Moment-Resisting Connections In Laminated Veneer Lumber (LVL) Frames Andy van Houtte Product Engineer-LVL Nelson Pine Industries Nelson, NZ Andy Buchanan Professor of Civil Engineering Peter Moss Associate Professor Andy van Houtte completed a Master of Engineering (Civil) degree last year. He now works as a product engineer for Nelson Pine Industries who are manufacturers of LVL in Nelson, New Zealand. University of Canterbury Christchurch, New Zealand Summary This paper describes an experimental investigation into moment-resisting joints in laminated veneer lumber (LVL) beams formed with steel bars bonded into the LVL beams with epoxy resin. The objective of this research was to provide information about the bending and shear strength of such connections. Nine portal frame knee joints were tested with a variety of connection details and LVL sizes in order to determine the behaviour and failure modes associated with each when subjected to cyclic loadings. The testing showed that very stiff joints can be fabricated using epoxied steel rods but there is difficulty developing the full strength of the LVL, especially in the opening mode. This technology requires further refinement before being recommended for widespread use, but the results to date show potential and research is continuing. Keywords: Connections, laminated veneer lumber, LVL, moment-resisting connections, epoxy bonded steel rods, failure, portal frames 1. Introduction Over the last 20 years, there has been an increasing interest in the use of steel threaded rod epoxied into glulam timber members in order to fabricate moment-resisting joints. The use of epoxied steel rod technology for timber connections began in Denmark about 1980. Since then many studies have been carried out in Europe to investigate the reliability of this jointing [1]. This connection method has been studied at the University of Canterbury with the aim of testing the European concept in New Zealand conditions. Townsend [2] and Deng [3] tested the tensile strength of single steel rods bonded into glulam, parallel to the grain, using epoxy resins. Korin et al [4] extended this work to test the tensile strength of single and multiple rods. In a study investigating the application of these joints in seismic frames, Fairweather [5] carried out tests on several moment-resisting connections and obtained hysteresis loops that showed good ductility and excellent strength, provided that the connected ends of glulam beams were cut square. Buchanan and Fairweather [6] describe the development of a moment-resisting beam-column joint that uses epoxied rods bolted to a steel hub.one common feature of this research on tensile and bending capacity of glulam beams with glued-in rods has been occasional shear splitting failure of the glulam. A solution to this problem is to glue small diameter transverse steel rods into the glulam across the potential lines of failure. Large joints of this type have been used in several large scale applications including buildings for the Sydney 2000 Olympic Games. Another potential problem with glulam has been the possibility of beam fracture at a cross section near the ends of the glued-in rods. The objective of this study was to investigate the suitability of epoxied threaded steel rod connections to provide moment-resisting knee joints in laminated veneer lumber (LVL) portal frames, similar to those already developed for glulam. To achieve this, pull-out tests were carried out on a large number of small specimens, and a small series of cyclic moment tests were carried out on full-scale prototypes. The differences between LVL and glulam were expected to be the
much higher axial tensile strength of LVL parallel to the grain, but unknown effects of the different properties perpendicular to the grain due to the gluing together of many thin veneers. The LVL was all manufactured from New Zealand grown Radiata pine by Carter Holt Harvey Ltd who provided the test specimens. 2. Small scale pull-out tests 2.1 Parallel to grain tests A series of small scale pull-out tests were carried out in order to compare the performance of epoxied connections in LVL with those previously carried out using glulam [2-6]. These tests showed that the tensile strength of threaded rods epoxied into LVL members parallel to the grain is excellent, similar to the performance of glulam members. It was found desirable to use self-tapping screws to prevent premature splitting of the LVL in the radial direction of the veneers, as shown in Figure 1. These screws increased the strength by up to 25% by reducing the possibility of splitting in the veneers parallel to the glue-lines. Splitting in the tangential direction (across the glue-lines) was not often observed in the tests. The LVL specimens never failed in axial tension at the ends of the rods as was sometimes observed with glulam, because of LVL s greater strength than glulam in tension parallel to the grain. Self-tapping stainless steel screws Screw has to be threaded through this region to prevent splitting Figure 1 Cross-section showing self-tapping screws near the end of the specimen For all pullout tests parallel to the grain, the failure surface was observed to be mainly in the wood itself, with almost no failure at the wood-to-epoxy interface. Based on the pull-out tests [7], the strength of the rods epoxied into LVL is given by Equation 1 being the line of best fit to the experimental results. F 0.85 1.885 x 10 4 E 2 L fs 15 (1) where F = pull-out force (kn), E = hole diameter (mm), L = embedment length (mm), and f s = LVL characteristic shear stress (MPa) (4.6 Mpa for the LVL used in these tests based on the manufacturer s literature). Equation 1 gives similar results to Deng s formula [3] for glulam provided the latter is scaled by a 0.85 reduction factor. Using this formula, an embedment depth of 300mm will develop the tensile strength of a 20mm diameter threaded steel bar with yield strength of 650MPa.
