Experimental and Numerical Studies on the Traditional Penetration Mortise–Tenon Connection Reinforced by Self-Tapping Screws

: In this paper, the mechanical performance improvement of traditional penetration mortise– tenon (PMT) connections reinforced by self-tapping screws (STSs) were experimental and numerical studied. Four unreinforced penetration mortise–tenon connections and four STS-reinforced connections were experimentally investigated under monotonic and low-frequency cyclic loading. Besides, the numerical model of PMT connections reinforced by STSs with different diameters and yield strength were established and analyzed.Their failure modes, rotational stiffness, moment-resisting capacity, ductility, and seismic behavior were studied. The experimental results indicated that the STSs reinforcements could enhance the moment-resisting capacity of the penetration mortise–tenon connection. Compared with unreinforced connections, the initial stiffness of STS-reinforced connections increased by 75% with the moment-resisting capacity increasing by 69%, respectively. Furthermore, the STSs reinforcement method can effectively restrict the pull-out of tenon during the whole loading process. Numerical simulation results showed that the yield strength of STS has little effect on the moment-bearing behavior of the connection. The diameter of STS signiﬁcantly inﬂuences the connection performance, and the diameter of 8 mm is suggested.


Introduction
Mortise-tenon (MT) connections are widely used in historical and traditional wooden structures in China and other Southeast Asian countries to connect beams and columns. Penetration mortise-tenon connections, featured by a beam without cross-sectional reduction inserted into a column with rectangular or circle section, is one of the most typical types of MT connections. Its construction is shown in Figure 1. The tensile-resisting capacity of the PMT connection mainly depends on the friction of the contact surfaces between the mortise and tenon and is very limited. Thus, tenon joints pulling out ( Figure 2) and even completely separating from the mortise are prone to occur under axial loads, which severely weaken the integrity and safety of the wooden structure. In addition, the PMT connection can transmit moment by the wood compression perpendicular to grain and friction of the contacted surface between mortise and tenon joints under lateral loads such as seismic load [1]. However, due to manufacturing errors and wood shrinkage, there are often initial gaps between mortise and tenon, resulting in low initial rotational stiffness of the connection, which leads to connections being usually the weakest link in a wooden structure [2,3]. Therefore, it is important to restrict the tenon pull-out and improve the moment resistance behavior of PMTs for structural safety.  Knowledge of the mechanical properties of PMT connections is necessary to enhance their moment resistance and limit the pulling out of the tenon. Thus far, some scholars have carried out many experimental tests, theoretical studies, and finite element analyses for PMT connections. Xu [5] carried out low-frequency cyclic loading on connections and found that the tenon pulling out was related to the controlled displacement of each loading cycle, but also to the number of cycles of the loading point on the connection. Considering the influence of initial gaps on the performance of PMT connections, Chang et al. [6,7] conducted monotonic loading tests on PMT connections used in the wooden structure in Taiwan and found that the initial gaps reduced the rotational stiffness of the connections. The Nuki connection used in Japan is similar to the PMT connections except for the existence of wedges for fasteners. Guan et al. [8] performed low-frequency reversed cyclic loading on Nuki connections and established the relationship of moment and rotation. They also studied the influence of oversize on the behavior of the Nuki connections. The 3D nonlinear finite element analysis indicated that the different oversize wedges introduced various tightness to the connection and proved that the wedges had significant effects on the racking resistance of the Nuki connection. Tanahashi et al. [9,10], Chang et al. [6], Xie et al. [11], and Yang et al. [12] established the theoretical model of the moment rotation relationship of PMT connections based on different assumptions. Briefly, the poor moment-resisting capacity and ease of pulling out of the tenon are the drawbacks of PMTs, which affects significantly the mechanical behavior of the whole structure.
To enhance the resisting capacity and limit the tenon pulling out of mortise-tenon connections, many methods of reinforcement have been proposed. Some scholars firstly suggested strengthening the PMT connection by installing flat steel on it. Xie et al. [13] conducted a low-frequency cyclic loading test on wooden frames characterized by MT connections strengthened with flat steel. The results indicated that flat steel reinforcement was suitable for hidden MT connection with obvious insufficient strength or stiffness. However, Pan et al. [14] found that flat steel reinforcement would cause the redistribution   Pull-out of tenon [4].
Knowledge of the mechanical properties of PMT connections is necessary to enhance their moment resistance and limit the pulling out of the tenon. Thus far, some scholar have carried out many experimental tests, theoretical studies, and finite element analyse for PMT connections. Xu [5] carried out low-frequency cyclic loading on connections and found that the tenon pulling out was related to the controlled displacement of each load ing cycle, but also to the number of cycles of the loading point on the connection. Consid ering the influence of initial gaps on the performance of PMT connections, Chang et al [6,7] conducted monotonic loading tests on PMT connections used in the wooden struc ture in Taiwan and found that the initial gaps reduced the rotational stiffness of the con nections. The Nuki connection used in Japan is similar to the PMT connections except fo the existence of wedges for fasteners. Guan et al. [8] performed low-frequency reversed cyclic loading on Nuki connections and established the relationship of moment and rota tion. They also studied the influence of oversize on the behavior of the Nuki connections The 3D nonlinear finite element analysis indicated that the different oversize wedges in troduced various tightness to the connection and proved that the wedges had significan effects on the racking resistance of the Nuki connection. Tanahashi et al. [9,10], Chang e al. [6], Xie et al. [11], and Yang et al. [12] established the theoretical model of the momen rotation relationship of PMT connections based on different assumptions. Briefly, the poo moment-resisting capacity and ease of pulling out of the tenon are the drawbacks o PMTs, which affects significantly the mechanical behavior of the whole structure.
To enhance the resisting capacity and limit the tenon pulling out of mortise-tenon connections, many methods of reinforcement have been proposed. Some scholars firstly suggested strengthening the PMT connection by installing flat steel on it. Xie et al. [13 conducted a low-frequency cyclic loading test on wooden frames characterized by MT connections strengthened with flat steel. The results indicated that flat steel reinforcemen was suitable for hidden MT connection with obvious insufficient strength or stiffness However, Pan et al. [14] found that flat steel reinforcement would cause the redistribution Knowledge of the mechanical properties of PMT connections is necessary to enhance their moment resistance and limit the pulling out of the tenon. Thus far, some scholars have carried out many experimental tests, theoretical studies, and finite element analyses for PMT connections. Xu [5] carried out low-frequency cyclic loading on connections and found that the tenon pulling out was related to the controlled displacement of each loading cycle, but also to the number of cycles of the loading point on the connection. Considering the influence of initial gaps on the performance of PMT connections, Chang et al. [6,7] conducted monotonic loading tests on PMT connections used in the wooden structure in Taiwan and found that the initial gaps reduced the rotational stiffness of the connections. The Nuki connection used in Japan is similar to the PMT connections except for the existence of wedges for fasteners. Guan et al. [8] performed low-frequency reversed cyclic loading on Nuki connections and established the relationship of moment and rotation. They also studied the influence of oversize on the behavior of the Nuki connections. The 3D nonlinear finite element analysis indicated that the different oversize wedges introduced various tightness to the connection and proved that the wedges had significant effects on the racking resistance of the Nuki connection. Tanahashi et al. [9,10], Chang et al. [6], Xie et al. [11], and Yang et al. [12] established the theoretical model of the moment rotation relationship of PMT connections based on different assumptions. Briefly, the poor moment-resisting capacity and ease of pulling out of the tenon are the drawbacks of PMTs, which affects significantly the mechanical behavior of the whole structure.
To enhance the resisting capacity and limit the tenon pulling out of mortise-tenon connections, many methods of reinforcement have been proposed. Some scholars firstly suggested strengthening the PMT connection by installing flat steel on it. Xie et al. [13] conducted a low-frequency cyclic loading test on wooden frames characterized by MT connections strengthened with flat steel. The results indicated that flat steel reinforcement was suitable for hidden MT connection with obvious insufficient strength or stiffness.
However, Pan et al. [14] found that flat steel reinforcement would cause the redistribution of internal forces of members and increase the internal forces of some adjacent members while improving the rotational stiffness of connections. Improper reinforcement of flat steel may lead to the failure of other members. Some scholars [15,16] introduced more flexible fiber materials than flat steel as connection reinforcement. The tested results indicated that the CFRP reinforcement was an effective method to eliminate the tenon pulling out and improve the moment-resisting capacity as well as rotational stiffness. However, Yan et al. [17] carried out low-frequency cyclic loading tests on a scaled mortise-tenon wooden fame model and found that the CFRP reinforcement reduced the dissipated energy of the MT connections. Afterward, efforts to apply other reinforcement materials in MT connections such as metal dampers [14,18] and mild steel [19] were reported. However, most of these reinforcement methods need sufficient installation space and are not suitable for practical application. For real structures, infilled walls are often arranged near connections, and there is no sufficient operated and installed space for these reinforcements.
In this paper, an STS-reinforcing method was proposed for its convenient installation, low requirement of space, and cost efficiency. In addition, the performance improvement of PMT connections reinforced by STSs was evaluated by experimental and numerical studies. The failure modes, rotational stiffness and moment resistance capacity of connections were investigated. Besides, the influence of the STS parameters on the was analyzed. Based on the experimental and numerical investigation, a suggestion of a new STS-reinforced method was proposed and evaluated.

