Construction and Building ELSEVIER Construction and Building Materias 1(0)407-417 MATERIALS www.elsevier.com/locate/conbuildmat Bolted joints in glulam and structural timber composites Tim J.Davis*,Peter A.Claisse School of the Built Environment.Coventry University.Coventry.CVI 5FB.UK Received 15 November 1;received in revised Abstract evouta Standard Th oden that nd characersatedat this programme high-ten r similar performance to Keyords Timber:Bolts:Composites 1.Introduction The UK im orts a significant tion of its co sites A of thi use of resin-injected dowelled joints has also been grown resources through various industry and govern- assessed. ment initiati mthod that has the potea products include small pieces of wood bonded in a 2.Research significance umbaldehyde-based resin-known as parallel snded more commo y,t removal of strength-reducing defects such as knots.A standardised bolted conne ction.The work shows the relative strength and stiffness of the reconstituted wood materials when used with this jointing system.The deformations within the highly stressed areas of a me e con chanically fastened joint.The results presented in this currently a draft for developm ment within the UK.It requires characteristic material data in order to facili- 838-485 responding author.Tel +4-247688-485:x+442476 the design tdavis ntry.ac.uk(T.J.Davis). 0Eevier Scne Ld.ll r reed
Construction and Building Materials 14 2000 407 Ž . ]417 Bolted joints in glulam and structural timber composites Tim J. DavisU, Peter A. Claisse School of the Built En¨ironment, Co¨entry Uni¨ersity, Co¨entry, CV1 5FB, UK Received 15 November 1999; received in revised form 1 August 2000; accepted 20 August 2000 Abstract The widespread adoption of the European design code for timber structures EC5 will facilitate a number of design options Ž . previously unsupported by British Standards. This code uses design equations that need characteristic material data, which exists for solid timber and some sheet materials, but not for the structural timber composites that were evaluated in this research. In this programme high-tensile steel black bolts have been used with solid timber, glulam and two commercially available structural timber composites } MicrolamTM and ParallamTM. The results suggest that the timber composites offer similar performance to high-density timbers in line with EC5 design guidance. Q 2000 Elsevier Science Ltd. All rights reserved. Keywords: Timber; Bolts; Composites 1. Introduction The UK imports a significant proportion of its construction timber but is keen to better utilise its homegrown resources through various industry and government initiatives. One method that has the potential to use the available resources efficiently is to produce reconstituted wood products. Existing commercial products include small pieces of wood bonded in a formaldehyde-based resin-known as parallel strand lumber PSL , and, more commonly, thin plies bonded Ž . into a laminate-laminated veneer lumber LVL . These Ž . composite materials offer reduced variability and the removal of strength-reducing defects such as knots. A possible cause for concern is the fact that the reconstitution of the wood may give rise to internal voids which will cause stress concentrations and hence increased deformations within the highly stressed areas of a mechanically fastened joint. The results presented in this U Corresponding author. Tel.: q44-2476-888-485; fax: q44-2476- 838-485. E-mail address: t.davis@coventry.ac.uk T.J. Davis . Ž . paper are taken from a larger investigation into the performance of mechanically fastened joints in structural timber composites. As part of this programme the use of resin-injected dowelled joints has also been assessed. 2. Research significance This paper presents results in order to show the comparative performance of glued laminated timber Ž . glulam and two structural timber composites utilising a standardised bolted connection. The work shows the relative strength and stiffness of the reconstituted wood materials when used with this jointing system. The application and relative merits of the composites are discussed. The new European design code, EC5, is currently a draft for development within the UK. It requires characteristic material data in order to facilitate timber design. There is currently a lack of available data for the design of joints, particularly with structural timber composites. 0950-0618r00r$ - see front matter Q 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 9 5 0 - 0 6 1 8 0 0 Ž . 00044-1
