Development of Limit States Design Method for Joints with Dowel Type Fasteners Part 2: Comparison of Experimental Results with European Yield Model

Transcription

1 Manufacturing & Products Project No. PN (Part 2) Development of Limit States Design Method for Joints with Dowel Type Fasteners Part 2: Comparison of Experimental Results with European Yield Model

2 2004 Forest & Wood Products Research & Development Corporation All rights reserved. Publication: Development of Limit States Design Method for Timber Joints with Dowel Type Fasteners Part 2 Comparison of Experimental Results with European Yield Model (EYM) Predictions The Forest and Wood Products Research and Development Corporation ( FWPRDC ) makes no warranties or assurances with respect to this publication including merchantability, fitness for purpose or otherwise. FWPRDC and all persons associated with it exclude all liability (including liability for negligence) in relation to any opinion, advice or information contained in this publication or for any consequences arising from the use of such opinion, advice or information. This work is copyright and protected under the Copyright Act 1968 (Cth). All material except the FWPRDC logo may be reproduced in whole or in part, provided that it is not sold or used for commercial benefit and its source (Forest and Wood Products Research and Development Corporation) is acknowledged. Reproduction or copying for other purposes, which is strictly reserved only for the owner or licensee of copyright under the Copyright Act, is prohibited without the prior written consent of the Forest and Wood Products Research and Development Corporation. Project no: PN (Part 2) Researchers: I. Smith, G.C. Foliente M. Syme, R. McNamara and C. Seath CSIRO Building, Construction and Engineering PO Box 56, Highett, Victoria 3190 Forest and Wood Products Research and Development Corporation PO Box 69, World Trade Centre, Victoria 8005 Phone: Fax: info@fwprdc.org.au Web:

3 Development of Limit States Design Method for Joints with Dowel Type Fasteners Comparison of Experimental Results with European Yield Model Prepared for the Forest & Wood Products Research & Development Corporation by I. Smith, G.C. Foliente, M. Syme, R. McNamara and C. Seath The FWPRDC is jointly funded by the Australian forest and wood products industry and the Australian Government.

4 Contents EXECUTIVE SUMMARY 4 INTRODUCTION 7 EXPERIMENTAL PROGRAM 9 Background 9 Joint Tests 12 Nailed Joints 13 Screw Joints 23 Coach Screw Joints 24 Bolted Joints 25 Material Tests 26 Embedment Properties 26 Fastener Properties 32 Bending Tests 32 Tension Tests 34 RESULTS AND DISCUSSION 37 Overview 37 Fastener Tests 38 Wood Embedment Tests 39 Joint Tests 38 Joint Capacity Comparisons Experiment vs Predictions 44 Nailed Joints 44 Screw Joints 45 Coach Screw Joints 46 Bolted Joints 47 CONCLUSIONS AND RECOMMENDATIONS 57 Conclusions 57 Recommendations 58 REFERENCES 59 APPENDIX A. Summary of Responses to Connector Survey 61 APPENDIX B. EYM and EYM-Based Equations 65 APPENDIX C. Notations 69 APPENDIX D. Experimental Data Timber and Fastener Properties 70 APPENDIX E. Experimental Data Complete Joint Properties 95 APPENDIX F. Calculation of Predicted Yield Capacities Using EYM-based Equations 116 APPENDIX G. Plot Comparisons of Joint Strength Prediction vs Experiment 126 3

5 Executive Summary The goal of this project is to develop a limit states design procedure for timber joints. Part 1 (Foliente and Smith 2000) reviewed current design practice in Australia, USA, Canada and Europe, and identified issues and experimental parameters that need to be considered in order to apply the so-called European Yield Model (EYM) for joints with dowel-type fasteners. The present report presents details of the experimental program that includes tests on complete joints, embedment specimens and fasteners, and compares experimental results with predictions from EYM-based equations. There have been several extensive series of experiments to validate EYMs for European and North American wood species and fasteners, and there is general confidence in the method as a design tool. Focus in the present study is on validation tests of specific relevance to Australian construction practice. The schedule of tests was devised with the aid of a Connector Survey carried out by the PRI (Appendix A) and feedback from representatives of the PRI. Attention is focused on joints in Radiata Pine and Slash Pine, nails, wood screws, coach screws and bolts. Wood used had density values corresponding with joint categories JD3 through JD5 defined in AS1649. The survey found that structural designers emphasise use of two-member joints, except in the case of bolts for which threemember arrangements are also common. Concern is principally with jointing relatively thin timber members, and although they were not used exclusively, such arrangements were emphasised in the test program. The experimental program included a total of 360 complete joint system tests, 500 wood embedment strength tests and 73 fastener strength tests (bending and tension). The specimens for wood embedment and fastener tests were matched with those used in the complete joint system tests. The strength data obtained from wood embedment and fastener tests were used to calculate the EYM predicted strength for complete joint systems. These predictions were then compared with the actual experimental data from the 360 complete joint system tests. Predicted joint capacity was calculated using three methods (Appendix B): (1) original EYM equations by Johansen (1949) (referred to as Original EYM ); (2) EYM-based approximations by Whale et al. (1987) (referred as Simplified-1 ); and (3) EYM-based approximations by Blass et al. (1999) (referred to as Simplified-2 ). For wood and coach 4

