Lumber preparation and panel manufacturing

CLT testing specimens were manufactured from 25 mm thick, 178 mm wide, and 3.1 m long surfaced lumber made from yellow poplar (Liriodendron tulipifera), red oak (Quercus rubra), and red maple (Acer rubrum) hardwood species were acquired from the northeast West Virginia Appalachian region. All boards were graded as No. 2B and No. 3B Common according to the National Hardwood Lumber Association (NHLA 2014) grading rules. All lumber was also subjected to visual strength regrading based on the “Standard Grading Rules for Northeastern Lumber” published by the Northeastern Lumber Manufacturers Association (NELMA 2024.) to select the longitudinal and transverse layers of the CLT panel. For the longitudinal layers, No. 2 graded lumber was selected, although No. 3 graded lumber was used for the transverse layers. The visually strength-graded lumber was conditioned for three months in the conditioning room to achieve a uniform moisture content of 12 ± 1 percent. After conditioning, the dynamic modulus of elasticity (MOE_d) was measured in flatwise orientation using a transverse vibration equipment (Metriguard Model 340 Transverse Vibration E-Computer). The average properties of the laminae are given in Table 1.

Table 1.Density and dynamic flexural properties of the laminae.

View Fullscreen

Before manufacturing the three-layer CLT panels, the lumber was surfaced on all four sides to a net dimension of 19 mm × 153 mm (T×W, thickness and width, respectively). For CLT fabrication, 1-component polyurethane (1C-PUR) (Henkel LOCTITE HB X602 PURBOND) with its corresponding primer (LOCTITE PR 3105 PURBOND), and phenol-resorcinol formaldehyde (PRF) (Dynea Aerodux 185+Hardener HRP.150) adhesives were used. In total, eight 57 mm thick, 915 mm wide, and 3,050 mm long panels were manufactured. Table 2 shows the eight panel configurations and the manufacturing parameters. Each type of CLT panel was cut into three beams, each measuring 300 mm wide and 1,722 mm long, which were then subjected to bending tests. The beams were subsequently shortened from 1,722 mm to 1,257 mm, then to 869 mm, then to 586 mm, and finally to 297 mm, according to the specific span-to-depth ratios described in the static bending test method section. The length reduction was applied symmetrically at both ends of the specimens. In all cases of the hybrid configurations, the yellow poplar laminates were used in the middle (transverse) layer.

Table 2.CLT panel configurations and manufacturing parameters.

View Fullscreen

Measurement of bending and shear properties of CLT panels using the multiple regression method according to the ASTM D198 procedures

Using the apparent modulus of elasticity (E_app) values from the variable span bending tests on a timber beam, the shear modulus (G) and the effective modulus of elasticity (E_eff) can be obtained through linear regression analysis. Although the CLT has an orthotropic composite structure, the method can measure only the global effective shear modulus (G_eff). The elastic deflection of a prismatic shear modulus specimen under a single-center point load is: (1) $$\Delta =\hspace{0.17em}\frac{P{l}^3}{48E{I}_{\mathit{app}}}\hspace{0.17em}=\hspace{0.17em}\frac{P{l}^3}{48E{I}_{\mathit{eff}}}\hspace{0.17em}+\hspace{0.17em}\frac{Pl}{4\kappa G{A}_{\mathit{eff}}}$$where EI_app (N-mm²/m of width) is the apparent bending stiffness, EI_eff (N-mm²/m of width) is the effective bending stiffness, GA_eff (N/m of width) is effective shear stiffness, P is the concentrated load applied at midspan, $l$ is the span (mm), and $\kappa$ is the shear coefficient equal to 5/6 for a rectangular section.

Removing like terms and solving for the inverse apparent bending stiffness and inverse effective bending stiffness produces Eq. 2: (2) $$\frac{1}{E{I}_{\mathit{app}}}\hspace{0.17em}=\hspace{0.17em}\frac{1}{E{I}_{\mathit{eff}}}\hspace{0.17em}+\hspace{0.17em}\frac{1}{G{A}_{\mathit{eff}}}\left(\frac{12}{\kappa l^2}\right)$$Eq. 2. can be expressed as a linear function by substituting $y=1/E{I}_{\mathit{app}}$ and $x=12/\kappa l^2$ equations of a line, $y=mx+b$. The slope of the line is the reciprocal of the effective shear stiffness (GA_eff), and $y$-intercept is the reciprocal of the effective bending stiffness (EI_eff), which is illustrated in Figure 1.

