Visual grading was developed to ensure the quality of commercialized lumber. The Northeastern Lumber Association Manufacturers (NELMA 2013) grading rules are the standard methods used for yellow-poplar (Liriodendron tulipifera) lumber when graded for structural purposes according to a national design specification (American Wood Council 2018). Mohamadzadeh and Hindman (2015) tested two yellow-poplar cross-laminated timber (CLT) panel layouts with the same National Hardwood Lumber Association (NHLA) visual grade (No. 2 Common) but different defect presence to find higher bending properties in the panels produced with boards with fewer defects, proving the importance of defect sorting. Another way to evaluate lumber is using nondestructive evaluation (NDE) techniques, including mechanical tests (e.g., proof loading) that do not damage the lumber. In particular, NDE via proof loading allows calculating a board's static modulus of elasticity (MOEs). Visual classification of the board can be a good predictor of wood properties, and is cost effective. However, when designing tall wood structures such as CLTs, NDEs can provide additional insurance to the architectural designs. When comparing visual grading and machine rating the lumber, machine stress rating identifies and quantifies a direct measurement of mechanical properties (Green et al. 1993).
Based on data from Azambuja et al. (2022), the distribution of MOEs from a population of low-grade yellow-poplar showed the potential of increasing panel mechanical properties, such as bending stiffness (MOE), by sorting boards based on NDE; this methodology was one of the paper’s final recommendations. Sorting boards by their MOE in outer layers is common in glue-laminated (glulam) beams. Janowiak et al. (1997) sorted boards in the making of red maple (Acer rubrum) glulam beams and were able to improve the bending strength and MOE of the beams by dividing the beam into the outer and core sections and placing stress-rated boards with higher MOE and defect size in the outer section of the beam. Sorting based on NDE to ensure the mechanical properties of composites was tested by Cunha and Matos (2011). The authors tested glulam beams comprising two groups: randomly selected and boards sorted by dynamic MOE (MOEd) as determined by NDE using a Metriguard stress wave timer. The research results concluded that the bending properties differed statistically between groups, suggesting the efficiency of MOEd sorting of the boards. Moody et al. (1993) produced yellow-poplar glulam beams by selectively placing boards based on their MOEs in the outer layers of the beams. These results suggest that selecting and sorting yellow-poplar boards by their MOE improved the sorting methods of glulam beams. Hernandez et al. (1997) added glass-fiber plastic to reinforce yellow-poplar glulam beams. They found that layup reinforced in the two bottom layers increased the bending stiffness, while reinforcement in the two outer layers increased bending strength. Similar sorting could also be used to increase the mechanical properties of CLT panels.
Therefore, this research aimed to produce and test CLT panels produced using solely low-grade yellow-poplar (NHLA 2A and below, and NELMA Below Grade [B.G.]) sorted based on NDE.
Material and Methods
A summary of the methodology is presented in Figure 1. The boards used in this study were from a population of yellow-poplar graded NHLA No. 2A, No. 2B, No. 3A, and No. 3B, based on NHLA (2014). Additionally, the boards used for the panel production were graded NELMA B.G., meaning boards that did not achieve any visual structural grade according to NELMA rules. These B.G. boards were sorted based on their MOEs obtained by nondestructive proof-loading evaluations performed previously. The five-layer panels were made from boards that had MOEs above 1.65 by 106 psi in the two outside layers, and boards with MOEs between 1.2 by 106and 1.65 by 106 psi (8,273 and 11,376 MPa) were used in the three inner layers of the panel. This range of MOEs values was selected because the 1.65 by 106 value was about the top 40 percent of the B.G. boards, and 1.2 by 106 psi was the minimal stiffness requirement of the American National Standards Institutes/The Engineered Wood Association (ANSI/APA) PRG 320-2019 (ANSI/APA 2020). The choice of using the top 40 percent was made to study the possibility of using all the boards available in the population, except for boards with MOEs below 1.2 by 106 psi, to produce five-layer CLT panels. This research CLT panel set composition was labeled Y.P. and is presented in Figure 1 (4).
Citation: Forest Products Journal 74, 4; 10.13073/FPJ-D-24-00029
The panel production started by surfacing two wide sides of the selected boards, which were laid on an assembly table for adhesive application prior to pressing. The adhesive used was Franklin Advantage EP-950, a two-part adhesive (acrylic-based emulsion polymer isocyanate system, EPI and H-200, a diphenylmethane diisocyanate, MDI, hardener). Details of the parameters used are presented in Table 1.