6 screws, we also used the equations proposed in the US National Design Specifications (NDS) for wood construction (AF&PA 1997; 1999) (referred to as NDS-Screw ). Based on study of the static load behaviour of laterally loaded joints with one or two nails, one wood screw, one or two coach screws, and one bolt, we made the following conclusions: Capacities of joints with two nails or two small diameter coach screws are twice the capacities of similar joints with one fastener. This conclusion should not be extrapolated to joints with large dowel type fasteners. EYM-based equations predict the so-called yield capacity of a joint made with Australian pine. This corresponds to the stage where either the wood beneath the fastener has reached its bearing (embedment) capacity throughout the length of penetration within one or more of the members, or one or more plastic hinges have developed in the fastener. For engineering purposes, this is a point beyond which a joint has sustained significant irreversible damage and residual strength can be substantially impaired. Yield capacity rather than ultimate capacity should be the basis of future design values for joints made with Australian pine. All EYM-type models considered (Original, Simplified-1, Simplified-2, and the NDS- Screw model) produce conservative estimates of a joint s capacity, relative to its experimentally determined ultimate strength. This applies to joints made with nails, wood screws, coach screws or bolts. The Original, Simplified-1 and the NDS-Screw models provide acceptable estimates of the yield capacity for joints made in Australian pine. Predicted yield capacities based on the Simplified-2 model are in some instances highly conservative. This occurs when one of the two interpolation equations governs. Although, this has the simplest set of equations among those we considered, its adoption and use might not be justified by its highly conservative predictions for some cases. For bolted joints, not all of the theoretical failure modes occur in practice because rotational restraint at the ends of the fastener is not accounted for in models. Restraint is provided by the head, or the nut and associated washer. This prevents the fastener from adopting deformed shapes assumed for some modes. Suppression of modes that do not occur is justified during failure predictions for joints with coach bolts. 5

7 On the basis of these findings and observations, we recommend the following: Yield capacity rather than ultimate capacity should be the basis of future design values for dowel type fastener joints in Australian pine. The Original and the simplified version due to Whale et al. (Simplified-1) may be regarded as the most appropriate choices from amongst the EYM-type models considered. Some hypothetical failure modes should be suppressed for joints made with bolts. The two models yield numerically equivalent predictions for all types of joint considered in this project. As calculations are more streamlined, it seems most appropriate to choose the Simplified-1 model. This would mirror current practice in Canada. Emphasis in the next phase of the project should principally be on development of code rule for design of Australian pine joints made with dowel type fasteners (nails, wood screws, coach screws and bolts). Some attention should also be given to experimental confirmation of the methods for assigning capacities to multiple-bolt joints that load members parallel to grain, and tension joints where bolts load one or more of the members at an angle other than parallel to the grain. 6

8 Introduction The goal of this project is to develop a limit states design procedure for timber joints. Part 1 reviewed current design practice in Australia, USA, Canada and Europe, and identified issues and experimental parameters that need to be considered in order to apply the socalled European Yield Model (EYM) for joints with dowel-type fasteners (Foliente and Smith 2000). Based on that literature review and results of a Connector Survey conducted by the Pine Research Institute (PRI) on behalf of CSIRO (see Appendix A for a summary of survey results), we developed an experimental test program to: 1. Validate the applicability of the EYM equations in design of timber joints with Australian Pine species; and 2. Establish a database of material properties needed in order to use EYM-based design equations for joints. The positioning of these objectives in the overall logic for developing new Australian design procedures is shown in Fig. 1. This report presents details of the experimental program that includes tests on complete joints, embedment specimens and fasteners, and compares experimental results with predictions from EYM-based equations. Alternative predictions are considered based on the original EYM equations as developed by Johansen (1949) (referred to as Original EYM ) and derivatives proposed by Whale et al. (1987) (referred as Simplified-1 ), Blass et al. (1999) (referred as Simplified-2 ), and the US National Design Specifications (NDS) for wood construction (AF&PA 1997; 1999) for wood and coach screws only (referred to as NDS-Screw ). Conclusions and recommendations are made for the next phase of work that should encompass model simplification, and development of design values and procedures (see Fig. 1). 7

9 2 MATERIAL PROPERTIES Embedment strength Nail yield moment EUROPEAN YIELD MODEL MODEL SIMPLIFICATION 1 VERIFICATION TESTS DERIVATION OF CHARACTERISTIC PROPERTIES CODE RULES AND LANGUAGE Figure 1. Overall logic of joint design development and the objectives of the experimental program (encircled) 8

10 Experimental Program Background EYM-type models are solely intended as a means of predicting the short-term static strength of joints having one fastener. Strength of multiple-fastener joints will be addressed in the next phase of the project. Therefore the experimental program has focussed on collection of material properties that enter EYM models, and validation tests aimed at proving their applicability to joints in Australian grown pine. EYM models are predicated on the assumption that both the fastener and the wood foundation on which it bears (members) have a perfect rigid-plastic response, and that the failure capacity can be fully determined from equilibrium calculations involving only normal forces applied to the fastener by the wood. The calculations take no account of strength enhancing influences of restraint at the end, or ends, of the fastener generated by features such as the head of the fastener; nor inter-member friction; nor string action due to axial force in the fastener. Friction between members can only develop once any gap between adjacent members has closed, and if there is tension in the fastener at the joint plane. Conventional wisdom is that inter-member friction cannot be relied on in service, and thus measures are taken during tests to avoid its influence, e.g. nailed joints are made with an initial gap at each joint plane, bolts are not tightened with a spanner (AS1649: SAA 1998a). Despite efforts to prevent it, inter-member friction commonly does occur during tests, but only once large levels of slip are achieved (slip = relative movement of adjacent members at the fastener location). Large displacement effects lead to significant axial force in a fastener, and therefore a string effect can substantially inflate the joint strength. Again this only happens at large slip. Extent of any enhancement of strength over and above that predicted by EYM theory depends upon the type of fastener and the geometric arrangement of members. Figure 2 illustrates possible experimental load-slip responses, and how they relate to the EYM prediction. 9

11 Load Friction between members and string effect at large slip Yield capacity EYM prediction Bearing failure in member(s) (no friction between members or string effect) Slip Figure 2. Experimental load-slip responses and their relationship to the EYM prediction (schematic) What EYM-type models are expected to predict is the so-called yield capacity of a joint. This corresponds to the stage where either the wood beneath the fastener has reached its bearing (embedment) capacity throughout the length of penetration within one or more of the members, or one or more plastic hinges have developed in the fastener. For engineering purposes, this is a point beyond which a joint has significant irreversible damage and residual strength can be substantially impaired. The yield and ultimate (maximum) capacities can be quite different for many types of joint. It is well accepted in other countries, e.g. Canada, European States and the USA, that the yield capacity rather that the ultimate capacity is the parameter of interest. Ability to predict yield capacity is what is evaluated in this report. Also the extent to which ultimate capacity can exceed the yield capacity is assessed. Within experiments and EYM model predictions various modes of failure are possible depending upon the number of members, wood species, fastener diameter (size) and the thickness of the members. Theoretical failure modes are shown in Figures 3 and 4. When applying various versions of the EYM theory (Appendix B) the capacity for each of the possible modes is calculated. The governing mode is that with the lowest predicted yield capacity. There are two pieces of experimental information that can be used to assess predictions. These are whether the predicted and observed capacities are in agreement, and whether the correct failure mode is predicted. Some subjectivity is involved in assigning the failure mode following a test. This is because tests are continued to levels of slip well above those at which the yield capacity is reached, in order to determine the ultimate capacity. 10