View Figure

• Download as Powerpoint
• Download Figure

When a single-center point load is applied, the apparent modulus of elasticity can be computed using the following: (3) $$MO{E}_{\mathit{app},c}\hspace{0.17em}=\hspace{0.17em}\frac{P{l}^3}{48I\Delta }$$By applying a four-point bending condition, the apparent modulus of elasticity can be computed using the following formula: (4) $$MO{E}_{\mathit{app},f}=\hspace{0.17em}\frac{23P{l}^3}{108b{d}^3\Delta }$$where P and Δ correspond to the load/deflection (N/mm) pair in elastic zone, $l$ is the length of the span (mm) and I is the moment of inertia (mm⁴).

The overall effective shear modulus (G_eff) of the panel can be expressed as (5) $$G_{\mathit{eff}}=\hspace{0.17em}{G}_{\mathit{eff}}/A$$where A is the gross cross-sectional area subjected to shear forces (mm²).

Due to the orthogonal structure of CLT, the shear strength is not evenly distributed among the layers. The equivalent shear strength can be calculated as (Ma et al. 2021b): (6) $$f_{s,eq}=\frac{P_{\mathit{max}}}{2·A}$$where P_max is the peak of the force, A is the gross cross-sectional area subjected to shear forces (mm²).

Prediction of stiffness properties of CLT based on Shear Analogy method

The method has been adopted by the ANSI/PRG 320 (APA 2019) product standard, which provides the most precise analytical design values for CLT (Karacabeyli and Gagnon 2019). The value of EI_eff for the shear analogy method is predicted as (7) $$E{I}_{\mathit{eff}}={\displaystyle \sum }_{i=1}^n{E}_i{b}_i\frac{h_i^3}{12}\hspace{0.17em}\hspace{0.17em}+\hspace{0.17em}{\displaystyle \sum }_{i=1}^n{E}_i{b}_i{h}_i{z}_i^2$$where E_i is the modulus of elasticity of the laminate in the ith layer (MPa), b_i is the CLT width in the CLT major strength direction (mm), h_i is the thickness of laminate in the ith layer, z_i is the distance between the center point of the ith layer and the neutral strength direction, and n is the number of layers in the CLT.

The effective shear stiffness GA_eff for the major strength direction can be calculated as (8) $$G{A}_{\mathit{eff}}=\frac{a^2}{\left[\left(\frac{h_1}{2G_1{b}_0}\right)\hspace{0.17em}+\hspace{0.17em}\left({\sum }_{i=2}^{n-1}\frac{h_i}{G_i{b}_0}\right)\hspace{0.17em}+\hspace{0.17em}\left(\frac{h_n}{2G_n{b}_0}\right)\right]}$$where (9) $$a={\displaystyle \sum _{i=1}^{n}h-}\frac{h_1}{2}\hspace{0.17em}-\hspace{0.17em}\frac{h_n}{2}$$where

G_i is the shear modulus of the laminate in the ith layer (MPa), G₁ is the shear modulus of the first layer of CLT (MPa) G_n is the shear modulus of the laminate in the nth layer (MPa).

The dynamic modulus of elasticity (MOE_d) of the lumber was used to calculate the design values. Regarding the orientation of the lumber in the panel layup, MOE_d was applied for the longitudinal (parallel) major direction, and MOE_d/30 for the perpendicular (transverse) layers. The shear stiffness (G) of the lumber in the parallel layer was assumed to be MOE_d/16, whereas in the perpendicular layer it was assumed to be MOE_d/16/10 (Cherry et al. 2019).

Static bending tests of CLTs

Single-center point and four-point bending tests were used to determine the apparent modulus of elasticity (MOE_app) and the shear strength (f_s) in the major direction of the CLT according to ASTM D198 (ASTM 2015). The load direction for each method was flatwise and four different span-to-depth ratios were applied. The testing details of the four beam lengths are shown in Table 3.

Table 3.Bending test setup parameters.

View Fullscreen

Three span-to-depth ratios (22:1, 15:1, 10:1) were applied to determine the MOE_app for the regression model, and in the case of the last setup (5:1 span-depth ratio) the test was carried out until failure. After each stiffness measurement on a certain span, the specimens were cut to the shortened span plus a 6 mm overhang on each side. The overhang beyond the span was minimized, as the shear capacity may be influenced by the length of the overhang (ASTM D198) (ASTM 2015). From each type of CLT beam, three specimens were subjected to stiffness measurements using four-point and center-point bending setups. The loading and support conditions of CLTs is shown in Figure 2.

