Experimental and numerical evaluation on the behavior of single-shear timber connections using lag screw

. In current design procedures, the capacities of single-shear timber connections with lag screws are determined based on their governing failure modes, which are assumed to occur independently. In this study, the capacity and failure modes of single-shear timber connections with lag screws are evaluated experimentally and numerically. Six specimens of Red Meranti wood and lag screws with different penetration lengths are tested to obtain their lateral load capacities and failure modes. Nonlinear 3D finite element analyses using Abaqus are conducted to simulate the test condition. The experimental and numerical results are evaluated in terms of connection failure modes and capacities. Moreover, the result will be compared with NDS 2018 predictions. The study results show that the connection capacities between experimental tests and numerical analysis are similar if the minimum penetration requirement of NDS 2018 is satisfied. It can also be seen that the dominant failure modes, both from experimental and numerical results, match with those predicted using NDS 2018, although they are slightly combined with the other failure modes.


Introduction
Timber has been widely used in construction.Orta et al. [1] reported that timber has been utilized in western architecture as a result of the growing of social awareness in natural resources sustainability and energy conservation.Consequently, the potential of using timber in high-rise building structures becomes the main consideration in many studies.
Connections are crucial parts in structures, particularly in timber or steel structures, and, in many cases, determine the overall capacity of the structures.
Mechanical connections are more preferable in timber structures.Several mechanical connection types, such as split rings and shear plates, timber rivets, structural framing connections, and dowel-type connections, rely on metal-to-wood bearing mechanism for lateral load transfer and friction mechanism for withdrawal load transfer.These connections use fasteners e.g., nails, bolts, wood screws, and lag screws [2].
In cases where the connection is loaded laterally, failure can occur due to yield failure of the fasteners or bearing failure, which can occur in the main member or the side member.The connection failure modes and capacities are determined by the fasteners size and strength, the connected members thickness, and the timber materials specific gravity, which has been validated experimentally by Sutanto et al. [3].
The behavior of dowel-type timber connections have also been investigated numerically.Chen et al. [4] modeled dowel-type timber joints as two-dimensional plane stress elements, in which the lateral deformation *Corresponding author: marchelhartono2@gmail.com is neglected.Oudjene et al. [5] modeled screwed wood connections where the screws were modeled using onedimensional beam elements, while the timber members were modeled using three-dimensional solid elements.This modeling method has a limitation in simulating the contact interaction between the nails and the surrounding timber material.
For design purpose, The National Design Specification (NDS) for Wood Construction 2018 [6] categorizes six failure modes in single-shear timber connections, as illustrated in Fig. 1.Each failure mode is assumed to occur independently.The lowest calculated load that causes one of the failure modes is the governing capacity of the connection.This paper investigates the behavior of single-shear timber connections with lag screws under lateral loadings.Experimental tests and numerical analyses using Abaqus are conducted to validate the assumption which is made in predicting the failure modes and capacities of the dowel-type connections under lateral loadings as described in NDS 2018.

Experimental test
The experimental test is carried out using a Universal Testing Machine (UTM) in accordance with ASTM D5652-15 Standard [7], with a modification by applying compression instead of tension forces.The test set up, the loading scheme, and members dimensions are shown in Fig. 2-4.The tested main and side members are made of Red Meranti (Shorea spp.) wood.Based on the preliminary calculation using yield limit equations of NDS 2018, it was determined to use three variations of lag screw penetration to generate three failure modes.Table 1 shows each test specimen's specific gravity (G), moisture content (MC), and lag screw penetration length.The connections use 9.53-mm-diameter lag screws with a yield strength of 895.41 MPa and yield bending strength of 527.44 MPa.

Numerical modeling and analysis
Finite element analysis software Abaqus was used for the numerical analysis.All parts are modeled using solid elements (C3D8).The timber material elastic properties, refer to a study conducted by Pranata et al. [9], as shown in Table 2.The inelastic behaviour of the timber material is taken as if the members are subjected to compression parallel to the grain, as shown in Fig. 5.
The modulus of elasticity of steel used for the lag screws is 200000 MPa.As the lag screws are modeled in full nominal diameter, the strength of the material is modified to compensate the reduced cross-sectional area of the threaded parts as shown in Fig. 6.
The models' boundary conditions are defined to simulate the experimental tests.The bottom surfaces of the main members are restrained.Uniform downward displacements are applied incrementally at the top surface of the side members.The contact between the surface of the main member and the side member is modeled as 'hard-contact', where the friction is neglected.The 'hard-contact' contact interaction between the surface of the holes and the lag screws use a penalty-type tangential behavior with a friction coefficient of 0.4, as recommended by Mirianon et al. [10].The 'hard-contact' interaction prevents the surfaces from penetrating each other.Fig. 7. shows the numerical model in Abaqus software.The IIIm failure mode, which is designated to occur in S-02 and S-05 (Fig. 10.), and the IV failure mode, which is designated to occur in S-03 and S-06 (Fig. 11.) are proven both experimentally and numerically.It can be seen that the deformation and the yielding locations of the screws, and also the damage part of the main and side members, comply with the NDS 2018 predictions.However, the screw parts which should theoretically

Connection capacity evaluation
The lateral load versus displacement of the specimens are plotted in Fig. 13.The lateral load capacity of the connections are also evaluated in terms of maximum loads that can be resisted.The capacity values are also compared with the capacity predicted using NDS 2018 as shown in Table 4.It can be concluded from the plots and the tabular values, that experimental tests and numerical simulations generally result in higher capacity values compared to NDS 2018 prediction, which ranged from 33.2% to 63.6%.Different results are obtained for S-01 and S-04.These specimens' failure modes are combined with a more dominant withdrawal mode which requires better friction and material modeling, especially an accurate perpendicular to grain strength.Meanwhile, the material properties that were used are only the parallel to grain strength.

Fig. 8 .
are found to have small rotations as shown in Fig. 12.This indicates that the failure modes may not occur independently.The typical damage of main members of S-01 and S-04: (a) Experimental; (b) Numerical.

Fig. 9 .
Fig. 9. Sectional view of S-01 and S-04 models at final stages: (a) showing all parts; (b) showing lag screws only.

Table 2 .
Timber material modeling parameters.

Table 3 .
Numerical and experimental test at their final stages.

Table 4 .
Connection capacity.Based on the numerical analysis and experimental test results, it can be concluded that finite element analyses can effectively be used to simulate the behaviour of timber connections with lag screws under lateral loading based on the similarity with experimental results.The dominant failure modes of the connections obtained from experiments and numerical simulations comply with NDS 2018 predictions, although they might be combined with other minor failure modes.The capacity of the connections predicted using NDS 2018 give reasonably conservative values compared with experimental and numerical results.The experimental test facilities and software license are provided by the Department of Civil Engineering, Parahyangan Catholic University (UNPAR).