40s TJ.Davis,P.A.Claisse/Co and Building)407-417 s.Souther Wood type Parallam pine(SS) ending parallel to grain ion 23159 80 328 30 10500 12750 12400 wood was used for solid timber and glulam samples in this research r to glue line/wide face of strand 3.Literature review 3.1.The use of structural timber composites performed in the 1930s during which the mode of action of a wide range of timber connectors tO P difficult to-date re from the natural forest resource These materials research.of timber joint design in the were introduced to the UK construction industry in the arly and U New jointing ystms structural eith significant ly higher vestiga be seen in Table 1.the benefits of reconstitution nts [ol UK timber design is currently going through a major appro the rigin Bs5268I990G artial co are h on bending strength).The modulus of elasticity the permissible stress approach used by BS5268:Part 2 argely un [11]In anticipation of the introduction of the new gover ed omp te R(TRADA an De [121 of tice for timher ioints and established research data that the design grade is recommended as SC5.This appar ent restriction on design of joint ura design cod The r ne general lack of was of the Structural timber composites are currently used in of EC5 and promoting the use of timber within Europe the UK predominantly to provide the lit more the STEP/EUROFORTECH initiative produced a sig- men he ISA sh 121 and medium span highway bridges [31 Several es tablished sources for r the mechanical properties of solid species The authors tound no pub 3.3.Bolted connections in plain timber he sites othe Nails.bolts and dowels (whether utilising an interfer stresses s in British Board of Agrement (BBA)certifi- ence fit or some form of resin bonding)are all exam cates[6,7刀. ougn a late dowe Iray 3.2.Timber jointing systems and their desigr Several state-of-the-art reviews of mechanically fas- rations.This work formed the basis of the empirical tened jointing systems have been performed,usually as design data for UK permissible stress design codes
408 T.J. Da¨is, P.A. Claisse rConstruction and Building Materials 14 2000 407 ( ) ]417 Table 1 Grade stresses of structural timber composites, Southern pine and European Whitewooda Property Wood type 2 Ž . all values in Nrmm Whitewood Southern Parallam Microlam Ž. Ž. Ž . SS pine SS 2.1 E b Bending parallel to grain 7.5 9.6 16.8 16.2 Tension parallel to grain 4.5 5.8 14.8 10.1 Compression parallel to grain 7.9 10.2 15.1 14.3 c Compression perpendicular to grain 2.1 2.5 3.6r2.8 4.9r3.0 b Shear parallel to grain 0.82 0.98 2.2 1.9 Modulus of elasticity in bending mean 10 500 12 500 12 750 12 400 Ž . a Southern pine is the source species for the composites, European Whitewood was used for solid timber and glulam samples in this research. b When loaded as a joist. c Parallelrperpendicular to glue linerwide face of strand. 3. Literature review 3.1. The use of structural timber composites Structural timber composites were introduced to the USA in the late 1980s in order to provide high-quality structural timber that was proving difficult to obtain from the natural forest resource 1 . These materials w x were introduced to the UK construction industry in the early 1990s and offered significantly higher grade stresses than either solid softwood timber or glulam. As can be seen in Table 1, the benefits of reconstitution are an increase in permissible stresses of between approximately 50 and 200% on the original solid timber. This results in the reconstituted products being assigned a strength class of SC7 to BS 5268:1990 based Ž on bending strength . The modulus of elasticity is . largely unaffected and since it is deflection that usually governs the design of timber beams, the composites appear to be best utilised in axially loaded structures such as trusses. However, for joints in such structures, the design grade is recommended as SC5. This apparent restriction on the design of joints in structural wood composites was one of the main reasons for initiating this programme of research. Structural timber composites are currently used in the UK predominantly to provide the more highly stressed elements in timber-framed buildings 2 al- w x though their