View Figure

• Download as Powerpoint
• Download Figure

An Instron universal testing machine with an integrated load cell (300 kN, 1% sensitivity) was used for all tests. The deflection of the neutral axis at midspan was measured using a linear voltage differential transducer (LVDT) (±51 mm, 0.25% sensitivity). To calculate the apparent modulus of elasticity (MOE_app), the P and Δ pairs were recorded between 20 percent and 40 percent deflection in the linear elastic zone. Figure 3 shows the experimental setup for shear test of the hybrid red oak (ROYP) CLT.

View Figure

• Download as Powerpoint
• Download Figure

Finite element model setup

The CLT configurations and experimental setup described in the previous sections were utilized as a basis for creating a finite element (FE) model of the CLT panels subjected to center-loading, as shown in Figure 2. The FE models focused on three spans, L = 574, 857, and 1245 mm, subjected to small prescribed central displacements applied to determine the stiffness in the initial nearly linear response. The models were created with the Simulia/Abaqus simulation suite (Dassault Systemes Simulia Corp. 2023) and included the following characteristics:

1. Parts and assembly. The CLT model had two top and bottom longitudinal panels and a series of transversal panels whose number varied depending on the span length. Figure 4(a) illustrates the configuration corresponding to L = 1245 mm.

2. Interactions. The top and bottom longitudinal panels are linked to transverse panels via a cohesive surface interaction that simulates the presence of the structural adhesive. The center line of the top longitudinal panels is coupled to a reference point (RP) created to simulate the transversal center line of load application. These interactions are illustrated in Figure 4(b).

3. Load and boundary conditions. The CLT panel is assumed to have pinned boundary conditions on one end and roller boundary conditions on the other end. A concentrated downward displacement is applied to the RP and drives the deflection of the beam. The magnitude of the prescribed displacement varies depending on the span length, and the values adopted for the simulations are summarized in Table 3. The position of the prescribed displacement and boundary conditions are shown in Figure 4(b).

4. Elements and mesh. All the parts are discretized using continuous three-dimensional, 8-node, linear hexahedral elements (C3D8R) with reduced integration and hourglass control. The resulting meshed configuration for a CLT beam with L = 1245 mm is shown in Figure 4(c). A control node at the center of the bottom longitudinal panel was used to monitor the deflection produced by the prescribed displacements on the different span lengths.

5. Material properties. The layers of the CLT beam are considered orthotropic materials, and their material parameters are defined for longitudinal and transversal directions. The material orientations corresponding to each panel direction are shown in Figure 4(a). The orthotropic engineering constants corresponding to each type of wood adopted in this study were extracted from the Wood Handbook (Forest Products Laboratory, 1999) and summarized in Table 4. Note that values G₂₃ for red oak and red maple are not reported in the Wood Handbook (Forest Products Laboratory, 1999). Average Geff's values, obtained from the regression analysis, were adopted instead. Moreover, the surface bonding interactions are based on the Cohesive Zone Model (CZM) available in Abaqus (Dassault Systemes Simulia Corp. 2023), in which the cohesive parameters were extracted from Tran et al. (Tran et al. 2014, Tran et al. 2015) and included the following values: Initial elastic behavior in Mode I is K_n = 4.5 MPa/mm, and for Mode II is K_t = 30 MPa/mm.

6. Analyses. The Abaqus/Standard solution solver, including the option to capture potential geometric nonlinearities, was implemented to calculate the response of the different CLT beam configurations. The total reaction force resulting from an imposed vertical displacement applied at the RP was extracted as the primary output for the analyses.

View Figure

• Download as Powerpoint
• Download Figure

Table 4.Material properties calculated from values reported by the Wood Handbook (Forest Products Laboratory, 1999).

View Fullscreen

Statistical analysis

All statistical analyses were conducted using TIBCO® Data Science/Statistica 13 software. A factorial analysis of variance (ANOVA) was performed to assess the significance of factors affecting effective bending and shear stiffness, with the significance level set at α = 0.05. Before ANOVA, the assumptions of homogeneity of variances and normality were evaluated using Bartlett's Chi-Square test and a normal plot of the raw residuals, respectively.

Bending modulus and stiffness

Figure 5 shows the results of linear regressions of the apparent bending stiffness (EI_app) obtained from different span bending tests for the homogeneous red oak (HRO) type of CLT glued with PRF resin. All eight types of CLT samples were similarly analyzed to obtain the regression line equations and the corresponding coefficients of determination (r²) indicated on the graph. The average r² value of all 24 regressions was 0.978, ranging from 0.888 to 1.0. This average coefficient value is slightly higher than the r² value of 0.926 reported by Hindman and Bouldin, who used a similar multiple regression method (Hindman and Bouldin 2019).

