Experimental analysis on strength and failure modes of wood beam-column connections

Abstract

This research experimentally examined the strength, failure modes, and behaviors of dowel-bearing and fiber-bearing wood beam-column connections and explored the effects of cyclic loading on the strength, failure modes, and behaviors of those connections. Base on limited numbers of exploratory laboratory tests (6 preliminary tests in total), the authors observed that the typical bolted connection (dowel-bearing type wood beam-column connection with fiber-bearing surfaces) showed good behavior (large peak moment) under the monotonic loads, and the tenon joint connection (fiber-bearing wood beam-column connection) showed good behavior under cyclic loads. The cyclic property of loading reduced the strength of the dowel-bearing type wood beam-column connections, but increased the strength of fiber-bearing type wood beam-column connections. More importantly, the authors identified a possible location of safety concern in current national design specifications (NDS) standards for the typical bolted connection (dowel-bearing connection with fiber bearing surface) under cyclic loading because the tested value was smaller than the NDS calculated value. But, because of the small amount of tests conducted, no final conclusion can be drawn based on those preliminary observations yet. A large number of repetitive laboratory tests should be conducted.

Keywords

wood / connection / dowel-bearing / fiber-bearing

Introduction

The wood beam-column frame is a commonly used structural system for residential buildings in the United States, and was the most popular building structure system in ancient Asia, such as in ancient China, Korea, Japan, etc. The typical wood beam-column frame is an assembly of wood beams and columns connected by nails, bolts, metal straps, or proprietary connectors, categorized as dowel-bearing connections. Some wood connections do not need a connector, such as tenon and mortise connection. The tenon joint connections are popular in furniture application in the US and in building construction in ancient Asia. The connection details for ancient Asia’s wood connections, such as types, dimensions, shapes, and functions, were summarized by Li [1] and Ma [2]. The tenon joint connection is categorized as wood fiber-bearing connections.

A great number of studies regarding the wood beam-column frames and wood connections have been carried out recently. Song and Lam [3] explored the stability capacity and lateral bracing force of wood beam-column frames subjected to biaxial eccentric compression loading. By taking the nonlinear parallel-to-wood-grain stress–strain relationship, size and stress distribution effects of wood strength, shear deformation, and the P-Delta effect of compression load into account, a numerical analysis model based on the column deflection curve method was developed. The model was verified by the biaxial eccentric compression tests of wood beam-column frames. Xing et al. [4] investigated the dynamic properties of conventional beam-column timber frame structure under successive damage through a full-scale, three-story conventional timber structure. A three-dimensional finite element model was used for predicting its natural frequency. Both the simulated and the experimental results showed a similar trend in most test cases. Damage sensitivity, as well as the influence of temperature and humidity on the natural frequency was also examined. The results from this study could benefit the research of structural health monitoring. O’Loinsigh et al. [5] examined the performance of full-scale multi-layered timber beams with composite action achieved with welded-through wood dowel connections. Different multi-layer beam designs, where the timber layers were interconnected with welded wood dowels providing interlayer shear resistance, were tested in bending with different dowel densities. The results demonstrated that the multilayered timber sections by wood dowels were structurally efficient and do not require non-wood based joining agents such as nails or adhesive. Using densified veneer wood (DVW) reinforcement and expanded tube connectors, a great moment transferring capacity of timber connections can be achieved. Leijten and Brandon [6] proved that, when certain conditions were fulfilled, two connections in series had the same rotational stiffness as one while the bending moment capacity increases. The rotational stiffness of two DVW reinforced connections joined in series by a steel plate in a splice and column-beam type of connection. The test results confirmed the isotropic properties of this connection type.

This study examines experimentally the strength, failure modes, and behaviors of the commonly used dowel-bearing and fiber-bearing wood beam-column connections and explores the effects of cyclic loadings. As an internationally and widely accepted design manual for the wood structures, national design specifications for wood construction (NDS) [7] describes procedures calculating the reference and adjusted force/moment strength values and predicting the failure modes of both types of wood connections (dowel-bearing and fiber-bearing connections). Through limited number of exploratory laboratory tests (6 preliminary tests in total) on the dowel-bearing and fiber-bearing wood beam-column connections and comparison of the preliminary test results with the NDS calculated values and predicted failure modes, the researchers of this study try to either validate the safety conservativity and/or identify possible safety concerns for wood connections in current standards.

Wood beam-column connection tests

Test specimen

Monotonic and cyclic tests on three wood beam-column connection configurations were conducted: 1) typical bolted wood beam-column connections (Fig. 1), 2) simplified bolted wood beam-column connections (Fig. 2), and 3) the tenon joint wood beam-column connections (Fig. 3). The configurations 1) and 2) are the dowel-bearing type connections, and the configuration 3) is a fiber-bearing type connection. Two assemblies were built for each configuration, one for the monotonic pushover test and the other for the cyclic test. Each assembly used two No. 2 kiln dried 38 mm × 234 mm (2″ × 10″) southern pine lumbers as beam members and one No. 2 air-dried 140 mm × 140 mm (6″ × 6″) cypress timber as the column member.

