Seismic Retrofit Technique Using Plywood and Common Nails for Connections of Low-Rise Timber Frame Construction

Abstract

Since the Japanese Building Standards Act was revised in 2000, the installation of steel timber connectors (STCs) to reinforce timber frame (TF) connections has been mandated for new-build TF houses in Japan. However, for the TF houses built before then, more than 40% do not have sufficient STCs and are considered earthquake-prone. This study proposed a simple and easy seismic retrofit technique for such earthquake-prone existing houses. The proposed technique can reinforce TF connections using only plywood and common nails and achieve equivalent performance to the benchmark STCs used in Japanese TF houses, such as the “CP-T type” or the combination of “VP type” and “BP type” STCs. Experiments were performed to compare the proposed technique with the STCs using pullout and full-scale in-plane cyclic tests. The experimental results showed that the proposed technique had a higher seismic performance than the STCs, which was particularly excellent in displacement ductility to prevent a collapse of the TF house without damaging the timbers. The proposed technique will be accepted by many carpenters when retrofitting earthquake-prone existing houses because simple and easy.

1. Introduction

Wooden construction has long been used and is still being used worldwide [1,2,3]. In particular, low-rise buildings (mainly detached houses) are often built with wood [4,5,6,7]. Globally, such wooden houses are built mainly by platform frame construction [8]. However, in Japan, they are often built with traditional timber frame (TF) construction, another post-and-beam construction [9]. Figure 1 shows the typical framing of a common Japanese detached house. Timbers of 105 mm width and height are often used for posts and beams, and those with a width of 30 mm and height of 105 mm are often used for braces.

According to the Japanese government, there are more than 26 million units of wooden housing stock in Japan, and more than 40% are considered earthquake-prone [10]. In 2000, after the Great Hanshin Earthquake, the Japanese Building Standards Act was revised [11]. Since then, sufficient steel timber connectors (STCs) have been installed for new-build TF houses to reinforce the connections against earthquakes. However, many existing TF houses built before then did not have sufficient STCs; they were mostly held together with nails. Therefore, it is necessary to reinforce the TF connections of such houses to increase their safety and extend their lives.

Figure 2 shows typical TF connections of the Japanese TF houses [9]. Until the 1980s, they were mortise and tenon with common nails, mostly added with large staple nails. In the 1980s, STCs such as “CP-T type”, “VP type” and “BP type” have been used for TF connections. Subsequently, these initial STCs have been the benchmark for developing new STCs, and improved designs of STCs are used for new-build TF houses now. Installing these STCs was mandated in 2000.

The TF seismic performance has long been studied in Japan [12,13], one of the most earthquake-prone countries. Until the early 1980s, many TF houses were built with mud or plaster-infilled earthen walls, one of the traditional Japanese wall styles. This wall composition is similar to what is known as “half-timbering” globally, and there are many such historic wooden buildings in Japan too. Many studies on such historic buildings focus on the strength of earthen walls [14,15,16,17] or techniques of reinforcing connections [18]. However, they are not for common detached houses of small scale and low historical value built after the 1980s, when insulated walls became the mainstream. Takagi et al. [19] proposed a seismic retrofitting technique for the houses built after the 1980s that covers the existing mortar exterior walls with steel cladding, using the shear capacity of the existing mortar exterior. Nevertheless, this technique is unsuitable for connection reinforcement.

Many studies on TF seismic performance worldwide, except in Japan, are related to retrofitting historic buildings [20,21,22,23]. Parisse et al. [24] stated that they are categorized into two. One is the “timber strong-backs”, which use TFs for masonry reinforcement [25,26,27]. However, few masonry houses can apply this type in Japan. The other one is the “half-timbered type”. Much research on the seismic performance of half-timbered types is concerned with reinforcing the TF connections [28,29,30,31]. They have shown that the response of the TFs depends essentially on the seismic resistance of the connections. Poletti et al. [31] reported the pullout and in-plane cyclic test results on reinforcing techniques for the connections of half-timber buildings, such as Pombalino buildings in Portugal. The tests were performed on four techniques, i.e., self-tapping screws, glass-fiber-reinforced polymer sheets, steel plates, and near-surface-mounted steel rods. All these techniques improve the dissipative and load-bearing capacity. However, retrofit techniques for common Japanese houses after the 1980s entail easy installation using local carpentry resources.

Other studies on seismic retrofitting for TFs include constructing shear walls by applying board materials. Huang et al. [32] proposed a retrofitting technique applying oriented strand board, Sudo et al. [33] proposed a medium-density fiber-board technique, and Dutu et al. [34,35] proposed a technique applying timbers at regular intervals in a diagonal sheathing arrangement. However, these techniques are unsuitable for connection reinforcement.