Ten repetitions of five-layer panels were produced with dimensions of 3.75 inches deep, 18 inches wide, and 120 inches long (95.25 mm by 457.2 mm by 3,048 mm). From each panel, specimens were prepared and tested, including one flatwise bending following American Society for Testing and Materials (ASTM) D198 (ASTM 2015), three shear blocks following ASTM D905 (ASTM 2010), and three cyclic delamination following ASTM D1101 (ASTM 2013). The bond evaluation specimens were taken from three positions (both ends and center) to assess the panel production, hence the three specimens from each panel. Additionally, the theoretical effective bending stiffness (EIeff) and the effective bending strength (FbSeff) of the panels were calculated based on formulas published in the CLT Handbook (Karacabeyli and Gagnon 2019). Additionally, to evaluate the effects of sorting and the panel composition the results were compared to Azambuja et al. (2023), since the research used similar methods, only differing in the boards’ structural visual grade. The author used boards graded No. 2 and No. 3 in accordance to NELMA visual grades. Finally, the software used for data management was Microsoft Excel 365, and statistical analyses were conducted using RStudio (version 3.6.3).
This study was a follow-up from Azambuja et al. (2023), and additional information regarding the methodology can be found in Azambuja (2022) and Azambuja et al. (2022).
Results and Discussion
Bending results
Table 2 presents the results from the third point flatwise bending in the major direction. The table shows the results of the ultimate load, the Fb, the MOE, and the failure modes from testing.
The average MOE and Fb results of Y.P. were higher than Azambuja et al. (2023) reported, specifically 12 percent and 17 percent higher for average MOE and average Fb, respectively. The differences between the two layups were their board's MOEs and NELMA grades, as Azambuja et al. (2023) utilized No 2 and No 3 NELMA-graded lumber. Based solely on the NELMA grade, CLT panels using B.G. lumber would generally be expected to be lower in strength and stiffness than those using No. 2 and No. 3 lumber. However, these results were not the case for this study. The ability of B.G. lumber to produce CLT panels with higher strength and stiffness was due to the defect presented in this population and the composite panel configuration. Azambuja et al. (2022) reported the defects found in low-grade yellow-poplar (NHLA 2A, 2B, 3A, and 3B) and reported that the most common defect was knots, followed by splits. However, the distribution of defects from B.G. lumber consisted of 43 percent of splits, 24 percent of knots, 11 percent of shake, 10 percent of wane, and 12 percent of other defects. Specifically, for these 10 panels and considering the cut layout presented in Figure 1, 40 boards composed the two outer layers of the set of panels, and their defect distribution was 50 percent splits, 20 percent knots, 20 percent shakes, 5 percent wane, and 5 percent slope of grain. This defect characterization could explain the mechanical differences between these layups. Based on these results, it is interesting to review structural visual grades when boards are used to produce CLT panels since boards with splits as their limiting defect could be used in CLT production.
A comparison between the bending results coefficient of variances of this research and Azambuja et al. (2023) is presented in Figures 2 and 3. The values ranged approximately from 3,000 psi (20.7 MPa) and 9,000 psi (62 MPa) in bending strength and 1.2 by 106 psi (8,274 MPa) and 1.9 by 106 psi (13,100 MPa) in bending stiffness. The comparison shows a higher variance in strength and a lower variance in stiffness compared to the other research. This variance could be explained by the differences in board selection between the two studies, nondestructive testing and visual structural grade.
Citation: Forest Products Journal 74, 4; 10.13073/FPJ-D-24-00029
Citation: Forest Products Journal 74, 4; 10.13073/FPJ-D-24-00029
The theoretical effective bending stiffness and bending strength and experimental results are shown in Table 3. The theoretical bending strength was not calculated as there are no published Fb values for NELMA B.G. lumber. The difference between calculated and experimental values is due to, among other reasons, the safety coefficients that are applied to the theoretical values and theoretical calculations, minimizing the effect of minor direction forces.
Table 4 shows the calculated characteristic value results. Results from Azambuja et al. (2023) and ANSI/APA PRG 320-2019 (2020) are compared. The data comparison showed that values from the current layup (Y.P.) were 19 percent greater than the highest Fb value in ANSI/APA PRG 320-2019 (2020). The MOE values from the Y.P. panels were only below the layups E1 and E4.
Mohamadzadeh and Hindman (2015) found that the presence of the defect would decrease the bending strength of panels from the same NHLA visual grade. Although considering that Azambuja et al. (2023) had better visual grades than the tested in this research, this indicates that sorting by MOEs can override the defects differences between NELMA No. 2 and B.G.