The typical bolted wood beam-column connection (configuration 1) has been widely used in wooden deck construction in the United States. The procedures described in the DCS [8] were used to build the assemblies. Two 12.7 mm (1/2 in) diameter through bolts were used as the connectors. The pair of bolts was placed at 114 mm (4-1/2 in) apart aligned with the centerline of the post. A 140 mm by 234 mm (6″ × 10″) notch was cut in the post so that the beam can sit flush with the 140 mm × 140 mm (6″ × 6″) post with enough fiber-bearing strength to help support the load.

For the simplified bolted wood beam-column connection (configuration 2), instead of notching the post, the beam was set flush on the post, and the 38 mm × 234 mm (2″ × 10″) beams were attached to the side surfaces of the 140 mm × 140 mm (6″ × 6″) post. Two 12.7 mm (1/2 in) diameter and 127 mm (5 in) long lag bolts were used on each side. The loads from the beam were carried only by the shear strength of the lag bolts.

The tenon joint wood beam-column connection (configuration 3) consisted of a male part, the tenon, in the beam and a female part, the mortise, in the column. The specimen assembly for the tenon joint connection in this study was a simplified version of Ma [2]. It should be noted that the fabrication procedure of the ten on joint connection is much more complicated and takes much longer time than the bolted connections.

Test setup

All the 6 preliminary laboratory experiments were conducted on a 2.5 m (8′2″) × 2.5 m (8′2″) adaptable steel testing frame. Four W16 × 67 hot-rolled steel members (406 mm high and 100 kg/m weight) were used for the frame beams and columns. The testing frame was equipped with a 155.7 kN (35 kips) hydraulic actuator with a ±127 mm (5 in) stroke. An 89.0 kN (20 kips) universal compression-tension load cell was attached to the actuator for the force measurement. A steel rod at the bottom of the load cell was used to prevent the out-of-plane displacement of the test sample, and a pin connection at the top of the steel rod released the bending moment from the arm of the hydraulic actuator.

Figure 4 shows the test setups for (a) a monotonic test and (b) a cyclic test. In Fig. 4(a), a supporting bar made by a 25.4 × 25.4 × 2.85 mm (1 × 1 × 1/8 in) hollowed steel tube was bolted to the testing frame to help control the horizontal and vertical displacements of the column. Four 12.7 mm (1/2 in) diameter anchor bolts with standard cut washers were used to attach the test sample to the testing frame and the supporting bar. In Fig. 4(b), a supporting truss instead of the supporting bar and two Simpson Strong-Tie S/HD10S hold-downs instead of four bolts were used to hold the column in position, controlling the uplifting displacement during the cyclic tests. The loadings of the tests were applied vertically on the top of the beams.

Two 32 mm (1-1/4 in) inner diameter washers were used at each loading contact point to better distribute the force on the surface of the wood beam, and a 229 mm (9 in) position transducer was employed to measure the vertical displacement of the loading point on the beam (Fig. 5). The transducer was mounted on the steel testing frame. The applied force and the vertical displacement were measured and recorded simultaneously during the test.

Test protocol

Both the 3 monotonic and the 3 cyclic tests were conducted under the displacement control procedure. The loading rate for the monotonic pushover tests was 6.35 mm/min (2.5 in/min) until failure. The CUREE (Consortium of Universities for Research in Earthquake Engineering) protocol, in accordance with the Method C in ASTM E2126 [9], was chosen for the cyclic tests. A constant cycling frequency of 0.2 Hz in the CUREE loading history was used for all the cyclic tests in this study. The CUREE basic loading history including 40 cycles is shown in Fig. 6. Because of the vertical position offset in setting up the test samples, three different specified displacement amplitudes were used in this study: 75 mm (2.95 in) for the typical bolted connection, 76 mm (3 in) for the simplified bolted connection, and 66 mm (2.60 in) for the tenon joint connection. Table 1 summarizes the test matrix.