This study proposes a simple and easy seismic retrofit technique only using plywood and common nails to reinforce TF connections with the equivalent performance to the benchmark STCs used in Japanese TF detached houses, such as the “CP-T type” or the combination of “VP type” and “BP type” STCs. The advantage of the proposed technique is the labor saving of the retrofit compared to the STCs’ installation, with equivalent seismic performance.

2. Materials and Methods

Figure 3 shows the proposed retrofit technique, “Plywood and Nails” (P and N), for reinforcing TF connections. This technique uses only plywood and common nails to reinforce the post-beam and post-beam-brace connections of the existing framing. The installation is performed from the exterior side. The procedure comprises the following two steps: (1) removing existing cladding materials only around 300 or 500 mm of the beams to expose the reinforcing frames and (2) applying a 300 or 500 mm width and 12 mm thickness of plywood to cover the connections and fastening it using “CN50” common nails.

Experiments were performed to compare whether the proposed technique has equivalent performance to the typical STCs. Notably, it was compared with the “CP-T type” STC using pullout tests on the post-beam connection, as well as compared with the combination of “VP type” and “BP type” (VP-and-BP-type) STCs using full-scale in-plane cyclic tests on the post-beam-brace connection.

Since the Japanese Building Standards Act was revised in 2000, plywood has been widely used in Japanese new-build TF houses to ensure wall shear capacity in place of braces, because installing plywood can use general nail guns and is much easier than hammering STCs’ nails at TF connections. Additionally, many specifications of plywood shear walls in the Act use 50 mm-length nails; therefore, many Japanese carpenters use nail guns for CN50. For those reasons, the 12 mm-thickness of the plywood and CN50 conforming to the Japanese Agricultural Standard: “Plywood” [36], and the Japanese Industrial Standard: JIS A 5508 “Nails” [37], respectively, were determined as the materials for the P and N retrofit technique.

2.1. Pullout Tests on the Post-Beam Connection

Specimens were designed to model a post-beam connection of TF construction. Figure 4 shows the details of the P and N specimen. Plywood was fastened to the post and beam with 8 nails each, i.e., 16 nails of CN50 in total. The nailing pattern was set with 50 mm nail spacing and 25 mm edge distance. Figure 5 shows the details of the CP-T-type specimen. CP-T-type STC was fastened to the post and beam with five nails each, i.e., ten nails of ZN65 in total. The post and beam timbers used were Sakhalin fir and had a width and height of 105 mm, conforming to the Japanese Agricultural Standard: “Sawn Lumber” [38], and the post was set at the center of the beam by mortise and tenon. Additionally, all nails conformed to the Japanese Industrial Standard: JIS A 5508 “Nails” [37].

Table 1. Density and water absorption of the timbers for the pullout test specimens.

Specimen	Parts	Dimension 1 [mm]	Density 2 [g/cm3]	Water Absorption 3 [%]
P and N	Post	105 × 105	0.35	6.5
	Beam	105 × 105	0.31	6.7
CP-T type	Post	105 × 105	0.32	7.9
	Beam	104 × 105	0.32	7.4

¹ Cross-sectional dimensions of the timbers. ² Density of the timbers. ³ Water content of the timbers as a percentage of weight.

presents the density and water absorption of the timbers used in the pullout test specimens. The densities of the Sakhalin fir timbers were generally close, and water absorption satisfied less than 20% of Japanese regulations.

Figure 6 shows the test setup and instrumentation of the pullout test. A specimen was fixed to the testing machine foundation by steel bolts, and the head of the specimen post was attached to the flexible jointed pullout jig of the testing machine. The loading was monotonic, with a 3 mm/min pullout speed. Displacement transducers (DTs) were placed on either side of the post to measure the displacement between the post and beam at 20 points/s sampling intervals.

2.2. Full-Scale In-Plane Cyclic Tests on the Post-Beam-Brace Connection

Specimens were designed to model a full-scale TF construction wall, with 910 mm post to post and 2730 mm beam to beam. Figure 7 and Figure 8 show the details of the specimens. As in the pullout test, Sakhalin fir was used for all timbers. The dimensions of the post and beam timbers were 105 mm in width and height, and those of the brace and stud were width 30 mm and height 105 mm. The post and stud timbers were set at the beam by mortise and tenon.

Figure 7 shows the P and N specimen. Figure 7b shows the nailing pattern of CN50 to the post-beam connection, basically the same as in the pullout test specimen. Figure 7c shows the CN50 to the post-beam-brace connection, basically the same nailing pattern but with a zigzag pattern at the brace nailing.

