Evaluation of shear performance of timber-timber composite joints

Abstract

The mechanical performance of timber composite floors is influenced by the degree of composite action between the components. In this study, the shear strength performance of cross-laminated timber and glued laminated timber composite floors based on the joining method was evaluated by push-out test. Eight types of timber-timber composite joints were evaluated using three different methods: lag screw joints, glued-in rod joints using fully threaded bolts and glass fiber reinforced plastic, and hybrid joints. Strength characteristics were derived to make theoretical predictions on the load-carrying capacity of the joints. The results showed that the glued-in rod joints were superior to the lag screw joints, with slip coefficients and ductility measured as 10 times and 2.5 times higher, respectively. The reliability of the strength characteristics of the glued-in rod joints was remarkably different depending on the presence or absence of anti-adhesive tape applied to the timber-to-timber joint surface. The load capacity of the hybrid joint, which combines mechanical and glued-in rod joining methods, was 47% higher than that of the lag screw joint and 38% higher than that of the glued-in bolt joint. In the European Yield Model modified to estimate the load capacity of joints, the rope effect and the yield moment of the fasteners had a remarkable impact on the predicted load capacity.

Full Article

Evaluation of Shear Performance of Timber-Timber Composite Joints

Yo-Jin Song,^a Seong-Yeob Baek,^b and Soon-Il Hong ^b,*

The mechanical performance of timber composite floors is influenced by the degree of composite action between the components. In this study, the shear strength performance of cross-laminated timber and glued laminated timber composite floors based on the joining method was evaluated by push-out test. Eight types of timber-timber composite joints were evaluated using three different methods: lag screw joints, glued-in rod joints using fully threaded bolts and glass fiber reinforced plastic, and hybrid joints. Strength characteristics were derived to make theoretical predictions on the load-carrying capacity of the joints. The results showed that the glued-in rod joints were superior to the lag screw joints, with slip coefficients and ductility measured as 10 times and 2.5 times higher, respectively. The reliability of the strength characteristics of the glued-in rod joints was remarkably different depending on the presence or absence of anti-adhesive tape applied to the timber-to-timber joint surface. The load capacity of the hybrid joint, which combines mechanical and glued-in rod joining methods, was 47% higher than that of the lag screw joint and 38% higher than that of the glued-in bolt joint. In the European Yield Model modified to estimate the load capacity of joints, the rope effect and the yield moment of the fasteners had a remarkable impact on the predicted load capacity.

DOI: 10.15376/biores.19.3.4984-5002

Keywords: Timber-timber composite (TTC); Push-out test; Glued in; Cross-laminated timber (CLT); Glued laminated timber (GLT)

Contact information: a: Institute of Forest Science, Kangwon National University, Chun-Cheon 200-701 Republic of Korea; b: Department of Forest Biomaterials Engineering, Kangwon National University, Chun-Cheon 200-701 Republic of Korea; *Corresponding author: hongsi@kangwon.ac.kr

INTRODUCTION

Using wood is increasingly recognized as a sustainable alternative in construction, in line with national carbon reduction policies. Engineered wood products, such as cross-laminated timber (CLT) and glued laminated timber (glulam), are known for their high specific strength, which makes them easy to handle during construction and allows for a high level of prefabrication (Izzi et al. 2018). Despite these advantages, timber floors have difficulty spanning unsupported lengths due to the relatively low modulus of elasticity of wood compared to concrete and steel. Increasing the thickness of the slab, such as CLT, is the easiest solution to control deflection and vibration, but excessive increases in slab thickness are not economically and structurally efficient. Hence, researchers have attempted to extend the span of timber floors over the past few years with studies on steel-timber composite (STC), timber-concrete composite (TCC), and timber-timber composite (TTC) floors (Hassanieh et al. 2017; Baek et al. 2021; Nie and Valipour 2022; Owolabi and Loss 2022; Song et al. 2022). In particular, TTC floors use more wood than STC or TCC floors. In addition, TTC flooring can also reduce the use of steel or concrete, which can significantly reduce the self-weight of the structure. The structural performance of timber-timber composite flooring is strongly influenced by the mechanical properties of the shear joints between the slab and the beam (or joist) (Nie et al. 2021). Therefore, in research on composites, the joint performance and behavior are usually analyzed through push-out tests on symmetrical joints, and the structural performance of the composite floor is predicted and analyzed through theoretical and finite element models based on the joint test results. In this process, the bending strength of the fasteners, the embedment strength of the wood, and the pullout strength are evaluated to derive strength properties for modeling because researchers have confirmed that the strength of TTC shear joints significantly impacts the maximum load-carrying capacity of TTC floors.