Dowel bearing and fiber bearing connections in NDS

The adjusted strength values and failure modes of the tested wood beam-column connections were estimated using the equations in NDS [7]. The yield limit (Z) equations were used to calculate the reference bearing force values (by setting the reduction term R_d equal to 1.0) of each fastener for the dowel-bearing connections (configurations 1 and 2 in this study). Four yielding modes are defined in NDS [7]. Mode I is that the fastener crushes the framing members (main or the side member) when the dowel-bearing stress passes the yield limit of the framing members. Mode I includes I_m for crushing in main member and I_s for crushing in side member. Mode II is that the fastener rotates and crushes the outside surface of each member when the dowel-bearing capacity of both framing members (beam and column) goes beyond their yield limit. Mode III is that a plastic hinge forms in the fastener close to the shear plane. The formation of the plastic hinge is followed by crushing in a frame member. Mode III also includes III_m for crushing in main member and III_s for crushing in side member. Mode IV is that two plastic hinges form near the shear plane of the connection. The yield limit equations for the yielding modes I through IV are listed as Eq. 11.3-3 through 11.3-6 in NDS [7].

The adjusted bearing force value of the dowel-type fastener is equal to the reference bearing force value (by yield limit equations) of the fastener multiplied by applicable adjustment factors, such as the wet service factor, size factors, etc. Based on the adjusted bearing force value of one single dowel-bearing fastener, equilibrium equations were then built using the free body diagrams in Fig. 7(a) and (b) to calculate the adjusted bending moment value that the connection can carry. In Fig. 7(a) and (b), the forces P₁ and P₂ represent the adjusted bearing forces of the single dowel-type fastener calculated by the yield limit equations. When P₁ and P₂ for the typical US connection (Fig. 7a) were calculated, in order to obtain a more accurate estimation, the origin point was assumed at the maximum wood bearing point. This assumption was based on the observation on the tested specimen.

For the tenon joint connection, forces P₁ and P₂ in Fig. 7(c) were calculated differently because no fastener was used. Since fiber yielding on the bearing surfaces was the failure mechanism for this type of connection, the reference compression (perpendicular to grain) design value equation (NDS 2012 [7] Tables 4A through 4F) were used in the calculation. The bearing area of 33.87 cm² (based on the observation on the tested specimen)and the effect of loading direction (the angle between the force and the grain) were taken into account in this analysis.

Results and discussion

M-θ curves, the bending moment (M) vs. the angular deformation (rotation) (θ) between the beam and column member, were used to show the test results of the monotonic and the cyclic pushover tests. The M-θ relationship has been widely utilized in commercial software, such as in SAP2000, RISA, and DRAIN, to depict the properties of a connection. The bending moment, M, was calculated as the tested vertical load multiplied by the horizontal distance between the loading point and the surface of the 140 mm × 140 mm (6″ × 6″) timber post. The vertical load applied on the top of the beam was measured by the load cell attached to the hydraulic actuator. The rotation, θ, was defined as the ratio of the vertical deflection of the loading point over the horizontal distance between the loading point and the post surface. The vertical deflection at the loading point was measured by the position transducer.

Figure 8 presents the resulting M-θ curves for (a) monotonic pushover tests and (b) cyclic pushover tests. Monotonic tests have commonly been used to simulate the cases of static gravity loading, such as dead loads, live loads, snow, rain, and construction loads, etc., or to simulate the uni-directional lateral loading cases, such as wind loads. The cyclic pushover tests were used to simulate bi-directional dynamic loading cases, such as seismic load cases, and the bi-directional wind loads.

The tested peak moments (shown in Fig. 8) and the adjusted bending moment design value (by NDS 2012 [7]) are compared in Tables 2 through 5.

Table 2 shows a comparison of the tested peak moments between the bolted connections and the tenon joint connection for both monotonic and cyclic tests. The positive moment (M+) in the table represents the clockwise direction moment. The percent differences in the table were calculated as bolted connection over the tenon joint connection (\mathrm{\%}Diff={\displaystyle \frac{bolted-tenon}{tenon}}). It can be seen from Table 2 that under the monotonic loading, the typical bolted connection showed a higher peak moment than that of the tenon joint connection, while the simplified bolted connection showed a lower peak moment compared to the tenon joint connection. The possible reason is that the typical bolted connection has a wood fiber bearing in addition to the connection through the bolts. During the monotonic loading, because both fiber-bearing (between the beam and the column) and dowel-bearing (between the members and bolts) affect the strength of connection, the peak moment was increased compared to the tenon joint connection. Since the simplified bolted connection only relied on the dowel-bearing strength between the bolts and the wood members, its peak moment was lower compared to the tenon joint connection. For the cyclic loading, the tenon joint connections showed larger peak moments than those of the bolted connections in both clockwise and counterclockwise directions. The reason is that the tenon joint connection relied on the fiber-bearing between the wood components, while the bolted connections depended on the dowel-bearing between the wood components and the metal bolts. The yielding on the fiber-bearing contact surface was recoverable during the cyclic loading, but not for the dowel-bearing contact surfaces.

Table 3 shows a comparison of tested peak moments between the cyclic tests and the monotonic tests. Same as in Table 2, the positive moment (M+) in Table 3 represents the clockwise direction moment, and the percent differences were calculated by \mathrm{\%}Diff={\displaystyle \frac{cyclic-monotonic}{monotonic}}. As shown in Table 3, the bolted connections (both typical and simplified) under cyclic loading gave lower peak moments compared to the monotonic loading with a reductionofaround10% for most cases (9.5% for simplified bolted connection of both positive and negative moments, 14% for the typical bolted connection of the positive moment), except for the typical bolted connection in the negative direction (about 52%). However, for the tenon joint connection, the results were opposite. The cyclic loading showed higher peak moments (in a range of 11.7%–20.2%) than that of the monotonic loading, because of the recoverable yielding on the fiber-bearing contact surfaces as well. This indicates that tenon joint connections (fiber-bearing type connections) perform better under bi-directional wind or earthquake loading.

Tables 4 and 5 show the comparisons between the tested peak moments and adjusted bending moment values (calculated through NDS 2012 [7]). Tables 4 and 5 compare the monotonic and cyclic test results, respectively. The ratios ({\displaystyle \frac{test}{NDS}}) are also presented in Tables 4 and 5, which indicate “safety margins” for the connections. For the negative moment of the typical US connection in Table 5, the ratio was smaller than 1.0, indicating that the connection could not reach the NDS provided adjusted strength value. This preliminary test result points out a possible safety concern here. But, conclusion cannot be drawn based on a single test result. A large number of repetitive laboratory tests should be conducted to validate the finding. This will be a future research topic for the research team.

Figures 9–11show the comparisons of the tested yielding modes with the NDS [7], predicted yielding modes for the bolted connections and the tenon joint connections, respectively. It can be seen that the yielding modes predicted by NDS [7] were comparable with the tested results.

Conclusions

Six full-scale preliminary monotonic and cyclic pushover tests were conducted on three commonly used wood beam-column connections: typical bolted wood connection, simplified bolted wood connection, and tenon joint connection. The tested results were compared with adjusted strength values and predicted yielding modes from NDS [7]. The following preliminary observations were drawn based on the limited test and analysis data:

1) Among the three tested beam column connection configurations, the typical bolted connection (dowel-bearing type connection with fiber bearing surfaces) behaved best (largest tested peak moment) under monotonic loads, while the tenon joint connection (fiber-bearing type connection) behaved best under cyclic loads.

2) The dowel-bearing type connections had lower tested peak moment under cyclic loadings than that under monotonic loading, and the fiber-bearing type connection had higher tested peak moment under cyclic loadings than that under monotonic loading. This observation indicated that the fiber-bearing type connection performed better for the structures under the bi-directional wind or earthquake attack.

3) Most of the tested peak moments were larger than the adjusted moment strength values calculated by NDS [7] with an exception of the typical bolted connection (dowel-bearing connection with fiber bearing surface) under cyclic loading. This exception pointed out a possible location of safety concern in current standards. Large numbers of tests are needed to validate the safety concern and propose a solution.

4) NDS [7] accurately predicted the failure (yielding) modes for both the dowel bearing and wood fiber bearing type of connections,

All the above preliminary observations cannot result in any proposable conclusion yet because of the small number of tests. A large number of repetitive laboratory tests should be conducted to validate the findings.

References

[1] Li J. Ying Zao Fa Shi. Royal Press, Kaifeng, China (Song Dynasty), 1100 (in Chinese)

[2] Ma B J. Construction Technology of Chinese Ancient Timber Architectures. 2nded. Beijing: Science Press, 2003 (in Chinese)

[3] Song X, Lam F. Laterally braced wood beam-columns subjected to biaxial eccentric loading. Computers & Structures, 2009, 87(17–18): 1058–1066

[4] Xing H, Xue S, Zong G. Dynamic properties of conventional beam-column timber structure under successive damage. Journal of Asian Architecture and Building Engineering, 2011, 202(1): 195–202

[5] O’Loinsigh C, Oudjene M, Ait-Aider H, Fanning P, Pizzi A, Shotton E, Meghlat E M. Experimental study of timber-to-timber composite beam using welded-through wood dowels. Construction & Building Materials, 2012, 36: 245–250

[6] Leijten A J M, Brandon D. Advances in moment transferring dvw reinforced timber connections- Analysis and experimental verification, Part 1. Construction & Building Materials, 2013, 43: 614–622

[7] NDS. National Design Specification for Wood Construction ASD/LRFD. American Forest & Paper Association, Washington, DC, 2012

[8] DCS. Deck Construction Specifications of the Building Department of the City of Lancaster, Lancaster, OH, 2004

[9] ASTM E2126. Standard Test Methods for Cyclic (Reversed) Load Test for Shear Resistance of Walls for Buildings, American Society for Testing and Materials, West Conshohocken, PA, 2007

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg