Bolted Wood-Steel and Wood-Steel-Wood Connections: Verification of a New Design Approach

M. Mohammad Research Associate, Department of Civil Engineering, Royal Military College of Canada, P.O. Box 17000, STN Forces, Kingston, Ontario K7K 7B4

J.H.P. Quenneville Associate Professor, Department of Civil Engineering, Royal Military College of Canada, P.O. Box 17000, STN Forces, Kingston, Ontario K7K 7B4

(Word Count = 6504) ABSTRACT: This paper covers the verifications tests carried out at the Royal

Military College of Canada (RMC) on wood-steel-wood and wood-steel bolted connections. Thirty groups of specimens were tested. Specimens configurations were selected in such a way to include fundamental brittle and ductile failure modes cases. Comparisons between experimental results and predictions from proposed equations developed from steel-wood-steel bolted connections are given. Proposed design equations were found to provide better predictions of the ultimate loads than current CSA standard O86.1 design procedures especially for bearing. However, row shearout predictions seem to over-estimate the strength. An adjustment using the reduced (effective) thickness concept is therefore proposed. Experimental observations on specimens that failed in row shear-out indicated that shear failure occurred over a reduced thickness. Stress analysis confirms findings on the reduced thickness. The research program is described in this paper along with the results and the proposed design equations for wood-steel-wood and wood-steel bolted connections loaded parallel-to-grain.

Key words: wood-steel-wood, wood-steel, bolt, connection, strength, failure, design, thickness.

BACKGROUND In timber structures, different types of connectors and connections configurations are used. Bolted timber connections however, are one of the most popular types used in North America. Steel plates are used to connect timber

structural members to secure a proper transfer of forces from one structural member to another. Two steel side plates are often used to connect members and are often referred to as steel-wood-steel connections (SWS). However, due to some structural and architectural requirements, a single steel plate may be used. Steel plates could be inserted within the timber member or could be installed between two timber members. Those kind of connections are known as wood-steel-wood connections (WSW). Occasionally, a steel plate is used to transfer the load from one single member to another. This is usually used in light timber structures and are referred to as wood-steel connections (WS). The problem of predicting the strength of multi-bolted connections is a well known one. For the last fifty years, the strength of connections that failed in a ductile fashion has been understood and predicted well by the European engineering community. As a result of this, some European wood design codes emphasize the importance of using many small diameter fasteners instead of few large diameter ones so as to obtain the ductile failure modes. In North America however, the engineering community has been slow to adopt the European Yield Model (EYM) and the normal practice in bolted connection design is to use fewer large diameter bolt to save on fabrication costs. The current design equations in the Canadian design code (CSA 1994), are 3

based on work by Johansen (1949) and further modified by Larsen (1973) and were first introduced into the Canadian design code in the 1989 edition (CSA 1989). In this design approach, failure is assumed to be governed by bearing (crushing of wood) and/or bending of the bolts (Mode I and Mode III according to the European Yield Model (EYM)). This assumption results in a ductile failure mode for connections. This is not always the case even when minimum requirements for spacing, end distances and edge distances are met. Consequently, the failure modes that show brittle failure (which are typical in connections with multiple fasteners) had to be addressed by modifying the EYM. However, when using large fasteners, brittle failure modes such as splitting, row shear-out, tearing and a combination of tearing and shear-out (known as a group tear-out) are the norm. This has been confirmed by test results from various sources (Yasumura et al. 1987, Mass et al. 1988, Mohammad et al. 1997, Quenneville and Mohammad 2000), and especially for multiple fastener connections and connections with low slenderness ratio (l/d) fasteners. These modes of failure can not be predicted by the EYM, resulting in discrepancies between design strength values and actual experimental ones. The current design model in the Canadian code for bolted timber connections (CSA 1994) assumes that connections will fail in a ductile manner. To account for situations where the connections show a brittle behaviour (generally connections with multiple bolts), modifications factors were introduced. Test results conducted on double shear steel-wood-steel bolted connections using 12.7 mm or 19.1 mm bolts have shown that the current Canadian design approach leads to conservative design strength values. Connections resistances as calculated from the O86.1-94 design 4

code were found to be as low as one third of the experimental values (Quenneville and Mohammad, 2000). This leads to the connections being over-designed. It is recognized that there is a need for a fundamental design method for bolts. The most desirable approach would be similar to the one used for other construction materials (i.e. steel), where a two-step process is utilized. The first step would be to check yield failure in the bolt, and is calculated by multiplying the capacity of one bolt in the connection times the number of bolts. The second step consists of checking the failure around the bolt, and is calculated by determining bearing and the combined tension and shear capacity of wood. This step depends on the joint configuration, spacing, end distances, etc.