Figure 8 shows the VP-and-BP-type specimen. Figure 8b shows the STCs installation to the post-beam-brace connection. The VP-type STC was fastened to the post and beam connection with four nails each, i.e., eight nails of ZN90 in total. The BP-type STC was fastened to the post, beam, and brace with three, four, and three ZN65 nails, respectively. Moreover, a 12 mm-diameter square neck bolt was added to fasten the brace.

Table 2. Density and water absorption of the timbers for the full-scale in-plane cyclic test specimens.

Specimen	Parts	Dimension 1 [mm]	Density 2 [g/cm3]	Water Absorption 3 [%]
P and N	Post (Right)	105 × 104	0.32	6.7
	Post (Left)	103 × 105	0.34	6.2
	Beam (Girder)	104 × 105	0.34	6.9
	Beam (Sill)	104 × 105	0.33	6.2
	Brace	105 × 30	0.32	6.5
VP and BP type	Post (Right)	105 × 103	0.36	6.7
	Post (Left)	102 × 105	0.34	6.8
	Beam (Girder)	104 × 105	0.34	6.9
	Beam (Sill)	104 × 104	0.36	6.7
	Brace	105 × 30	0.36	6.2

¹ Cross-sectional dimensions of the timbers. ² Density of the timbers. ³ Water content of the timbers as a percentage of weight.

presents the density and water absorption of the timbers used in the full-scale in-plane cyclic test specimens. The densities of the Sakhalin fir timbers were generally close, and water absorption satisfied less than 20% of Japanese regulations.

Figure 9 and Figure 10 show the test setup and instrumentation of the full-scale in-plane cyclic test and the test setup of the P and N specimen, respectively. The vertical load was applied to the upper beam by suspending 180 kg of steel weights, 90 kg each on the left and right sides. The servo-hydraulic actuator applied horizontal displacement to the upper beam with a maximum displacement and load of 600 mm and 50 kN, respectively. The upper beam and the servo-hydraulic actuator were connected with steel bolts via the flexible jointed jig. Rigid casters were set in front and behind the upper beam as lateral restraints to prevent the out-of-plane displacement of the specimen. The lower beam was fixed to the steel foundation by steel bolts at two points, and steel jigs were set at both cut ends as horizontal restraints. The horizontal displacement of the upper beam was measured at the beam center with a laser DT. The displacements of the lower post-beam connections were measured vertically and horizontally with DTs. The applied forces were positive- and negative-alternating cyclic forces with the cyclic test schedule shown in Figure 11.

Table 3. Amplitudes of each cyclic step.

Step	Amplitude [mm]	Shear Deformation Angle [rad]
1	6	1/450
2	9	1/300
3	14	1/200
4	18	1/150
5	27	1/100
6	36	1/75
7	55	1/50
8	91	1/30

shows the amplitudes of each cyclic step, where the shear deformation angle is defined as the ratio of the amplitude to the distance between the upper and lower beam (2730 mm) [39]. The number of cycles on each step was set at one.

3. Results and Discussion

The seismic performances of the proposed technique and typical STCs were discussed according to the pullout and full-scale in-plane cyclic test results. The discussion is performed using bilinear diagrams and seismic performance parameters obtained from the load–displacement or envelope curves of the test results.

Figure 12 shows the method to obtain the bilinear diagrams. They were obtained from the equivalence of areas that were made between the curve and bilinear diagram [29], where $F_{max}$ denotes the maximum load, $d_u$ denotes the ultimate displacement as the displacement at 80% of the $F_{max}$, $F_y$ denotes the yield load obtained from the equivalence of the areas, and $d_y$ denotes the yield displacement as the displacement at the intersection point of the linear elastic (slope $K$) and $F_y$ line.

The initial stiffness $K$ was obtained from Equation (1) according to ISO 21581 (2010) [40]:

$$K=\frac{0.3 F_{max}}{l_{40\% F_{max}}-l_{10\% F_{max}}}$$

(1)

where $l_{40\% Fmax}$ and $l_{10\% F_{max}}$ denote the displacements at 40% and 10% of the $F_{max}$, respectively. Moreover, the displacement ductility $\mu$ was defined as the ratio of $d_u$ to $d_y$ [29]:

$$\mu =\frac{d_u}{d_{y }}$$

(2)

3.1. Pullout Tests

Figure 13 shows the load–displacement curves and their bilinear diagrams. Moreover, the benchmark diagram of the CP-T type, including the safety rate, is added in Figure 13. This benchmark diagram set in the Detailed Seismic Evaluation and Retrofit of Wooden Houses [41] by the Japan Building Disaster Prevention Association (JBDPA) is the performance that the CP-T type should have at the minimum. The proposed technique aims for the equivalent performance to the benchmark STCs, but the performance ($F_y$) is finally evaluated as the same as the benchmark STCs’ performance for the actual seismic design of existing TF houses.

