STRUCTURAL CHARACTERISTICS OF BEAM-COLUMN CONNECTIONS USING COMPRESSED WOOD DOWELS AND PLATES Zhongwei Guan*1, Kohei Komatsu2, Kiho Jung2 and Akihisa Kitamori2
ABSTRACT: Steel dowels and plates are widely used in modern timber joints, which are difficult to be recycled. Also due to large difference on stiffness between the steel and the timber, there is a poor compatibility in the joint that produces less integrity. Therefore, a desired joint should be made with non-metallic fasteners, ideally timber fasteners. In this paper, 3-D nonlinear finite element models have been initially developed to simulate structural behaviour of beam-column connections subjected to racking loading conditions. All models developed are validated against experimental results. Then using validated models, characteristics of the beam-column connections are thoroughly investigated in terms of load carrying capacities of individual dowels made of compressed wood. In the numerical models, timber and compressed wood are modelled as orthotropic linear elastic materials in tension, and as elastoplastic materials in compression in the embedding areas. Various contact conditions within the joints are modelled. Moment-rotation relationships of the joints are simulated with reasonably good correlation to the corresponding experimental results. Based on the structural characteristics obtained, recommendations are given on dowel patterns and geometrical conditions of the constituent members. KEYWORDS: compressed wood, dowel, joint, finite element, non-metallic fastener timber.
1 INTRODUCTION 123 Although steel dowels and plates have been used in modern timber joint systems for a few decades, some problems are still remained, i.e. it is difficult to recycle them. More importantly, due to large difference on stiffness between the steel and the timber, there is a poor compatibility in the joint that leads less integrity. Therefore, a desired joint should be made with nonmetallic fasteners, ideally timber fasteners. The idea to use wood fasteners in construction is not new. There are still some outstanding traditional temples and shires using wood fasteners in China and Japan which were built a few hundred even a thousand years ago. However, the way to produce such traditional wood fasteners is complex, which needs high skills and is not mass productive. Also, hardwood used in the traditional wood fasteners has high stress relaxation feature. This 1
Zhongwei Guan, Department of Engineering, Brodie Tower, University of Liverpool, Liverpool L69 3GQ, UK. Email:
[email protected] 2 Kohei Komatsu, Research Institute for Sustainable Humanosphere, Kyoto University, Uji, Kyoto, Japan. Email:
[email protected] 2 Kiho Jung, Research Institute for Sustainable Humanosphere, Kyoto University, Uji, Kyoto, Japan. Email:jungkiho@ rish.kyoto-u.ac.jp 2 Akihisa Kitamori, Research Institute for Sustainable Humanosphere, Kyoto University, Uji, Kyoto, Japan. Email:
[email protected]
leads joint loose so that joints need to be fastened on a regular basis. The use of compressed wood fasteners could overcome above problems, as compressed wood is an engineered product which has much better mechanical properties by controlling density and moisture content through manufacturing processes. There are major advantages to use compressed wood fasteners against steel fasteners, such as reduction of CO2 emission, good environmental impact, better compatibility, tight fitting, better recycle ability, etc. Although traditionally produced compressed wood could offer higher density and better mechanical properties, however such enhancement was not stable and of shortterm duration. An unfortunate reason is that untreated compressed solid wood and veneer tend to undergo irreversible “springback” or recovery from compression when exposed to moisture. This prevented it to be used to produce structural members. To eliminate springback wood should be pressed in conditions that cause sufficient flow of the lignin. There have been research developments since 1960s to tackle the problem. Series of breakthrough research outcomes have been produced since 1990s. A compressed wood product without resin treatment is Staypak [1], which is produced by compressing wood at a moisture content equal to or below the service one. However, due to the thermoplastic nature of the lignin, also because the moisture content of the wood is only slightly less after