Proposed Design Equations for Steel-Wood-Steel Connections In an attempt towards developing a more rational approach to determine connections design strength, a set of equations has been developed by Quenneville to predict the ultimate strength of connections based on the actual failure modes and mechanisms observed during tests (Quenneville and Mohammad, 2000). Design equations were derived so that specified strength values for materials as listed in O86.1-94 could be used. Failure modes covered in these design equations were row shear (RS), group tear-out (GT) and bearing (B). The connection strength (pu) would be the minimum of puRS, puGT and puB. Design equations proposed by Quenneville (1998) are given below.

Row shear-out: [1] puRS = 2 nr ns t w Jr f'v MIN(e,sb ) N 5

Group tear-out: [2] puGT = ns ( 2t w f'v MIN(e,sb ) N) + ( t w (nr - 1)(sr - (d + 2))ftg )

Bearing: [3] puB1 = 0.8 fw d t w nr N

[4]

puB2 = 0.8 fw d

ns nr N

fy 1 t w 1 fs + 5 d 6 (fw + fs ) fw fy 2 fs 3 (fw + fs ) fw MIN(e, s b ) , 0.25 tw

[5]

puB3 = 0.8 fw d ns nr N

where,

f' v = f vO86.1 MAX 1.085 0.085 N

Validation tests on SWS bolted connections were carried out to compare predictions from proposed equations with those of the O86.1-94 values (Quenneville and Mohammad, 2000). A reasonably good agreement was found especially for row shear-out and group tear-out. However, proposed design equations for WSW and WS bolted connections were not validated and there was a pressing need to carry out some extra connection tests to ensure that changes in bolted connection design equations were well justified and that they have been validated for practical situations. So, the main objective of this research project was to verify if the proposed design equations are valid for WSW and WS connections with single or multiple bolts. This will be accomplished by comparing predictions from the proposed design equations with experimental results. These tests are also required to complement the 6

results database already available at the RMC.

MATERIALS AND TEST PROCEDURES

Specimens Thirty groups of 10 replicates each were used in this study. Details of the groups are given in Table 1. Specimens were made of either glulam or lumber members. Glulam specimens consisted of either a single member (130mm wide) or two members (80mm wide each) with a steel plate in the middle (WSW connections). The reason for choosing the two sizes of glulam members was to compare the response of the connections, using a single wood member with a slot in the middle or when using two separate members. Glulam specimens were either Spruce-Pine (S-P) grade 20f-EX or Douglas fir 20f-EX and were either 80mm x 190mm or 130mm x 190mm. A slot of 10 mm wide was made in the center of the 130mm x 190mm glulam members to accommodate a 9.5mm steel plate using a chain saw rigged to a cutting table (referred to in Table 1 as Insert). Other groups were fabricated with two members of 80mm x 190mm sandwiching a 9.5mm steel plate. One group was fabricated with a single 80mm glulam member and a steel side plate (group 12). Lumber specimens were made of either one, two or three members nailed together and were bolted to a steel plate (WS connections). Specimens made with lumber were S-P-F, No. 2 and better 38mm x 140mm. These groups were fabricated using two or three lumber members nailed together using 90mm nails. Two types of bolts were used in this study, 12.7mm and 19.1mm. No particular effort to have matched group was made for these specimens. However, glulam billets were purchased in 7

different batches. Upon delivery, billets were cut and pieces were distributed to alternate groups. Glulam and wood members were stored in a conditioning chamber to attain a 12% equilibrium moisture content (EMC). Detailed connections configurations are listed in Table 1. Although 10 physical specimens were tested

per group, 20 connections were tested (2/specimen).

Test Set-up and Procedures A typical test set-up is shown in Figure 1. All bolts were finger tight to allow a self-alignment. Specimens were loaded parallel to grain and were fabricated with identical connection configurations at each end. A universal loading machine (MTS) was used to apply the load. A monotonic tension load was applied through the central steel plate (WSW) or through the side steel plate (WS). Four linear variable

displacement transducers (LVDTs) were used to record the slip of the wood side member(s) with reference to the steel plate (two at each end). A data logging system was used to record the machine load and slip from the four LVDTs. An initial pre-load of about 1.0 kN was applied to the specimens. The test was displacement driven at a rate of 0.9mm/min. (0.035in/min.) in accordance with ASTM standard D07.05.02 (ASTM 1988). Tests were stopped upon failure, when the load dropped with no recovery.