In terms of initial stiffness, yield load, and ultimate displacement, the P and N performed better than the CP-T type and the benchmark.

Table 4. Seismic performance parameters from the bilinear diagrams of the pullout tests.

Specimen	${\mathit{F}}_{\mathit{y}} \left[\mathbf{kN}\right]$	$\mathit{K}$ [kN/mm]	$\mathit{\mu }$ [-]
P and N 1	11.62	4.20	9.98
P and N 2	11.37	3.60	10.48
CP-T type	9.68	1.87	4.12
Benchmark of CP-T	7.40	1.21	1.64

presents the seismic performance parameters. A significant difference was observed in the $K$ and $\mu$ values; the values of P and N were about twice and more than twice as high as those of the CP-T type, respectively. All failure modes were pulling out of the nails.

3.2. Full-Scale In-Plane Cyclic Tests

Figure 14 shows envelope curves and their bilinear diagrams. Moreover, the benchmark load–displacement curve (monotonic tensile test) of the VP and BP type, including the safety rate, is added in Figure 14. This benchmark curve was also set in the Detailed Seismic Evaluation and Retrofit of Wooden Houses [41] by the JBDPA and is the performance that the VP and BP type should have at the minimum. The performance ($F_y$) of the proposed technique is finally evaluated as the same as the benchmark STCs’ performance for the actual seismic design of existing TF houses.

Because the proposed technique strengthens the connections against tensile load, the curves were evaluated under tensile load. The initial stiffnesses of the P and N and VP and BP types were generally close, but there was a significant difference in ultimate displacement because the tests were terminated when the brace of the VP-and-BP-type specimens buckled in the compression load, and such buckling was not observed in the P and N specimens. The initial stiffness of the benchmark is above them because of the not-cyclic nature of the monotonic tensile test.

Table 5. The seismic performance parameters from the bilinear diagrams of the full-scale in-plane cyclic tests.

Specimen	${\mathit{F}}_{\mathit{y}} \left[\mathbf{kN}\right]$	$\mathit{K}$ [kN/mm]	$\mathit{\mu }$ [-]
P and N 1	4.21	0.092	4.26
P and N 2	4.06	0.096	5.74
VP and BP type 1	3.49	0.077	1.14
VP and BP type 2	3.70	0.098	1.50
Benchmark of VP and BP	2.84	0.169	5.49

presents the seismic performance parameters. The $F_y$ and $\mu$ values of the P and N are about 15% and 3–4 times higher than those of the VP and BP type, respectively. The $\mu$ value of the benchmark bilinear diagram is also as good as the P and N under tensile load. However, there is some risk of brace buckling under compression load.

Figure 15 shows the ultimate failure of post-beam-brace connections in the P and N; CN50 pulled out of the braces. However, there was no damage to TFs such as posts, beams, and braces. The P and N is better for reducing the risk of brace buckling with excellent ductility.

4. Conclusions

This study proposed a simple and easy seismic retrofit technique, the proposed technique, for TF connections using only plywood and nails, equivalent to the benchmark STCs in Japanese detached houses. Experiments were performed to compare the proposed technique with a CP-T-type STC using pullout tests on a post-beam connection and a VP-and-BP-type STC using full-scale in-plane cyclic tests on a post-beam-brace connection. From the experimental results, the following conclusions were obtained:

• In the pullout test results, the proposed technique had about 18% higher yield load, about twice the initial stiffness, and more than twice the displacement ductility of the CP-T-type STC. All failure modes were pulling out of the nails.
• In the full-scale in-plane cyclic test results, the proposed technique had about 15% higher yield load and 3–4 times higher displacement ductility than the VP-and-BP-type STCs. The initial stiffnesses were generally close.
• In the ultimate failure of the connections in the full-scale in-plane cyclic test results of the proposed technique, there was no damage to TFs; only CN50 nails pulled out of the braces. The brace of VP-and-BP-type specimens buckled in the compression load.

Those results show that the proposed technique is particularly excellent in displacement ductility than the STCs, which was to prevent a collapse of the TF house without damaging the timbers. The proposed technique will be accepted by many carpenters when retrofitting earthquake-prone existing houses because it is simple and easy.