Research on joints has been mainly focused on mechanical joints with screws or bolts, and less frequently, glued-in rod joints, where holes are drilled in the timber to insert bolts or steel rods and filled with adhesive or grout. These have been studied, but not enough (Yagi et al. 2016; Chiniforush et al. 2021; Hammad et al. 2022; Zhang et al. 2023). Adhesive joints are considered to have very high load capacity and stiffness. However, their high brittleness and unreliable joint behavior are reasons that mechanical joints, such as nails, screws, or dowel-type fasteners, are often chosen for TTC joints instead.

This study evaluated the shear performance of CLT-GLT composite joints with lag screw joint, glued-in dowel joint, and hybrid joint. Eight types of TTC joints were tested for push-out strength based on the type of timber joint, depth of lag screw penetration, the diameter of the fully threaded bolt, diameter, and presence of threads on the GFRP rod. Additionally, a report on the improved constructability of the glued-in dowel method was included. Strength characteristics derived from embedment tests on the fastener’s shear strength and timber materials (CLT and GLT) were compared between a model applying the European Yield Model (EYM) and a model applying the formulas of Eurocode 5 (EN1995-1-1 2010). The estimated load capacity and slip coefficient of the models applying EYM were compared with the results of push-out tests.

EXPERIMENTAL

Material Properties

CLT and GLT

Both CLT and GLT were fabricated from kiln-dried larch (Larix kaempferi (Lamb.) Carr.) lumber. The average air-dried specific gravity and average air-dried moisture content of the larch lumber pieces were 0.57 (S.D: 0.05) and 11.9% (S.D: 1.4), respectively, and their dimensions were 30 mm (t) × 120 mm (w). The GLT, the beam element of the CLT-GLT composite floor, was laminated in eight layers, and its strength class was 10S-30B (symmetrical combination) according to KS F 3021 (2018). The floor element, CLT, was laminated in five layers and had a strength class of C-E12-E10 (E12 for the longitudinal layer, E10 for the horizontal layer) according to KS F 2081 (2021). Both timbers were laminated with phenol-resorcinol formaldehyde adhesive (PRF). Timber dimensions and density (ρ), moisture content (MC), bending strength (), and elastic modulus (E) are shown in Table 1.

Table 1. Mechanical Properties of GLT and CLT (Bending strength and elastic modulus are characteristic values)

Fastener

Six different types of fasteners were used in the CLT-GLT joint, including lag screws, fully threaded bolts, and glass fiber-reinforced plastic (GFRP), depending on the diameter, length, and threaded or unthreaded, as illustrated in Fig. 1(a). Lag screws were used in lengths of 200 and 250 mm with a diameter of 10 mm. Fully threaded bolts were used in diameters of 10 and 12 mm and length of 250 mm. The GFRP rods had the same diameter and length of 12 and 250 mm, respectively, threaded or not. The bending tests on them were performed with a three-point load, as indicated in Fig. 1(b). The yield load was determined by the 5% offset method, and the yield moment was calculated from . The plastic yield moment was estimated using the following equation for the plastic cross-sectional modulus s,

(1)

where d_in is the fastener’s root diameter. Table 2 summarizes the bending strength characteristics of the fasteners.

(a)                                                                                               (b)

Fig. 1. Mechanical fasteners and setup of three-point bending test

Table 2. Characteristics of Bending Strength of the Fasteners

Push-out Test of CLT-GLT Joints

Configuration of CLT-GLT joints

The push-out specimens for the shear strength evaluation of CLT-GLT joints were made of eight types, as presented in Fig. 2. The specimens consisted of a symmetrical structure (CLT-GLT-CLT) with two CLT slabs and one GLT beam joined by four fasteners to ensure uniform load transfer and avoid eccentric loads at the joints. The distance between the fasteners was 300 mm. The end distance, edge distance, and spacing of the fasteners satisfied ETA-11/0190 (DIBT 2013).

Fig. 2. Outline of geometry, dimensions (all dimensions in mm), and configuration of the CLT-GLT push-out specimens

The specimens were made from three jointing methods: mechanical joint (L10) by lag screws, glued-in rod joint (GB, GG) in which a dowel-type fastener was inserted by machining a hole in the joint and filling the gap between the hole and the dowel with glue, and hybrid method (HGB) in which a glued in rod joint was constructed using fully threaded bolts on the GLT beam, and a nut was used on the CLT floor. In the case of lag screw joints, the CLT and GLT were pre-drilled 2 mm smaller than the diameter of the lag screws to prevent breakage of the lag screws and misalignment of the GLT beams during the process of joining the CLT and GLT. For the glued-in rod joints, the holes drilled in the CLT and GLT were 4 mm larger than the diameter of the dowel and filled with epoxy (adhesive layer: 2 mm). For some of the specimens (GB10-250-3 and 4, GB12-250-3 and 4, GG(N)10-250, GG(T)12-250) with glued-in rod joints, anti-adhesive tape was applied to the contact surface of the CLT and GLT to prevent the CLT and GLT from being bonded by the adhesive that leaked out during adhesive filling (i.e., to prevent adhesive bonding behavior during the push-out test). For the hybrid method, a hole with a diameter of 25 mm and a depth of 10 d (d: diameter of the dowel) was machined in the GLT and filled with adhesive, and a hole 2 mm larger than the diameter of the fully threaded bolt was machined in the CLT to improve the constructability of the joint installation. In the specimen naming convention, the first part (before -) refers to the type and diameter of the fastener, and for GFRP rods, information about the presence or absence of threads is included in parentheses. The second part (after -) indicates the length of the fastener, and the third part (after the second -) represents the number of repetitions. More information about the push-out specimens is given in Table 3.

Table 3. Designation, Nominal Geometry/Dimensions of the Fasteners, and Details of the TTC Joints

Instrumentation and loading procedure

Push-out tests on CLT-GLT joints were performed in a vertical load-testing machine with a maximum capacity of 300 kN. The load was applied to a cross-section of the GLT, with the direction of the load parallel to the fiber direction of the GLT. The slip between CLT and GLT was measured using two displacement transducers (CDP-50, Tokyo Sokki Kenkyujo, Japan) installed at the same height as the fasteners at the front and back of the joint, as detailed in Fig. 3. The loading protocol for the test was divided into two phases according to EN 26891 (1991) (Fig. 4). In the first step, the specimen was loaded to 40% of the expected maximum load for the first 120 s, and the load was held for 30 s. The load was then reduced from 40% to 10% of the expected maximum load and held for another 30 s. In the second step, the specimen was reloaded until failure. Based on the test results, the maximum load values for subsequent specimens were modified and, if necessary, the loading procedure was redefined.

Fig. 3. Outline of the push-out test setup

Fig. 4. Loading protocol of push-out test (BS EN 26891 1991)

Embedment Tests

Specimens

Eight types of embedment test specimens (two for each type of specimen) were fabricated, as enumerated in Table 4, depending on the type of timber (CLT, GLT), jointing method, and fastener type and size. The specimens were half-hole type according to ASTM D5764 (2007). The glued-in rod specimens had a 2-mm-thick epoxy adhesive layer between the timber and the fastener, while the hybrid specimens had a 7.5-mm-thick adhesive layer between the GLT and the bolt. In the specimen naming convention, the first part (before -) refers to the timber type, and the second part (after -) is the bonding method and the diameter of the fastener.