Experimental Program
To evaluate the reinforcing effect of STSs on the mechanical performance of PMT connections, two series of connections were fabricated-namely, unreinforced connections denoted as series U, and connections reinforced by self-tapping screws denoted as series S. All of the specimens had the same materials and same dimensions. There were four replicates in each series. Three of them were tested under monotonic loading, and the other one was tested under low-frequency cyclic loading. The information of all the tested PMT connection specimens is listed in Table 1.

Materials
All of these wooden specimens were fabricated by Chinese fir (Cunninghamia lanceolata) from Guizhou Province in southeastern China. Columns were made of logs, and beams were made of plain sawn timber with a class grade of grade Ia according to GB50005-2003 Code for Design of Timber Structures [20]. The physical and mechanical properties were tested according to timber material test standard [21][22][23][24][25], as summarized in Table 2.
The partially threaded self-tapping screws were used as reinforcements in the STSreinforced connections, with a length of 200 mm and a diameter of 6 mm. A total of 20 samples were used for each test. MV is the mean value, COV is the coefficient of variation. M.C is moisture content, ρ is density of wood, f t is the tensile strength parallel to grain, f m is the bending strength, E m is the flexure modulus of elasticity, f c, 0 is the compressive strength parallel to grain, and f c, 90, n is the compressive strength perpendicular to grain along radial direction.

