CHARACTERISTICS OF LUMBER-TO-LUMBER FRAMING CONNECTIONS IN LIGHT-FRAME WOOD STRUCTURES

CHARACTERISTICS OF LUMBER-TO-LUMBER FRAMING CONNECTIONS IN LIGHT-FRAME WOOD STRUCTURES Andi Asiz 1*, Lina Zhou 1, Ying Hei Chui 1 ABSTRACT: Connections are arguably the most important part of light-frame wood structures (LFWS). They hold structural components together and transfer forces from one component to another. The majority of connections in LFWS are sheathing-to-lumber frame and lumber-to-lumber framing connections. Accurate prediction of structural response from analytical or numerical models of LFWS depends largely on accurate characterization of connection behaviours, which are normally non-linear. The main connection property required for analytical or numerical model of LFWS is -slip response. Most of the connection test data available in the literature have been used for deriving connection design values (yield, ultimate, ductility), and there is limited amount of connection deformation data for use in whole-building numerical models. The main objective of this study is to test various nail lumber-to-lumber framing connections used in LFWS to generate input properties into 3-d numerical models. Tests were conducted on roof-to-wall connection, internal-to-external wall connection, and wall to-floor connection. Loadslip responses under applied in different directions were determined. Furthermore, this paper discusses the slip response and failure mechanism of each type of connection, and their implication on 3-d system responses. KEYWORDS: Light-frame wood structures, nailed connections, -slip responses, connection model, 3-d building model, system response 1 INTRODUCTION 123 Connections are arguably the most important part of light-frame wood structures (LFWS). They hold structural components together and transfer forces from one component to another. The majority of connections in LFWS are composed of nail connections, which in general can be divided into two parts: sheathing-tolumber frames connections; and lumber frame-to-lumber frame connections. The response of nail connection under applied is normally non-linear and depends on various parameters such as wood moisture content, wood species, ing direction (parallel or perpendicular to wood grain) and fastener geometry and configuration. Accurate predictions from analytical or numerical models of LFWS depend largely on the availability of accurate connection properties, which are normally derived from connection tests. The main connection property required for numerical modelling is the -deformation (slip) response. Several data points in the curve are needed to represent the non-linear response until a peak or post-peak is reached. In the numerical model using finite element, the connection property is normally represented by non- 1 Faculty of Forestry and Environmental Management, University of New Brunswick, Fredericton, New Brunswick, CANADA, E3B-6C2. (*Corresponding author s e-mail: aasiz@unb.ca) linear spring composed of axial, shear and rotational degrees of freedom. Fig.1 shows an illustrative example of roof-to-wall connection of LFWS that can be represented by this type of spring behaviour. A (a) typical one-storey residential LFWS Toe-nail connection Typical roof truss Truss bottom chord Double top-plate (b) roof-to-wall toe-nail connection (detail A)

connection modeled as spring (link) element with: two lateral/shear stiffness properties withdrawal/uplift stiffness property (c) spring element model slip 2.1 TEST DESCRIPTION The first part of the test program was to study the behaviour of roof-to-wall connection subjected to static. Common method of connecting roof to wall components is by toe-nailing the (38mm x 89mm) bottom chord element of roof-truss to the double topplate (2-) of wall using 3-8d or 2-16d common nails, Fig. 1b [6]. In this study, since the wall frame system consisted of (38mm x 140mm) lumber including the double top-plate, 3-16d common nails (diameter=4.1 mm, length= 89 mm) were used. The test was designed to accommodate independent ing in each major direction of the toe-nail connection, one in uplift and two in shear (Fig. 2). The