View Figure

• Download as Powerpoint
• Download Figure

Table 5 summarizes the flexural properties of the different types of CLT panels, based on the regression method and the shear analogy method. The coefficients of variation (COV) for the different types of CLTS ranged from 5.9 percent to 15.6 percent. This variability can be influenced by the variability in the actual Modulus of Elasticity (MOE) of the raw material, the performance of the adhesive, and the presence of local defects in the lumber. Since the CLTs were manufactured from No. 2 and No. 3 structural grade lumber, which can contain several defects such as knots, these imperfections could have a more significant impact, especially in the case of short-span bending tests. However, because of the composite structure, the CLT is less sensitive to the defects’ effect than the solid timber.

Table 5.Flexural properties of the eight types of CLT panels.

View Fullscreen

In all CLT configurations, the average bending stiffness determined by tests (EI_eff,reg) exceeded the theoretical values predicted by the shear analogy method(EI_eff,an). The average difference between the regression EI_eff,reg value of the specimens and the computed EI_eff,an from the shear analogy method was 8.9 percent, with a minimum of 4 percent and a maximum of 13.7 percent (see Table 5). The differences between the tested and predicted values in previous studies are highly inconsistent; therefore, the average difference of 8.9 percent can be considered a moderate value. For instance, Sikora et al. reported a 27 percent higher test value for three-layer CLT manufactured from 20 mm thick Irish Sitka spruce (Sikora et al. 2016). In the case of the 105 mm thick, three-layer red maple and white ash CLT, the shear analogy method underestimated the experimental apparent EI values by 24.1 percent and 23.9 percent, respectively (Crovella et al. 2019). On the contrary, Hematabadi et al. found that the average EI_eff value calculated by the regression method was 9 percent less than that obtained using the shear analogy method for three-layer CLT produced from 20 mm thick poplar lumber (Hematabadi et al. 2020). In addition to Table 5, the average bending stiffness EI_eff values of the specimens, categorized by method and applied adhesive, are shown in Figure 6.

View Figure

• Download as Powerpoint
• Download Figure

As shown in Figure 6, the types of adhesives applied did not produce a statistically significant effect on the average EI_eff values across all types of CLT. However, homogenous specimens (HRM and HRO) bonded with PUR resin exhibited slightly higher stiffness than those bonded with PRF resin. The average difference in the homogeneous red maple specimens was 7.6 percent, and in the homogeneous red oak samples, it was 4.1 percent. In the case of hybrid samples (RMYP and ROYP), an inverse trend was observed: PUR-bonded samples demonstrated lower average EI_eff values. Specifically, hybrid red maple-yellow poplar (RMYP) samples bonded with PRF showed a 5.6 percent higher average stiffness, and red oak-yellow poplar hybrids (ROYP) demonstrated a 9.1 percent increase in stiffness. Given that the variation between adhesive types remains below 10 percent, all data were consolidated into one single dataset for further analysis.

Because shear deflection plays a major role in the total deflection (bending + shear) of short-span supported CLTs, it should be calculated to ensure accurate estimates of serviceability performance (Hindman and Bouldin 2019).

Figure 7 illustrates an important change in the E_eff/MOE_app ratio across four different span-to-depth ratios for the four different lay-up configurations. The relationship between the apparent MOE and the span-to-depth ratio has been well-studied for several CLT types. However, the dependence of the E_eff/MOE_app ratio on the span-to-depth ratio has received less attention. This ratio can offer more insights into the presence of shear deflection in a given span-to-depth bending load. In general, as the span-to-depth ratio increases, the E_eff/MOE_app ratio for each type of CLT decreases, and the differences between the homogeneous and hybrid CLT also decrease. At a 30 l/d ratio, the average E_eff/MOE_app ratio was 1.09, ranging from 1.07 to 1.15, which is nearly identical to the values obtained for solid timber at an 18 l/d ratio (Nocetti et al. 2013). These findings support the conclusion that the influence of shear deformation in the CLT beam is greater than that in the solid structural timber.