RESULTS AND DISCUSSION

Test results are given in Table 2. Figure 2 shows typical load-slip relationships for all specimens in group 5 (ductile) and group 9 (brittle). The ultimate strength 8

values for each group was determined and the lower 5th percentile value was calculated using a two-parameter Weibull distribution based on a 75% confidence level (ASTM 1994). Test results were adjusted for a normal duration of loading by dividing by a factor of 1.25 (CWC 1995). Calculated values for connection are presented in Table 2 for comparison. O86.1-94 values represent the lateral strength resistances as determined from Clause 10.4.4 (CSA 1994). Predictions using the equations proposed by Quenneville and modes of failure observed for each group are given as well. It should be noted that although 10 physical specimens were tested per group, 20 connections were tested (2/specimen). The 10 ultimate values are the lower resistance of each 10 pair of connections.

Comparison Between Selected Groups Connections configurations (i.e. loaded end distance, thickness, number of shear planes ..etc) have a significant effect on their ultimate strength. Generally increasing the end distance from 5d to 10d has increased the ultimate strength considerably. Specimens in group 5 with an end distance of 5d have a lower 5th percentile value compared to those in group 6 with an end distance of 10d by a factor of 0.66. The 5th percentile value for specimens in groups 7 and 8 with equal end distances (5d) are found to be the same. When comparing the ultimate strength of specimens in groups 5 and 10 with exactly the same configurations, except that group 10 was fabricated with two members sandwiching a steel plate, it is evident that their 5th percentile strength values are similar (25.7 kN compared to 28.9 kN). The same could be found when 9

comparing specimens in groups 7 and 8. This indicates that for this type of connections, using two members sandwiching a steel plate or using a single one with a steel plate in the middle does not have any significant influence on the 5th percentile strength of the connections. This is quite interesting, since the cumulative thickness of wood in group 8 is less than that of group 7 (120mm compared to 180mm).

Observations on Failure Modes Generally, two dominant types of failures were observed in all specimens. These were row shear-out and bearing. Splitting was observed but was not as significant. Almost all WSW connections specimens fabricated with 12.7mm bolts failed due to wood bearing. Yielding of the bolts was observed as well. Connections fabricated with a single bolt and an end distance of 5 times the bolt diameter (5d) and with a single 130mm wood member, exhibited considerable crushing prior to failure. However, the final failure was mainly in row shear. Group 6 with a single bolt and an end distance of 10 times the bolt diameter failed mostly in bearing. Splitting failure which were observed after significant deformation, were followed by localized row shear-out failure. Group 7, with two members and two bolts in a row, failed in row shear-out, however, few specimens failed in splitting. Group 8, with two bolts in a row and a spacing and end distance of 5d, failed mostly in row shear-out. Hardly any signs of bearing were observed. The failure scenario for these few specimens was as follows: splitting developed first, resulting in a sudden drop in the load, followed by failure in row shear. Failure in WS connections specimens fabricated with a single glulam member (group 12) was characterized by a row shear-out failure in one row 10

followed by failure in the second. Bearing deformation was obvious in few specimens. Groups of specimens in WS connections fabricated with 2 or 3 lumber members failed ultimately in row shear-out, however, they exhibited considerable amount of bearing deformation prior to failure. Localized drops in the load corresponded with row shear-out failure in the individual members, with the member adjacent to the steel plate being the first to fail, followed by the second or the middle (in the case of 3 members connections).

Comparison Between O86 Predictions and Experimental Values Predictions using the current design code (CSA 1994) were found to be conservative compared to the validation tests results (the lower 5th percentile). Excluding the groups for which the O86.1-94 predictions are zero, the ratio between O86.1-94 values and the experimental results was found to be between 0.53 to 0.89 (with the exception of group 23 and 24) with an average of 0.73. It can be seen in Figure 3 that the O86.1-94 predictions when plotted against the 5th percentile experimental, lie below the 45o line, thus considered to be conservative. One reason for such discrepancies between O86 values and those of the experimental ones could be attributed to the axial tensioning force that develops in the bolt once the plastic hinge is developed. This axial force reinforces the connection and results in the connection sustaining higher loads than anticipated. In fact the axial force may even alter the mode of failure completely in some cases (i.e. where row shear-out strength is not much higher than bearing strength of wood). The influence of the axial force was obvious especially in group 6, where almost all tested specimens exhibited that effect. This was observed in the load-slip curve as a 11