ing direction here is referred to the wood grain direction of the roof-truss. For example, shear perpendicular means the is applied perpendicular to the wood-grain of the bottom chord lumber. No combined s were applied in this test. As shown in Fig. 2, symmetrical test arrangements were employed to avoid rotational movements induced by eccentric (un-symmetric) ing. Roof-truss bottom cord () Toe-nail (d) typical slip response (lateral directions) Fig. 1: Connection model in LFWS analysis Most of the connection test data available in the literature have been used for deriving connection design values (yield, ultimate, ductility), and there is limited amount of connection -deformation data for use in whole-building (3-D) numerical models. The main purpose of this connection test program is to fill gaps in input data to 3-D numerical models of LFWS, pertaining specifically to non-linear -deformation (slip) response under static. After reviewing and examining previous connection test programs in the literature, it was observed that majority of test data available are for nail connections in wood shear wall sub-assemblies (e.g. [1-4]), and the following connection tests are needed to complement those existing data: Roof-to-wall connection: toe-nail joint between bottom chord of roof truss and double top plate, Internal-to-external wall connection: nail joint between internal wall-stud and external wall-stud frames, and Wall-to-floor connection: nail joint between bottom plate and floor joist. In this study, the test set-ups were developed based on ASTM D1761 Testing Mechanical Fasteners in Wood [5] with adjustments in boundary conditions and application. Although in the numerical model only the average response of -slip is needed, all related properties, including failure mechanism, ultimate, yield, and ductility, were examined. Wall double top-plate () (a) toe-nail connections with three ing scenarios double top-plate 3-16d toe-nails Load (b) uplift test double double 2 ROOF-TO-WALL CONNECTION (c) shear parallel 3-16d toe-nails

double ¼ steel plate double 3-16d toe-nails (d) shear perpendicular Fig. 2: Roof-to-wall toe-nail connection tests Wood specimens included (roof truss) and (wall double top-plate) S-P-F lumber of No 2 grade or better. The specimens were conditioned in environmental chamber at 65% RH and 20 0 C for adjustment of the lumber moisture content to 12 percent. The specific gravity (SG) was computed from the dimensions and weight of the specimen at time of test and moisture content. These lumber pieces were all cut into 300 mm length. The lumbers were connected using 3-16d common nails using toe-nailing technique. To avoid wood splitting, pilot holes were drilled into the lumber of the specimen assembly at 30 degrees at marked locations using a drilling tool. Plastic sheets were inserted between the member and s to prevent excessive friction that could influence the slip response of the connection (Fig. 2b). Once the specimen was assembled in a universal test machine, compression ing mode was applied to the component, except for the shear-parallel applied on the element (Fig. 2b), at a rate of 3 mm/minute until failure was reached or until excessive separation (gap) between the connected elements was detected. Two displacement transducers (LVDTs) were attached symmetrically with respect to the centre line of ing to record slip between the two connected components. The connection slip is determined using the average value of the two LVDTs. 10 specimens were tested for each ing direction. 2.2 TEST RESULTS Failure mode provides important information regarding whether the connection can be appropriately modelled using the non-linear spring element. If the failure mode is brittle (e.g. wood splitting), the connection -slip response is generally linear and can be adequately represented by the initial (linear) connection stiffness. As was expected, test results indicated that the failure mode for the toe-nail connections was ductile, indicated by nail withdrawal for the uplift case and excessive bending of nails for the lateral (shear) cases. No splitting of wood was observed in the test. Table 1 shows a summary