Specimen Design
Based on the dimensions measured by the field survey, full-scale PMT connection specimens were fabricated. The details of the specimens are shown in Figure 3. The beam members were 180 mm × 55 mm in cross section and 1620 mm in length, and the column members were 220 mm in diameter and 1500 mm in length; the mortise on each column was 180 mm × 55 mm. Four STSs were arranged on each STS-reinforced connection, and the scheme is shown in Figure 3. A total of 20 samples were used for each test. MV is the mean value, COV is the coefficient of variation. M.C is moisture content, ρ is density of wood, ft is the tensile strength parallel to grain, fm is the bending strength, Em is the flexure modulus of elasticity, fc, 0 is the compressive strength parallel to grain, and fc, 90, n is the compressive strength perpendicular to grain along radial direction.
The partially threaded self-tapping screws were used as reinforcements in the STSreinforced connections, with a length of 200 mm and a diameter of 6mm.

Specimen Design
Based on the dimensions measured by the field survey, full-scale PMT connection specimens were fabricated. The details of the specimens are shown in Figure 3. The beam members were 180 mm × 55 mm in cross section and 1620 mm in length, and the column members were 220 mm in diameter and 1500 mm in length; the mortise on each column was 180 mm × 55 mm. Four STSs were arranged on each STS-reinforced connection, and the scheme is shown in Figure 3.

Loading Test Setup and Data Measurement
The test loading device is shown in Figure 4. The bottom end of the column was fixed to a steel beam by constraining its translation and rotation. The bottom end of the column was inserted into a U-shaped slot on the steel beam and pressed tightly by eight bolts. The steel beam was fixed to the floor by two anchor bolts. The top end of the column was applied a vertical load of 10 kN by jack1 with a loading capacity of 30 kN and a stroke range of 300 mm, to simulate the practical load. In the monotonic test loading, the vertical load leading to connection rotation was applied on the free end of the beam by Jack 2 with a loading capacity of 30 kN and a stroke range of 300 mm. Jack 2 was placed directly on the top face of the beam. In the low-frequency cyclic loading, the free end of the beam was connected to a vertically electro-hydraulic servo actuator through a hinge. The hydraulic actuator had a maximum loading capacity of 250 kN and a stroke range of 250 mm in two directions.

Loading Test Setup and Data Measurement
The test loading device is shown in Figure 4. The bottom end of the column was fixed to a steel beam by constraining its translation and rotation. The bottom end of the column was inserted into a U-shaped slot on the steel beam and pressed tightly by eight bolts. The steel beam was fixed to the floor by two anchor bolts. The top end of the column was applied a vertical load of 10 kN by jack1 with a loading capacity of 30 kN and a stroke range of 300 mm, to simulate the practical load. In the monotonic test loading, the vertical load leading to connection rotation was applied on the free end of the beam by Jack 2 with a loading capacity of 30 kN and a stroke range of 300 mm. Jack 2 was placed directly on the top face of the beam. In the low-frequency cyclic loading, the free end of the beam was connected to a vertically electro-hydraulic servo actuator through a hinge. The hydraulic actuator had a maximum loading capacity of 250 kN and a stroke range of 250 mm in two directions.   Three 200 mm range linear variable differential transformers (LVDTs) and two 30 deg range inclinometers were used to measure the displacement and rotation angle, respectively.The transducer arrangement is exhibited in Figure 4. Considering the rotation center is the center of the mortise, the moment applied to the connection can be calculated according to Equation (1).
where F is the vertical force recorded by Jack 2 or hydraulic actuator; L is the horizontal distance between the loading point and the rotation center. The rotation angle can be calculated according to Equation (2) or Equation (3). In order to obtain the pull-out amount of tenon, two displacement transducers (D2 and D3) were attached to the top and bottom surface of the beam to record the relative displacement between mortise and tenon. The pull-out value of tenon can be calculated according to Equation (4).
where D 1 is the vertical displacement of loading point measured by displacement transducer D1; D 2 and D 3 are the displacements measured by displacement transducer D2 and D3, respectively. G 1 and G 2 are the rotation angle of the beam and column, respectively. They are measured by inclinometers G1 and G2, respectively.

Loading Program
The first step of the loading procedures in both the monotonic and cyclic tests was the application of an axial compression force with 10 kN on the top of the column, simulating roof load, which was kept constant during the whole test. In the subsequent loading steps, a vertical force was applied on the free end of the beam, generating a moment on the connection.
The monotonic loading tests were performed in control load and at each 4 kN increment. A cyclic test was performed in displacement-driven according to EN 12512 [26], as illustrated in Figure 5. θ y is the yield rotation, which is determined by the monotonic test. The test program included one cycle in displacement 0.25θ y and 0.5θ y , followed by three cycles in displacement 0.75θ y , 1.00θ y , 2.00θ y , 4.00θ y , etc. until the specimen failed. The vertical loading stopped under one of the following conditions: (1) the applied load dropped to 80% of the peak load; (2) the loading device reached its range, or the displacement and rotation measurement was out of range of the sensors; (3) the tested specimen failed.

Loading Program
The first step of the loading procedures in both the monotonic and cyclic tests was the application of an axial compression force with 10 kN on the top of the column, simulating roof load, which was kept constant during the whole test. In the subsequent loading steps, a vertical force was applied on the free end of the beam, generating a moment on the connection.
The monotonic loading tests were performed in control load and at each 4 kN increment. A cyclic test was performed in displacement-driven according to EN 12512 [26], as illustrated in Figure 5. θy is the yield rotation, which is determined by the monotonic test. The test program included one cycle in displacement 0.25θy and 0.5θy, followed by three cycles in displacement 0.75θy, 1.00θy, 2.00θy, 4.00θy, etc. until the specimen failed. The vertical loading stopped under one of the following conditions: (1) the applied load dropped to 80% of the peak load; (2) the loading device reached its range, or the displacement and rotation measurement was out of range of the sensors; (3) the tested specimen failed.