discontinuity in the envelope around 55 kN, followed by an increase in the capacity of the connection due to the development of a plastic hinge and the axial tensioning force in the bolt. This effect is more pronounced in SWS connections due to the anchorage provided by the steel side plates as compared to wood side members. It is not surprising that O86 predictions for bearing based on the EYM under-estimate the failure load, since the EYM does not take into consideration the axial tensioning force that develops in the bolt. That explains the higher values for the experimental tests compared to O86.1-94 predictions. Other reasons could be attributed to group and loaded end distance modification factors (JG and JL) used in the calculations of the O86.1-94 values. These factors are very restrictive resulting further in underestimating the capacity of bolted connections.

VALIDATION OF PROPOSED EQUATIONS (QUENNEVILLE, 1998) FOR WSW AND WS CONNECTIONS

In order to verify if the proposed design equations provide some reasonable accuracy for WSW and WS connections, strength values calculated using the proposed equations are compared with experimental values (5th %). Figure 3 suggests a reasonable correlation between the experimental and the minimum predicted values which are based on failure due to bearing (B), indicating that the bearing equation (which are based on the EYM) are adequate for predicting bearing

12

failures. However, as can be seen in Figure 4, where row shear-out failure governs, row shear-out predictions (RS) were found to be considerably higher than the 5th percentile as determined from tests, unlike RS predictions for SWS connections (Quenneville and Mohammad 2000). The following discussion describes a theory proposed by Jorissen (1998) that could explain such a discrepancy between predictions using the proposed equations for row shear-out and those obtained from tests for WSW and WS. It is based on the stress distribution underneath the bolts and the observed failure patterns during the tests.

Analysis In the new design equations for row shear-out predictions of bolted WSW connections proposed by Quenneville (1998), the row shear-out failure was assumed to occur over the full thickness of the wood member (tw was taken as equal to the full thickness of the wood side member). This assumption is valid for connections with rigid type of fasteners, where the embedment stress is usually assumed to be uniformly distributed over the full timber thickness. This was validated with visual inspection of failed specimens, where the shear failure plane occurred across the full timber thickness (see Figure 5-a)). Good agreements were found between predictions from Quenneville proposed equation for row shear strength and those from validation tests for SWS connections (Quenneville 1998). However, row shear-out predictions for WSW and WS connections calculated over the full thickness of the wood side members were found to be high, compared to the validations tests (the lower 5th percentile). Further inspection of failed WSW and WS connections revealed that row shear-out failure did not occur over the full 13

thickness of the wood side member of the WSW connection, but over a reduced thickness which was smaller than the wood member full thickness (tw), as can be seen in Figure 5-b). Considering that similar type of fasteners (12.7 mm or 19.1 mm bolts) were used in both types of connections and with more or less thicknesses of wood members, the influence similar

of material properties was

eliminated. This difference in the row shear failure pattern could be attributed to the assumed embedment stress distribution along the fastener length. A different embedment stress distribution clearly takes place in connections with a middle steel plate, compared to those with steel side members. Embedment stress distribution for rigid dowel type fasteners is assumed to be uniform along the fastener length, Figure 6-a), and c). Figure 6-b) and d) shows the assumed embedment stress distribution along the fastener for connections with nonrigid dowel type fasteners and with steel side plates or middle steel plate. Unlike connections with rigid type of fasteners, a uniform embedment stress distribution is assumed only over a specified length y. Now, for connections with a middle steel plate, this assumption results in crack propagation near the shear planes over a length ye, which is assumed to be slightly bigger than y. This means that the shear force (F) is acting over a reduced thickness which is less than the full thickness (tw) of the wood side member (Jorissen 1998). Jorrisen refers to this specified thickness as the effective thickness. The result is a lower than anticipated shear strength for the connection with a middle steel plate. In an attempt to verify this theory, the thickness across which row shear-out failure plane took place was measured for all groups that exhibited row shear-out failure and the mean value for each group is presented in Table 3, column 3 (except for those groups fabricated with 2 or 3 lumber members, 14