12 Above mentioned discrepancies between model assumptions and the real behavior contribute to difficulty of assigning a failure mode. Members can split prior to the end of a test, and whether that represented the true failure or was a development of it is a moot point. Whether or not the fastener developed any plastic hinges is a good guide to the failure mode in the context of EYMs. Figure 3. Yield modes for two-member joints Mode I s Mode I m Mode III Mode IV Mode I Mode II Mode III Mode IV [ plastic hinge ] [ plastic hinge ] Figure 4. Yield modes for three-member joints Two-member joints are typically not symmetric. When fasteners such as nails are used penetration is commonly not the same for the head-side and point-side members. For twomember joints with coach screws, variation in the fastener cross-section along its length means the arrangement can never be symmetric. Although bolts might join members of the same thickness, differences in the arrangement at the head and nut ends mean that symmetry is never truly attained (this is especially true when coach bolts are used as in tests reported here). Recognition of these points is important for interpretation of experimental data. There have been several extensive series of experiments to validate EYM s for European and North American wood species and fasteners, and there is general confidence in the method as a design tool (see report on Part 1 of this project). Focus here is on validation tests of specific relevance to Australian construction practice. The schedule of tests was devised following the Connector Survey in Appendix A and feedback from representatives of 11

13 the PRI. Attention is focused on joints in Radiata Pine and Slash Pine, nails, wood screws, coach screws and coach bolts (simply referred to as bolts in this report). Wood used had density values corresponded with joint categories JD3 through JD5 (AS1649: SAA 1998a). The survey found that structural designers emphasize use of two-member joints, except in the case of bolts for which three-member arrangements are also common. Concern is principally with jointing relatively thin timber members, and although they were not used exclusively, such arrangements were emphasized in the test program. For nailed joints it is commonly accepted that the ultimate capacity is independent of the direction relative to the grain at which nails load the wood members (see report on Part 1 of this project). Based on that presumption, only arrangements where nails load members parallel to grain were tested. By contrast, it is known that when larger diameter fasteners such as bolts are used, joint capacity is quite sensitive to the direction relative to the grain at which the fastener loads each member (Smith et al. 1988). The test program was designed to encompass bolted joints in which all members were loaded parallel to the grain, and others where bolts loaded one or two members perpendicular to grain and the other member parallel to grain. The sensitivity of wood screw joints to the direction at which screws load the members is expected to be moderate, based on previous work (AF&PA 1999). Coach screw joints are expected to exhibit similar sensitivity to bolted joints to loading direction (AF&PA 1999). Testing joints with wood screws or coach screws was primarily necessary to assess the unique effects that the head, tapering (coach screws only) and the thread type have on load carrying capacity. It was judged only necessary to test arrangements where both members are loaded parallel to grain with such fasteners. As is detailed in the following sections, all testing was carried out in accordance with applicable Australian standards where they exist, and failing that, in accordance with internationally accepted methods. Joint Tests The schedule of joint tests is given in Tables 1 to 4. Minimum spacing, end and edge distances, and penetrations used are those specified in AS (SAA 1997a), Table 5. All tests were carried out following the provisions of AS1649 Timber Method of test for mechanical fasteners and connectors Basic working loads and characteristic strengths (SAA 1998a). Specimens were manufactured from commercially produced Australian pine lumber, pine plywood (wood screw joints only), glued-laminated-timber with 33mm thick 12

14 Radiata Pine laminates in the case of joints made with a 20mm diameter bolt, or treated pine for 12mm coach screws. The sample of timber to be used for each joint/fastener test was initially selected employing density grading of each member to ensure the appropriate joint group classification was met. At the time of fabrication the moisture content was about 12, 12, 8, 10, and 14 percent for Radiata Pine, Slash Pine, plywood, glued-laminated timber and treated pine, respectively. Specimens were cut so as to avoid occurrence of knots in locations receiving fasteners. Where all members were of the same material, member densities within a joint specimen were not matched to realistically reflect site practice. As noted in Tables 1 to 4, Slash Pine that was expected to meet the density requirements of joint categories JD5 or JD4 actually met those of JD3. This does not impinge significantly on the usefulness of data as its prime purpose is validation of EYM-type models. All specimens were kept in a controlled environment of 20 ± 2 C and 50 ± 5 percent relative humidity after fabrication, during testing and following testing. The number of test replicates was 10, which corresponds to the minimum requirements of ASTM D1761 (ASTM 1997b) and AS1649 (SAA 1998a). Specimens were loaded in tension, except for some bolted joints. Figure 5 shows how typical two-member and three-member tension specimens were loaded. The loading arrangements are designed to minimize eccentricity in the specimen, and avoid the need for lateral restraint as in some other methods. Nailed Joints All joints with nails were two-member arrangements. Most types of specimen (joint codes N1 to N7) had two nails in a row. This was done because structural nailed connections, unlike those made with most other dowel type fasteners, rarely have one fastener. This creates the possibility that parallel to grain cracks formed when nails are driven interact to weaken a joint. Joint with codes N8 to N10 had only one nail. Codes N8 to N10 corresponded to codes N3 to N5 respectively, except for the number of nails used. Comparison of results for these cases is intended to assess the effect of nail interaction. Nails were either bullet head bright mild steel with diamond points (AS2334: SAA 1980) that were hand driven with a 20-oz hammer, or flat head diamond point machine driven cartridge nails manufactured by SENCO. The latter were driven using a pneumatic nailing gun. A 0.8mm gap was created between members using metal shims that were removed immediately after nailing. Although nails were inserted perpendicular to surfaces of headside members, no unusual precautions were taken to ensure this. Similarly, no special precautions were taken or to avoid deviation of nails from a straight path through joint members. The aim was to reflect good site practice, but no better. Specimens were tested 13

