**3.1 Tension testing**

The tension specimens were thin rectangular slabs made in compliance with ASTM D3039 (ASTM, 1998). Fabrication utilized the FDM T12-nozzle, providing an individual layer or slice height of 0.1778 mm. The specimen thickness of 2.6 mm was subsequently achieved with a total of 15 layers (Figure 4).

Fig. 4. SEM image displaying 15 layer thickness

A summary of the tension test results for the four raster orientations is displayed in Table 2. The mean ultimate and yield strengths (0.2% offset) were largest for the longitudinal (0°) raster orientation, 25.72 and 25.51 MPa respectively, and weakest for the transverse (90°) raster orientation, 14.56 and 14.35 MPa respectively. The mean ultimate strength of the 90° specimens represented only 56.23% of that of the 0° raster specimens, followed by the 45º specimens at 61.45% and the +45º/-45º specimens at 74.09%.


Table 2. Tension test results

A one-way analysis of the variance (ANOVA) was completed in order to consider the equivalence of the population means for the four raster orientations. The results, appearing in Table 3, include a calculated F-test statistic of F(3,16) = 490.98 and a *p*-value of 0.0001, indicating a significant difference between some or all of the mean ultimate strength (UTS) values associated with the four raster orientations at a level of significance of = 0.05. The coefficient of determination associated with this analysis was R2 = 0.9886.


Table 3. ANOVA results comparing mean ultimate tensile strengths of 4 raster orientations

Further analysis in the form of post hoc comparisons was performed to determine which raster orientations differed in mean UTS. Tukey's method (Montgomery, 2009), creating a set of 95% simultaneous confidence intervals for the difference between each pair of means, indicated that the difference was significant for all pairwise comparisons of mean UTS values (Table 4). The difference between the mean UTS of the longitudinal rasters (25.72) and that of the transverse rasters (14.56) was the most significant. These results confirm that raster orientation has a significant effect on the tensile strengths of the FDM specimens. Tensile strength is thus verified to be affected by the directional processing and subsequent directionality of the polymer molecules, signifying an anisotropic property.


\* Mean difference is significant at the 0.05 level

164 Mechanical Engineering

The tension specimens were thin rectangular slabs made in compliance with ASTM D3039 (ASTM, 1998). Fabrication utilized the FDM T12-nozzle, providing an individual layer or slice height of 0.1778 mm. The specimen thickness of 2.6 mm was subsequently achieved

A summary of the tension test results for the four raster orientations is displayed in Table 2. The mean ultimate and yield strengths (0.2% offset) were largest for the longitudinal (0°) raster orientation, 25.72 and 25.51 MPa respectively, and weakest for the transverse (90°) raster orientation, 14.56 and 14.35 MPa respectively. The mean ultimate strength of the 90° specimens represented only 56.23% of that of the 0° raster specimens, followed by the 45º

*Strength* 

(MPa), *Std Dev* (MPa), *Std Dev* (MPa), *Std Dev*

Longitudinal (0°) 25.51, *0.73* 25.72, *0.91* 987.80, *19.98* Diagonal (45°) 15.68, *0.27* 16.22, *0.27* 741.78, *20.28* Transverse (90°) 14.35, *0.08* 14.56, *0.05* 738.77, *7.91* Default (+45°/-45°) 18.90, *0.53* 19.36, *0.39* 768.01, *33.31*

A one-way analysis of the variance (ANOVA) was completed in order to consider the equivalence of the population means for the four raster orientations. The results, appearing in Table 3, include a calculated F-test statistic of F(3,16) = 490.98 and a *p*-value of 0.0001, indicating a significant difference between some or all of the mean ultimate strength (UTS) values associated with the four raster orientations at a level of significance of = 0.05. The

*Mean Effective Modulus* 

**3. Results and discussion** 

with a total of 15 layers (Figure 4).

Fig. 4. SEM image displaying 15 layer thickness

Table 2. Tension test results

specimens at 61.45% and the +45º/-45º specimens at 74.09%.

*Raster Orientation Mean Yield Strength Mean Ultimate* 

coefficient of determination associated with this analysis was R2 = 0.9886.

**3.1 Tension testing** 

Table 4. Post hoc Tukey HSD multiple comparisons of mean tensile strengths

The quantitative data analysis was followed by detailed physical inspection of the specimens at both macro and microscopic levels. Macroscopically, the fracture patterns of the specimens varied somewhat as a function of the raster orientation of the twodimensional layers and the resulting weakest path for crack propagation (Figure 5). The 90º specimens failed in the transverse direction and the 45º specimens failed along the 45º line. The 0º specimens failed primarily in the transverse direction, although there was some fiber pullout and delamination intermittently evident as well. The +45º/-45º specimens broke at intersecting fracture paths along ±45º, resulting in a saw-tooth fracture pattern across the specimen width. It is likely that fracture paths controlled by weak interlayer bonding are affected by the residual stresses that result from the volumetric shrinkage of the polymer layers during solidification and cooling. In addition, interlayer porosity and air gaps serve to reduce the actual load-bearing area across the layers, providing an easy fracture path.

In specimens with the longitudinal (0°) raster orientation, the molecules tend to align along the stress axis direction. This produces the strongest individual two-dimensional layers subjected to tension loading. During the testing of these specimens, stress whitening due to

Anisotropic Mechanical Properties of ABS Parts Fabricated by Fused Deposition Modelling 167

Under microscopic examination, the +45°/-45° specimens displayed multiple failures of individual raster fibers in both shear and tension (figure 8a). Failure occurred by the pulling and eventual rupturing of individual fibers whereby the material separated at a +45°/-45° angle relative to the tensile load, creating the saw-tooth like appearance evident at the macro-scale (figure 5). Failure of the 45° raster specimens was similar to that of the default +45°/-45° in that it was a brittle shear failure on each of the individual fibers at the microscopic level, as each raster was pulled in tension and failed at 45 degrees relative to the loading axis (figure 8b). Macroscopically, the samples also displayed a characteristic shear

(a) (b)

The air gap that forms during fabrication and remains present between fibers of the FDM specimens is a significant factor in considering tensile ultimate and yield strengths and comparing these properties to those of injection molded ABS specimens. Although the FDM machine setting indicated a desired air gap of 0.0 mm, the fiber geometry inherently causes the presence of triangular air voids as seen in the SEM image in Figure 9. These voids influence the effective tensile strengths and effective elastic moduli of the FDM parts by decreasing the physical cross-sectional area of material specimens. This is in part why the

Fig. 8. SEM image of fracture surfaces of (a) +45°/-45° and (b) 45° raster specimen

Fig. 7. SEM image of fracture surface of a 90° raster specimen

failure along the 45° line with the tensile load (figure 5).

Fig. 5. Failure modes of the specimens with each of the four raster orientations

craze formation and growth was observed to develop prior to reaching the yield stress. Failure occurred at whitened areas from plastic deformation where some evidence of localized fiber delamination was observed. The fracture surfaces were further analyzed with a scanning electron microscope, and displayed failure that was predominantly brittle in nature with localized micro-shearing on each fiber face (Figure 6). The tensile strength of these specimens is thus more heavily dependent on the strength of the ABS monofilament than specimens with fibers running at orientations other than 0° with the stress axis (Rodriguez et al., 2001).

Fig. 6. SEM image of fracture surface of a 0° raster specimen

In contrast, the specimens with transverse (90°) raster orientations did not display obvious crazing during testing, and failure occurred predominantly at the weak interface between layered ABS fibers (Figure 7). These specimens experienced brittle interface fracture. Weak interlayer bonding or some amount of interlayer porosity was evident in the failure of many of the specimens with raster orientations other than 0°, and appeared to be the cause of layer delamination along the fiber orientation during loading. The tensile strength of these specimens depended much more heavily upon the fiber-to-fiber fusion and any air gap resulting between the fibers, as opposed to the strength of the fibers themselves.

Fig. 7. SEM image of fracture surface of a 90° raster specimen

Fig. 5. Failure modes of the specimens with each of the four raster orientations

Fig. 6. SEM image of fracture surface of a 0° raster specimen

(Rodriguez et al., 2001).

craze formation and growth was observed to develop prior to reaching the yield stress. Failure occurred at whitened areas from plastic deformation where some evidence of localized fiber delamination was observed. The fracture surfaces were further analyzed with a scanning electron microscope, and displayed failure that was predominantly brittle in nature with localized micro-shearing on each fiber face (Figure 6). The tensile strength of these specimens is thus more heavily dependent on the strength of the ABS monofilament than specimens with fibers running at orientations other than 0° with the stress axis

In contrast, the specimens with transverse (90°) raster orientations did not display obvious crazing during testing, and failure occurred predominantly at the weak interface between layered ABS fibers (Figure 7). These specimens experienced brittle interface fracture. Weak interlayer bonding or some amount of interlayer porosity was evident in the failure of many of the specimens with raster orientations other than 0°, and appeared to be the cause of layer delamination along the fiber orientation during loading. The tensile strength of these specimens depended much more heavily upon the fiber-to-fiber fusion and any air gap

resulting between the fibers, as opposed to the strength of the fibers themselves.

Under microscopic examination, the +45°/-45° specimens displayed multiple failures of individual raster fibers in both shear and tension (figure 8a). Failure occurred by the pulling and eventual rupturing of individual fibers whereby the material separated at a +45°/-45° angle relative to the tensile load, creating the saw-tooth like appearance evident at the macro-scale (figure 5). Failure of the 45° raster specimens was similar to that of the default +45°/-45° in that it was a brittle shear failure on each of the individual fibers at the microscopic level, as each raster was pulled in tension and failed at 45 degrees relative to the loading axis (figure 8b). Macroscopically, the samples also displayed a characteristic shear failure along the 45° line with the tensile load (figure 5).

Fig. 8. SEM image of fracture surfaces of (a) +45°/-45° and (b) 45° raster specimen

The air gap that forms during fabrication and remains present between fibers of the FDM specimens is a significant factor in considering tensile ultimate and yield strengths and comparing these properties to those of injection molded ABS specimens. Although the FDM machine setting indicated a desired air gap of 0.0 mm, the fiber geometry inherently causes the presence of triangular air voids as seen in the SEM image in Figure 9. These voids influence the effective tensile strengths and effective elastic moduli of the FDM parts by decreasing the physical cross-sectional area of material specimens. This is in part why the

Anisotropic Mechanical Properties of ABS Parts Fabricated by Fused Deposition Modelling 169

pieces, yield stresses were analyzed for consistency and to allow for comparisons with that of the injection molded specimens. As a result, a one-way ANOVA was conducted to compare the effect of raster orientation on mean compressive yield strengths in 0º, 45º, 90º, and +45º/-45º conditions. The test indicated that raster orientation had a significant effect on mean compressive yield strength at the *p* < 0.05 level for the four conditions, F(3, 16) = 31.25, p = 0.0001. Post hoc comparisons using the Tukey HSD test indicated that this significance was limited to and specifically in regard to comparisons with the 45º diagonal condition. The mean yield strength for the 45º raster orientation (24.46 MPa) was significantly different (lower) than that of the other three raster orientations (Table 6), while all other paired comparisons indicated statistically insignificant differences in the mean yield strengths.

Inspection of the failed compression specimens provided additional evidence that the 45º raster specimens were significantly weaker in compression than the other raster orientations. The specimens ultimately separated into two or three pieces, following the displacement of the cylinder's top relative to its bottom, as seen in Figure 10. This distortion occurred as a result of the shearing or sliding along the 45º rasters as the specimens was subjected to an axial compressive load. The other three raster orientations displayed less distortion prior to failure and had mean compressive yield strengths that were significantly

> **Difference of Mean Yield Strength (***i-j***)**

Transverse -0.653 -2.279 0.973 Default 0.683 -0.943 2.309

Default -3.682\* -5.308 -2.056

Longitudinal 45-Degree 4.365\* 2.739 5.991

45-Degree Transverse -5.018\* -6.644 -3.392

Transverse Default 1.336 -0.290 2.962

Table 6. Post hoc Tukey HSD multiple comparisons of mean yield compressive strengths

Fig. 10. Photos of failed 45º raster specimens under compression loading

**95% Confidence Interval** 

**Lower Bound Upper Bound** 

larger than that of the 45º raster specimens.

\* Mean difference is significant at the 0.05 level

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

**Raster Orientation (***i***)** 

injection molded specimens displayed tensile strengths greater than that of any of the FDM parts, achieving a mean yield and ultimate tensile strength of 26.95 and 27.12 MPa respectively. The mean UTS achieved by the 0° raster specimens was closest to that of the injection molded specimens, representing 94.8% of its value.

Fig. 9. SEM image of air voids seen on fractured 0º raster specimen