Moody et al. (1993) used a similar sorting system to produce yellow-poplar glulam beams with outermost layers with a MOE of 2.0 by 106 psi. This value of MOE is above the values used in this research, which is 1.65 by 106 psi (11,376 MPa), indicating the potential for improving the yellow-poplar panels' composition of this research.
In this current study, B.G. lumber with high MOE values was placed in the outer layers. This sorting method likely resulted in CLT panels with higher strength and stiffness. The results are similar to those of Hernandez et al. (1997), who reported that increasing the resistance (via fiber-reinforcement) of the outer layers of glue-laminated, yellow-poplar beams improved bending strength and stiffness, showing the potential of reinforcing outer panel layers. Finally, the NDE showed more potential for use on grading boards than structural visual grading.
Bonding evaluations
Table 5 shows the results from cyclic delamination of the 10 panels. Seven out of the 10 panels presented delamination above 5 percent. Delamination above 5 percent was more frequently found in the outer areas of the panel. Ten out of 30 tested samples showed delamination above 5 percent, with one specimen from position 2, at the center of the panel. The average delamination of the positions was 6.4, 1.5, and 6.8 percent for positions 1, 2, and 3, respectively. The analysis of variance test showed no statistical difference between the positions 1, 2, and 3 (P = 0.14).
As expected, similar results were found by Azambuja et al. (2023), given both studies used the same methods and equipment. Some panels presented delamination into the ends, probably due to uneven pressure during the pressing of the panels. With the same delamination results and manufacturing parameters, such as the adhesive application, spread rate, and nominal pressure, it is safe to affirm that the bonding of the two sets of panels was similar. Therefore, there is no evidence that the change of board visual grade due to a defect affects the bonding delamination of the panel according to bond quality tests from ANSI/APA PRG 320-2019 (2020).
Table 6 shows the results of the shear block tests. The average percentage of wood failure in all the samples was 97 percent, above the standard requirement of 80 percent, and 95 percent of the samples presented wood failure of at least 74.5 percent, above the standard requirement of 60 percent. These results indicated that the glue bond quality based on shear block evaluation was satisfactory, contrary to the cyclic delamination evaluation.
The bonding parameters were based on product specifications and preliminary small-scale tests. In the preliminary samples, the results meet ANSI/APA PRG 320-2019 (2020) requirements. Even using the same parameters, the results of the full-scale panels did not meet the requirements, which can be attributed to laboratory limitations, which could be resolved in industrial production.
Summary and Conclusion
In this study, yellow-poplar boards graded NELMA B.G., mainly due to splits and boards not rated for minimum structural use requirements, were used to produce CLT panels. The boards that composed the panels were selected by their static MOE, and those with higher values (top 40%) within the population were placed in the outer two layers of the CLT panels.
The results of cyclic delamination showed delamination above the 5 percent requirement from ANSI/APA PRG 320-2019 (2020) in 10 out of 30 samples, mainly in the outer areas of the panel (9 out of 10 samples with delamination over 5%). The shear-block tests showed a result of an average wood failure of 97 percent in all the samples and a fifth percentile of 74.5 percent. These results highlight the stringency parameters in panel production and the severity of bond line evaluation of cyclic delamination and shear block. No bending specimen presented bonding failure, eliminating the possible influence of the bonding issues in bending results. The results of flatwise bending in the major direction showed an average of Fb of 6,682 psi (46.07 MPa) and MOE of 1.56 by106 psi (10,756 MPa) for the 10 CLT panels tested. The calculated Allowable Stress Design (ASD) reference design values results indicated that sorting the boards according to MOEs with outer layers with at least 1.65 by 106 psi (11,376 MPa)and core of at least 1.2 by 106 psi (8,273 MPa) can produce CLT panels that exceed listed Fb ASD values from all stress-rated layups (E) and MOE only lower than E1 and E4 from ANSI/APA PRG 320-2019 (2020).
The results from this study indicate that NELMA B.G. lumber downgraded due mainly to end splits has the potential to be used in the production of CLT panels if sorted by nondestructive MOEs. Specifically, the selection of boards based on nondestructive tests appeared to be more feasible than solely using NELMA visual grades.
Based on these findings, NDE and board sorting can allow material that would otherwise be rejected to be used in CLT production. This option is viable because of this type of panel configuration, where adjacent layers can minimize the board's individual defects. However, the majority of the B.G. lumber used in this research consisted of end splits as the limiting factor. More research is needed to see if similar findings occur when the limiting defect of B.G. lumber is another defect type (e.g., knots).