Table 4. Details of the Specimens and Size of Fasteners in Embedment Tests

Test setup and procedure

The embedment strength test was performed according to ASTM D5764 (2007), with the crosshead applying force to a fastener placed over the half-hole at a speed of 1 mm/min. The direction of the load was parallel to the fiber direction of the GLT and CLT (relative to the outermost layer). The yield load was determined by the 5% offset method, and the embedment strength was calculated from Eq. 2 below,

(2)

where f_h,5% is the embedment strength (MPa), F_y,5% is the 5% yield load (kN), d is the diameter of fastener (mm), and d is the thickness of timber (mm). The theoretical embedment strength, f_h,EC5 suggested by Eurocode 5 (EN 1995-1-1 2010), was estimated using the following Eq. 3,

(3)

where dis the diameter of the fastener (mm), and ρ_m is the density of timber (kg/m³), α is the penetration angle of the fastener with respect to the grains, k₉₀ is a correction factor that depends on the type of timber.

RESULTS AND DISCUSSION

Push-out Test Results

Failure mode

Fasteners in CLT-GLT joints were subjected to a single shear load, and six failure modes can be categorized according to the embedment strength of the CLT and GLT, the size and strength of the fastener, the thickness of the CLT, and the fastener embedment length in the GLT. Failure modes I and II were related to the embedment of the CLT or GLT around the fastener without significant plastic deformation of the shear fastener. In the case of glued-in rod joints, the epoxy adhesive layer of the CLT or GLT was destroyed. Failure mode Ⅲ involved the embedment of CLT and GLT (or epoxy adhesive layer). Failure modes Ⅳ and Ⅴ resulted in a single plastic hinge in the shear fastener, while failure mode Ⅵ resulted in two plastic hinges in the shear fastener. Reportedly, failure modes I through III were associated with brittle failure, while failure modes IV through VI were categorized as ductile failure (Chiniforush et al. 2021). In addition, for adhesive bonding, the brittle failure of the adhesive layer affects the brittle behavior of the joint (Nie et al. 2021). When the penetration depth of the lag screw in the GLT was 10 d (100 mm) (L10-250), two plastic hinges were observed in the fastener, which corresponded to failure mode Ⅵ, and when the penetration depth was shortened to 5 d (50 mm) (L10-200), it was related to failure mode Ⅳ or Ⅴ. In the glued-in bolt joints (GB10-250, GB12-250) and hybrid joints (HGB10-250), a plastic hinge occurred in the bolt after the adhesive layer was destroyed. These joints were subjected to failure mode Ⅵ regardless of the diameter of the bolt or the thickness of the adhesive layer. Glued-in GFRP rod joints (GG(N)10-250, GG(N)12-250, GG(T)12-250) were also associated with failure mode VI. Most of these joints saw the GFRP cut due to increased slip after the maximum load, which was observed more in the threaded GFRP (GG(T)12-250). For the unthreaded GFRP rod, the slip of the joint increased, as it was easily pulled out of the timber by the low bond shear strength with the adhesive layer.

Fig. 5. Failure modes of TTC joints

Load-slip behavior

The load-slip curve for the push-out test is shown in Fig. 6. The slip in the load-slip curve is the average value of the strain measured by the four displacers. The behavior of the load-slip curves for the lag screw joint specimens was close to bilinear. The specimens with glued-in rod joints without anti-adhesive tape (GB10-250-1 and 2, GB12-250-1 and 2, GG(N)12-250) exhibited a near linear load-slip behavior with little slip up to the maximum load, after which the load dropped remarkably due to the breakdown of the adhesive layer between the two timber surfaces. This brittle behavior has been commonly observed in adhesive joints (Nie et al. 2021; Hammad et al. 2022), which suffer from inconsistent load-slip behavior. In contrast, the load-slip curves of glued-in rod joints with anti-adhesive tape (GB10-250-3 and 4, GB12-250-3 and 4, GG(N)10-250, and GG(T)12-250) were similar to those of lag screw joints. The test results revealed that applying anti-adhesive tape could induce a constant load-slip behavior of the joint. For the hybrid joint (HGB10-250), the load-slip behavior was close to bilinear without any tape application.

Fig. 6. Load-slip curves of CLT-GLT composite joints

Strength and slip modulus ()

The maximum shear load and yield load of the joints, along with their corresponding displacements, as well as the slip coefficient and ductility index, are reported in Table 5. Various methods have been proposed to determine the yield strength characteristics of the joint (Muñoz et al. 2008). The values on the load-slip curve obtained from experimental tests were determined using the Yasumura and Kawai (Y&K) and 5% offset methods (Yasumura 1998). Among them, the 5% offset method was not suitable for determining the yield strength for specimens with adhesive bonding behavior, so the Y&K yield strength determination method was finally used.