compression than prior to pressing, there is considerable springback on the product if it is removed from pressing while still hot [2]. This prevented Staypak from industrial uses. Therefore, stabilisation of compressed wood becomes a key to open a door for its broad industrial applications, particularly structural uses. Compressed wood can now be produced with a density up to 3 times of its original one and with desired stiffness and strength. Due to greatly enhanced mechanical properties of compressed wood with necessary stabilised dimensions, the newly developed compressed wood increasingly attracts researchers to find its structural uses. For the past a few years, there have been many studies on its mechanical properties, but limited research on its structural uses as fasteners for connecting soft timber components. Zhou et.al [3] investigated about bending creep behaviour of hot-press wood under cyclic moisture condition and found that the thickness swelling increased with moisture cycle, which led to increase in the dimension of hot-press specimen by the end of cyclic moisture sorption. Studies were also carried out on structural timber and glulam in compression perpendicular to grain and the corresponding stress states [4]. Heger et al. [5] studied about mechanical and durability performance of thermohydro-mechanically densified wood and found a diminution of the mechanical properties of THM treated wood at temperatures higher than 180°C, which might be the maximum temperature practical for processing. Kobujima et al [6] investigated the bending properties of compressed Japanese cedar (Cryptomeria japonica D. Don), the specimens being compressed in the radial direction with ratios (the deformation to the initial thickness) of 33% and 67%. Adlam [7] also studied effects of relatively low compression ratios of 13% and 22% on the mean MOE and MOR of densified wood. Yoshihara and Tsunetamtsu, [8] examined the bending and shear properties of compressed wood and showed that Young‟s modulus increased with increasing compression ratio. They also used tension tests to investigate the elastic properties of Sitka spruce (Picea sitchensis Carr.) compressed wood with various compression ratios [9]. Jung et al. [10] studied timber joint systems using compressed wood fastener. The result shown that the joint with compressed wood dowels and plates have enhanced mechanical properties such as pull out strength and rotational performance. Hassel et al. [11] undertook studies on the performance of a wooden block shear wall which utilizes compressed wood as a connecting element in place of the traditional metal connector. After absorbing moisture, the compressed connector recovered partly its radial dimension and filled the gaps with adjacent block. Jung et al. [12] carried out research on applications of compressed wood (CW) made of Japanese cedar, as a substitute for high density hardwood, to make shear dowels. CW with its annual ring radial to loading direction (0 o) had a unique double shear performance characteristic, and showed good properties as a dowel material with its strength and rich
ductility. However, CW with its annual ring tangential to loading direction (90o) and maple exhibited brittle failure. When the density of base member increased, its stiffness, yield load, and maximum load exhibited proportional improvement with different inclination. To date, there is hardly any numerical models being developed to assist optimise structural performance of timber connections using compressed wood fasteners. Testing the proposed portal frame joints is very time consuming, on top of material and manpower costs. When conducting experimentally-based research, tests need to cover as many scenarios as possible, such as stacking configurations, materials, geometries, loading and boundary conditions, etc. Therefore, the optimisation process is likely to be very expensive and hugely time consuming. In contrast, the development of computer models using finite element (FE) analysis is a relatively quick and inexpensive process, especially in cases where there is access to supercomputer facilities. In such circumstances, only a limited number of materials measurement tests and structural tests are required for validation purposes. Once computer models are verified against typical tests covering extreme cases and possible an intermediate case, systematicallydesigned parametric studies can be undertaken using validated numerical models [13-16]. In this study, 3-D nonlinear finite element models have been initially developed to simulate structural behaviour of beam-column connections subjected to racking loading conditions. All models developed are validated against experimental results. Then using validated models, characteristics of the beam-column connections are thoroughly investigated in terms of load carrying capacities of individual dowels made of compressed wood. In the numerical models, timber and compressed wood are modelled as orthotropic linear elastic materials in tension, and as elasto-plastic materials in compression in the embedding areas. Strain hardening of the compressed wood is taken into account in the modelling. Complex contact conditions of the joints are simulated, which cover those between the compressed wood dowel and the compressed wood plate, the dowel and the timber (column and beam), the timber and the plate, the timber beam and the timber column. Different contact algorithms are used to simulate small slide, finite slide and possible separation between contact pairs. Structural behaviour of the joints in terms of the moment-rotation relationship is simulated with reasonably good correlation with the corresponding experimental results. Based on the structural characteristics obtained, recommendations are given on dowel patterns and geometrical conditions of the constituent members.