2.2 Perpendicular to grain tests A number of tests were also carried out with the rods glued in perpendicular to the grain of the LVL. Again, similar results were obtained to those previously tested in glulam. The pullout strength was found to be approximately 65% of that in tension parallel to the grain. For all pullout tests perpendicular to the grain, the failure surface was observed to be mainly at the wood-to-epoxy interface, because the liquid epoxy did not appear to have been able to penetrate the end-grain of the wood in the sides of the drilled holes. This type of failure raises doubts about the long-term performance of such connections, especially if there is fluctuating moisture content in the wood. For the knee joint testing, this observation led to the decision to provide nuts with large washers at the ends of the rods which were drilled into holes perpendicular to the grain. This was done so that there would be some protection against total collapse if there was any pullout failure after many years of loading in-situ. 3. Knee joint tests Nine portal frame knee joints were tested, including several different threaded rod connections and LVL sizes in order to understand the behaviour and failure modes associated with each. 3.1 Joint specimens tested The basic knee joint specimen comprised a column member with the beam sitting on top of it and held in place by an inner and outer set of 2 or 3 threaded rods embedded in the column and extending into the beam (see Figure 2). A summary of the joint specimens tested is given with the results summary in Table 1. The rods embedded into the column terminated in threaded couplers set flush with the top of the column. The rods through the beam were then screwed into the coupler before being epoxied Figure 2 Typical joint details into the beam. Unless otherwise noted in Table 1, the steel rods went right through the full depth of the beam and were epoxied in place with steel washers and nuts on the protruding ends (as an additional safety measure for long-term loading). In some of the tests the inner rods did not extend the full depth of the beam in order to enhance the flexural strength in the closing mode. The required depth of embedment into the column was derived from the small scale pull-out tests. Stainless steel self-tapping screws were provided through the thickness of the LVL near the couplers to prevent radial splitting. 3.2 Test procedure The specimens were tested in an upright position and loaded using a double acting hydraulic ram. They were loaded cyclically using three cycles of loading at load levels of 25%, 50%, 75%, 100% and 125% of the design moment, or until failure. The design moment was calculated from the pullout strength of the rods in the column, which was fabricated to be stronger than the design moment for the given portal frame size. The pull-out loads were calculated from Equation (1) derived from the pull-out tests [7]. If the knee joint failed under the opening moment, as it did in several cases, the cycling continued under a closing moment only. 4. Results A summary of the test results is given in Table 1 where the design values are shown (in italics) along with the separate results for the opening and closing parts of the cyclic loading as well as a brief description of the expected and actual failure modes.
4.1 Typical failure modes Typical failure modes are illustrated in Figure 3. In the opening mode, the typical tensile failure in the beam at the front rods is shown in Figure 3a. One joint failed by pullout of the front rods from the column as seen in Figure 3b, but this was because insufficient embedment depth was provided. In the closing mode, none of the specimens were tested to their capacity in an undamaged state because the recorded failure was after an earlier partial failure in the opening mode. In these cases the failure was often in tension in the beam near the front rods. Of the three tests which failed by pull-out of the back rods as shown in Figure 3c, two exceeded the design strength and the third had insufficient embedment depth. In the cases where the front rods only went part way through the beam, the beams first failed in the opening mode as in Figure 3a, then on applying a closing moment the beams split as shown in Figure 3d. (a) Tensile failure in beam at front rods (b) Pullout of front rods (c) Pullout of back rods (d) Split through middle of beam Figure 3 Typical failure modes 5. Discussion In several of the tests the design strength of the joint, based on a tensile failure of the beam at the front rods, was equalled or exceeded in the closing mode but the joints generally failed well short of this in the opening mode. The significance of this for design depends on whether the opening or closing mode is likely to govern the design of a building using this type of connection. Buildings in high snow load areas are likely to be governed by the closing mode, whereas wind loading is likely to govern the opening mode.