The parameter K_s, which remarkably impacts the short-term settlement of CLT-GLT composite floor structures, was calculated using the following equation, through a straight line with a slope penetrating between 10% and 40% of F_max on the load-slip curve (EN-26891 1991),

(4)

where 0.1F_max and 0.4F_max are 10% and 40% of F_max and v_0.1 and v_0.4 are slips corresponding to 0.1F_max 0.4F_max.

The ductility of the joint was quantified using Eq. 5, as reported by Smith et al. (2006),

(5)

where µ is a ductility index, v_max is a slip corresponding to F_max, and v_y is a slip corresponding to F_y that was determined by the above-mentioned Y&K method. According to Smith et al. (2006), it is classified as ductile if the ductility index µ >6 and brittle if µ< 2.

For the lag screw joint, when the penetration depth into the GLT was 5 d, the average F_max was 45.7 kN. Increasing the penetration depth of GLT to 10 d resulted in a 97% increase in the average F_max to 90.2 kN. The increase in penetration depth resulted in a 31% increase in K_s and improved the ductility of the joint.

Joints without applied anti-adhesive tape may exhibit a very high F_max , making them appear as superior joints. However, due to minimal slip, it was extremely challenging to derive strength characteristics from the load-slip curve. Within the same type of joint, the variation in slip response to loads was remarkable, depending on the test specimen, resulting in low reliability of strength characteristics. Anti-adhesive tape considerably reduced the effect of load increase due to adhesive bonding, resulting in a relatively lower F_max measurement while increasing the reliability of the derived strength characteristics. Therefore, the analysis of the glued-in rod joints described below compared specimens where the anti-adhesive tape was not applied.

Comparing fasteners of the same diameter confirmed that the glued-in bolt joint method could increase the K_s of the CLT-GLT joint over 10 times in GB10-250 and L10-250. The joint exhibited a very high ductility level. Increasing the diameter of the bolt from 10 to 12 mm resulted in a 24% increase in average F_max and a 41% increase in K_s confirming that increasing the diameter effectively enhances the joint’s performance, which contradicts the common belief that bonding by adhesives leads to high strength but low ductility (Hassanieh et al. 2017).

The reliability of the strength characteristics of joints using GFRP rods, namely GG(N)12-250, was low due to adhesive failure. Comparing the performance of joints based on the increase in diameter of GFRP rods was deemed impractical. Increasing the diameter of the GFRP rod from 10 mm to 12 mm and adding a threaded insert to the GG(T)12-250 resulted in a 70% increase in F_max but K_s decreased 42%, which did not effectively enhance the performance of the joint, contrasting with the results of joints using bolts. Nevertheless, GG(T)12-250 was still effective in the composite of CLT-GLT because of its higher K_s and µ compared to L10-250.

The average F_max of GB12-250 was 121.4 kN, while the average F_max of GG(T)12-250 was 86.3 kN, implying that the bolt exhibited 41% higher shear performance. In the glued-in rod jointing method, using bolts increased K_s 178% compared to GFRP rods. This outcome is closely related to the difference in strength (MOR and MOE) between bolts and GFRP rods.

The average F_max of the hybrid joint was measured at 132.7 kN, the highest among the glued-in rod joints tested, except for the specimen where the adhesive failure occurred. This is an average F_max that is 47% higher than the L10-250 joint with lag screw of the same diameter, and 38% higher than the GB10-250 joint with glued-in bolt. Considering the challenging nature of constructing glued-in rod joints, hybrid construction methods can be a rational choice for improving the constructability of CLT-GLT composite floors. However, the hybrid method exhibited minimal impact on the joint’s K_s and ductility. In the case of K_s the loose clearance between the bolt hole in the timber and the bolt diameter, as well as increased slip at the joint, were suspected to cause the bolt head to easily be sucked into the CLT interior, leading to destruction modes. The ductility of the bonding area did not improve due to the increased brittleness in proportion to the thickened adhesive layer.

Table 5. Characteristic of CLT-GLT for Push-out Test