2 EXPERIMENTAL WORK ON THE BEAM-COLUMN CONNECTIONS In this research, E60-grade Japanese cedar glulam was used for column and beam, with cross-sectional dimensions of 120mm×120mm and 120mm×240mm,
respectively. Dowel was made by compressing Japanese cedar (Crytomeria japonica D. DON) in the radial direction until 30% of its original thickness was reached at a temperature of 130℃ for duration of 30minutes to obtain a density of about 1000 kg/m3. No fixation treatments, such as steaming, chemical agent or resin, were applied. All boards selected for compressing had flat annual growth rings and were without knots, split, and pith. The initial moisture content (MC) was approximately 12% prior to the compression process. For the fabrication of the dowels, the initial dimensions of wood pieces were 15mm×15mm, which were then processed into round shape with final diameter of 12mm. The processes to make the compressed wood plates were almost the same as those for the dowel, with only difference on the compression rate. The dimensions of the plate were 80mm×580mm×14mm for the column-to-beam joints. There are total 14 compressed wood dowels of 12 mm in diameter used to link beam/CW plates and column/CW plates. Figure 1 shows the experimental apparatus of the rotation test for the column-beam joint. A quadratic-link steel frame system was used to load the joint with a moment, which was connected by a steel pin with a span of 1000 mm span. Each specimen was set up on the frame and was jointed with a steel pin (Ø22 mm), as shown in Figure 1. The rotational deformation of the joint was applied by this steel frame, which was controlled by a hydraulic actuator. The loading schedule was determined through a step displacement with angles of the steel frame of 1/300, 1/200, 1/150, 1/100, 1/75, 1/50, 1/30, and 1/15 rad. At each step, three loading cycles were applied. The relative displacements between the plate and each member were measured by displacement transducers for estimating accurate rotational angles against the corresponding applied load measured by a 50-kN load cell. 120
100
500 #1
120
Load Cell 50kN
500
#6 #7 #2
240
1000
#4
#3
100
500
#5
120
100
500
500
Figure 1: Apparatus for rotational test on CPD columndouble-beam joint
Figure 2 shows typical failure mode of the joint, i.e. splitting and embedding on the top and the bottom boundary of the beam – column interface area. Therefore, the numerical modelling developed needs to simulate such failure features, on top the loaddisplacement relationship.
Figure 2: A typical failure mode of the column – double beam joint
3 DEVELOPMENT OF FINITE ELEMENT MODELLING There are two different timber materials in the joint, i.e. soft wood for the column and the beam and compressed wood for the dowels and plates. For both materials in tension, orthotropic elasticity would be an appropriate constitutive relationship. Equation (1) shows such the relationship used for all members in the tension zones. 11 1 E1 E 22 12 1 33 13 E1 12 0 13 0 23 0
21 E2
31 E3
1 E2
32 E3
0
23 E2
1 E3
0
0
11 0 0 22 0 0 33 0 0 12 1 G13 0 13 1 G 23 23
0
0
0
1 G12
0
0
0
0
0
0
0
(1) However, for timber under compression, especially under high contact stresses, elasto-perfect plasticity would be a proper model [14, 17]. Commercial finite element code Abaqus offers a good tool to tackle the prescribed problems [18]. Material properties used in the modelling were shown in Table 1. In order to simulate the behaviour of compressed wood, elasto-plasticity with strain hardening was used, which was corresponding to the test results. Equation (2) shows elasto-plastic relationship used in the modelling. d{ } [D]ep d{ }
(2) where [ D]ep is elasto-plastic matrix, which is dependent on the elastic matrix , the yield function and the hardening function (for perfect plasticity hardening is zero). The corresponding total stresses and plastic strains are shown as follows.
Total stress (Mpa): 22.0, 25.6, 29.9, 34.3, 39.0, 44.2, 49.1 Plastic strain (mm/mm):0.0, 0.0127, 0.0153, 0.0232, 0.0262, 0.0353, 0.0445 Table 1: Material properties
Component Beam&column CW dowel CW plate GLR 560 590 530
EL 10100 25500 22600
GTR 35 139 111
LT 0.020 0.020 0.020
ER 160 1170 918
LR 0.408 0.408 0.408
ET 390 2050 2650
GLT 470 1860 1480
RL 0.030 0.030 0.030
* N / mm2 for all modulus Embedding strength of 6 MPa perpendicular to the grain of compressed wood plate was implemented to the specific compressive zones. Mesh generation is shown in Figure 3, with loading and boundary conditions. However, in order to model racking behaviour of the joint accompanied by large displacement, four rigid beams were attached to two beams through pin joints. This is also shown in Figure 3.
to overcome numerical singularity problems. The most challenging tasks are to deal with various contact surface pairs, some with finite sliding and some with small sliding. There are total 64 contact pairs in the modelling, which are comprised from the following interfaces: The compressed wood dowel – the timber, The compressed wood dowel – the compressed wood plate, The beam – the column, The compressed wood plate – the beam, The compressed wood plate – the column Due to various contact features in the joint, in some contact pairs, such as those between the compressed wood plate and the column, the beam and the column, finite slippage was allowed. In contact pairs formed between the dowel and the plate, the dowel and the timber (beam and column), only small slide was necessary.