Non-Reinforced Connections
Three unreinforced connection replicates under monotonic loading suffered similar damage. During the whole loading test, they successively experienced tenon rotation, contact extrusion, the gradual intensification of tenon and mortise, and increasingly severe, embedded plastic deformation until the bending failure of the beam member. In the later stage of loading, the wood fiber fracture at the embedded zone and wood fiber warp was also observed. The failure modes of unreinforced penetration mortise-tenon connections under monotonic loading and cyclic are given in Figure 6. The failure mainly occurred on beam members, severe local crushing, fracture, and shear failure of wood fiber at the tenon zone contacted to mortise due to the embedded compression perpendicular to grain and shear. For specimens U1 and U3, the beam failure occurred at the inclined section formed by connecting the top and bottom embedded areas of the tenon. In particular, for specimen U2, due to the presence of knots, the surface wood fibers near the knot were broken firstly and then prorated to the bottom compression side near the contact area between mortise and tenon ( Figure 6a). Moreover, the beam end of specimen U2 split along the initial crack. Compared with severe damage of beam element, the columns of all three specimens were subjected to only slight damage, as shown in Figure 6a. The column was subjected to local compression failure due to stress concentration and slight plastic deformation parallel to grain when in contact with the tenon.

Non-Reinforced Connections
Three unreinforced connection replicates under monotonic loading suffered similar damage. During the whole loading test, they successively experienced tenon rotation, contact extrusion, the gradual intensification of tenon and mortise, and increasingly severe, embedded plastic deformation until the bending failure of the beam member. In the later stage of loading, the wood fiber fracture at the embedded zone and wood fiber warp was also observed. The failure modes of unreinforced penetration mortise-tenon connections under monotonic loading and cyclic are given in Figure 6. The failure mainly occurred on beam members, severe local crushing, fracture, and shear failure of wood fiber at the tenon zone contacted to mortise due to the embedded compression perpendicular to grain and shear. For specimens U1 and U3, the beam failure occurred at the inclined section formed by connecting the top and bottom embedded areas of the tenon. In particular, for specimen U2, due to the presence of knots, the surface wood fibers near the knot were broken firstly and then prorated to the bottom compression side near the contact area between mortise and tenon ( Figure 6a). Moreover, the beam end of specimen U2 split along the initial crack. Compared with severe damage of beam element, the columns of all three specimens were subjected to only slight damage, as shown in Figure 6a. The column was subjected to local compression failure due to stress concentration and slight plastic deformation parallel to grain when in contact with the tenon.
Permanently embedded deformation perpendicular to grain was also observed in connection specimen under low-frequency reversed cyclic loading (Figure 6b). However, compared with that of specimens under monotonic loading, the deformation of the specimen under cyclic loading was much smaller, which was caused by a continuous change in the contact area of tenon and mortise due to pull-out of the tenon. Eventually, the connection specimen U4 failed with pulling out of the tenon.

Reinforced Connections
For the STS-reinforced connections, the initial slight compressive deformation perpendicular to grain under monotonic loading also occurred at later than that of unreinforced connections, and at the rotation around 3.7° (mean value of series S). Moreover, the pulling out of the tenon occurred at unreinforced connections was restricted in STS-reinforced connections.
The failure mode of STS-reinforced connections under monotonic loading is shown in Figure 7. As can be seen, the excessive compressive deformation perpendicular to grain presented in unreinforced connections was significantly reduced due to the reinforcement of self-tapping screws. It should be noted that screws 1 and 2 near the vertical loading point on the end of the beam were bent slightly, and screws 3 and 4 away from the vertical loading point were not bent in connection specimen S1, under monotonic loading. This Permanently embedded deformation perpendicular to grain was also observed in connection specimen under low-frequency reversed cyclic loading (Figure 6b). However, compared with that of specimens under monotonic loading, the deformation of the specimen under cyclic loading was much smaller, which was caused by a continuous change in the contact area of tenon and mortise due to pull-out of the tenon. Eventually, the connection specimen U4 failed with pulling out of the tenon.

