The wood species used in the manufacture of the CLT came from the city of Riberão Branco - SP, which are: pino ellioti (Pinus elliottii Engelm.), eucalipto (Eucalyptus grandis W.Hill ex Maiden), toona (Toona ciliata M.Roem.) and lazcar (Acrocarpus fraxinifolius Arn.). The structural adhesive used was a two-component polyurethane (AG101) produced by KEHL^®, with the same chemical properties described by Silva et al. (2021).

Five levels of bonding pressure (0,1 MPa, 0,4 MPa, 0,7 MPa and 1,3 MPa) were tested for each wood species, totalizing twenty CLT boards with dimensions of 500 mm x 50 mm x 75 mm (total of 3 layers with lamellae of thickness equal to 25 mm each) manufactured.

The lamellae used were obtained from commercial dimensions planks (3000 mm x 140 mm x 75 mm). These planks were planed and cut-to-length until they reached a cross section of (110 x 50) mm² and a length of 500 mm. The lamellae were conditioned in an environment with controlled temperature and humidity until they reached a moisture content of 12 %. On the day of manufacture of the CLT, the lamellae were planed until they reached sectional dimensions of (100 x 25) mm².

The lamellae were cleaned with compressed air, in order to remove dirt and dust from the planning procedure, and thus, it was ensured that the surfaces were clean before applying the adhesive. The two-component polyurethane adhesive was spread to the surfaces of the lamellae with the use of a brush, in the proportion of 200 g/m², as indicated by Gomes (2018).

The position of the lamellae (outer and central layers) on the board panel was determined according to their visual and mechanical grading, as the ones with best performance and visual grading were positioned on the outer layers. The pressing of the boards were carried out in a mechanical press with pressure controlled by torque wrench.

The pressing step was carried out at room temperature with monitoring and compensation of the applied pressure, whenever necessary, in the first 24 h of gluing in order to maintain the initial value of pressure applied to the boards. Finally, CLT was sanded in a belt sander (grain 120) and had their edges sectioned to remove excess adhesive. Some of the CLT manufacturing procedures are shown in Figure 1.

Figure 1: CLT
manufacturing procedures: (a) Adhesive application; (b) CLT pressing; (c) CLT board after sanding; (d) CLT specimens cutting.

The delamination tests were performed according to the European standard EN 14080 (2013), for outdoor environments, and also based on the recommendations of Betti et al. (2016), who suggested the maximum dimensions of the specimens of (80 x 80) mm².

The delamination tests (Figure 2) were performed in an autoclave at room temperature with the application of vacuum and pressure cycles. The test specimens (80 x 80) mm², obtained from the CLT boards, were subjected to the test cycle consisting of: initial vacuum of 80 kPa for 30 minutes; pressure application of 550 kPa for two hours; removal of the autoclave and drying in a drying oven with air and temperature circulation control (65 °C) until their mass reached 100 % to 110 % of its initial value.

Before the pressure and vacuum cycles in the autoclave, all of the delamination specimens had their masses measured as well as all the lengths of the glue lines. Once the delamination tests were finished, the lengths of the glue line openings of the specimens were measured using a caliper with sensitivity of 0,01 mm.

The percentages of delamination for each species and pressure levels evaluated were compared with the limit of 10 % by EN 14080 (2013) for total delamination of the specimen. The delamination percentages were also compared with the limits of 4 % (for hardwoods, i.e. toona (Toona ciliata M.Roem.), lazcar (Acrocarpus fraxinifolius Arn.) and eucalipto (Eucalyptus grandis W.Hill ex Maiden) and 6 % (for softwood, i.e. pino ellioti (Pinus elliottii Engelm.) recommended by the draft standard ABNT NBR PN 02: 126.10-001-5 (2021).

Figure 2: (a)
CLT specimen before testing; (b) Autoclave; (c) Specimens drying after pressure cycles; (d) CLT specimen with delamination.

The glue line shear test also followed the methodology recommended by standard EN 14080 (2013), and the specimens had the same dimension as the delamination tests (Figure 2a). The dimensions used for the specimens in the glue line shear tests considered the shear area as well as the capacity of the test metallic device, which was made specifically for this purpose.

The determination of the shear strength in the glue line was carried out in the fabricated metallic device coupled (Figure 3) to a universal testing machine (INSTRON EMIC - 300 kN), ensuring that the failure of the test piece, in the glue line, occurred with a minimum time of 20 seconds.

Figure 3: (a)
Glue line shear test in progress; (b) Glue line shear strength test proposed by EN 14080 (2013).

It was not possible to visually identify the percentage of failure at wood and adhesive (bonding regions) after the shear tests on the glue line, since the polyurethane adhesive becomes translucent after curing. Therefore, a digital image method was required for identifying the amount of wood in the shear broken surface of the bonding region. ImageJ software (Rasband 2021) was used for image processing, after scanning (Figure 4) of the broken bonding regions.

The scanned image was separated into three parts, according to the color spectrum, which were in red, green and blue. It was observed that only the blue color of the spectrum clearly highlighted the region of the image that contained the adhesive. Therefore, only this image was used to determine the percentage of adhesive in the bonding region. Finally, a threshold was performed using the Otsu algorithm, as it presents a better division of the regions with adhesive in relation to the other available algorithms.

Figure 4: ImageJ
method for determining the percentage of failure at wood and adhesive on the broken surface by shearing the glue line.

The percentage of white pixels was counted and considered to be the percentage of wood failure. Special attention should be paid when using this image methodology, because in some cases, regions containing knots or cracks (which have darker tones) can be identified as regions with adhesive.

The comparison of all results was performed by analysis of variance (ANOVA) according to the Tukey test, with a 5 % level of significance. Tests of normality of data and homogeneity of variance were performed using the Shapiro-Wilk and Bartlett tests, respectively. All analyzes were performed with the use of software R version 4.0.4 (2021).

Table 1 lists the average results of the densities of the CLT specimens, with the analysis of variance and the standard deviation (in parentheses). Lower case letters equal in horizontal and upper-case letters in vertical indicated that there was no significant difference (p-value > 0,05).

Table 1:
Density of CLT specimens.

The mean values of the densities of the CLT samples ranged from 353 kg/m³ to 490 kg/m³, and these values differed significantly for each species, where lazcar (Acrocarpus fraxinifolius Arn.) and toona (Toona ciliata M.Roem.) showed higher and lower density, respectively.

The minimum density required for the production of CLT, in the United States and Canada, is 350 kg/m³ (Grandmont et al. 2019). In Brazil, ABNT NBR 7190 (1997) suggests that lamellae with 12 % apparent density between 400 kg/m³ and 750 kg/m³ should be used.

The delamination observed for the CLT specimens for all pressure tested are shown in Figure 5.

Figure 5: Total
delamination of the CLT specimens and standards limits.

The degrees of total delamination varied according to the wood species, being significantly lower in the CLT panels of pino ellioti (Pinus elliottii Engelm.) and lazcar (Acrocarpus fraxinifolius Arn.). On the other hand, the different pressure levels applied in the manufacture of CLT panels, on tested specimens, did not affect delamination values.

There was a trend of decreased percentage of total delamination, in the pressure range from 0,4 MPa to 1,0 MPa, for CLT made with lazcar (Acrocarpus fraxinifolius Arn.) and eucalipto (Eucalyptus grandis W.Hill ex Maiden). pino ellioti (Pinus elliottii Engelm.) CLT. The total delamination remained practically constant between the pressures from 0,7 MPa to 1,0 MPa. For the CLT made with lazcar (Acrocarpus fraxinifolius Arn.), there was a tendency of increase in total delamination from the applied pressure from 0,4 MPa to 1,3 MPa.

Faria et al. (2019), using the same delamination method, did not observe delamination in “Glulam elements” manufactured with toona (Toona ciliata M.Roem.) and polyurethane adhesive, indicating that the fiber direction of the lamellae in the CLT bonding may affect delamination performance.

The stresses generated due to the swelling and shrinkage of the wood in CLT elements are higher than in Glulam elements, due to restriction of its dimensional variation of the orthogonally arranged layers (Grandmont et al. 2019). This increase in stresses can cause a higher level of delamination compared to Glulam samples. Besides, the bonding of wooden elements arranged orthogonally to each other may not be as effective as the bonding of elements arranged in the same direction as the fibers.

According to Oliveira (2018), the CLT with three layers are not watertight, which may have caused a high percentage of delamination in the specimens. On the other hand, according to Betti et al. (2016) and Grandmont et al. (2019), the levels of retraction and swelling of the wood generate stresses in the glue lines, which are proportional to the sample size. Thus, the high delamination values are also justified by the sample size being larger than those generally used in Glulam specimens.

Figure 6 shows the relationship between the pressure level applied in the manufacture of the panels and the characteristic value of the shear strength in the glue line, for each of the species / pressure combinations.

Figure 6:
Shear strength in the glue line of CLT specimens.

Unlike the delamination tests, the results of shear strengths in the glue lines were influenced by the manufacturing pressure used. CLT panels bonded with pressures of 0,7 MPa obtained higher glue line shear strength compared to boards manufactured with other pressure values (0,1 MPa, 0,4 MPa and 1,3 MPa). CLT bonded with pressure of 1,0 MPa obtained intermediate results of shear strength on the glue lines.

According to Vilguts et al. (2015), the pressure required for the production of CLT must be at least 0,6 MPa, in order to promote an adequate binding of the glue lines. On the other hand, all the wood used in making the panels, in this research, presented apparent densities, for the 12 % moisture content, less than 500 kg/m³. In this context, the revision project of the Brazilian standard for wooden structures ABNT NBR 7190 (1997) recommends to use a bonding pressure of 0,7 MPa, for wooden elements glued with structural adhesives (Glulam and CLT). The percentages of failure at wood, in the bonded region of the CLT specimens, are shown in Figure 7.

Figure 7: Percentage
of rupture in the wood in the bonded region.

No relationship was observed between the glue line shear strength and the percentage of failure at wood in the bonded region, which the values varied from 35 % to 60 %. On the other hand, denser woods lazcar (Acrocarpus fraxinifolius Arn.) and eucalipto (Eucalyptus grandis W.Hill ex Maiden) showed a higher percentage of failure at wood. The fact that the adhesive is translucent may have contributed to the variability of the results obtained for the determination of the failure percentage at wood. It becomes clear that the use of a dye (when using an adhesive without distinctive tone after its curing) is recommended to better visualize the glue regions, and as long as it does not impair the adhesive bonding performance.

In general, there was no relationship between the density of the species and the quality of the bonding in the manufacture of the CLT, both in the delamination tests and in the shear bond strength tests in the glue line. Other factors may have been predominant in the results obtained, such as the anatomy and the chemical compounds present in each wood species (Vick 1999).

Three types of specimen failures were obtained experimentally after the shear tests of the glue lines (Figure 8), as follows:

Type I failure: Separation of the bonded region (the only ones used in the analysis of the results, ranging from 8 to 24 samples per treatment, totaling 75 % of all the glue lines analyzed);

Type II failure: Failure by compression perpendicular to the grain in one of the layers, that is, there was no separation of the lamellae in the bonded region (totaling 19 % of the glue lines analyzed);

Type III failure: Failure due to incorrect positioning of the specimen at the test device (totaling 6 % of the glue lines analyzed).

Figure 8: Types
of failures observed on specimens after the shear tests of the glue lines.

^♠Corresponding author: jvf.silva@unesp.br