LVL PSL N
6.3.2 Bending MOE of LVL and PSL
Table 6.11 provides the statistical summary of the measured flatwise and edgewise bending MOE of LVL and PSL. Pairwise t-tests failed to show statistically significant differences between the flatwise and edgewise MOE, for either of the composite types. Another t-test indicated a statistically significant difference between the MOE of LVL and PSL. Analysis of variance on LVL’s bending MOE provided the within- and between panel variance for LVL, which was 0.634 GPa2 and 0.456 GPa2, respectively (for the calculations see APPENDIX B.)
As Table 6.11 shows, LVL has a higher bending MOE than does PSL, in spite of the higher average densification of the latter. This may be due to several factors, such as the imperfect orientation of the strands, strand damage during hot pressing, or weaker gluelines in PSL, caused by its less regular structure.
Figures 6.17 a and b show the relationship between the edgewise and flatwise bending MOE of LVL and PSL, respectively. As the low r2 values indicate, the correlation between the two parameters is relatively weak. This fact can be explained by the laminate theory. As the orientation of a beam changes, the distance of some
Table 6.11 – Flatwise and edgewise bending MOE of LVL and PSL (N=20)
LVL PSL
Eedge (GPa)
Eflat (GPa)
Eedge (GPa)
Eflat (GPa)
Mean 13.22 13.36 12.82 12.57
s 1.21 0.72 0.75 0.57
Min 11.85 12.20 11.18 11.64
Max 15.90 15.16 13.81 13.92
Eflat (GPa)
11 12 13 14 15 16
E edge(GPa)
11 12 13 14 15 16
E (GPa)
11 12 13 14 15
E edge(GPa)
11 12 13 14 15
Eedge = 0.551 Eflat - 5.901 r2 = 0.171
a.
b.
Eedge = 0.990 Eflat - 0.002 r2 = 0.342
constituents from the neutral plane changes, too, therefore their importance in determining the beam’s overall bending MOE increases or diminishes. In some beams, stronger components are more deterministic in edgewise application, while weaker strands gain importance in flatwise orientation, or vice versa. It is probable that the slope of the regression line in PSL, which differs from unity considerably, is a result of this weak association, rather than a reflection of physical reality.
6.3.3 Orthotropy of shear strength
Table 6.12 contains the summary statistics of the measured shear strength of LVL and PSL at every combination of load and strand/layer orientation that the study incorporated. The examined composites exhibited similar tendencies of out-of-sheared-plane failure, as did solid wood. Shear strength results reported in Table 6.12 should, therefore, be treated as apparent values.
Because the test did not include every angle combination, general tendencies are hard to observe in the data. Shear strength values typically decreased with increasing load orientation (ϕ’). When load was applied in the longitudinal (x) direction, peak shear strength was experienced at 75° strand or layer orientation (θ’), in both LVL and PSL.
This peak is especially pronounced in LVL. A similar tendency is observable in solid yellow-poplar timber (see Table 6.2 and Figure 6.3 a.)
Statistical data analysis included two-way ANOVA for the two composite types.
Results indicated significant load and layer/strand orientation effect in both materials.
The incomplete statistical design did not allow the assessment of the interaction effect.
Table 6.12 – Summary statistics of the experimentally determined orthotropic shear strength of LVL and PSL
The effectiveness of the three prediction models presented in section 4.1.1, was evaluated, using r2 analysis (Equation 6.1.) Table 6.13 shows the r2 values associated with the three models. This table also indicates the n values used in the Modified Hankinson formula (Equation 4.7), which were established by curve fitting, using the whole experimental database. Figure 6.18 shows the prediction of the orthotropic tensor theory and the modified Hankinson formula for both composite types in a three-dimensional Cartesian coordinate system, along with the experimental mean values, to facilitate visual appraisal.
The results show that the orthotropic tensor theory and the modified Hankinson formula are suitable for describing the orthotropy of wood based composite lumber. The latter fits experimental values, because it can predict high shear strength values at ϕ = 15°, which occur in LVL and PSL, as well as solid wood. Both model provided relatively low r2 values, when predicting the shear strength of LVL. Figure 6.18 reveals the reason for this; the models were unable to estimate the high shear strength values experienced in the longitudinal direction, at θ’=75°. Apart from this point, both models provide excellent fit to the experimental shear strength of LVL.
Table 6.13 – Coefficients of determination provided by the various prediction models Modified Hankinson
formula Species Orthotropic
tensor theory r2
Quadratic formula
r2 n r2
LVL 0.68 0.66 2.17 0.69
PSL 0.59 0.55 2.74 0.79
n ! the power in the Modified Hankinson formula
2 Load orientation (
ϕ
a.
- LVL - Orthotropic tensor theoryb.
- LVL - Modified Hankinson formula1 Load orientation (
ϕ Load orientation (
ϕ Load orientation (
ϕ ϕ' o)
a.
- PSL - Orthotropic tensor theoryb.
- PSL - Modified Hankinson formulaPredicted shear strength Experimentally measured values
The predictions of the Quadratic formula were, again, inferior to those of the other two models. It appears that, while this formula might be best to describe the theoretical shear strength of wood and composite lumber, the other theories approach apparent shear strength values, measured by traditional methods, better.
6.3.4 Orthotropy of compression elasticity
In the duration of the compression tests (up to 0.1 in displacement), very few specimens failed or reached the stress plateau indicated in Figure 6.4 b. For this reason, compression strength is unavailable for the composites. Specimens with sloping grain that had been re-glued to provide the necessary specimen length, did not show apparent signs of shear dislocation along the glueline. Specimens that did reach failure were sheared inside the composites, rather than along this interface. This lead to the conclusion that displacement data was not significantly influenced by the practice of re-gluing, and that collected data should produce valid MOE values.
Table 6.14 provides the summary statistics of the compression MOE of LVL and PSL in the six measured directions. Similarly to the elastic properties of solid wood, composite MOE dropped considerably between ϕ’ = 0° and ϕ’ = 45°, but showed little change as load orientation increased further. Moreover, measured data indicates some negative correlation between layer/strand orientation (θ’) and compression MOE at ϕ’ = 45° and ϕ’ = 90°. The low number of measurement points, however, does not warrant broad conclusions based on this observation.
Table 6.14 - Summary statistics of the compression MOE of the two composites
The ANOVA procedure could not be directly applied to composite MOE data, because of the severe inequality of variances. A logarithmic transformation made the variance of the different groups more homogeneous. Two-way Analysis of Variance, executed using the transformed data, showed that variations both in load and in strand/layer orientation caused statistically significant differences in the compression MOE of LVL and PSL (see APPENDIX B.)
The compression MOE of LVL and PSL in the longitudinal (x) direction is significantly higher than that of their raw material (yellow-poplar.) The slightly better performance of LVL was expected, because of some densification present in this composite. In PSL, strands are not aligned perfectly in the longitudinal direction, which reduces the MOE value, but the high level of densification improves the MOE significantly. The particular composite used in this study also includes 25% southern yellow pine material (see section 5.3.1), which is likely to have improved the MOE
In the other directions, the composites’ MOE values are either comparable to or worse than those of yellow-poplar. This was unexpected, since the same tendencies outlined above should lead to higher MOE values in these directions.
6.4 Composite geometry