15 not less than 24 hours after fabrication, which allowed some relaxation of pressure displaced fibers exert on nails following driving. This is common practice for test on driven fasteners and is expected to lead to a slightly lower, but more realistic, estimates of joint capacity than tests carried out immediately after nails are driven. Figure 6 shows a typical specimen. Specimens were loaded in tension within INSTRON universal test machines, models D and K. Loading was displacement controlled with an initial rate of cross-head movement of 1.25mm +/- 25 percent per minute. The rate of displacement was increased to 2.5mm per minute after a joint slip of at least 12.5mm or after a post yielding response had been well established. This allowed observation of the ultimate capacity in an elapsed time of about 8 to 10 minutes. Slip in the joint was measured using two Mitutoyo digital displacement gauges (reading to 0.01mm) each side of the fastener, Figure 7. Three synchronous data streams were recorded representing cross-head load, slip (average of the two) and elapsed time using a computer based data acquisition system at a sampling frequency of 75 readings per second. 14

16 Table 1. Joint configuration for nails Joint Type 1 Configuratio n 1 A A A B Joint Code N1 N2 N3 N4 N5 N4S N5S N6 N7 Fastener 2 2.8φ x 65 long 2.8φ x 65 long 3.15φ x 65 long 3.05φ x 75 long (MD) 3.05φ x 75 long (MD) 3.05φ x 75 long (MD) 3.05φ x 75 long (MD) 3.75φ x 75 long 4.5φ x 100 long Timber Members 3 t x w A = 35 x 90 B = 35 x 90 A = 35 x 90 B = 35 x 90 A = 35 x 90 B = 35 x 90 A = 45 x 90 B = 45 x 90 A = 45 x 90 B = 45 x 90 A = 45 x 90 B = 45 x 90 A = 45 x 90 B = 45 x 90 A = 37.5 x 90 B = 45 x 90 A = 45 x 90 B = 45 x 90 N Comment s 3 10 JD5 10 JD4 10 JD5 10 JD5 10 JD Slash Pine Target: JD5 Actual: JD4 Slash Pine Target: JD4 Actual: JD3 10 JD5 10 JD5 B B N8 3.15φ x 65 long A = 35 x 90 B = 35 x JD5 2 N9 3.05φ x 75 long (MD) A = 45 x 90 B = 45 x JD5 N φ x 75 long (MD) A = 45 x 90 B = 45 x JD4 Notes: 1 End and edge distances and fastener spacing follow the minimum specified in AS (SAA 1997a), unless otherwise noted under Comments. 2 All nails are bullet head bright mild steel except for machine driven flat head (MD). 3 Radiata Pine timber members unless otherwise noted under Comments. If target joint group is not different than actual (from experiment, see Appendix D), only one classification is given. 15

17 Table 2. Joint configuration for wood screws Joint Type Configuration 1 Joint Code S1 S2 3 B S2S Notes: A A S2SE S3 Fastener 2 Type 17 No. 8 x 40 Shank φ = 3.4 Type 17 No. 10 x 50 Shank φ = 3.8 Type 17 No. 10 x 50 Shank φ = 3.8 Type 17 No. 8 x 40 Shank φ = 3.4 Type 17 No. 14 x 115 Shank φ = 5.2 Timber Members 3 t x w A = 9 x 90 plywood B = 35 x 90 A = 18 x 90 B = 35 x 90 A = 18 x 90 B = 35 x 90 A = 18 x 90 B = 35 x 90 A = 65 x 90 B = 44 x 90 N Comment s 3 10 JD5 10 JD Slash Pine JD5 Slash Pine Target: JD5 Actual: JD4 10 JD4 1 End and edge distances and fastener spacing follow the minimum specified in AS (SAA 1997a), unless otherwise noted under Comments. 2 All screws were machine driven, no pre-drill. 3 Radiata Pine timber members unless otherwise noted under Comments. If target joint group is not different than actual (from experiment, see Appendix D), only one classification is given. 16

18 Table 3. Joint configuration for coach screws Joint Type Configuration 1 Joint Code Fastener 2 Timber Members 3 t x w N Comment s 3 A 4 CS1 B 6φ x 65 coach screw zinc plated A = 20 x 90 B = 45 x Target: JD5 Actual: JD4 Notes: A CS2 5 CS2S B CS3 CS4 6φ x 65 coach screw zinc plated 6φ x 65 coach screw zinc plated 12φ x 130 coach screw zinc plated 20φ x 200 coach screw galvanized A = 20 x 90 B = 45 x 90 A = 20 x 90 B = 45 x 90 A = 35 x 90 B = 90 x 90 A = 60 x 90 B = 200 x JD5 10 Slash Pine Target: JD5 Actual: JD4 10 JD4 10 Target: JD5 Actual: JD4 1 End and edge distances and fastener spacing follow the minimum specified in AS (SAA 1997a), unless otherwise noted under Comments. 2 Pre-drilled for shank and thread. Largest proprietry washers were used. 3 Radiata Pine timber members unless otherwise noted under Comments. If target joint group is not different than actual (from experiment, see Appendix D), only one classification is given. 17

19 Table 4. Joint configuration for bolts Joint Type Configuration 1 Joint Code Fastener 2 Timber Members 3,4 t x w N Comment s 3 BSSI 8φ x 90 long zinc plated A = 35 x 90 B = 35 x JD4 6 A BSS2 8φ x 90 long zinc plated A = 35 x 90 B = 35 x JD5 B BSS2S 8φ x 90 long zinc plated A = 35 x 90 B = 35 x Slash Pine Target: JD5 Actual: JD4 7 A B BSS3 BSS4 BDS1 16φ x 90 long galvanized 20φ x 150 long galvanized 8φ x 130 long zinc plated A = 35 x 90 B = 35 x 90 A = 60 x 100 B = 60 x 200 A = 35 x 90 B = 45 x 90 C = 35 x JD Gluelaminated Pine (JD4) Target: JD4 Actual: JD5 8 B BDS2 8φ x 130 long zinc plated A = 35 x 90 B = 45 x 90 C = 35 x JD5 A C BDS3 16φ x 130 long galvanized A = 35 x 90 B = 45 x 90 C = 35 x Target: JD4 Actual: JD5 BDS4 20φ x 240 long galvanized A = 60 x 90 B = 60 x 90 C = 60 x Gluelaminated Pine (JD4) 18