Test No. Test description: size of LVL; embedment depth into top of column, bar no. and sizes 89x800; large washer plate, no epoxy in beam 300 embedment, Table 1 Comparison of knee joint stresses Load ratio Failure at Moment failure (knm) Max Wood Stress (MPa) Pullout force per rod (kn) Steel stress (MPa) Failure mode 1 Design 243 35.0 187 830 Fracture in beam at front rods Opening 0.60 146 21.1 113 500 Fracture in beam at front rods 4xM20 rods Closing 0.63 152 22.0 117 521 LVL Compression failure 2 89x800, insitu epoxy Design 243 35.0 187 830 Fracture in beam at front rods repair of Test 1. Opening Specimen previously failed 300 embedment, 4xM20 Closing 0.79 193 27.8 148 658 Fracture in beam at front rods 3 89x800; Design 243 35.0 187 830 Fracture in beam at front rods front dowels epoxied half Opening 0.50 122 17.6 94 418 Fracture in beam at front rods depth of beam 300 embedment, 4xM20 Closing 0.95 230 33.2 177 787 Splitting of beam at back rods 4 89x800; Design 243 35.0 187 830 Fracture in beam at front rods front dowels epoxied 2/3 Opening 0.49 120 17.3 92 410 Fracture in beam at front rods depth of beam 300 embedment, 4xM20 Closing 1.04 252 36.3 194 860 Splitting of beam at back rods 5 63x600; Design 108 42.0 109 760 Fracture in beam at front rods 250 embedment Opening 0.75 81 31.3 81 566 Fracture in beam at front rods 4xM16 rods Closing 1.11 120 46.4 121 840 Pullout of back rods 6 105x1000; Design 448 35.0 264 813 Fracture in beam at front rods 400 embedment, Opening 0.67 300 23.4 176 544 Pullout of front rods 4xM24 rods Closing 0.89 400 31.3 235 726 Pullout of back rods 7 63x600; Design 108 42.0 109 760 Fracture in beam at front rods 300embedment, Opening 0.71 77 29.9 78 541 Fracture in beam at front rods 4xM16 rods Closing 1.20 130 50.5 132 913 Pullout of back rods 8 105x1000; Design 448 35.0 187 576 Fracture in beam at front rods 450 embedment, Opening 0.83 372 29.1 155 479 Fracture in beam at front rods 6xM24 rods Closing 1.00 449 35.1 187 577 Testing reached capacity of jack no failure of joint 9 63x600; Design 108 42.0 79 545 Fracture in beam at front rods 300 embedment, 6xM16 Opening 1.09 118 45.7 85 593 Fracture in beam at front rods (front rods taped) Closing 0.98 106 41.2 77 535 Fracture in beam at front rods The first test, where no epoxy was used to bond the rods into the beam, the joint showed considerable deformation and was not able to reach the design strength. When the joint was repaired in-situ by epoxy injection, the beam did show some improvement in performance with the much stiffer joint, though the prior failure in the closing mode prevented testing in the opening mode. In the tests where the front rods did not extend all the way through the beam, high flexural strength was developed in the closing mode as usual, but lower flexural strength in the opening mode because the partially embedded front rods caused a longitudinal split in the beam near the ends of the rods. With the smaller member size in the fifth test, it was found that greater embedment than predicted from the small scale tests was required in order to prevent pull-out of the rods. The embedment in the column was increased from 250 mm to 300 mm for the seventh test. The closing strength of the joint was much greater than the LVL beam strength for both tests. A typical opening moment failure occurred in both tests wherein a tensile failure occurred in the beam at the front rods. In the case of the sixth test, a larger size member was used along with larger diameter rods. However, the embedment of the rods in the column was insufficient as the rods pulled out in both
the opening and closing modes. This was rectified in the eighth test by increasing the embedment length and using three rods instead of two. This prevented pull-out of the rods and the rods did not reach their yield strength. Since many of the beams developed a tensile fracture at the front rods during loading in the opening mode, it was concluded that there was possibly a combined stress problem at the bottom of the beam where the fracture was initiated. This would be caused by the combined effect of the longitudinal flexural stress and an orthogonal tensile stress produced by the tension in the glued rods. A possible simple solution to this combined stress problem was to provide plastic tape on the front bars for the bottom quarter of the beam, to transfer the high bond stresses away from the critical region. This was tested successfully in Test 9, but the results of only one test are not sufficient to draw conclusions on this proposal. 6. Summary And Conclusions On the basis of the tests described in this paper, the following conclusions can be drawn: Large LVL portal frames can be constructed with knee joints consisting of steel rods epoxied right through the beam member and into the end grain at the top of the column. The knee joint with no epoxy in the beam section exhibited excessive deformations. The epoxied knee joints are not able to develop their full flexural strength in the opening mode because of premature fracture of the beam at the cross section where the front rods are located. In the closing mode, the knee joints are able to develop their full flexural strength, provided that the area of the hole is subtracted when calculating the cross section properties at the critical section. More testing is required before recommending this type of joint for commercial design or construction. Future research will investigate alternative forms of knee joint, including those which only use epoxied rods glued in parallel to the grain of the wood. 7. References [1] Riberholt, H., Glued bolts in Glulam, Dept of Structural Engineering, Technical University of Denmark. Series R, No 228, 1988. [2] Townsend, P.K., Steel dowels epoxy bonded in glue laminated timber, Research Report 90/11, Department of Civil Engineering, University of Canterbury, Christchurch, New Zealand, 1990. [3] Deng, J.X., Strength of epoxy bonded steel connections in glue laminated timber, Research Report 97/4, Department of Civil Engineering, University of Canterbury, Christchurch, New Zealand, 1997. [4] Korin, U, Buchanan, A.H. and Moss P.J., Effect of bar arrangement on the tensile strength of epoxied end bolts in glulam, Proc. Pacific Timber Engineering Conf., Rotorua, New Zealand. pp 217-224, 1999. [5] Fairweather, R.H., Beam column connections for multi-story timber buildings, Research Report 92/5, Department of Civil Engineering, University of Canterbury, Christchurch, New Zealand, 1992. [6] Buchanan, A.H. and Fairweather, R.H., Seismic Design of Glulam Frame Structures, Bulletin of the New Zealand National Society for Earthquake Engineering, Vol.26, No.4, 415-436, 1993. [7] Van Houtte, A., Innovative connections in laminated veneer lumber using epoxied steel rods, Master of Engineering Thesis, Department of Civil Engineering, University of Canterbury, Christchurch, New Zealand, 2003.