where the thickness of a single member was considered as being the experimental one). These values were found to be smaller than the full thickness of connections wood members (column (2)). The mean ratio between the measured thickness and full thickness of the wood member was found to be 0.65, Table 3. In SWS connections with non-rigid fasteners, though the embedment stress distribution is also assumed to be uniform over a length y, smaller than the member full thickness, different stress distribution is associated with this type of connections, see Figure 6-b). This distribution does not seem to influence the propagation of the cracks near the shear planes. The development of infinite number of plastic hinges in the fastener (unlike the case for WSW connections) leads to nearly uniform stress distribution underneath the fastener. Failure usually occurs across the full thickness for the type of fasteners used in the validation study of bolted SWS connections. This may not necessarily be the case for connections with higher slenderness ratio, where row shear failure is not dominant normally. This may lead to the conclusion that the effective thickness theory could be applicable only for connections with a middle steel plate. Adjustment to the wood side members thickness (tw) may be necessary to account for that phenomenon in order to achieve better predictions for row shear-out in WSW . The following discussion describes how to determine the effective thickness (ye) across which the row shear-out failure takes place for WSW and WS connections, based on the stress distribution along the fasteners, for rigid and nonrigid dowel type fasteners.

Effective Thickness 15

Johansens Yield Model for double shear symmetrical connections with a single internal steel plate and assuming a rigid dowel type fastener (see Figure 6-a) and 6-c)) for failure Mode I, is given by the following equation: F = d t w fw

[6]

For Mode II (Figure 6-d), the failure mode is given by the following equation: [7] F = d y fw

Using Eq. [7], the length y can be determined as follows:

[8]

y=

F d fw

Additional stress analysis given by Jorissen (1998) indicated that for connections with more slender dowel type fasteners, cracks propagate near the shear planes over a thickness ye, which is assumed to be slightly bigger than y (see Figure 6-d). For connections with rigid dowel type fasteners, it can be assumed that y = ye = tw. According to Jorrisen, the value ye was determined using linear interpolation where y < ye < tw and was given by the following equation:

[9] where,

y e = 1+ C y

tw y y tw

Cy is a constant (0 < Cy < 1.0), which was calculated based on the following equation: Cy = 0.3 sb e 16

[10] For connections with a single fastener, Cy was taken as equal to 1.0. Since ye = tw for rigid dowel type of fasteners, Eq. [10] does not influence the calculation for rigid dowel fasteners. Eq. [10] was derived empirically using test results (Jorissen 1998). Measured values of the effective thickness (ye) across which the row shearout failure took place were found to be comparable with those calculated using Eq. [9], except for connection made with lumber (WS), in which, due to the presence of discontinuity, the row shear failure occurred at the interface between the two wood laminates. From Table 3, the mean ratio of the measured effective thickness and that of the member thickness was found to be 0.65. However, excluding the WS group measurements brings the mean value up to 0.8, which corresponds well with the computed effective thickness calculation (0.85). It should be noted that the effective thickness approach should only be applied to determine the strength of the row shear-out failure mode. For bearing failure modes I, II and III, the member thickness must be used in design calculations. Group tear-out calculations should also be based on the thickness of the wood members since for group tear-out, failure must occur over the entire thickness. Partial tension failure is unlikely to occur. Test observations for group tear-out failures confirm that requirement. To determine the appropriate row shear-out equation, predictions using equation [1] were plotted against the 5th % as determined from tests for the original group. A linear regression analysis was carried out to determine the best fit line. A modification factor of 0.8 was found to be appropriate for row shear-out predictions in 17

WSW and WS connections. In order to account for the reduced thickness and to simplify the proposed RS equation, a reduction factor could be introduced in the proposed row shear-out equation [1] for WSW and WS connections. This leads to the following equation:

[11]

puRS = 2nr ns (0.8 t w ) Jr f' v MIN(e, s b ) N

Using the calculated effective thickness (ye ) instead of the full thickness (tw ) in the row shear-out equation (Eq. [11]), leads to lower predictions for row shear-out. In Table 4, group tear-out and bearing predictions calculated based on the full thickness are shown together with predictions based on the calculated effective thickness (Column (6), Table 4). Row shear-out predictions based on the effective thickness, provide a better agreement with the 5th percentile from tests, where the row shear-out failure controlled. In Figure 7, the 5th % values determined experimentally are plotted against model predictions including the modified row shear-out predictions. A better agreement was found between test results and row shear-out predictions (see Table 4). This should not be understood as being only an empirically derived or a simple curve fit exercise. It is based on the stress analysis described earlier and on the laboratory observations. Based on the above discussion, it is evident that introducing a factor of 0.8 in the row shear-out proposed design equations for WSW and WS connections leads to better predictions. Using the full thickness of the wood side members in WSW and WS connections tend to overestimate the row shear strength. 18

CONCLUSIONS

Based on the validation tests of the proposed design equations for WSW and WS bolted connections, it can be concluded that:

1.