use in the USA has extended into short and medium span highway bridges 3 . Several es- w x tablished sources for the mechanical properties of solid wood species exist 4,5 . The authors found no pub- w x lished source for the mechanical properties of the structural wood composites other than the grade stresses in British Board of Agrement BBA certifi- ´ Ž . cates 6,7 . w x 3.2. Timber jointing systems and their design Several state-of-the-art reviews of mechanically fastened jointing systems have been performed, usually as a result of the introduction of new design codes that rely on existing research data. The first significant review was performed in the 1930s during which the mode of action of a wide range of timber connectors was established. The American Society of Civil Engineers ASCE provided a much needed up-to-date re- Ž . view 8 which details the design rules, and supporting w x research, of timber joint design in the USA, Canada and UK. New jointing systems that utilise structural adhesives are being investigated that show enhanced structural performance over mechanically fastened joints 9 . w x UK timber design is currently going through a major period of change as a result of the introduction of the draft EC5 10 and BS 5268: Part 1, which are both w x partial coefficients limit states design codes rather than the permissible stress approach used by BS 5268: Part 2 w x 11 . In anticipation of the introduction of the new codes the Timber Research and Development Association TRADA conducted a review 12 of design prac- Ž . w x tice for timber joints and established research data that was needed to support joint design to the new EC5 design code. The review highlighted the general lack of research data available, particularly for the new structural timber composites. In support of the introduction of EC5 and promoting the use of timber within Europe the STEPrEUROFORTECH initiative produced a significant review of timber engineering including jointing systems 13 . A similar work is available for US design w x standards 14 . w x 3.3. Bolted connections in plain timber Nails, bolts and dowels whether utilising an interfer- Ž ence fit or some form of resin bonding are all exam- . ples of mechanical fasteners that form timber joints through a laterally loaded dowel action. Trayer 15 w x performed an extensive research programme into bolted joints involving several wood species and joint configurations. This work formed the basis of the empirical design data for UK permissible stress design codes
TJ.Davis,PA Claisse /Construction and Building Materials 14(00)407-417 Johansen [16]supplemented Trayer's work and devel- 4.Laboratory testing programme oped theoretical equations for predicting the yield load The solid timber was commercially available Euro- me into the load-embedment response ne doweledonThey d that thereood ture the glulam joint samples using a resorcinol-fo correlation between the embedment response and den- maldehyde resin. sity of wood. tec The LVL and PSL structural wood composites wer that have been integrated into current testing standards [22]that were adopted for this research pro crolamTM LVL Grade 2.1E.They are both forms of gramme.Wilkinson compared design strengths based reconstituted wood:Parallam is made from strands of the American design while Mi Intrallam Parallam Glulam Microlam Fig.1.Glulam and the family of structural timber composites produced by TrusJoist MacMillan
T.J. Da¨is, P.A. Claisse rConstruction and Building Materials 14 2000 407 ( ) ]417 409 Johansen 16 supplemented Trayer’s work and devel- w x oped theoretical equations for predicting the yield load and ultimate load of doweled joints that now form the basis of design in EC5 17 . Whale and Smith 18 wx wx extended this work by performing an extensive testing programme into the load-embedment response of doweled joints. They concluded that there was a good correlation between the embedment response and density of wood. Similar experimental techniques have been employed on a range of investigations 19 w x ]21 that have since been integrated into current testing standards 22 that were adopted for this research pro- w x gramme. Wilkinson compared design strengths based on Johansens’ equations with the American design standard 23 . w x 4. Laboratory testing programme The solid timber was commercially available European WhitewoodrRedwood, visually graded SS C24 to Ž EC5 , obtained from a local timber merchant. Selected, . conditioned, pieces of this wood were used to manufacture the glulam joint samples using a resorcinol-formaldehyde resin. The LVL and PSL structural wood composites were both products of the American company TrusJoist w x MacMillan 24 -Parallam PSL Grade 2.0E and Mi- TM crolamTM LVL Grade 2.1E. They are both forms of reconstituted wood: Parallam is made from strands of Douglas Fir or Southern Pine timber bonded together with a phenol-formaldehyde PF adhesive, while Mi- Ž . Fig. 1. Glulam and the family of structural timber composites produced by TrusJoist MacMillan
uction and Building Materials1 000)407-417 crolam is made from laminated veneers of Southern samples were made from solid timber,glulam and dhesive. Theay-up o n Fig. ess in thi ch hae a Ld ratio of 3.67.The been fully described previously 21 In summary,all experimental arrangement,showing the sample dimen sions and positioning of LVDTs for this series of tests n order to remov of which are:initial loading to 40%of the estimated maximum load(F),approximately the working load. 5.Results of the testing programme of the tested using ot sample an09n289 est connector LVDT position to o I test nt of 44 mm 100mm HT teel bolts) Section A-A Fig.2.Loading rig and LVDT positions for recording movement of the test joint
410 T.J. Da¨is, P.A. Claisse rConstruction and Building Materials 14 2000 407 ( ) ]417 crolam is made from laminated veneers of Southern Pine timber bonded, again bonded with a PF adhesive. The lay-up of these materials is shown in Fig. 1. The testing programme utilised in this research has been fully described previously 21 . In summary, all w x wood was conditioned for 3 weeks prior to the test, which was performed in accordance with BS EN 26891 in order to give a load-deformation response. This involves a multi-stage loading regime, the key elements of which are: initial loading to 40% of the estimated maximum load Ž . F , approximately the working load, est of the joint; removal of load to 10% of F and finally est loading to failure. A total of 26 samples were tested using 12-mm diameter Grade 8.8 bolts in a 13-mm diameter hole. Six samples were made from solid timber, glulam and Parallam and 8 samples from Microlam. The samples had a nominal thickness of 44-mm, which resulted in an embedment response a L Ž . rd ratio of 3.67 . The experimental arrangement, showing the sample dimensions and positioning of LVDTs for this series of tests, is reproduced in Fig. 2. The actual loading rig, with the LVDTs removed for clarity, is shown in Fig. 3. 5. Results of the testing programme A typical load-slip response resulting from the test, and the identification of the derived characteristics, is Fig. 2. Loading rig and LVDT positions for recording movement of the test joint
TJ.Davis,PA Claisse /Construction and Building Materials 14(00)407-417 Fig.3.Loading rig with LVDTs removed for clarity
T.J. Da¨is, P.A. Claisse rConstruction and Building Materials 14 2000 407 ( ) ]417 411 Fig. 3. Loading rig with LVDTs removed for clarity
412 TJ.Dais,PA.Claisse/Constnuction and Building Materials 14(00)47-417 25 20 10 0.4 F.m Stirness K. slip st Fnee 0.1 Fast 0.00 1.00 200 3.00 4.00 5.00 6.00 7.00 8.00 Joint Slip (mm) Fig4.Typical load-slip respon ofjointitestandidcnticeationofcalculhiedtetparameters in Fig.4.The following 0h. the base wood ma the maximum load in for all but the Parallam samples.The coefficient of kN achieved by the variation (CV)of a set of results has been denned as the its Th the standa ted area of the fastener; Ki,the initial stiffness of the joint in kN/mm d from a are comparable although some solid samples exhibited and lip respons extreme (in exc 10 mm at failure).I K,the stiffness of the joint in kN/mm,determined be comeffect on npatentdcu from a line r regre n analy 0I stiffness.and slip at maximum load was much less response during th variable in the glulam.The Microlam samples per- spe The initial stiffness for the joint,representing 'bed- bited les 6 ad (orin?品 slip at ne jomnt,and able).The Parallam samples wer ticularly disap 11 The embedment strength is calculated according to the pointing,exhibiting the widest variations in results,but relationship: this is due to the presence of internal voids (visible during rge varia ton in 人-器 a Its of s ples P-B-1 and P-B-3 which had a low than average density. where d is the diameter of the bolt(12 mm),and t is Fig.7 shows the embedment strength of all the the thickness of the e【est sample(44士1mm samples plotted against their de ty.Th result pana aneanhipse plywood. For solid timberf=0.082(1-0.01d)p givingf=0.072p