20 Table 4. Joint configuration for bolts cont BDS5 8φ x 130 long zinc plated A = 35 x 90 B = 45 x 90 C = 35 x JD4 Notes: A B C BDS6 9 BDS6S BDS7 BDS8 8φ x 130 long zinc plated 8φ x 130 long zinc plated 16φ x130 long galvanized 20φ x 240 long galvanized A = 35 x 90 B = 45 x 90 C = 35 x 90 A = 35 x 90 B = 45 x 90 C = 35 x 90 A = 35 x 140 B = 45 x 140 C = 35 x 140 A = 60 x 200 B = 60 x 90 C = 60 x JD Slash Pine Target: JD5 Actual: JD4 Target: JD4 Actual: JD5 Gluelaminated Pine (JD4) 1 End and edge distances and fastener spacing follow the minimum specified in AS (SAA 1997a), unless otherwise noted under Comments. 2 Coach bolts used with the largest available proprietary washers. 3 Radiata Pine timber members unless otherwise noted under Comments. If target joint group is not different than actual (from experiment, see Appendix D), only one classification is given. 4 Length of horizontal members in perpendicular configuration and distance between supports as per AS 1649 requirements Table 5. Minimum spacing, end and edge distances, and penetrations used are those specified in AS (SAA 1997a) Fastener type End distance (grain) Edge distance ( grain) Spacing in row (grain) Loaded edge dist. ( grain) Unloade d edge dist. ( grain) Main member penet n Side member penet n Nail 20D 5D 20D - - > 10D >10D Wood 10D 5D 10D - - >10D >7D screw Coach screw 7D 2D >3D JD3 >7D JD4 >8D JD5 >10D Bolt 7D 2D - 4D Note: Only those values relevant to the test program are given. Where D is the shank diameter of the fastener. 19

21 Figure 5. Arrangements for testing typical joints in tension Figure 6. Typical nailed joint 20

22 Figure 7. Measurement of slip in a nailed joint Figure 8. Typical screw joint prior to test 21

23 Figure 9 Typical coach screw joint prior to test (a) two member joint (b) three member joint Figure 10. Typical bolted joints prior to test 22

24 Figure 11. Measurement of slip in three-member bolted joint (two digital displacement gauges: one either side) Screw Joints Screw joints were all two-member arrangements with one screw. Four wood-to-wood and one plywood-to-wood arrangement were tested. Screw lengths ranged from 40 to 115mm. The plywood-to-wood arrangement employed a 40mm screw (joint code S1). The plywood member was oriented so that the screw loaded it parallel to the grain in the face veneers. Type 17 steel screws (AS3566: SAA 1988) were used throughout and were installed by machine without pre-drilled lead holes. A 0.8mm gap was created between members using metal shims that were removed immediately after screws were inserted. Screws were installed perpendicular to the face of the head-side member. Specimens were tested not less than 24 hours after fabrication which allowed some relaxation of pressure displaced fibers exert on screws following driving. This is expected to lead to a slightly lower, but more realistic, estimate of joint capacity than a test carried out immediately after screws are driven. Figure 8 shows a typical specimen prior to testing. Specimens were loaded in tension within a INSTRON universal test machine, model K (equipped with an A grade load cell) Loading was displacement controlled with an initial rate of cross-head movement of 1.25mm per minute +/- 25 percent. The rate of displacement was increased to 2.5 mm per minute 23

25 after a joint slip of at least 12.5mm, or a post yielding response had been well established. This allowed observation of the ultimate capacity after an elapsed time of about 8 to 10 minutes. Slip in the joint was measured using two Mitutoyo digital displacement gauges (reading to 0.01mm), as for nailed joints. Three synchronous data streams were recorded representing cross-head load, slip (average of the two) and elapsed time using a computer based data acquisition system at a sampling frequency of 75 readings per second. Coach Screw Joints Coach screw joints were two-member arrangements. Most specimens had one coach screw (joint codes CS2 to CS4). Code CS1 specimens had two coach screws in a row, but otherwise they were the same as code CS2 specimens. This allowed assessment of any negative effects of fastener interaction. Coach screws were zinc plated or galvanized steel and manufactured in accordance with AS/NZS1393 (SAA 1996). They were inserted into pre-drilled lead holes created using a drill press. Lead holes for the shank where not exceed by 1mm or 10 percent of the shank diameter, whichever was less. The diameter drilled for the threaded portion did not exceed the root diameter of the screw (Cl (iv), AS1720.1: SAA 1997a). Washers were installed beneath the heads of coach screws. These were plated flat round washers. Corresponding fastener diameters and washers dimensions were 6mm (16.5 O.D x 1.2 thickness), 12mm (29.0 x 2.4 thickness) and 20mm (39.0 x 3.2 thickness). While this does not meet the requirements of AS (SAA 1997a), they were the largest commercially available. All joints were done up tightly and then slightly backed off (slightly loosened) to avoid pre-tensioning in the fastener. Specimens were tested not less than 24 hours after fabrication which allowed some relaxation of pressure displaced fibers exert on coach screws following installation. Figure 9 shows a typical specimen prior to testing. Specimens were loaded in tension within one of several universal test machines, utilizing proprietary grade A load cells. Loading was displacement controlled with an initial rate of cross-head movement of 1.25mm per minute +/- 25 percent. The rate of displacement was increased to 2.5mm after a joint slip of at least 12.5mm, or a post yielding response had been well established. This allowed observation of the ultimate after a total elapsed time of about 8 to 10 minutes. Slip in the joint was measured using a combination of two digital gauges as for nailed joints. Three synchronous data streams were recorded representing cross-head load, slip (average of the two) and elapsed time using a computer based data acquisition system at a sampling frequency of 75 per second. 24