Current Canadian design code (O86.1-94) leads to over-designed WSW and WS bolted glulam connections, especially with multiple bolts, where it underestimates the failure loads.

19

2.

Proposed design equations for WSW bolted connections (Quenneville and Mohammad, 2000) provide better predictions of the ultimate loads than current design procedure.

3.

Improved predictions for row shear-out can be achieved if the effective thickness principle is used. A modification factor of 0.8 was found to be suitable for row shear-out strength predictions of WSW and WS connections.

ACKNOWLEDGMENT

Funding from the Academic Research Program, from the Military Engineering Research Group (MERG) of the Royal Military College of Canada and from the Canadian Wood Council (CWC) is greatly appreciated. The authors wish

20

to express their gratitude to Mr. Lee and Ocdt Carriere who conducted the experimental tests.

REFERENCES

American Society for Testing and Materials (ASTM). 1988. Standard test methods for mechanical fasteners in wood. Standard D1761-77, ASTM, Philadelphia, PA.

American Society for Testing and Materials (ASTM). 1994. Standard specification for computing the reference resistance of wood-based materials and structural connections for load and resistance factor design. D5457-93, ASTM, Philadelphia, PA.

Canadian Standards Association. 1989. Engineering design in wood ( limit states design). Standard O86.1-M89. Canadian Standard Association, Rexdale, ON.

Canadian Standards Association. 1994. Engineering design in wood ( limit states design). Standard O86.1-94. Canadian Standard Association, Rexdale, ON.

21

Canadian Wood Council. 1995. CSA Commentary, Wood Design Manual, Canadian Wood Council, Ottawa, ON.

Johansen, K.W. 1949. Theory of timber connectors. IABSE Journal, No. 9. pp.249262.

Jorissen, A. 1998. Double shear timber connections with dowel type fasteners. Ph.D. Thesis, Technical University of Delft, Delft, The Netherlands.

Larsen, H.. 1973. The yield load of bolted and nailed joints. Structural Research Laboratory, Technical University of Denmark, IUFRO Division 5, p.14.

Mass, D.I., Salinas, J.J. and Turnbull J.E. 1988. Lateral strength and stiffness of single and multiple bolts in glued-laminated timber loaded parallel to grain. Engineering and Statistical Research Centre, Research Branch, Agriculture Canada, Report No. C-029, Ottawa. ON.

22

Mohammad, M., Smith, I. and Quenneville, J.H.P. 1997. Bolted timber connections: investigations on failure mechanism. Proceedings of IUFRO S5.02 Timber Engineering Conference, Copenhagen, Denmark.

Quenneville, J.H.P. 1998. Predicting the failure modes and strength of brittle bolted connections. Proceeding of the 5th World Conference on Timber Engineering (WCTE), Montreux, Switzerland, 2:137144.

Quenneville, J.H.P. and Mohammad, M. 2000. On the failure modes and strength of steel-wood-steel bolted timber connections loaded parallel-to-grain. Canadian Journal of Civil Engineering (in press).

Yasumura M., Murota T. and Nakai H. 1987. Ultimate properties of bolted joints in glued-laminated timber. Report to the Working Commission W18-Timber Structures. Dublin, Ireland.

23

List of Symbols

B Cy

= bearing strength, N = a constant 24

d e F ftg fv fw

= bolt diameter, mm = end distance, mm = applied load resisted by one side of the wood member, N = specified strength in tension parallel to grain at the gross section, MPa = specified strength in shear parallel to grain, MPa = embedment strength of wood member, MPa = 63G (1-0.01d), for parallel to grain loading

GT Jr

= group tear-out strength, N = factor for number of rows = 1.0 for 1 row, or for 1 bolt per row = 0.8 for 2 rows, (2 or more bolts in a row) = 0.6 for 3 rows, (2 or more bolts in a row)

N nr ns RS sb sr tw y ye

= number of bolts in a row = number of rows = number of shear planes = row shear-out strength, N = bolt spacing in the row, mm = row spacing, mm = thickness of the wood side member, mm = thickness along which the embedment stress is assumed to be uniform, mm. = effective thickness, mm. 25

Table 1. Summary of specimens configurations.

Group Wood Type
1)

Specimen size . (mm) .

nr

Type e

sb

sr

Mean COV 5 % (kN) (%) (kN) 35 24 37 30 26 39 61 61 107 29 69 28 8 13 15 22 106 96

th

. (mm) . 19.1 12.7 19.1 12.7 19.1 19.1 19.1 19.1 19.1 19.1 19.1 19.1 12.7 12.7 19.1 19.1 19.1 19.1 1 1 1 1 1 1 1 1 2 1 2 2 1 1 1 1 2 2 1 1 1 1 1 1 2 2 2 1 1 1 1 2 1 2 2 2 WSW 134 (7d) WSW 89 (7d) N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

S-P G S-P G S-P G S-P G S-P G S-P G S-P G S-P G S-P G S-P G S-P G S-P G SPF L SPF L SPF L SPF L S-P G D-fir G

2)

2@80x152 2@80x152 2@80x190 2@80x190 130x190 130x190 2@80x190 130x190 2@80x190 2@80x190 2@80x190 1@80x190

57 46 68 56 42 65 103 95 48

8 16 15 13 8 8 9 6 9 8 6 12 14 14 12 22 8 13

WSW 191 (10d) N/A WSW 127 (10d) N/A Insert 95 (5d) WSW 95 (5d) Insert 95 (5d) WSW 95 (5d) WSW 95 (5d) WSW 95 (5d) WS WS WS WS WS 95 (5d) 63 (5d) 64 (5d) 95 (5d) 95 (5d) N/A 95 (5d) 95 (5d) 95 (5d) N/A N/A N/A N/A 64 (5d) N/A 95 (5d) 95 (5d) 95 (5d) Insert 191 (10d) N/A

95 (5d) 181 95 (5d) 106 95 (5d) 52 N/A N/A N/A N/A 14 25 29 48

3)

2@38x140 2@38x140 2@38x140 2@38x140 130x190 130x190

Insert 95 (5d) Insert 95 (5d)

95 (5d) 175 95 (5d) 181

26

19 20 21 22 23 24 25 26 27 28 29
1)

D-fir G S-P G S-P G S-P G SPF L SPF L SPF L SPF L S-P G D-fir G S-P G

2@80x190 2@80x190 130x190 130x190 2@38x140 2@38x140 3@38x140 3@38x140 130x190 130x190 2@80x190

19.1 19.1 19.1 19.1 12.7 19.1 12.7 19.1 12.7 12.7 12.7

2 2 2 2 1 1 1 1 2 2 2

2 2 1 2 2 2 2 2 2 2 2

WSW 95 (5d) Insert 95 (5d) WS WS WS WS

95 (5d) N/A

95 (5d) 225 95 (5d) 88 24 47 16 26

13 11 11 6 16 23 18 19 12 6 11

120 141 49 147 12 20 8 11 64 89 80

WSW 191 (10d) 191 (10d) 95 (5d) 243 Insert 191 (10d) 191 (10d) 95 (5d) 227 127 (10d) 127 (10d) N/A 191 (10d) 191 (10d) N/A 64 (5d) 96 (5d) 64 (5d) 96 (5d) N/A N/A

Insert 127 (10d) 127 (10d) 64 (5d) 115 Insert 127 (10d) 127 (10d) 64 (5d) 140 WSW 127 (10d) 127 (10d) 64 (5d) 142

30 D-fir G 2@80x190 12.7 2 2 WSW 127 (10d) 127 (10d) 64 (5d) 143 5 93 Both S-P and D-fir glulam were from 20f-EX grade and SPF lumber was No. 2 and better Structural light Framing. 2) Glulam 3) Lumber

Table 2. Validation tests results and predictions using O86.1-94 and proposed equations (Quenneville, 1998).
5 % test
th 1)

Group (1) 1

O86.1 Ratio 94 (4)/(2)

Proposed Design Equations

3)

pu Min. puRS PuGT PuB1 puB2 puB3 puB4 puB5 .. (kN) .. 71 71 55 30 39

(2) (3) (4) .. (kN) .. 35 23 0.65

Observed failure mode B/RS

--

--

30

27

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
1) 2)

24 37 30 26 39 61 61 107 29 69 28 8 13 15 22 106 96 120 141 49 147 12 20 8 11 64 89 80 93

13 30 17 21 28 2) 0 0 0 0 0 0 0 0 0 0 0 0 0 125 0 115 16 30 0 0 57 60 57 60

0.53 0.82 0.56 0.80 0.71 0 0 0 0 0 0 0 0 0 0 0 0 0 0.89 0 0.78 1.33 1.5 0 0 0.89 0.67 0.72 0.65

49 94 68 38 65 94 65 150 52 105 52 10 18 14 25 105 119 173 232 76 140 31 38 29 41 124 141 186 212

49 94 68 38 65 94 65 244 52 203 101 10 18 14 25 179 188 259 296 151 201 31 38 29 41 171 200 234 274

39 55 39 41 41 110 82 219 55 110 55 18 36 25 50 164 183 244 219 82 164 36 50 54 75 118 131 157

17 30 17 28 28 61 56 122 30 61 31 10 16 15 29 111 119 131 122 56 111 16 29 20 34 59 64 67

18 39 18 39 39 78 78 156 39 78 57 17 34 28 57 156 165 165 156 78 156 34 57 34 57 72 75 72 75

----------49 16 32 24 48 ------32 48 36 52 -----

----------39 9 18 19 38 ------18 38 18 38 -----

17 30 17 28 28 61 56 122 30 61 31 9 16 14 25 105 119 131 122 56 111 16 29 18 34 59 64 67 73

B/RS B B B/RS B/RS RS RS RS/S RS RS RS B RS RS/B RS RS RS RS B RS RS/B RS/B RS/B RS/B RS/B B B B B

175 73

Adjusted for the normal duration of loading by dividing over a factor of 1.25. No values are given for groups with an end distance below the 7d limit (O86.1-94). 3) From Quenneville (1998).

28

Table 3. Comparison between measured and calculated effective thicknesses.

Wood Measured Effective Ratio Ratio member effective thickness (3) / (2) (5) / (2) thickness Thickness Eq. [9] (2) (3) (4) ----0.8 0.9 0.6 0.8 0.7 0.8 0.9 0.5 0.5 0.5 0.5 0.5 0.9 0.9 0.7 -(5) (mm) 68 80 68 80 58 58 55 52 55 68 68 76 67 54 76 71 52 48 50 55 (6) 0.85 1.00 0.85 1.00 0.97 0.97 0.69 0.87 0.69 0.85 0.85 0.95 0.88 0.71 1.00 0.93 0.87 0.80 0.63 0.69 .. (mm) .. 1) 80 -80 -80 -80 -60 48 60 57 80 49 60 42 80 58 80 62 80 75 80 41 76 38 76 38 76 38 76 38 60 54 60 54 80 56 80 --

Group (1) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

29

21 22 23 24 25 26 27 28 29 30 Mean STD
1)

60 60 76 76 114 114 60 60 80 80

54 -38 38 38 38 -----

0.9 -0.5 0.5 0.33 0.33 ----0.65 0.19

58 52 73 57 80 60 60 60 72 66

0.97 0.87 0.96 0.75 0.70 0.53 1.00 1.00 0.90 0.83 0.85 0.13

information not available.

Table 4. Test results and predictions using O86.1-94 and proposed design equations.
Group 5th % test O86.1-94
2) Mod.

PuGT PuB

puRS

1 2 3 4 5 6 7

... (kN) ... 35 23 71 30 24 13 49 17 37 30 94 30 30 17 68 17 26 21 38 28 39 28 65 28 1) 61 94 61 0

56 39 3) 75 3) 54 30 52 75

30

8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
1) 2)

61 107 29 69 28 8 13 15 22 106 96 120 141 49 147 12 20 8 11 64 89 80 93

0 0 0 0 0 0 0 0 0 0 0 0 125 0 115 16 30 0 0 57 60 57 60

65 244 52 203 101 10 18 14 25 179 188 259 296 151 201 31 38 29 41 171 200 234 274

56 122 30 61 31 9 16 14 25 105 119 131 122 56 111 16 29 18 34 59 64 67 73

52 120 42 84 42 8 14 11 20 84 95 138 3) 186 61 112 3) 25 3) 30 3) 23 33 3) 99 3) 113 3) 149 3) 170

O86.1 has no provision for an end distance less than 7d. RS values multiplied by a factor of 0.8. 3) Bearing failure governs.

31

LIST OF FIGURE CAPTIONS

Fig. 1. Fig. 2. Fig. 3. Fig. 4. Fig. 5. Fig. 6. Fig. 7.

Typical specimen in testing apparatus. Typical load-slip envelopes that exhibit ductile and brittle behaviour. Comparison between test results and O86.1-94 or proposed equations predictions. Comparison between test results and row shear-out predictions using proposed design equations. Comparison between row shear-out failure in bolted connections: a) SWS; b) WSW or WS connections. Comparison between embedment stress distribution in SWS and WSW or WS connections. Comparison between test results and O86.1-94 or modified row shear-out predictions.

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47