412 T.J. Da¨is, P.A. Claisse rConstruction and Building Materials 14 2000 407 ( ) ]417 Fig. 4. Typical load-slip response of joint test and identification of calculated test parameters. shown in Fig. 4. The following parameters were obtained from the load-slip response: v Fmax , the maximum load in kN achieved by the joint, and the corresponding slip in mm; v f , the embedment strength in Nrmm2 , defined as h Fmaxrprojected area of the fastener; v K , the initial stiffness of the joint in kNrmm, i determined from a linear regression analysis of the load-slip response after any initial slip and 0.4 F ; est and v Ks, the stiffness of the joint in kNrmm, determined from a linear regression analysis of the load-slip response during the reloading stage 0.1 to 0.4 F . est The initial stiffness for the joint, representing ‘bedding-in’ following joint fabrication, is 60]70% of the reload working stiffness of the joint, and is used to Ž . determine non-recoverable deformation of the joint. The embedment strength is calculated according to the relationship: Fmax f hs dt where d is the diameter of the bolt 12 mm , and Ž . t is the thickness of the test sample 44 Ž . "1 mm . A summary of the embedment strength and stiffness results is shown in Figs. 5 and 6, respectively. On these graphs all of the samples are shown in order to indicate the spread of the data. Other test results, including modes of failure are listed in Tables 2]5 for each of the base wood materials. The embedment strengths were very consistent with low coefficients of variation for all but the Parallam samples. The coefficient of variation CV of a set of results has been defined as Ž . the ratio of its mean to the standard deviation Ž . sny1 of the results. The joint stiffness values, however, are much more variable-both within a material group, and between groups. The glulam and solid timber results are comparable although some solid samples exhibited extreme ductility in excess of 10 mm at failure . It can Ž . be concluded that the glueline had no apparent detrimental effect on joint strength or stiffness, in fact the stiffness, and slip at maximum load was much less variable in the glulam. The Microlam samples performed consistently better in both strength and, especially, stiffness terms. They also exhibited less slip at maximum load although it was four times more vari- Ž able . The Parallam samples were particularly disap- . pointing, exhibiting the widest variations in results, but this is due to the presence of internal voids visible Ž during fabrication , the relatively large variation in . sample density, and the influence of the relatively poor results of samples P-B-1 and P-B-3 which had a lower than average density. Fig. 7 shows the embedment strength of all the samples plotted against their density. These results follow a linear trend consistent with the embedmentdensity relationships given in EC5 for solid timber and plywood. For solid timber fh,0,k k s0.082 1Ž . y0.01d r giving fhs0.072 r
TJ.Davis,PA Claisse /Construction and Building Materials 14(00)407-417 80 50 % 0 Solid wood Microlam Paralam Base wood material Fig.5.Embedment strength results (columns represent test samples in sequential order). (1-001p and for ply 6.Load-slip responses and modes of failure No comparable equation for the structural timber All but two samples exhibited a splitting mode of composites is offered due to the low number of sam- failure (Fig.8).This took the form of local crushing of ples no the wood fibres at the bearing interface followed by orted here. the remaining two samples exhibited a block-shea 0 50 3 Base wood material Fig.6.Joint stiffness results,initial stifnessKand stiessK
T.J. Da¨is, P.A. Claisse rConstruction and Building Materials 14 2000 407 ( ) ]417 413 Fig. 5. Embedment strength results columns represent test samples in sequential order . Ž . and for plywood fh,0,k k s0.11 1Ž . y0.01d r giving f s0.097r. h No comparable equation for the structural timber composites is offered due to the low number of samples not being statistically significant. Similarly, average rather than characteristic values for the joint properties are reported here. 6. Load-slip responses and modes of failure All but two samples exhibited a splitting mode of failure Fig. 8 . This took the form of local crushing of Ž . the wood fibres at the bearing interface followed by transverse perpendicular to the grain tension failure, Ž . the remaining two samples exhibited a block-shear Fig. 6. Joint stiffness results, initial stiffness K and stiffness K . i s
414 TJ.Davis,PA.Claisse /C d Building Materials 14(00)407417 bolted joints timber Initial Stiffness Maximum Slip at Embedmen Wood (N/mm) () ad(m) 279 183 193 30 i68 "Moisture content at test =11t2%. mode of failure.The high-tensile steel bolts were gen- Fig.9.On initial loading the joints exhibited a non-lin- e permanent eormaom None non-recoverable, Within three levels,i.e. nd 04 slip graphs for the tests are shown in imber extreme ductility (Fig.9a) rised results for bolted joints Wood (kN/m (kN/mm) load (m e Splitting 21 18 "Moisture content at test 12+1%. Table4 Wood (kN/mm) 20.7 890645 185806 00004500 三 182 25.5 0.3 27 6 .01 Moisture content at test 7+1%
414 T.J. Da¨is, P.A. Claisse rConstruction and Building Materials 14 2000 407 ( ) ]417 Table 2 a Summarised results for bolted joints in solid timber Sample Initial Stiffness Maximum Slip at Embedment Wood Failure number stiffness K K load F maximum strength f density mode i s max h 2 3 Ž . Ž . Ž. Ž . Ž . Ž . kNrmm kNrmm kN load mm Nrmm kgrm S-B-1 13.0 17.9 18.3 3.08 34.7 490 Splitting S-B-2 11.3 21.7 17.9 2.79 33.9 490 Splitting S-B-3 11.1 20.7 21.0 3.28 39.8 500 Splitting S-B-4 6.41 10.8 19.3 4.61 36.5 510 Splitting S-B-5 11.7 16.6 21.7 3.83 41.0 510 Splitting S-B-6 8.91 22.4 16.8 2.93 31.9 470 Splitting Ave. 10.4 18.3 19.2 3.42 36.3 490 CoV 0.23 0.24 0.10 0.20 0.10 0.03 a Moisture content at tests11"2%. mode of failure. The high-tensile steel bolts were generally unaffected, confirming the desired embedment response, although the three strongest Parallam samples did cause noticeable permanent deformation. None of the glulam samples failed at the glueline, the transverse splitting occurred in the adjacent wood. Typical load-slip graphs for the tests are shown in Fig. 9. On initial loading the joints exhibited a non-linear, non-recoverable, ‘bedding-in’ response. Within working stress levels, i.e. on reloading, the joints gave a linear load-slip response but on loading beyond 0.4Fest the response was non-linear up to the maximum load, which occurred at a slip of 2]4 mm. Some of the solid timber samples exhibited extreme ductility Fig. 9a Ž . Table 3 a Summarised results for bolted joints in glulam Sample Initial Stiffness Maximum Slip at Embedment Wood Failure number stiffness K K load F maximum strength f density mode i s max h 2 3 Ž . Ž . Ž. Ž . Ž . Ž . kNrmm kNrmm kN load mm Nrmm kgrm G-B-1 11.7 20.0 19.9 3.60 37.6 500 Splitting G-B-2 15.5 24.2 20.0 3.59 37.8 510 Splitting G-B-3 18.4 23.0 22.4 2.88 42.4 550 Splitting G-B-4 8.38 22.1 21.4 3.79 40.5 510 Splitting G-B-5 13.6 21.0 19.4 3.20 36.7 500 Splitting G-B-6 12.0 20.5 22.8 3.80 43.2 510 Splitting Ave. 13.5 22.1 20.6 3.41 39.0 510 CoV 0.28 0.08 0.06 0.11 0.06 0.04 a Moisture content at tests12"1%. Table 4 a Summarised results for bolted joints in microlam Sample Initial Stiffness Maximum Slip at Embedment Wood Failure number stiffness K K load F maximum strength f density mode i s max h 2 3 Ž . Ž . Ž. Ž . Ž . Ž . kNrmm kNrmm kN load mm Nrmm kgrm M-B-1 19.2 20.7 28.0 3.49 54.2 650 Splitting M-B-2 13.8 20.0 30.5 3.24 59.0 640 Splitting M-B-3 18.4 31.8 26.8 2.08 51.9 650 Splitting M-B-4 9.01 18.6 28.0 4.14 54.3 640 Splitting M-B-5 30.4 36.7 24.8 1.35 48.0 630 Splitting M-B-6 26.1 38.1 25.1 1.35 49.9 690 Splitting M-B-7 24.1 40.8 29.0 1.62 57.5 700 Shear M-B-8 25.6 66.1 26.0 1.96 51.6 660 Splitting Ave. 18.2 25.5 27.6 2.86 53.5 640 Cov 0.44 0.32 0.07 0.39 0.07 0.01 a Moisture content at test 7"1%
nd Building Materials14(200)407-417 nitial (kN/mm) 207 0 474 76 品 33 74 .09 "Moisture content at test 8+1%. although the majority behaved in a similar manner confirmed.An actual embedment strength-density re lationship is not offered but is likely to be similar to 9c).Although this behaviour did occur in some of the rial is more homogeneous.This is misleading however. since the natural wood samples used in this experiment more so than would be the 7.Discussion o in t of strength and stifmess but showed a lower joint sli The positioning of a bolted connection on the glue at maximum load than solid timber and glulam.The effect on the Parallam samples were in joi structur com neir re re gene and ga materials would suggest a higher strength and this is were present in the bearing zone of the joint and this is AA -024 40 30 550 500 66 700 Density (kg/m) Fig.7.Embedment strength plotted against wood density
T.J. Da¨is, P.A. Claisse rConstruction and Building Materials 14 2000 407 ( ) ]417 415 Table 5 a Summarised results for bolted joints in parallam Sample Initial Stiffness Maximum Slip at Embedment Wood Failure number stiffness K K load F maximum strength f density mode i s max h 2 3 Ž . Ž . Ž. Ž . Ž . Ž . kNrmm kNrmm kN load mm Nrmm kgrm P-B-1 5.28 12.3 21.7 5.09 41 660 Splitting P-B-2 16.9 21.0 38.7 4.29 73 780 Splitting P-B-3 19.2 38.0 21.1 1.84 41 680 Splitting P-B-4 20.0 20.9 36.6 4.64 71 760 Splitting P-B-5 13.0 18.6 32.8 3.77 62 800 Splitting P-B-6 15.7 20.8 42.9 5.26 81 760 Combined Ave. 14.9 22.2 30.2 3.93 58 740 Cov 0.40 0.43 0.28 0.32 0.27 0.09 a Moisture content at test 8"1%. although the majority behaved in a similar manner to the glulam samples Fig. 9b . A number of the Ž . Microlam samples were noticeably brittle in their response-failing suddenly at the maximum load Fig. Ž 9c . Although this behaviour did occur in some of the . Parallam samples the majority exhibited similar loadslip responses Fig. 9d to solid wood and glulam. Ž . 7. Discussion The positioning of a bolted connection on the glueline appears to have no detrimental effect on the performance of the joint. The structural timber composites give an increased embedment strength relative to solid timber and glulam. The higher density of these materials would suggest a higher strength and this is confirmed. An actual embedment strength]density relationship is not offered but is likely to be similar to that for bolted joints in plywood. The composites showed a wider variation in response than the solid timber samples, especially given that the actual material is more homogeneous. This is misleading however, since the natural wood samples used in this experiment were carefully selected } more so than would be the case in stress-graded timber for construction. Microlam gave the best overall performance in terms of strength and stiffness but showed a lower joint slip at maximum load than solid timber and glulam. The Parallam samples were in many ways disappointing. Their results were generally the most variable and gave lower joint stiffness values than their density would suggest. Visual inspection showed that internal voids were present in the bearing zone of the joint and this is Fig. 7. Embedment strength plotted against wood density
416 TJ.Dais,PA.Claisse/Constnuction and Building Materials 14(00)47-417 the likely explanation for the performance of the sam- ples during testing.This is at odds with Parallam's eme als th ed ir timber connections to EC5 is largely dependent on 8.Conclusions E6sasigaRme Acknowledgements positioning of a bolted connection on the glueline of a
416 T.J. Da¨is, P.A. Claisse rConstruction and Building Materials 14 2000 407 ( ) ]417 Fig. 8. Local crushing and longitudinal splitting failure of a Microlam sample. the likely explanation for the performance of the samples during testing. This is at odds with Parallam’s performance in flexure where it offers better strength and stiffness than Microlam. 8. Conclusions Bolted connections in structural timber composites appear to give strength and stiffness properties at least comparable to a solid timber of similar density. The positioning of a bolted connection on the glueline of a glulam member does not detrimentally affect the joint’s performance. The internal voids contained within parallel strand lumber appear to produce a lower, more variable, performance in bolted connections than their performance in flexure would suggest. This behaviour needs further investigation since the embedment strength of wood materials that is used in the design of timber connections to EC5 is largely dependent on wood density. The presence of these internal voids, not visible during fabrication, would not be taken into account under current EC5 design rules. Acknowledgements The authors gratefully acknowledge the support of the EPSRC and its staff for this project