26 Bolted Joints Bolted joints were either two-member or three-member arrangements, with most being three member arrangements. Half the specimens had all members loaded parallel to grain while the other half had one or two members loaded perpendicular to grain. All types of specimen had only one bolt. Bolts were steel coach bolts manufactured in accordance with AS/NZS 1390 (SAA 1997b). They were inserted into oversized holes drilled with a drill press the diameter of holes being generally 10 percent larger than the shank diameter. A combination of spade and twist bits was used. Members within a specimen were drilled separately. Holes were drilled perpendicular to the surface of head-side members with an accuracy that reflects good site practice. For types of joint in which not all members are loaded parallel to grain the bolt was inserted so that the head was located in a member to be loaded perpendicular to grain (joint codes BSS3, BSS4 and BDS5 to BDS8). Washers were installed under nuts but not heads. Corresponding fastener diameters and washers dimensions were 8mm (19.9 O.D., 1.7 thickness), 16mm (34 O.D., 2.9 thickness), 20mm (39 O.D., 3.2 thickness). No gap was left between members at the time of fabrication but nuts were only made finger tight. This meant that there was no inter-member friction during initial phases of tests. Specimens were tested not less than 24 hours after fabrication for consistency with tests on specimens with other types of fastener. Figure 10 shows typical specimens prior to testing. Specimens with all members loaded parallel to grain were loaded in tension, while specimens with a member(s) loaded perpendicular to grain were loaded in compression. The length of the horizontal member(s), i.e. that (those) loaded perpendicular to grain, was 3W +150mm (where W = width of the vertical member), and the clear distance between its supports was 3W (AS1649; SAA1998a). For two member joints loaded in compression, the horizontal member had to be restrained against rotation about its axis for stability. Tests were performed using INSTRON universal test machines, models D and K. Loading was displacement controlled with an initial rate of cross-head movement of 1.25mm per minute +/-25 percent. The rate of displacement was increased to 2.5 mm after a joint slip of at least 12.5mm, or a post yielding response had been well established. This allowed observation of the ultimate capacity after an elapsed time of about 8 to 10 minutes. For twomember joints slip was measured using two digital displacement gauges (reading to 0.01mm), as for joints with all other types of fastener. In the case of three-member joints slip was averaged across the two joint planes, Figure 11. Three synchronous data streams were recorded representing cross-head load, slip (average value for two joint planes in the case of three-member joints) and elapsed time using a computer based data logging system at a sampling frequency of 75 per second. 25

27 For all types of joint and each replicate, the failure mode was recorded and density and moisture content estimated for each member. Moisture content was determined by use of standard resistance moisture meters with the prongs being inserted close to the fastener position. Density was calculated based on the mass and volume at the time of test. Data for each specimen was post-processed to determine the density at 12% moisture content, initial secant stiffness, and ultimate load, and the average load-slip responses for specimens representing each joint code. Material Tests Material tests determined embedment properties and apparent yield strengths for each type of fastener. Embedment specimens were cut from the same supplies of material as used in joint tests for each combination of fastener type and size, and direction of loading relative to the grain (direction of the face grain in the case of plywood). Properties of fasteners were determined for a random sample drawn from the same batches as in joint tests. Materials tests are detailed below. Embedment Properties Tests conformed to the provisions of European Standard EN 383 (CEN 1995b). The intent in embedment tests is to load a wood member in such a manner that the fastener is pushed through wood in a pure translation mode in order to measure the apparent bearing/embedment strength. This is achieved using a thin wood member (embedment specimen) sandwiched between two rigid steel plates, Figure 12. It is possible to test specimens where the fastener is embedding parallel to grain either in compression or tension. A tension test was chosen because that agrees closely with the stress state within joint members (some bolted joints are an exception). For cases where a fastener is embedding perpendicular to grain a compression embedment test was used. By keeping the gaps between the steel side plates small, and the embedment specimen thin, bending deformation in the fastener is negligible, which achieves the objective of a pure translation movement of the fastener. The target is that specimens be no more than two fastener diameters thick (Rodd et al. 1987, Whale and Smith 1989). This is not always possible with panel products with a layered construction. Such products are tested in full thickness to properly conserve the effect of the layering. A small gap ( 2mm) must be left either side of the specimen to avoid friction, and to accommodate slight localised expansion immediately beneath the fastener as face fibers in the specimen are displaced laterally. Two embedment 26

28 apparatuses were built especially for the project based on designs recommended by Rodd et al. (1987) and Whale and Smith (1989). One apparatus was built for nails, wood screws, small diameter coach screws and small bolts, and another for large coach screws and large bolts. Fastener Embedment specimen Dummy (reaction) end Figure 12. Embedment test (schematic) 27

29 (a) (b) (c) Figure 13. Embedment test apparatus and test arrangement: (a) apparatus for testing small specimens (load parallel to grain) (b) apparatus for testing large specimens (load parallel to grain) (c) apparatus for testing large specimens (load perpendicular to grain) Geometry of embedment specimens is defined in Figure 14. The schedule of tests and dimensions of embedment specimens are given in Table 6. There were 10 test replicates for nails and bolts, i.e. for fasteners with what are assumed uniform and circular cross-sections. For wood screws and coach screws, i.e. fasteners with a significant threaded portion that bears on the wood, there were 10 replicates representing the unthreaded shank plus 10 replicates representing the threaded portion. Fasteners were inserted in the wood in exactly the same way as in associated joint tests, except for machine nails. Using a machine to drive nails into thin pieces splits wood and is not practical, therefore machine nails were driven using a hammer. Specimens were tested not less than 24 hours after fabrication. Specimens were loaded within a universal test machine. Loading was displacement controlled with rate of cross-head movement of 2mm per minute. A test was discontinued once it was clear that the maximum load had been attained. Usually this occurred at a displacement less than 5mm. Displacement of the specimen relative to the steel side plates was measured using two Mitutoyo digital displacement gauges with an accuracy of 0.01mm, one mounted either side of the specimen. Three synchronous data 28