of the test results, which includes peak (P u ), yield (P y ), and slip at yield (δ y ), slip at peak (δ u ), and ductility ratio for each direction. Yield was computed as the intersection point between the -slip curve and a line having the same slope as initial -slip curve and is offset horizontally by 5% of the fastener diameter [5]. Displacement at yield was also determined to investigate the ductility response of the connection, which is defined as the ratio between displacements at peak and yield. Maximum s were taken as peak values irrespective of the associated displacements. In general, it can be noted that the variability in the connection response under uplift was higher compared to those of the lateral s. Also, as was anticipated, the peak values are lower in the uplift case compared to those of the lateral s. Table 1: Roof-to-wall toe nail connection test result SG P u, P y, δ u, kn kn mm Uplift 0.42 2.45 2.07 4.19 (0.05) (0.29) (0.29) (0.38) Shear 0.42 4.73 2.95 6.14 parallel (0.05) (0.05) (0.16) (0.11) Shear 0.42 3.29 2.13 8.38 perp (0.05) (0.06) (0.17) (0.34) Note: values in parenthesis are C.O.V. Load-slip curves provide important information on the overall response of the connection under. Fig.3 shows the average responses of -slip curves for each ing direction determined using curve-fitting leastsquare method. An overall observation is that the connection should be modelled as ductile (non-linear) in the three directions. To be useful as input in the numerical models, several data points in the curve need to be selected in a way that closely reflect the characteristics of the curve. For example, at low level small interval between the data points in the horizontal axis (slip) is needed to capture the linear response indicated by an abrupt change in the vertical axis () values. Table 2 summarizes the data points selected for each ing direction that were used for the numerical modelling input. Only positive -slip responses were plotted, and it is presumed that in the negative directions the -slip response is symmetrical for the shear tests (see Fig. 1d). For the axial direction, when negative (compression mode) occurs, deformation curve for compression parallel to the grain of wood was used [4]. Load Slip (mm) Fig. 3: Average -slip response of the roof-wall toenail connection δ y, mm 0.78 (0.24) 0.83 (0.42) 1.7 (0.4) Shear parallel uplift Duct. 5.83 (0.51) 8.12 (0.26) 5.6 (0.7) Shear perpendicular

Table 2: Numerical data points used as input for numerical model slip/def (mm) Shear Perp* Shear Parallel* Uplift 0 0.01 0.52 0.02 0.5 1.62 4.56 1.69 1 2.98 6.22 2.10 1.5 4.03 7.07 2.21 2 4.77 7.68 2.27 3 5.53 8.61 2.40 4 5.79 9.09 2.37 5 5.96 9.25 2.34 6 6.17 9.35 2.30 7 6.40 9.41 2.29 8 6.55.33 2.27 9 6.58 9.11 2.26 10 6.50 8.88 2.25 11 6.42 8.70 2.21 12 6.37 8.44 2.15 13 6.33 8.10 2.08 14 6.28 7.88 2.06 was obtained. Ten specimens were used for each ing direction. Internal wall studs (a) Interior-to-exterior walls connection OSB-to-stud spring model OSB Int-to-ext wall spring model plasterborad 16d nail 12 (305 mm) o.c. Exterior wall plasterboard Note: *denotes the overall connection response with 3-16d common nails toe-nailing on each side of the topplate (symmetrical arrangement). In the numerical model, this implies the s are half of these values recorded. 3 INTERIOR-TO-EXTERIOR WALLS CONNECTION 3.1 TEST DESCRIPTION Fig 4a illustrates one of few ways in connecting (38mm x 89mm) interior wall-frame to (38mm x 140mm) exterior wall-frame of LFWS, and Fig 4b illustrates its connection modelled as spring element. In the test, again symmetrical ing arrangement was built by connecting two- studs to a stud with one-16d common nail on each side (Figures 6.4c, d). Plasterboard of thickness 3/8 (9.5mm) was inserted between these two studs. Two lateral (shear) s were applied with respect to the wood grain direction in the stud, one perpendicular and another parallel. No axial direction (with respect to the nail position) was applied, since there are some previous data available that can be used to closely represent the nail withdrawal behaviour between studs (e.g. [3-4]). To reduce rotation during the perpendicular ing, the plasterboard was extended beyond the side surface area and frictionless plastic material was inserted (Fig 4b). The lumber used was SPF grade No. 2. Both lumber sizes (24 and ) were cut into the same lengths of 300 mm. Note that in practice nailing between interior and exterior studs were applied at a spacing between 12 and 24 (300-600mm). Two LVDTs were attached to record the slips between the -stud and the plasterboard and between the plasterboard and the stud. Loading was applied at the same rate as the previous toe-nail test, 3 mm/minute until failure was observed or until excessive slip or big separation between connection components (b) connection spring model (c) perpendicular ing test 16d nail Gypsumboard Frictionless plastic 16d nail Gypsum wallboard (d) parallel ing test OSB Frictionless plastic material Fig. 4: Testing of interior-to-exterior walls connection 3.2 TEST RESULTS The majority of the test results indicated that ductile failure mode in the form of excessive bending of nail in combination with wood and plasterboard crushing was

Load observed when the was applied parallel to the wood grain. Few specimens failed in brittle mode indicated by wood splitting in member when the was applied perpendicular to the grain. Table 3 summarizes the test results including peak (P u ), yield (P y ), and slip at yield (δ y ), slip at peak (δ u ), and ductility ratio for each ing direction. Fig. 5 shows the average -slip responses determined using least-square curve fitting method. The -slip responses plotted were the total slips recorded between the and studs, which were assumed by simply adding the two slip signals recorded with two LVDTs. In general, ductility ratios determined were higher than those obtained from the previous toe-nail connection test, meaning that the connection response needs to be modelled non-linearly. It can also be noted that the connection stiffness at low level between the parallel and perpendicular are almost the same. Table 4 shows the data points representing connection -slip response used as input into the numerical models. Again, for the negative directions, the -slip responses were assumed symmetrical to that of the positive ing direction. Table 3: Test result for interior-to-exterior walls connection SG P u, P y, δ u, kn kn mm Perp. 0.42 1.77 0.88 11.7 (0.05) (0.10) (0.14) (0.23) Par. 0.42 1.60 1.02 14.5 (0.05) (0.10) (0.06) (0.38) Note: values in parenthesis are C.O.V. Slip (mm) Fig. 5: Load-slip for the Interior-to-exterior walls connection δ y, mm 1.7 (0.46) 2.1 (0.22) Shear perp Shear parallel Duct. 9.5 (0.53) 7.4 (0.55) Table 4: Numerical data points used as input for numerical model slip/def (mm) Shear Perp* Shear Parallel* 0.0 0.06 0.26 0.5 1.42 1.03 1.0 1.83 1.50 1.5 2.01 1.80 2.0 2.17 2.01 3.0 2.49 2.30 4.0 2.72 2.53 5.0 2.88 2.71 6.0 3.03 2.83 7.0 3.15 2.91 8.0 3.24 2.98 9.0 3.33 3.04 10.0 3.40 3.09 11.0 3.43 3.12 12.0 3.44 3.14 13.0 3.44 3.14 14.0 3.42 3.13 15.0 3.37 3.11 16.0 3.30* ) 3.10* ) Note: *) In the non-linear modeling analysis with control, one data point after the peak point is reached is the minimum requirement to complete the -slip path due to numerical instability reaching post peak analysis. This is not the case if displacement control is used. 4 WALL-TO-FLOOR CONNECTIONS 4.1 TEST DESCRIPTION The third part of the test program was conducted to obtain the connection response between wall and floor frames. Fig 6a shows a typical external wall-to-floor connection using 2-16d common nails spaced at 300 mm along the bottom wall plate. Fig 6b illustrates the modelling strategy for the wall-to-floor connection involving lumber studs, sheathing, and nails. No wall sheathing and its nails are shown here. The bottom plate () and 5/8 (15.5 mm) OSB sub-floor were joined using 2-16d nails on each side of the 2x10 joist (Figs 6c, d). As the previous two connection tests, all lumber pieces used were SFP grade No 2 or better grade. These lumber pieces (s and 2x10s) were cut into the same length, 300mm. Loading rate was 3mm/minute until failure was reached or excessive separation between components was observed. Ten specimens were tested for each ing direction.

wall studs 2-16d common nails sub-floor Floor joist 4.2 TEST RESULTS Table 5 summarizes the results for the wall-floor connection test. Based on the ductility ratios and the total -slip curves (Fig 7), the connection responses are quite ductile, despite a few of the connection specimens failing in a brittle manner. This is due to small edge distance of nails drilled at the edge of 2x10 joist member creating a tendency of wood to split (Fig 8b). The ductile failure mode is the same as the previous two tests; excessive nail bending and wood crushing near the nail area (Fig.8a). Table 5: Test result for wall-to-floor connection (a) wall-floor connection studs bottom plate SG P u, kn Perp. 0.42 2.72 (0.05) (0.09) Par. 0.42 3.06 (0.05) (0.06) P y, kn 1.23 (0.07) 1.68 (0.06) δ u, mm 10.5 (0.27) 20.4 (0.14) δ y, mm 1.06 (0.29) 2.37 (0.36) Duct. 11.0 (0.5) 9.4 (0.29) Wall-floor connection spring model OSB sub-floor Shear parallel Shear perp Joists Load (b) connection model around the wall-floor connection 2-16d common nail 2x10 5/8 OSB Slip (mm) Fig.7: Average -slip curve of wall-to-floor connection (c) test of perpendicular ing 5/8 OSB 2-16d common nail (a) ductile 2x10 Joist (d) test of parallel ing Fig. 6: Testing of wall-to-floor connection (b) brittle (wood splitting) Fig. 8: Failure observed in wall-floor connection

Table.1: Data points as input for numerical modelling slip/def (mm) Perpendicular Parallel 0.0 0.41 0.22 0.5 1.88 1.60 1.0 2.66 2.48 1.5 3.10 3.03 2.0 3.40 3.38 3.0 3.87 3.82 4.0 4.29 4.15 5.0 4.64 4.48 6.0 4.88 4.79 7.0 5.04 5.06 8.0 5.19 5.27 9.0 5.34 5.42 10.0 5.49 5.53 11.0 5.61 5.61 12.0 5.67 5.69 13.0 5.68 5.77 14.0 5.64 5.84 15.0 5.59 5.92 16.0 5.54 5.97 17.0 5.47 6.01 18.0 5.35 6.04 19.0 5.10 6.05 20.0 4.59 6.04 A (a) whole building model B C (b) internal and external wall 5 CONNECTION RESPONSES IN 3-D MODEL To study the implication of the connection responses obtained from isolated laboratory tests to the overall (system) response of LFWS, a 3-D whole-building structural model was developed. In this study, a 9.3m x 14m (28ft x 42ft) one-storey LFWS was modeled using commercial FE software SAP2000 as shown in Fig 9. The arrangement followed the so-called platform frame construction method with selection of the structural members and other components based on construction guidelines from the Canada Mortgage and Housing Corporation [6]. Detailed numerical modelling techniques including types of element used, material and connection properties, boundary conditions applied, and numerical failure analysis can be seen in a companion paper [7]. The models developed in those papers have been updated with the connection properties obtained from the test conducted in this study. Load applied was only wind pressure derived based on ASCE7-05 [8]. The wind direction used is the critical one, acting perpendicular to the long axis of the structure leading to uplift pressure on its roof surfaces. Other wind data used were: wind with exposure category B; importance factor I=1.00; topographic factor K zt =1 (flat region); and wind directionality factor K d =0.85. The basic wind speed used was varied until the critical one causing failure to the structure was obtained. Fig 9c illustrates the deformed shape and contour of displacement magnitude of the structure under 125-mph wind. The blue colour indicates the most deformed region, which is in the peak of the roof structure. (c) deformed shape and displacement contour Fig. 9: 3-D finite element model wind After running the 3-D model extensively using various wind speeds, it was found that 125 mph was the crucial wind speed causing one of the roof-to-wall (toe-nail) connection reaching its peak. Figure 10 shows the connection response under this wind speed in selected locations as indicated in Fig 9. It can be seen that the uplift response of the toe-nail connection between the roof and wall is the most critical one compared to other type of connections, because the uplift peak was already reached. As expected, the shear response of the toe-nail connection is well below the yield limit, i.e. at the initial linear range (Fig. 9a). This is due to the lateral wind-ing configuration at the roof surface that almost cancelled out each other, in addition to the contribution of internal wall partitions that absorb this lateral [9]. As shown in Fig 9c, yield is almost reached for the wall-to-floor connection indicated in the structure.