Reinforced Connections
For the STS-reinforced connections, the initial slight compressive deformation perpendicular to grain under monotonic loading also occurred at later than that of unreinforced connections, and at the rotation around 3.7 • (mean value of series S). Moreover, the pulling out of the tenon occurred at unreinforced connections was restricted in STS-reinforced connections.
The failure mode of STS-reinforced connections under monotonic loading is shown in Figure 7. As can be seen, the excessive compressive deformation perpendicular to grain presented in unreinforced connections was significantly reduced due to the reinforcement of self-tapping screws. It should be noted that screws 1 and 2 near the vertical loading point on the end of the beam were bent slightly, and screws 3 and 4 away from the vertical loading point were not bent in connection specimen S1, under monotonic loading. This was because the local compression applied on screws 1 and 2 was larger than screws 3 and 4. was because the local compression applied on screws 1 and 2 was larger than screws 3 and 4. As demonstrated in Figure 8, the vertical load and moment were generated at the connection when the vertical load was applied at the free end of the beam. Under the action of moment M, the connection rotated around the geometric center of the four selftapping screws, and wood around the holes subjected to the forces F1, F2, F3, and F4, respectively. Under the action of vertical force F, the wood around the holes was subjected to the force Fv. Thus, the total forces FTi (i represents 1, 2, 3, and 4) acted on each dowel hole were obtained, as illustrated in Figure 8. The total force acted on dowel holes 1 and 2 was greater than holes 3 and 4. Therefore, the crushing of wood fiber around holes 1 and 2 was more server than that around the other two holes. In addition, there were compressed force Ft, c and friction force Ft, f acting on the tenon top surface, compressed force Fb, c, and friction force Fb, f acting on the tenon bottom surface, respectively. The PMT connections reinforced by STSs rely on the embedded compression and friction of tenon surface contacting with mortise but also the local compression of wood fiber surrounding STSs. Under cyclic loading, the pulling out of the tenon of STT-reinforced connections was restricted due to the presence of STSs. The plastic embedded deformation of the tenon, as well as combined bending and shear deformation of the STSs, were also observed. The connection specimen S4 failed early by beam fracture due to the presence of knots on the tenon top surface. As demonstrated in Figure 8, the vertical load and moment were generated at the connection when the vertical load was applied at the free end of the beam. Under the action of moment M, the connection rotated around the geometric center of the four self-tapping screws, and wood around the holes subjected to the forces F 1 , F 2 , F 3, and F 4 , respectively. Under the action of vertical force F, the wood around the holes was subjected to the force F v . Thus, the total forces F Ti (i represents 1, 2, 3, and 4) acted on each dowel hole were obtained, as illustrated in Figure 8. The total force acted on dowel holes 1 and 2 was greater than holes 3 and 4. Therefore, the crushing of wood fiber around holes 1 and 2 was more server than that around the other two holes. In addition, there were compressed force F t, c and friction force F t, f acting on the tenon top surface, compressed force F b, c , and friction force F b, f acting on the tenon bottom surface, respectively. The PMT connections reinforced by STSs rely on the embedded compression and friction of tenon surface contacting with mortise but also the local compression of wood fiber surrounding STSs. was because the local compression applied on screws 1 and 2 was larger than screws 3 and 4. As demonstrated in Figure 8, the vertical load and moment were generated at the connection when the vertical load was applied at the free end of the beam. Under the action of moment M, the connection rotated around the geometric center of the four selftapping screws, and wood around the holes subjected to the forces F1, F2, F3, and F4, respectively. Under the action of vertical force F, the wood around the holes was subjected to the force Fv. Thus, the total forces FTi (i represents 1, 2, 3, and 4) acted on each dowel hole were obtained, as illustrated in Figure 8. The total force acted on dowel holes 1 and 2 was greater than holes 3 and 4. Therefore, the crushing of wood fiber around holes 1 and 2 was more server than that around the other two holes. In addition, there were compressed force Ft, c and friction force Ft, f acting on the tenon top surface, compressed force Fb, c, and friction force Fb, f acting on the tenon bottom surface, respectively. The PMT connections reinforced by STSs rely on the embedded compression and friction of tenon surface contacting with mortise but also the local compression of wood fiber surrounding STSs. Under cyclic loading, the pulling out of the tenon of STT-reinforced connections was restricted due to the presence of STSs. The plastic embedded deformation of the tenon, as well as combined bending and shear deformation of the STSs, were also observed. The connection specimen S4 failed early by beam fracture due to the presence of knots on the tenon top surface. Under cyclic loading, the pulling out of the tenon of STT-reinforced connections was restricted due to the presence of STSs. The plastic embedded deformation of the tenon, as well as combined bending and shear deformation of the STSs, were also observed. The connection specimen S4 failed early by beam fracture due to the presence of knots on the tenon top surface. Figure 9 demonstrates the moment-rotation (M-θ) curves of the two series of PMT connection specimens under monotonic loading. It can be seen that the unreinforced connections display an almost linear relationship between moment and rotation during the initial loading phase, then show a nonlinear relationship with decreasing slope after the compression perpendicular to grain deformation occurred. The moment-rotation curves of all the three self-tapping screw-reinforced connections display a small slope before the rotation around 2 • , then exhibit an almost linear relationship with a larger slope, and then also show a nonlinear relationship with decreasing slope. This was because there were initial gaps between mortise and tenon in the connections, which led to a small rotational stiffness before the mortise and tenon contact with each other. With the increasing rotation, tenon came in contact with mortise, and the connection stiffness increased. After the tenon was pressed into the mortise and plastic compressive deformation perpendicular to grain occurred, the moment-rotation curve became nonlinear. Figure 9 demonstrates the moment-rotation (M-θ) curves of the two series of PMT connection specimens under monotonic loading. It can be seen that the unreinforced connections display an almost linear relationship between moment and rotation during the initial loading phase, then show a nonlinear relationship with decreasing slope after the compression perpendicular to grain deformation occurred. The moment-rotation curves of all the three self-tapping screw-reinforced connections display a small slope before the rotation around 2°, then exhibit an almost linear relationship with a larger slope, and then also show a nonlinear relationship with decreasing slope. This was because there were initial gaps between mortise and tenon in the connections, which led to a small rotational stiffness before the mortise and tenon contact with each other. With the increasing rotation, tenon came in contact with mortise, and the connection stiffness increased. After the tenon was pressed into the mortise and plastic compressive deformation perpendicular to grain occurred, the moment-rotation curve became nonlinear.

Experimental Results Analysis
(a) Unreinforced connection (Series U) (b) Self-tapping screw-reinforced connection (Series S)  Table 3. The initial stiffness and yield point of connection specimens were determined according to EN 12512 [26]. From Figure 10, the yield values are determined by the intersection between the two lines of the moment-rotation curves. The maximum moment-resisting capacity was the moment corresponding to the peak point. The failure rotation corresponded to the point at which the moment of the specimen dropped to 80% of the maximum moment. The ductility is defined as μ = θu/θy. As can be seen in Table 3, the initial stiffness and maximum moment-resisting capacity of unreinforced connections are 1.00~1.37 kN·m/° and 4.30~6.72 kN·m, respectively. The unreinforced connections had the lowest initial stiffness and maximum moment-resisting capacity, which indicated that the self-tapping-screw-reinforced method can improve the initial stiffness and moment-resisting capacity of PMT connections. Figure 11 shows normalized parameters of the unreinforced connection specimens and other derived parameters on the structural performance of connections. Compared with that of unreinforced connection (U), the initial stiffness of self-tapping screw connections (S) increased by 75% with the moment-resisting capacity increasing by 69%, respectively. In terms of ductility, the unreinforced connection specimens have the largest deformation ratio of 4.62, which implies that the unreinforced connection performed well in deformation. Even though the deformation ratio of self-tapping screw-reinforced connection specimens are lower than The initial stiffness K ini , yield rotation θ y , yield moment M y , maximum momentresisting capacity M max , failure rotation θ u , and ductility µ of the two series under monotonic loading are summarized in Table 3. The initial stiffness and yield point of connection specimens were determined according to EN 12512 [26]. From Figure 10, the yield values are determined by the intersection between the two lines of the moment-rotation curves. The maximum moment-resisting capacity was the moment corresponding to the peak point. The failure rotation corresponded to the point at which the moment of the specimen dropped to 80% of the maximum moment. The ductility is defined as µ = θ u /θ y . As can be seen in Table 3, the initial stiffness and maximum moment-resisting capacity of unreinforced connections are 1.00~1.37 kN·m/ • and 4.30~6.72 kN·m, respectively. The unreinforced connections had the lowest initial stiffness and maximum moment-resisting capacity, which indicated that the self-tapping-screw-reinforced method can improve the initial stiffness and moment-resisting capacity of PMT connections. Figure 11 shows normalized parameters of the unreinforced connection specimens and other derived parameters on the structural performance of connections. Compared with that of unreinforced connection (U), the initial stiffness of self-tapping screw connections (S) increased by 75% with the moment-resisting capacity increasing by 69%, respectively. In terms of ductility, the unreinforced connection specimens have the largest deformation ratio of 4.62, which implies that the unreinforced connection performed well in deformation. Even though the deformation ratio of self-tapping screw-reinforced connection specimens are lower than that of unreinforced connections, the deformation ratios are also higher than 3.38, which indicates well deformation capacity of this kind of reinforced connection [27]. that of unreinforced connections, the deformation ratios are also higher than 3.38, which indicates well deformation capacity of this kind of reinforced connection [27].    that of unreinforced connections, the deformation ratios are also higher than 3.38, which indicates well deformation capacity of this kind of reinforced connection [27].     Figure 12 shows the pulling out of the tenon of unreinforced connection specimen U4 under cyclic loading. It can be found that the pull-out amount gradually increased with the increase in rotation. Finally, the pull-out amount reached 144.5 mm. The pull-out amount of the STS-reinforced connection specimen S4 was 7.9 mm. This indicated that the STS reinforcement method can effectively restrict the pull-out of tenon during the whole loading process. Figure 12 shows the pulling out of the tenon of unreinforc under cyclic loading. It can be found that the pull-out amoun the increase in rotation. Finally, the pull-out amount reach amount of the STS-reinforced connection specimen S4 was 7.9 STS reinforcement method can effectively restrict the pull-ou loading process.

Numerical Simulation
To obtain stress (strain) distribution within the PMT con and investigating the influence of STSs parameters on the initi connection, 3-D nonlinear finite element models of the connec the commercial code ABAQUS.

Finite Element Modeling
The Specimen S1 serves as a reference for finite elemen and dimension of the beam are the same as that of specimen column of specimen S1 is inserted into the U-shaped slot on th ground. Therefore, this part can be regarded as a fixed constra as 1000mm in height and 220mm in diameter. The initial ga and tenon in specimen S1 is also considered in the numerica this model is 184mm in height and 55mm in width. In addition are simplified as plain round screws with the same diameter a Two materials are used for the connections. Columns an and STSs are made of steel. In the numerical simulation, wood material. The same modulus of elasticity is used for wood in The elastic modulus and yield strength of the wood are adop sults ( Table 2). The values of Poisson coefficients LR ν , LT ν , RT ν

Numerical Simulation
To obtain stress (strain) distribution within the PMT connection reinforced by STSs and investigating the influence of STSs parameters on the initial rotational stiffness of this connection, 3-D nonlinear finite element models of the connection were developed using the commercial code ABAQUS.

Finite Element Modeling
The Specimen S1 serves as a reference for finite element models. The construction and dimension of the beam are the same as that of specimen S1. The bottom end of the column of specimen S1 is inserted into the U-shaped slot on the steel beam fixed with the ground. Therefore, this part can be regarded as a fixed constraint. The column is modeled as 1000 mm in height and 220 mm in diameter. The initial gap of 2 mm between mortise and tenon in specimen S1 is also considered in the numerical model. Thus, the tenon of this model is 184 mm in height and 55 mm in width. In addition, four partial threaded STSs are simplified as plain round screws with the same diameter and length of the specimen.
Two materials are used for the connections. Columns and beams are made of wood, and STSs are made of steel. In the numerical simulation, wood is defined as an orthotropic material. The same modulus of elasticity is used for wood in tension and compression. The elastic modulus and yield strength of the wood are adopted based on the tested results ( Table 2). The values of Poisson coefficients V LR ,V LT ,V RT are 0.45, 0.51, 0.36, respectively. Steel is defined as isotropic elasto-perfectly plastic materials both in tension and compression. The elastic modulus and Poisson coefficient of steel are 2.1 × 105 MPa and 0.3, respectively. The yield strength of the steel of the reference model is 240 MPa.
Three-dimensional linear reduced integral solid element C3D8R is adopted for the finite element model. The grid size of critical parts of this model is about 10 mm, and that of other parts is about 20 mm or 30 mm. Figure 13 shows the mesh generation of the connection. During the beam rotating, there is a slide between surfaces in contact. Contact can occur in the interface between beam and column, STS and beam, STS and column. In the normal direction of the contact interface, the "hard" contact model is used to describe the relationship between the normal contact pressure and the gap (the distance between the two surfaces). When the gap between two surfaces is zero, the two surfaces are subject to contact constraints, and the contact pressure can be transferred between the contact surfaces. When the two surfaces are separated, the contact gap is greater than zero, the constraint between the contact surfaces is released, and the contact pressure is zero.
Contact causes friction in the tangent direction of the contact interface. The friction is simulated using the Coulomb stick-slip model. The friction coefficient between wood and wood is set equal to 0.35, the friction coefficient between STSs and wood is set equal to 0.3.
The boundary conditions of the numerical model are consistent with that of the tested connection specimen. The column bottom is fixedly connected. Reference point RP-1 is established above the center of the column top and coupled with the top surface of the column. A concentrated vertical load of 10kN is applied to this reference point. Reference point RP-2 is established and coupled to the loading area of the top surface of the beam. A downward vertical load is applied to this reference point in the same way as in the test. Figure 14 shows the moment-rotation curves of specimen S1 and the finite element model. It can be seen that the curves of the two coincide in the gap stage. When tenon and mortise contact and press tightly, the rotational stiffness of the two increases suddenly. However, the finite element model is stiffer than specimen S1 due to the non-consideration of the initial cracks, knots, and other defects presented in specimen S1 in the finite element model. Specimen S1 finally shows a load drop stage, while the finite element model does not experience load drop due to without simulating wood failure. Generally, the shape and trend of the two cures are basically the same when the rotation angle is within eight degrees. Table 4 shows the comparison of the rotational stiffness and moment-bearing capacity. K0 is rotational stiffness in the gap stage. Kini and Mmax are the same definitions in Section 4. The numerical result of Mmax is in good agreement with experimental observations. The numerical model is stiffer than the test specimen, but the relative error is within 15%, which indicates that the 3D-FEM model can predict well the stiffness and moment-bearing capacity of the PMT connection reinforced by STSs. Figure 15 shows the beam deformation of the numerical model. It is observed that the embedment deformation on the contact area between mortise and tenon. During the beam rotating, there is a slide between surfaces in contact. Contact can occur in the interface between beam and column, STS and beam, STS and column. In the normal direction of the contact interface, the "hard" contact model is used to describe the relationship between the normal contact pressure and the gap (the distance between the two surfaces). When the gap between two surfaces is zero, the two surfaces are subject to contact constraints, and the contact pressure can be transferred between the contact surfaces. When the two surfaces are separated, the contact gap is greater than zero, the constraint between the contact surfaces is released, and the contact pressure is zero.

Comparison of Numerical and Experimental Results
Contact causes friction in the tangent direction of the contact interface. The friction is simulated using the Coulomb stick-slip model. The friction coefficient between wood and wood is set equal to 0.35, the friction coefficient between STSs and wood is set equal to 0.3.
The boundary conditions of the numerical model are consistent with that of the tested connection specimen. The column bottom is fixedly connected. Reference point RP-1 is established above the center of the column top and coupled with the top surface of the column. A concentrated vertical load of 10 kN is applied to this reference point. Reference point RP-2 is established and coupled to the loading area of the top surface of the beam. A downward vertical load is applied to this reference point in the same way as in the test. Figure 14 shows the moment-rotation curves of specimen S1 and the finite element model. It can be seen that the curves of the two coincide in the gap stage. When tenon and mortise contact and press tightly, the rotational stiffness of the two increases suddenly. However, the finite element model is stiffer than specimen S1 due to the non-consideration of the initial cracks, knots, and other defects presented in specimen S1 in the finite element model. Specimen S1 finally shows a load drop stage, while the finite element model does not experience load drop due to without simulating wood failure. Generally, the shape and trend of the two cures are basically the same when the rotation angle is within eight degrees. Table 4 shows the comparison of the rotational stiffness and moment-bearing capacity. K 0 is rotational stiffness in the gap stage. K ini and M max are the same definitions in Section 4. The numerical result of M max is in good agreement with experimental observations. The numerical model is stiffer than the test specimen, but the relative error is within 15%, which indicates that the 3D-FEM model can predict well the stiffness and moment-bearing capacity of the PMT connection reinforced by STSs. Figure 15 shows the beam deformation of the numerical model. It is observed that the embedment deformation on the contact area between mortise and tenon.

Strength
To investigate the yield strength (fy) of STS on load-bearing behavior tion reinforced by STSs. Four finite element models with yield strength r 240MPa and 600MPa have been modeled. Figure 16a shows the moment-r ship of connection with different yielding strengths of STS. It can be seen th rotation curves of the four models have the almost same rotational stiffne resisting capacity. Figure 16b shows the normalized parameters of the num model and other derived parameters on the structural performance of models. The ratio is between 1.00 and 1.09, which indicates that the yield

Strength
To investigate the yield strength (fy) of STS on load-bearing behavior of PMT c tion reinforced by STSs. Four finite element models with yield strength ranging b 240MPa and 600MPa have been modeled. Figure 16a shows the moment-rotation re ship of connection with different yielding strengths of STS. It can be seen that the m rotation curves of the four models have the almost same rotational stiffness and m resisting capacity. Figure 16b shows the normalized parameters of the numerical re model and other derived parameters on the structural performance of other num models. The ratio is between 1.00 and 1.09, which indicates that the yield strength has little effect on the moment-bearing behavior of the PMT connection reinfor STSs.

