**3.4 Impact testing**

The impact study utilized an Izod test configuration with a notched specimen held as a vertical cantilevered beam as shown in Figure 14. In this position, the material was subjected to a load in the form of an impact blow from a weighted pendulum hammer striking the notched side of the specimen. The test measures the impact energy or notch toughness, and the results are expressed in energy absorbed per unit of thickness at the notch in units of J/cm. The impact energy absorbed by the specimen during failure is measured by calculating the change in the potential energy of the hammer. The change in potential energy is proportional to difference in the height of the hammer from its initial position to the maximum height achieved after impact.

Fig. 14. Izod impact test configuration

Impact tests were completed on 10 specimens with each of the four raster orientations. The mean impact energy results are displayed in Table 9. The absorbed energy was the highest for the longitudinal (0°) fiber orientation (2.989 J/cm) and the lowest for the transverse (90°) orientations (1.599 J/cm). The 45º and +45º/-45º default specimens broke with mean impact resistances between those of the 0° and 90° specimens.


Table 9. Impact test results

172 Mechanical Engineering

(a) (b)

rasters, from which the crack then splits to the right and left and eventually climbs, as seen in the shear failure of individual rasters in layers. Upon closer microscopic examination, it was observed that individual rasters showed shear failure with a clearly defined exaggerated shear lip on the top of each fiber; something that was not observed in the analysis of 90° raster specimens that failed in tension testing (Figure 7). In both the 45° and the 90° raster specimens, there is little localized plastic deformation before failure initiates

The impact study utilized an Izod test configuration with a notched specimen held as a vertical cantilevered beam as shown in Figure 14. In this position, the material was subjected to a load in the form of an impact blow from a weighted pendulum hammer striking the notched side of the specimen. The test measures the impact energy or notch toughness, and the results are expressed in energy absorbed per unit of thickness at the notch in units of J/cm. The impact energy absorbed by the specimen during failure is measured by calculating the change in the potential energy of the hammer. The change in potential energy is proportional to difference in the height of the hammer from its initial position to

Fig. 13. SEM image of fractured 90° flexural specimen magnified (a) 25X and (b) 400X

and fibers begin to break.

the maximum height achieved after impact.

Fig. 14. Izod impact test configuration

**3.4 Impact testing** 

The relative impact strengths of the four raster orientations correlated well with the tensile strength results. In addition, the variation of the impact strengths was smallest for the transverse orientation (Figure 15), coinciding with the variation of the tensile test results.

Fig. 15. Interval plot of impact test results

A one-way ANOVA was conducted to compare the effect of raster orientation on mean impact energies in 0º, 45º, 90º, and +45º/-45º conditions. There was a significant effect of raster orientation on impact energies at the *p* < 0.05 level for the four conditions, with F(3, 36) = 37.23, p = 0.0001. Post hoc comparisons using indicated that the mean impact energy for the 45º diagonal condition (2.339 J/cm) did not significantly differ from that of the +45º/- 45º condition (2.514 J/cm), applying a 95% confidence interval (Table 10). All other paired

Anisotropic Mechanical Properties of ABS Parts Fabricated by Fused Deposition Modelling 175

plane. Interlayer delamination was also evident and resulted in hinged specimens at failure. Figure 17 displays a 0° specimen where the slightly varied fiber length is evident, along with the formation of a hinge. The longitudinal specimens had the highest mean impact strength,

Weak interfaces running at ±45° to the crack front require relatively high amounts of energy as a result of a mixed fracture mode. As a result, a rougher texture was evident for the fracture surfaces of the default raster specimens, and several hinged as shown on the bottom of Figure 18a. The fracture patter of the 45° specimens, in contrast, fractured more consistently along the 45° maximum shear plane as shown on the top of Figure 18a. The macroscopic roughness of the fracture surface was the result of the varying fiber lengths,

(a) (b)

Fig. 18. (a) Fracture patterns of 45° & +45°/-45° specimens; (b) SEM image of impact

correlating well with the results of three-point bend tests.

Fig. 17. SEM image of impact fractured 0° raster specimen

evident in the SEM image of Figure 18b.

fractured 45° raster specimen


comparisons indicated statistically significant differences in mean impact energies. These results suggest that impact strength has anisotropic characteristics.

\* Mean difference is significant at the 0.05 level

Table 10. Post hoc Tukey HSD multiple comparisons of mean impact energies

Inspection indicated that the fracture patterns of the specimens varied as a function of the raster orientations. The longitudinal (0º) and transverse (90º) pieces fractured along a path oriented 90º to the length of the specimen. The fracture surface appeared smooth along the layers of each of the transverse specimens, all of which experienced clean and complete separation of the specimen into two discrete pieces. Weak interfaces parallel to the crack front affected the transverse raster specimens by providing a straightforward path for crack propagation, resulting in the least amount of energy absorption. Failure initiated from the side on the notch where the energy absorption was greatest, thus causing the rasters to fracture longitudinally along the length of the fiber through several layers until the energy adsorption decreased significantly and the sample broken in half (Figure 16a). Upon closer examination, it can be seen that individual fibers plastically deformed and twisted at the ends upon catastrophic failure, whereby halves of the samples separated (Figure 16b).

Fig. 16. SEM image of fractured 90° impact specimen magnified (a) 20X and (b) 160X

The fracture surfaces of the 0° raster specimens were macroscopically rougher, in contrast, displaying slightly jagged edges of individual broken fibers along the transverse fracture

comparisons indicated statistically significant differences in mean impact energies. These

Longitudinal 45-Degree 0.652\* 0.291 1.012

45-Degree Transverse 0.740\* 0.379 1.101

Transverse Default -0.916\* -1.276 -0.555

Inspection indicated that the fracture patterns of the specimens varied as a function of the raster orientations. The longitudinal (0º) and transverse (90º) pieces fractured along a path oriented 90º to the length of the specimen. The fracture surface appeared smooth along the layers of each of the transverse specimens, all of which experienced clean and complete separation of the specimen into two discrete pieces. Weak interfaces parallel to the crack front affected the transverse raster specimens by providing a straightforward path for crack propagation, resulting in the least amount of energy absorption. Failure initiated from the side on the notch where the energy absorption was greatest, thus causing the rasters to fracture longitudinally along the length of the fiber through several layers until the energy adsorption decreased significantly and the sample broken in half (Figure 16a). Upon closer examination, it can be seen that individual fibers plastically deformed and twisted at the ends upon catastrophic failure, whereby halves of the samples separated (Figure 16b).

(a) (b)

The fracture surfaces of the 0° raster specimens were macroscopically rougher, in contrast, displaying slightly jagged edges of individual broken fibers along the transverse fracture

Fig. 16. SEM image of fractured 90° impact specimen magnified (a) 20X and (b) 160X

Table 10. Post hoc Tukey HSD multiple comparisons of mean impact energies

**Difference of Mean Impact Energy (***i-j***)** 

Transverse 1.392\* 1.031 1.753 Default 0.473\* 0.116 0.837

Default -0.175 -0.536 0.184

**95% Confidence Interval Lower Bound Upper** 

**Bound** 

results suggest that impact strength has anisotropic characteristics.

**Raster Orientation (***j***)** 

\* Mean difference is significant at the 0.05 level

**Raster Orientation (***i***)**  plane. Interlayer delamination was also evident and resulted in hinged specimens at failure. Figure 17 displays a 0° specimen where the slightly varied fiber length is evident, along with the formation of a hinge. The longitudinal specimens had the highest mean impact strength, correlating well with the results of three-point bend tests.

Fig. 17. SEM image of impact fractured 0° raster specimen

Weak interfaces running at ±45° to the crack front require relatively high amounts of energy as a result of a mixed fracture mode. As a result, a rougher texture was evident for the fracture surfaces of the default raster specimens, and several hinged as shown on the bottom of Figure 18a. The fracture patter of the 45° specimens, in contrast, fractured more consistently along the 45° maximum shear plane as shown on the top of Figure 18a. The macroscopic roughness of the fracture surface was the result of the varying fiber lengths, evident in the SEM image of Figure 18b.

Fig. 18. (a) Fracture patterns of 45° & +45°/-45° specimens; (b) SEM image of impact fractured 45° raster specimen

Anisotropic Mechanical Properties of ABS Parts Fabricated by Fused Deposition Modelling 177

These results confirm that certain raster orientations have a significant effect on the tension-

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

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

(a) (b)

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

The mechanical properties of ABS specimens fabricated by fused deposition modelling display anisotropic behaviour and are significantly influenced by the orientation of the layered rasters and the resulting directionality of the polymer molecules. The presence of air

in anisotropic behaviour of FDM specimens under tension-fatigue loading.

**4. Conclusion** 

shows the smooth, brittle, tensile failure on each individual raster face.

the pattern indicates the existence of a dynamic transition of failure mode.

fatigue properties of the FDM specimens.