5.0 4.0 3.0 2.0 1.0 uplift shear uplift-test shear-test 0.0 0.0 2.0 4.0 6.0 deformation (mm) (a) A: roof-to-wall joint** 6 DISCUSSION In high wind areas, metal connectors (e.g. metal hurricane ties) are normally installed in addition to the toe-nail connections (Fig 11). Detail test programme of the connection test involving hurricane metal ties is given in [10]. Fig. 12 shows -slip comparison between the toe-nail connection and hurricane metal connections (with and without toe-nail). As can be seen, the uplift connection strength with hurricane metal fasteners is larger, but with decreasing ductility response. Also, a small variation with respect to the strength was observed, since the failure occurred at the hurricane metal connector. A slight increase (16%) in the connection strength was observed when toe-nails were also applied together with the hurricane metal connector. 4.0 3.0 2.0 shear 1.0 shear connect test 0.0 0.0 4.0 8.0 12.0 16.0 20.0 deformation (mm) 6.0 5.0 4.0 3.0 2.0 1.0 0.0 (b) B: internal-external connection* / shear shear test withdrawal withdrawal-test 0.0 5.0 10.0 15.0 20.0 deformation (mm) (c) C: wall-floor connection* // Notes: **: only shear response perpendicular to the longbuilding axis was plotted, since there is no wind pressure s applied in the parallel direction. * / : ditto * // : the withdrawal connection test data was quoted from [4]. Fig. 10: Selected connection responses: comparisons between the numerical mode and tests Fig 11: Metal hurricane tie in roof-to-wall connection Loa d 12 10 8 6 4 2 0 Hurricane tie & toe-nail Hurricane tie only 0 5 10 15 Slip (mm) Fig 12: Load-slip responses Toe-nail only In high seismic areas, metal connectors known as holddown are normally applied at the end of wall-segment areas to increase wall-racking resistance. From the failure analysis under wind conducted in this study, the racking force was small and well below its racking strength. This is in accordance with the initial assumption that racking wall force would not be dominant factor in wind failure analysis. However, this observation should not be extrapolated into multi-storey LFWS subjected to wind. 7 CONCLUSIONS From the connection tests conducted in this study, it can be concluded that the majority of the connection responses are ductile, indicated by excessive bending of

nail and wood crushing around it. Consequently, in modelling the connection response within LFWS, the -slip response needs to be represented as non-linear. The 3-D numerical model has shown that the connection responses in the structure under wind vary depending on the location. The most critical one is in the roof-to-wall toe-nail connection under uplift showing highly non-linear response up to the peak. Therefore, failure identification of LFWS should be initiated from this connection. Future tests and numerical modelling should investigate dependency of the connection response to the directions other than their orthogonal applications. ACKNOWLEDGEMENT Financial provided by Natural Resources Canada was greatly acknowledged. REFERENCES [1] Dolan J.D., Foschi R.O.: Structural analysis model for static s on timber shear walls. J. Struc. Eng., 117(3), 851-861, 1991. [2] Judd J.P., Fonseca, F.S.: Analytical model for sheathing-to-framing connections in wood shear walls and diaphragms. J. Struc. Eng., 131(20), 345-352, 2005. [3] Mi H.: Behaviour of un-blocked shear walls. MS thesis, University of New Brunswick, Canada, 2004, p220. [4] Winkel M.: Behaviour of light-frame walls subject to combined in-plane and out-of-plane s. MS thesis, University of New Brunswick, Canada, 2006, p191. [5] ASTM: Standard test methods for bolted connections in wood and wood base products. Standard D5652-95, American Standard for Testing and Materials, West Conshohocken, PA, USA, 2007. [6] CMHC: Canadian wood-frame house construction. Canada Mortgage and Housing Corporation. Ottawa, Canada, 2006. [7] Asiz A., Chui Y.H., Zhou L., Smith I., Zhou N.: Three-dimensional numerical model of progressive failure in wood light-frame buildings. In: Proc. 11 th World Conf. Timber Eng., Riva del Garda, Italy, June 20-24, 2010, p6, in press. [8] ASCE7-05: Minimum Design Load for Building and Other Structures. American Society of Civil Engineers, Reston, Virginia, US, 2005. [9] Asiz A., Chui Y.H., Smith I. Failure Analysis of Light Wood Frame Structures Under Wind Load. In: Proc. CIB-W18 Meeting 41, St. Andrews, Canada, 2008, Paper 41-15-5. [10] Asiz A., Chui Y.H., Zhou L., Smith I.: Development of Advanced System Design Procedure in Canadian Wood Design Standard. Final Report, Natural Resources Canada, Ottawa, 2009, p107.