% of the mean failure load was selected following a preliminary study focused on determining a reasonable compromise between the occurrences of excessively long fatigue lives and being too close to the static strength.

All fatigue tests were performed at room temperature. To eliminate any heating effects due to considerable strains, the fatigue loading was applied at a low frequency, 0.25 Hz. The specimen surface was observed during testing and the number of cycles to final failure was determined. The test results appear in Table 11. The 45º raster orientation fractured with the smallest mean number of cycles to failure (1312), representing only 26.7% of that of the +45º/-45º raster orientation (4916).


Table 11. Tension-tension fatigue test results

A one-way ANOVA was completed in order to consider the equivalence of the mean number of cycles to failure for the four raster orientations. The results, appearing in Table 12, provide a calculated F-test statistic of F(3,16) = 124.95 and a *p*-value of 0.0001, indicating a significant difference between some or all of the mean cycle values associated with the four raster orientations at a level of significance of = 0.05. The resulting coefficient of determination associated with this analysis was R2 = 0.9591.


Table 12. One-way ANOVA results for tension-tension fatigue testing

Tukey post hoc comparisons indicated that the difference between the mean number of cycles to failure was not significant for the pairwise comparisons of 0° with +45°/-45°, or 45° with 90° raster specimens. All other pairwise comparisons were significantly different. These results confirm that certain raster orientations have a significant effect on the tensionfatigue properties of the FDM specimens.

The failure modes of the specimens were similar to those for static tension testing (Figure 5), except that several of the 0° raster specimens fractured with a more uneven and almost toothed appearance during fatigue testing. This is shown in Figure 19a where the clusters of rasters have broken at various fiber lengths showing an erratic crack path most probably driven by the areas of weakest fiber bonds and voids between fibers. This SEM image also shows the smooth, brittle, tensile failure on each individual raster face.

The fracture surfaces of the +45°/-45° raster specimens, in contract, showed a mixed mode repeated failure of individual fibers by shearing and tension (Figure 19b). Upon close examination of the individual raster faces, failure initiation sites can be observed at multiple locations. In areas of closely bonded clusters of fibers, "river patterns" can be observed and are believed to occur at large crack growth rates. At the same time, patterns resembling "fish scales" are observed and are often an indication of small crack growth rates. This change of the pattern indicates the existence of a dynamic transition of failure mode.

Fig. 19. SEM images of fatigue fractured specimens with: (a) 0° rasters (b) +45°/-45° rasters

There was some level of correlation between the tension-tension fatigue results and the static tension test results. While the 0° raster orientation achieved the maximum tensile strength, the +45°/-45° specimens survived the most fatigue cycles to failure on average. However, the mean number of cycles to failure for the 0° raster orientation was not found to be statistically different than the +45°/-45° specimens at a level of significance of = 0.05.

Although these fatigue tests only serve the purpose of a pilot study, the results indicate that the directionality of the polymer molecules and the presence of air gaps and porosity result in anisotropic behaviour of FDM specimens under tension-fatigue loading.