Strength
To investigate the yield strength (f y ) of STS on load-bearing behavior of PMT connection reinforced by STSs. Four finite element models with yield strength ranging between 240 MPa and 600 MPa have been modeled. Figure 16a shows the moment-rotation relationship of connection with different yielding strengths of STS. It can be seen that the moment-rotation curves of the four models have the almost same rotational stiffness and moment-resisting capacity. Figure 16b shows the normalized parameters of the numerical reference model and other derived parameters on the structural performance of other numerical models. The ratio is between 1.00 and 1.09, which indicates that the yield strength of STS has little effect on the moment-bearing behavior of the PMT connection reinforced by STSs.

Diameter
To evaluate the diameter of STS on the load-bearing behavior of PMT connections reinforced by STSs, three PMT connections reinforced by STSs of diameter ranging between 6 and 10 mm have been modeled. Figure 17 shows the influence of the diameter of STS on connection. As STS diameter increases, the rotational stiffness in the gap stage increases slightly. The initial rotational stiffness and maximum moment of the numerical model with STS diameter of 8mm and 10mm are greater than that of the numerical model with STS diameter of 6mm. The moment-rotation curves of numerical model with STS diameter of 8mm is almost the same with that of numerical model with STS diameter of 10mm. It indicates that the rotational stiffness and moment-bearing capacity of the connection increase with the increase of STS diameter ranging from 6mm to 10mm. However, when the diameter reaches 8mm, the improvement is limited. The STS with a diameter of 8mm is suggested to be adopted as the reinforcement.

Conclusions
This paper presented the test results of two series (unreinforced and self-tappingscrew-reinforced) of full-scale traditional penetration mortise-tenon connections. The damage characteristics, failure modes, initial rotational stiffness and moment-resisting capacity of all the connections are investigated. The influence of diameter and yield strength of STS on the moment-bearing behavior of the connection is discussed. The main conclusions are as follows:

Diameter
To evaluate the diameter of STS on the load-bearing behavior of PMT connections reinforced by STSs, three PMT connections reinforced by STSs of diameter ranging between 6 and 10 mm have been modeled. Figure 17 shows the influence of the diameter of STS on connection. As STS diameter increases, the rotational stiffness in the gap stage increases slightly. The initial rotational stiffness and maximum moment of the numerical model with STS diameter of 8 mm and 10 mm are greater than that of the numerical model with STS diameter of 6 mm. The moment-rotation curves of numerical model with STS diameter of 8 mm is almost the same with that of numerical model with STS diameter of 10 mm. It indicates that the rotational stiffness and moment-bearing capacity of the connection increase with the increase of STS diameter ranging from 6 mm to 10 mm. However, when the diameter reaches 8 mm, the improvement is limited. The STS with a diameter of 8 mm is suggested to be adopted as the reinforcement.

Diameter
To evaluate the diameter of STS on the load-bearing behavior of PMT connections reinforced by STSs, three PMT connections reinforced by STSs of diameter ranging between 6 and 10 mm have been modeled. Figure 17 shows the influence of the diameter of STS on connection. As STS diameter increases, the rotational stiffness in the gap stage increases slightly. The initial rotational stiffness and maximum moment of the numerical model with STS diameter of 8mm and 10mm are greater than that of the numerical model with STS diameter of 6mm. The moment-rotation curves of numerical model with STS diameter of 8mm is almost the same with that of numerical model with STS diameter of 10mm. It indicates that the rotational stiffness and moment-bearing capacity of the connection increase with the increase of STS diameter ranging from 6mm to 10mm. However, when the diameter reaches 8mm, the improvement is limited. The STS with a diameter of 8mm is suggested to be adopted as the reinforcement.

Conclusions
This paper presented the test results of two series (unreinforced and self-tappingscrew-reinforced) of full-scale traditional penetration mortise-tenon connections. The damage characteristics, failure modes, initial rotational stiffness and moment-resisting capacity of all the connections are investigated. The influence of diameter and yield strength of STS on the moment-bearing behavior of the connection is discussed. The main conclusions are as follows:

Conclusions
This paper presented the test results of two series (unreinforced and self-tapping-screwreinforced) of full-scale traditional penetration mortise-tenon connections. The damage characteristics, failure modes, initial rotational stiffness and moment-resisting capacity of all the connections are investigated. The influence of diameter and yield strength of STS on the moment-bearing behavior of the connection is discussed. The main conclusions are as follows: (1) The unreinforced penetration mortise-tenon connections failed by the beam fracture under monotonic loading and the pull-out of tenon under cyclic loading; (2) The self-tapping-screw-reinforced method significantly improved the initial stiffness and moment-resisting capacity of unreinforced connections. The self-tapping-screw reinforcements could restrict the pull-out of the tenon during the whole loading process; (3) The yield strength of STS has little effect on the moment-bearing behavior of the connection. The diameter of STS significantly influences the connection performance, and the diameter of 8 mm is suggested; (4) The self-tapping-screw-reinforced methods have the advantages of cost effectiveness, easy implementation, and low demand for installation space. It is a reasonable reinforced method. Data Availability Statement: Not applicable.