4 RESULTS AND DISCUSSION Here a series of numerical modelling results are presented to assist evaluating structural behaviour of the double beam – column joints. First, the predicted moment-rotation relationship is plotted against the corresponding experimental results, which is shown in Figure 4. Very good correlation is obtained, in terms of the initial stiffness, the peak load and overall relationship. The deformation mode is also shown in the same figure that demonstrates large rotational deformations due to the horizontal racking.
pinned pinned pinned
pinned
Figure 3: Mesh generation of the joint
Since rigidity of the joint was underestimated by those pin joints in contrast with the test setting shown in Figure 1, some low stiffness springs (1 N/m) were connected to the column and compressed wood dowels
Figure 4: Moment-rotation relationships and deformation mode
Figure 5 shows the contour plot of the maximum principle stress (Pa) and a failed specimen. It can be seen that the predicted failure areas are coincident with the failure areas obtained from experimental work, i.e. the tensile splitting failure in the beam – column interface areas opposite to the high embedding areas. The maximum principle stress along the longitudinal direction of the column has reached 36 MPa, which is critical for Japanese cedar. Also large embedding
Figure 6: Longitudinal shear stress distribution on the column
Figure 7 shows the minimum principle stress distributions on the CW plates , which displays high contact stress regions. The maximum stress value is -63 MPa which is located adjacent to the high embedding deformation areas between the column and the beam. Deformed shapes of the dowels 7 and 8 are also shown in the figure to view the deformation modes of the mostly deformed dowels in the joint.
Figure 5: Comparison of the predicted failure mode and experimental failure mode
deformations are simulated reasonably well. In addition, the opening between the beam and the column is reproduced. The longitudinal shear stress can be critical. Figure 6 shows such stress distributions on the central area of the column. The contour plot demonstrates the maximum shear stress of about 9 MPa, which is on the critical value. If the area with such critical value is big enough there may be shear failure occurred.
Figure 7: The minimum principle stress distributions on the CW plates and the dowels 7 and 8
The total contact force on the individual dowels is shown in Figure 8, which combines both the normal and shear interactions. It is clearly seen that the dowel 7 and dowel 8 bear the largest contact forces, i.e. 27.6 and 26.7 kN respectively, whilst dowels 1, 3, 5, 10, 12 and 14 also carry high contact forces ranging from 19.8 to 23.4 kN. However, the contact forces on dowels 2, 4, 6, 9, 11 and 13 are much lower, from 1.0 to 5.7 kN, but they would pick up much more contact forces when racking the joint
in an opposite direction. Anyhow, contact forces on the dowels 7 and 8 remain the highest. Therefore, both dowels need to be reinforced by either increasing the dowel size or number of dowels, subjected to dimension increasing on the column. Also, it is interesting to see the sequence order of the dowel getting fully engaged. The dowels 7 and 8 pickup contact loads at 10% of the loading (before the loading is taken by the beam-column interface and the supports), then the dowels 1, 3, 5, 10, 12 and 14 get engaged at 25% of the loading, finally rest of the dowels start to contribute load carrying capacity at 50% of the loading.
beam and the CW plate. If CW dowels with much lower moisture content than the ambient MC are inserted to the joint, much the higher tight fitting will be anticipated due to moisture-dependent swelling of the dowels. Therefore more contribution from the contact shear force will be expected.
Figure 9: Contact shear forces on individual dowels
Figure 10 shows the total contact force between the beam and the column. There are 5 contact pairs on each side of the column. The contact forces on the right of the column are asymmetric to their counterparts on the left
2 4 6
8
10 12 14
1 3 5
7
9 11 13
Figure 8: Contact forces on individual dowels
Figure 9 shows the contact shear forces carried by individual dowels. As expected, the dowels 7 and 8 carry the highest contact shear forces, about 4 kN, whilst dowels 1, 3, 5, 10, 12 and 14 carry such forces in the range from 1.5 to 2.1 kN. Again, when racking the joint in an opposite direction the contact shear forces will be swapped between the group of dowels 1, 3, 5, 10, 12, 14 and the group of dowels 2, 4, 6, 9, 11, 13. By subtracting the contact shear forces from the total contact forces (see Fig. 8), it can be seen that the normal contact forces play much more important roles. However, this is related to how tight fitting between the dowel and the column, the
B-C-L1 B-C-L2 B-C-L3 B-C-L4 B-C-L5
B-C-R1 B-C-R2 B-C-R3 B-C-R4 B-C-R5
Figure 10: Total contact forces on the beam – column interface areas
of the column in terms of contact force locations due to rotation of the joint. Also from the figure, it can be seen that the total contact forces on the beam – column interface pick up at much earlier stage in comparison to such contact forces on the dowels. It is understandable as the beam gets into contact with the column quickly due to racking force before the dowels get fully engaged. The ratio of the contact area to the total contact surface area for the dowels 1, 2 and 7 are shown in Figure 11. The dowel 7 has the highest contact region as it bears the highest contact force (see Fig. 7). Variation of the contact areas on the beam and the CW plate is related to their geometric conditions and material properties (mainly stiffness). In general, contact force is dependent upon the corresponding contact area in the joint, which is reflected by Figs. 8 and 11.
Figure 11: Changing in contact area versus the loading percentage
5 CONCLUSIONS Non-linear finite element models have been developed to simulate structural behaviour of double beam – column joints. Orthotropic material properties of all members were implemented into the models. Both elasto-perfect plasticity and elasto-plasticity with non-linear hardening were used to model behaviour of soft wood and compressed wood, respectively. Very good correlation between the test results and the simulation in momentrotation relationship was obtained. The predicted failure mode of the joint is also correlated well with the failed specimen. The numerical models also produced information on the total contact forces, the shear contact force and the contact area of various contact pairs, which are useful to analyse detailed characteristics of the joints. The models developed may be used for further parametric studies to optimise the joint systems.
ACKNOWLEDGEMENT Authors sincerely thank the Research Institute for Sustainable Humanosphere of Kyoto University to support publication of the partial reserach output from a joint research project.
REFERENCES [1]. Soborg RM, Millett MA and Stamm AJ. Heatstabilized compressed wood (Staypak). FPL, Report No. 1580 (revised), 1962. [2]. Kollmann FP, Kuenzi EW and Stamm AJ. Principles of wood science and technology. Vol. II Wood based materials. Springer-Verlag New York Heidelberg Berlin, 1975. [3]. Zhou Y, Fushitani M, Sato K and Ozawa M. Bending Creep Behavior of hot-pressed wood under cyclic moisture change conditions, Journal Wood Science, 2000, 46, 423-430. [4]. Hoffmeyer P, Damkilde L and Pedersen TN. Structural timber and glulam in compression perpendicular to grain, Holz als und Werkstoff, 2000, 58, 73-80. [5]. Heger F, Groux M, Girardet F, Welzbacher C, Rapp AO and Navi P. Mechanical and durability performance of THM-densified wood, Workshop COST Action E22 „Environmental Optimisation of Wood Protection‟, Lisboa Portugal, March 2004, pp. 1-10. [6]. Kubojima Y, Ohtani T and Yoshihara. Effect of Shear Deflection on bending properties of Compressed Wood, Journal of Wood and Fiber Science, 2004, 36 (2), 210-215. [7]. Adlam R. Thermochemical Densification of Timber: Initial Investigation of Potential of Softwood Timber, Final report to Forest and Wood Products Research and Development Coporation, Project No. PN03.3802, Australia, 2005. [8]. Yoshihara H and Tsunematsu S. Bending and shear properties of compressed Sitka spruce, Wood Science Technology, 2007, 14, 117-131. [9]. Yoshihara H and Tsunematsu S. Elastic Properties of Compressed Wood Spruce with Respect to Its Section Obtained Under Various Compression Ratio, Forest Product Journal, 2007, 57(4), 98-100. [10]. Jung K, Akihisa K, Munekaju M and Komatsu K. Development Joint System using by compressed wood fastener, Proceedings of the 10th World Conference on Timber Engineering, Miyazaki, Japan, June2-5,2008. [11]. Hassel I, Berard P and Komatsu K. Development of Wooden Shear Wall-Improvement of Stiffness by utilizing elements of densified wood, Holzforschung, 2008, 62(5), 584-590. [12]. Jung K, Kitamori A and Komatsu K. Evaluation on Structural Performance of Compressed Wood as Shear Dowel, Holzforschung, 2008, 62(4), 461-467. [13]. Guan, ZW and Rodd PD. Hollow steel dowels – A new application for semi-rigid timber connections, Engineering Structures, 2000, 23(1), 110-119.
[14]. Guan ZW and Rodd PD. DVW – local reinforcement for timber joints, ASCE’s Journal of Structural Engineering, 2001, 127(8), 894-900. [15]. Guan ZW, Kitamori A and Komatsu K. Experimental study and FE modelling of Japanese “Nuki” joints, part one: inital stress states subjected to different wedge configurations, Engineering Structures, 2008, 30, 2032–2040. [16]. Guan ZW, Kitamori A and Komatsu K. Experimental study and FE modelling of Japanese “Nuki” joints, part two: racking resistance subjected to different wedge configurations, Engineering Structures, 2008, 30, 2041–2049. [17]. Guan ZW and Rodd, PD. Numerical modelling of timber connections locally reinforced by DVW discs. Structural Engineering and Mechanics, 2003, 16(4), 391-404. [18]. ABAQUS/Standard, User‟s Manual, Version 6.8, Hibbitt, Karlsson & Sorensen, Inc., 2009.