30 streams were recorded representing load, deformation (average of two readings) and elapsed time using a computer based data acquisition system at a sampling frequency of 75 per second. Subsidiary measurements determined the actual fastener shank diameter and the wood thickness for each specimen. Any special features of the failure were recorded. Moisture content at the time of test was determined using standard moisture resistance metres with the probes inserted near the fastener. Density was determined based on the mass and volume of the sample at the time of testing and subsequently adjusted to the value at 12 per cent moisture content. Data for each specimen was post-processed to determine the initial secant stiffness and the ultimate load. The logic employed for determining embedment strength is depicted in Figure 15. It was calculated as: S H = P ult / (d act t act ) where P ult is the ultimate (peak) load for the specimen, d act is the actual shank diameter, and t act is the actual specimen thickness at the fastener location. d e Live end L D Dummy end f w t Figure 14. Geometry of embedment specimens (Schematic: tension arrangement for embedment parallel to grain shown) 29

31 Table 6. Schedule of embedment tests, and dimensions of embedment specimens Nominal diameter, d & Type Joint category w L t e f D Nails in Radiata Pine 2.8 BH JD BH JD BH JD BH JD BH JD MD JD MD JD Nails in Slash Pine 3.05 MD JD5 target JD3 actual 3.05 MD JD4 target JD3 actual Nominal diameter, d Joint category w L t e f D Wood screws in Radiata Pine 3.4 JD JD JD Wood screws in Slash Pine 3.4 JD5 target JD3 actual 3.8 JD5 target JD3 actual Wood screws in 8mmPine plywood

32 Table 6 Ctd Nominal diameter, d Joint category w L t e f Coach screws in Radiata Pine 6.0 JD JD JD Coach screws in Slash Pine 6.0 JD5 target JD3 actual Nominal diameter, d Joint category Directi on of load w L 12.5 t e f Bolts in Radiata Pine 8.0 JD JD JD JD JD Glulam Glulam Bolts in Slash Pine 8.0 JD5 target JD3 actual D (m m) D (m m) Bearing Stress Embedment Strength Embedment Figure 15. Definition of embedment strength 31

33 Fastener properties Bending Tests (nails, wood screws, small coach screws, small bolts) Tests conformed to the recommendations of European Standard EN 409 (CEN 1995c). This is a bending test, the principle of which is that a short length of fastener is located at the reaction point of a cantilever beam the end of which is loaded, Figure 16. A special purpose test apparatus was built based on devices constructed at the University of Karlsruhe, Germany, and Forintek Canada Corp-West located in Vancouver, British Columbia, Figure 17. Essentials of the test method are that the nail is assumed to be subject to a constant moment in the length over which it forms the link between the lever and reaction mechanism. The actual operation is that the lever remains essentially horizontal while the reaction mechanism is rotated. Force applied to the lever and rotation of the reaction mechanism are recorded. The force applied to the end of the lever is multiplied by the lever s effective length to give the moment applied to the fastener. Tests are terminated at a rotation of about 45 unless there is a premature failure as can happen with wood screws or small or small diameter coach screws. Yield moment M Y is extracted from a moment-rotation plot as shown in Figure 18, and is equal to the peak value. The exposed length of fastener, l 2 = clear distance between the end of the lever and the reaction mechanism, varies as a function of fastener diameter. The test standard specifies it should be between d and 3d. Actual values are given in Table 7. The length of lever used in moment calculations was l 2. The number of test replicates was five. Short coach screws were positioned in the apparatus so that the yield moment was measured in the unthreaded portion of the length. Long coach screws were tested to determine M Y for both threaded and unthreaded portions of the length. Previous studies have shown that M Y for the threaded portion can be taken as 0.75 times the value for the unthreaded shank (AF&PA 1999). M!FL F Figure 16. Principle of bending apparatus for small diameter fasteners L Reaction Point Lever arm l 2 32

34 Figure 17. Bending apparatus for small diameter fasteners 33

35 Nail bending moment Yield Moment 0.0 Rotation reaction point (deg.) Figure 18. Interpretation of moment-rotation plots for small diameter fasteners Table 7. Values of l 2 in bending tests on small diameter fasteners Type of fastener Diameter, d l 2 Nail 2.8 BH BH BH BH MD 5 Wood screw Coach screw Bolt Tension Tests (large coach screws, large bolts) Tests conformed to the recommendations of AS/NZS (SAA 1995). Although it is a method intended for determination of tensile properties of bolts, it was deemed also appropriate for large coach screws. Fasteners were necked down to insure that yielding and failure occur in a controlled manner away from the threaded portion of the shank, Figure 19. This method was adopted for fasteners too large to be tested in bending. Table 8 gives nominal specimen dimensions and the number of replicates for each type of fastener. Specimens were loaded within a universal test machine, with A grade calibration. Loading was displacement controlled with rate of cross-head movement of 12mm per minute. A test was discontinued once it was clear that the ultimate load had been attained. Strain in the specimen was not measured. Subsidiary measurements determined diameter of the 34

36 unreduced shank and the diameter of the reduced section. Data for each specimen was post-processed to determine the yield stress and the ultimate (peak) stress. The logic employed for determining yield and ultimate stresses is depicted in Figure 20. Based on prior experience (Whale and Smith 1987a, Ehlbeck and Larsen 1993), the yield moment for a bolt or coach screw is estimated from the expression: M Y = (σ y + σ ult ) d shank 3 / 12 where σ y is the yield stress, σ ult is the ultimate stress and d shank is the diameter in the unthreaded portion of the length. In the case of coach screws, this value of M Y applies to the unthreaded portion of the length. Based on previous studies it was assumed that M Y for the threaded portion is 0.75 times the calculated value (AF&PA 1999). Figure 19. Typical tension test specimens for (a) large bolts, and (b) large coach screws 35

37 Axial Stress Ultimate stress, σ ult Yield stress, σ Y Cross-head movement Figure 20. Interpretation of tension test data for large bolts and large coach screws Table 8. Nominal dimensions and number of replicates of bolt and coach screw tension specimens Type of fastener Coach screw Bolt Diameter of unreduced shank, d Diameter in reduced region Length of bolt Length of reduced portion Number of replicates

38 Results and Discussions Overview This section presents the summary of results of the fastener tests, embedment tests and whole joint tests and compares the EYM predictions in relation to the joint test results. Complete details about measured data and experimental results, prediction calculations and comparative plots of joint strength capacities between experiments and predictions are given in Appendices D, E, F and G. Four models were used for prediction: Original EYM (Johansen 1949; Larsen 1979) Simplified-1 (Whale et al. 1987) Simplified-2 (Blass et al. 1999) NDS-Screw (AF&PA 1997; 1999) (for wood and coach screws only) As indicated in the previous section, all the specimens that were tested to obtain fastener yield moment and wood embedment strength were representative of those used in whole joint tests. Wood specimens for embedment strength tests were matched with the timber members in the corresponding joint code. Fastener Tests Fasteners were tested in bending (for small-diameter dowels) and tension (for largediameter dowels). Figure 21 shows typical fasteners before and after a bending test to determine yield moment, M Y. 37

39 Figure 21. Typical fasteners before and after a bending test to determine M Y Full experimental results for the fastener testing are given in Appendix D. As expected, the coefficient of variation (COV) of the fastener diameter for nails, screws, coach screws and bolts was very low, up to 0.8%, 0.6%, 0.2% and 0.2%, respectively. It should be noted that the fastener diameter given for screws, coach screws and bolts is the shank diameter. In the extreme case it was found that the average diameter of the shank of the 16-mm bolt was 14.7-mm an 8% reduction on the nominal diameter. The average measured diameter was used in the prediction calculations. The COV of the yield moment for the nails, screws, coach screws and bolts was up to 5.9%, 4.9%, 8.6% and 19.1%, respectively. The COV for yield moment for the 8-mm bolts that were measured directly was 2.4%. The yield moments for the 16-mm and 20-mm bolts were measured indirectly, using tension strength tests, as described in the previous section. To be consistent with the measurement of fastener diameter, the yield moment was that measured in the shank of the fastener. Wood Embedment Tests Full experimental results for the wood embedment testing are also given in Appendix D. The moisture content of the embedment test specimens was in the range 8.5% to 9.1% for wood specimens with nails, 8.2% to 13.6% for wood with screws, 8.5% to 20.1% for wood with coach screws and 8.2% to 13.5% for wood with bolts. Where the moisture contents were high, e.g., CSE3 and CSE4, the corresponding joint test specimens were also high. It follows that the prediction should be relatively high in relation to the joint test result for those 38

40 joint codes. This was in fact the case, as evidenced in the plot comparisons for joint codes CS3 and CS4 in Appendix G. To accord with the fastener tests, the embedment tests for wood with screws, coach screws and bolts were carried out with the shank of the fastener within the wood test specimen. The measured densities are given in Appendix D and the joint group classifications as defined in AS are tabulated in Tables 1, 2, 3 and 4 in the previous Section. The Radiata pine specimens were mostly in joint group JD5 with a small number in JD4. Slash pine generally fell into the JD4 joint group with one joint configuration (code N5S, Table 1) in JD3. The glued-laminated timber was JD4. Figure 22 shows an embedment test specimen for 50-mm type 17 No. 10 screw loading Radiata pine parallel to grain (top: unthreaded shank loading wood, bottom: thread loading wood). Joint Tests The full experimental results for the joint tests are given in Appendix E. The moisture contents and densities of the joint test specimens were well matched with the accompanying embedment tests. All types of joint that were investigated exhibit ductile load-slip behaviour under static load, Figures 29 to 32 (see also Appendix G). This is because failure mechanisms depend upon compression behaviour of wood and bending of steel fasteners. Despite the ductile behaviour, except for nailed joints there was often no well defined yield point. Although there was an initial gap between adjacent members of joints made with nails, wood screws or coach screws, those gaps typically closed due to large displacement effects soon after a post yield response was established. In nailed joints, the fasteners incrementally withdrew from point-side members as the slip in the joint became large. This prevented development of large axial force in nails at joint planes, and thus limited the extent of capacity enhancement due to inter-member friction. For other types of joint, threads (wood screws, coach screws), heads (wood screws, coach screws, bolts), and nuts and washers (bolts) prevented fasteners from withdrawing from members. (Figures 24 to 28 show typical joint failure, with observation comments.) Thus, substantial enhancements of capacities occurred due to inter-member friction and large displacement string effects. Gap closure and other large displacement effects were prime reasons why there were often indistinct yield points. 39

41 Because yield points were not always well defined, no attempt has been made to define their numerical values from test data. Subsequent discussion in this document on agreement between predicted and observed capacities is based on observed ultimate strengths and visual comparison of load-slip trends with predicted yield capacities. Figure 22. Embedment test specimen for 50mm type 17 No. 10 screw loading Radiata pine parallel to grain (top: unthreaded shank loading wood, bottom: thread loading wood) 40

42 Figure 23. Two-member joint with slender nails exhibiting mode IV failure (Note: Plastically deformed shape of the wood screw is symmetric. Members are drawn tightly together and the nails slide within members as the nails deform in bending.) Figure 24. Two-member joint with a slender wood screw exhibiting mode IV failure (Note: Plastically deformed shape of the wood screw is not symmetric. Members are drawn tightly together as the screw deforms in bending.) 41

43 Figure 25. Two-member joint with a slender coach screw exhibiting mode III s failure (Note: Partial rotational restraint at the head end of the fastener.) Figure 26. Two-member joint with a slender coach bolt exhibiting mode IV failure (Note: Plastically deformed shape of the bolt is approximately symmetric. The first thread on nut end of the bolt controls the location of the plastic hinge on that side of the joint plane.) 42

44 Figure 27. Three-member joint with a slender bolt exhibiting mode III s /IV failure (Note: that the plastically deformed shape of the bolt is not symmetric owing to lack of symmetry in the bolt itself. Partial rotational restraint at end of the bolt.) Figure 28. Three-member joint with a stocky coach bolt exhibiting mode III s failure (Note: Plastically deformed shape of the bolt is approximately symmetric despite the lack of symmetry in the bolt itself. Partial rotational restraint at end of the bolt.) 43