View Figure

• Download as Powerpoint
• Download Figure

Figure 8 shows the linear correlation between the apparent MOE at the 30 l/d span condition and the E_eff,reg value from the regression model. Data from all eight types of CLT are plotted together, without distinction by type. The r² value of the regression line is 0.735, which is slightly lower than the r² = 0.88 reported by Nocetti et al. in the case of solid timber (Nocetti et al. 2013). However, the low p-value (p < 0.00000) suggests a linear correlation. Since a linear relationship exists between E_eff and MOE_app, the regression line equation can be used to calculate the effective bending stiffness (EI_eff) from the apparent values.

View Figure

• Download as Powerpoint
• Download Figure

Effective shear stiffness

The effective shear stiffness (GA_eff) from the short-span bending test on CLT beams and the predicted values from the shear analogy model are shown in Table 6. The average GA_eff,reg value ranged from 1,945 to 3,354 kN/m, depending on the CLT type, with a coefficient of variation (COV) ranging from 5.7 percent to 29.4 percent. To complement the data presented in Table 6, the average GA_eff values of the specimens, categorized by method and applied adhesive, are shown in Figure 9.

View Figure

• Download as Powerpoint
• Download Figure

Table 6.Effective shear stiffness of the eight types of CLT.

View Fullscreen

Except for the homogeneous CLT configurations, the specimens glued with PUR adhesive exhibited greater shear rigidity. However, no significant effect of the adhesive type was found in the measured GA_eff among these four types of CLT panels. The factorial ANOVA results shows that lay-ups of the CLT panels made from red maple had no influence on the GA_eff. For the red oak configurations, the significance level of the CLT configurations was p < 0.05, indicating that the lay-ups significantly affected the GA_eff value (Table 7).

Table 7.ANOVA results of the CLT samples made from red oak, which were used to evaluate the influence of various factors on the effective shear stiffness.

View Fullscreen

Comparison of the results of the regression analyses with the predicted via Eq. 8 from the shear analogy model reveals that the analogy model underestimates the GA_eff value by an average of 74.8 percent. This difference is more moderate than that reported by Hematabadi et al. on poplar CLT, where the shear analogy model underestimated the effective shear stiffness by 510 percent compared with the regression method in the major strength direction (Hematabadi et al. 2020).

Shear modulus and shear strength of the short-span bending test

Detailed results of the shear tests are presented in Table 8. The average shear modulus (G_eff,reg) for the different types of CLT ranges from 108.2 MPa to 186.5 MPa, with the COV ranging from 8.9 percent to 29.44 percent. In contrast, the COV for the equivalent shear strength (f_s,eq) is narrower, ranging from 5.6 percent to 9.9 percent.

Table 8.Shear modulus and stiffness of the eight types of CLT.

View Fullscreen

Except for the lowest average value found in the hybrid red oak glued with PRF adhesive, the shear modulus is considerably higher than that of softwood CLT, and even higher than of the homogeneous sugar maple CLT, where Ma et al. reported values of 88.67 to 97.76 MPa, depending on the adhesive used (Ma et al. 2021a).

Regardless of the CLT type, three characteristic failure modes were observed. As an example, Figure 10 shows these failures, which occurred in the #1-#6 homogeneous red oak (HRO) CLT specimens. The most common failure was rolling shear in the transverse layer, as observed in specimens #3 and #5. For specimens #1, #2, and #4, delamination in the adhesive layer was the second most frequent failure mode. In many cases, both failure modes appeared simultaneously (e.g., in specimen #6), suggesting that in these cases, the adhesive strength was nearly equivalent to the rolling shear strength of the wood.

View Figure

• Download as Powerpoint
• Download Figure

The force-position curves, shown in Figure 11, also highlight the different failure patterns. In the case of rolling shear failure, multiple force peaks were observed, and the failure occurred gradually (specimens #3, and #5), whereas, in the case of pure delamination (specimens #1, #2, and #4), a significant and abrupt drop in force was noted. For specimen #6, delamination and rolling shear failure occurred together, although the force-displacement curve more closely resembled the pattern typically observed in pure delamination. Multiple shear cracks were developed through the transverse layers which propagated to the bonding line. In addition to the shear failure, tensile failure also occurred in the bottom layer, originating from a knot.

View Figure

• Download as Powerpoint
• Download Figure

After the load–deflection across four different spans was measured to obtain the stiffness values, the strength test was carried out at a 5:1 span-to-depth ratio until failure. The equivalent shear strength (f_s,eq) values for each CLT configuration were computed via Eq. 6 and are depicted in Figure 12.

View Figure

• Download as Powerpoint
• Download Figure

The difference in average shear strength between the homogeneous red maple CLT specimens bonded with the two different adhesives is 0.9 percent. In contrast, for the homogeneous red oak specimens, those bonded with the PRF adhesive exhibited an average shear strength 4.3 percent greater than those bonded with PUR adhesive. The strength results for the homogeneous CLT types are consistent with the findings of Crovella et al. who reported similar results for three-layer red maple (f_s = 3.0 MPa) and white ash (f_s= 3.1 MPa) CLT specimens (Crovella et al. 2019).

However, for both wood species, a significant difference was observed between the homogeneous and hybrid lay-up configurations. In the case of red maple, this difference averaged 30 percent, whereas for red oak, the average difference was 28 percent. This significant reduction in shear strength is attributable to the presence of the lower-strength yellow poplar. Even though the yellow poplar transverse layer reduced the shear strength of the hybrid lay-up configuration, the average shear strength of 2.51 MPa for the RMYP and 2.4 MPa for the ROYP specimens is significantly higher than that of the sugar maple–white spruce–sugar maple hybrid configuration, where Ma et al. (Ma, et al. 2021b) reported average equivalent shear strengths of 1.81 MPa and 1.7 MPa. Figure 13 shows the regression results for shear modulus and shear strength across four different CLT lay-up configurations.

View Figure

• Download as Powerpoint
• Download Figure

Significant differences can be observed between the correlation coefficients for the four types of CLT when both types of adhesives are considered together. The strongest correlation was found for homogeneous red oak CLT (r² = 0.45), whereas a smaller correlation (r² = 0.33) was observed in homogeneous red maple CLT. Both hybrid CLTs had correlation values of r² < 0.05. Previous studies revealed that the strength of the relationship is inconsistent, as it varies by wood species. For instance Nero et al. reported similarly low correlations (r² < 0.05) for CLTs made from Eucalyptus nitens, radiata pine, and Norway spruce (Nero et al. 2022). A slightly greater correlation was observed for poplar (Populus tremula L.), with an r² of 0.08 (Ehrhart and Brandner 2018). In contrast, European beech and birch species presented significantly higher correlations, achieving r² values of 0.63 and 0.55, respectively (Ehrhart and Brandner 2018). None of the regressions of the four types of CLT panels have a low p-value (p < 0.05), indicating that the linear correlation between the two properties is not statistically significant. Hence, the linear equations cannot be used to estimate one property from the other.

Simulation results

Finite element (FE) models were created to explore a parallel path for the determination of the initial stiffness of CLT panels subjected to static bending tests, as described in Section 2.4, with the parameters and geometric configurations summarized in Tables 3 and 4 and Figures 2 and 3, respectively. Considering that experimental determination on elastic properties can be expensive and time-consuming, the development of FE models that can be validated with fewer experimental iterations offers a path for a broader range of design exploration and material combinations. A critical aspect of the accuracy of the FE models is the material parameters selected to reproduce the mechanical behavior of the CLT laminates. In this study, a combination of orthotropic material properties reported in the literature (Forest Products Laboratory, 1999) and shear modulus values obtained from the regression methods presented in Section 3.3 are adopted to determine the elastic response of the CLT panels subjected to center-point bending.

Figure 14 compiles deflected shapes as well as vertical displacement (U2) contours corresponding to CLT panels with spans of 574, 857, and 1245 mm subjected to prescribed displacements in the vertical direction of 2, 4, and 5 mm, respectively. Figure 14 shows the beams deform uniformly without signs of delamination or discontinuities at the bonding interfaces. Figure 15 collects the load-deflection curves observed experimentally vs. those obtained from the FE simulations for all the CLT configurations. Results compiled in Figure 15 indicate an overall good agreement between experiments and simulation results, with a slight tendency to over-predict the slope of the load-deflection curves in the ROYP and RMYP hybrid configurations, and a mix of slight under-prediction and over-prediction in the case of the HRO and HRM homogenous configurations. The slight over-prediction is attributed to the elastic cohesive properties assigned to the boding interfaces, which account for the elasticity of the bonding interface. The cohesive zone model only included cohesive elastic properties without damage initiation parameters that could capture the debonding because of failure of the adhesive, which may have started occurring in some of the experimental specimens. This simplified approach is valid for small deflections in which the strength of the adhesive is not exceeded. Further evaluations will need to be conducted to determine the damage initiation parameters corresponding to the two types of adhesives implemented in this study so the FE models can capture the potential debonding that appears beyond the linear elastic range considered in the study. Despite this limitation, the FE models created based on available material properties offer an additional way to assess the stiffness of CLTs and potentially reduce the number of experimental evaluations.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure