Experimental and numerical study of nailed laminated timber elements for in plane and transverse loading

Transcription

1 Experimental and numerical study of nailed laminated timber elements for in plane and transverse loading Haller, Peer 1 SUMMARY Nailed laminated timber elements are used in housing construction for floor, wall and roof elements. They consist of boards - continuous or butt jointed - being assembled to solid elements. The present paper outlines the structural behaviour of nailed laminated elements for in plane and transversal loading. The influence of the arrangement of butt joints on the structural behaviour, especially with respect to bending stiffness, was examined for different load configurations. In addition, various wall elements were subjected to vertical and horizontal loading. Small specimens served to determine the stiffness in longitudinal and transverse direction for the numerical modelling of the structural behaviour. Finally, three different models are proposed to predict the structural response. It appeared that a grid model fitted with good accuracy in the experimental results for all joint and load configurations. INTRODUCTION - MATERIAL AND COMPOSITION Boards of equal height and width are nailed over the whole length to laminated elements being used in housing construction as solid wall, floor and roof elements. In general, the boards are continuous and correspond to the element s length, however, they can be butt jointed (fig. 1b) in different way in order to make use of shorter pieces and waste for economical reason. Figure 1a Nailing of laminated elements 1 Professor, Institut für Baukonstruktionen und Holzbau, Technische Universität Dresden, Mommsenstraße 13, 162 Dresden, Germany

2 Figure 1b Examples of laminated timber floors with different butt joint configurations, elements The elements consisted in kiln-dried spruce boards (section 28x12 mm, a=165 mm) and were manufactured in a specialized company by means of an automatic nailing machine (fig. 1a). The elements had been stored for four weeks in the laboratory before testing. The wall elements consisted without exception of continuous boards. They were either assembled in the above mentioned industrial process or by hand nailing with a gun in order to realize different nail patterns, partially in addition with gluing. For this, single boards and nails from the same manufacturer were used. LOAD BEARING BEHAVIOUR GENERAL REMARKS Structural behaviour of floors The structural behaviour of nailed laminated timber is determined by its orthotropic structure, especially, by the load in transverse direction being transferred with the help of mechanical fasteners (nails). Moreover, butt joints and their distribution are crucial for the structural behaviour of the floors. Structural behaviour of walls Wall elements are vertically laminated so that the load is transferred parallel to grain leading to an important load bearing capacity. The latter is primarily governed by the shear modulus of the wall which is determined by the stiffness of the assembly. Nailed laminated timber walls fail by buckling in plane due to the compliance of the nail joint. Only glued elements fail out of plane at considerably higher loads. In (Haller, Pannke 1998) some results from structural and physical tests on laminated elements were reported. EXPERIMENTAL STUDY ON FLOORS Test set-up of floors The load carrying behaviour of floors subjected to uniform loading was studied in a four-point-bending-test, point loads were applied either at the centre or at the edge (see also Haller, Pannke 1998) (figure 2). The bearings at both ends of the specimens were pinned. The distributed load was introduced via a stiff steel cross section embedded in sand for a more uniform distribution of the load. The point loads were introduced on a square of 15 x 15 cm corresponding to five board widths

3 Figure 2 Test set up for floors four-point bending under uniform load (left); point load at the edge (right) The influence of the longitudinal butt joint and its distribution on the bending stiffness was tested for uniform load in the four point bending test. In order to avoid the scattering of the element s properties, four specimens were produced in which different joint patterns were realized afterwards by means of a drill (diameter corresponds to the board s width) (see figure 1). The bending stiffness was determined for a given element before and after having completely separated the boards by means of the drill. To determine an eventual influence of a contact pressure in the longitudinal or butt joint, the drill-holes were closed with mortar. The test showed that regularly arranged butt joints reduce bending stiffness of about 2% and that a gaping joint does not have any significant influence. The loading of the specimens was done according to ISO In order to avoid rheological influences (creep) the specimen was loaded and unloaded until a constant deformation for the evaluation of the bending stiffness was achieved. Finally, the element was loaded until fracture. Experimental results on floor elements As the theoretical displacement of a plate is integrally determined over its entire length, an infinitesimally small joint does not have any influence on the displacement field, however, as the condition of continuity (discrete nailing!) is not fulfilled a difference is observed in the tests. It appears, that regular butt joints reduce the bending stiffness of about 2% and that randomly or irregularly distributed joints behave better. Point loads at the edge and at the centre showed poor distribution in transverse direction which could be improved by means of additional layers (Haller, Pannke 1998). Table 1 Structural properties of floors for different parameters MOE [MPa] Continuous Gap joint Reduction [abs.] Reduction [%] Mortar filled Reduction [abs.] Reduction [%] Failure load [kn] MOR [MPa] Element ,9 9124,5 2697,4 22,8 934,5 2481,4 21, 76, 2,5 Element ,5 7575,3 2252,2 22,9 85, 1777,6 18,1 56, 15,1 Element ,6 8219,9 3242,7 28,3 8892,5 257,1 22,4 63,3 17,1 Element 4 992, 7421,8 2498,1 25,2 8229,2 169,8 17, 73,5 19,8

4 EXPERIMENTAL STUDY ON WALLS Test set-up of floors Wall elements of structural size were investigated for vertical, horizontal and combined loadings (figure 3). Pure vertical loading was realized in a press for maconery, horizontal and combined loading in a special panel test rig. The anchorage of the wall was done by means of a steel plate nailed upon the lower left corner of the specimen. Apart from the influence of the joint stiffness, the assembly of two elements as well as their bracing with a metal band was investigated. The displacement at the horizontal loading point (upper left corner) was measured by means of a strain gauge. In addition, the displacement field of the entire element was enregistred with the help of a digital camera. Figure 3 Test set-ups for walls; vertical (left), horizontal and combined loading (right) Experimental results on wall elements Tests were realized on elements with continuous boards, different nailing and nailing in addition with gluing. The joint stiffness in the longitudinal direction has a decisive influence on the structural behaviour which appears in low buckling loads and large horizontal deformation for vertical and horizontal loading respectively. The shear stiffness of nailed laminated diaphragms may be insufficient so that a bracing is needed to take the horizontal loads. Metal strips or wooden panels are considered in the design to cope with low shear stiffness of nailed laminated timber. In the present study a simple metal strip was nailed to stiffen the wall. Table 4 summarizes the results from these tests. Table 2 Test results of wall elements for horizontal and vertical loading Element Material 1) Spruce 2) Spruce 3) Spruce 4) Poplar 5) Poplar 6) Spruce 7) Spruce 8) Spruce 9) Spruce Dimension [cm] HxBxD 25x1x8 25x1x8 24x1x8 25x1x8 25x86x8 248x1x8 248x1x8 245x1x8 242x1x8 Joint Dimension/Type of nail/ No. of shear planes 28x7 / smooth / 25 28x7 / smooth / 25 42x7 / screw. / 22 31x7 / smooth / 25 42x7 / screw. / 46 3x8 / smooth / 36 3x8 / smooth / 36 Additional gluing Additional gluing Load Load Moisture horizontal vertical [kn/m] [kn/m] [%] 358,5 11,6 27,2 311,6 11,7 39,9 457,3 11,4 354,7 12,2 55, 12,4 9,8 12, 13,7 25,2 1794,9 48,3 388,5

5 Experimental study on small elements The modelling of the load bearing behaviour requires the knowledge of the joint stiffness in longitudinal and transversal direction. In case of floors under points loads and walls under vertical uniform loading the nails are stressed perpendicular and parallel to grain respectively. This stiffness data was determined on small specimens using the same material and nail patterns as the full-scale elements. The specimens and the experimental arrangement to determine longitudinal and transversal stiffness data are shown in figure 4. Results are reported in tables 3 and 4. Figure 4 Specimen and test arrangement for joint stiffness parallel (above) and perpendicular to grain (below) Table 3 Experimental stiffness of small specimens parallel to grain Specimen [kn/m] C 1 C 2 C 3 C 4 C g No 1 756,8 1616,1 849,2 1311,5 1133,4 No 2 873,5 1272,5 953,2 1134,3 158,4 No 3 779, 819,9 626,7 111,6 831,8 No 4 844,6 968, 827,9 99,9 97,9 No 5 518,1 1421,9 825,9 72,9 867,2 Mean 959,7 Table 4 Experimental stiffness of small specimens perpendicular to grain Specimen [kn/m] C 1 C 2 C 3 C 4 C g No 6 648,5 936,9 577,3 114,1 825,7 No 7 586,4 612,2 397,4 1215,4 72,8 No 8 732, 124,1 72,3 1294,1 983,1 No 9 495,7 911,7 468,3 121,7 724,3 No 1 58,3 89,1 616, 817,6 726, Mean 792,4

6 NUMERICAL STUDY The modelling of the load bearing behaviour of laminated elements requires a standard test method as described before for instance in order to evaluate stiffness data for in plane and transversal loading. Moreover, the influence of butt joints and their distribution is crucial for the structural response. Hence, an appropriate model must reflect the composition of the element as well as the joint stiffness. The gap of the joints is not taken into account since its influence can be neglected. Three different model were considered: Figure 5 Numerical models spring-bar (left); beam-grid (centre); orthotropic plate (right) 1. Beam-spring model, were the nail is modelled as an elastic spring coupling two boards with a spring stiffness determined from small specimens. 2. Beam-grid model where the stiffness in transversal direction is modelled by means of a transversal beam coupled with the longitudinal boards. 3. An analytical model based on orthotropic plate theory It appeared, that model 1 does not predict satisfactorily the load distribution in transversal direction. Model 2 - beam-grid - provides good results and seems to be an appropriate model for the prediction of structural behaviour. Also, the continuous orthotropic beam theory fitted well in the experimental results (see figure 6). Figure 6 Transversal deflection of test results (dots) and beam-grid model (line); centre load (left) and edge load (right) CONCLUSION A series of specimens made of nailed laminated timber was tested in plane and in transverse direction. It appeared from the tests that the composition and the stiffness of the joints are decisive of the structural behaviour. The arrangement of the longitudinal or butt joints in a line reduces the bending stiffness of about 2% while an irregular distribution behaves better. The scattering of the section s strength and stiffness is low due to the large number of laminations leading to more homogeneous structural properties of the entire element. As concern the behaviour of the wall the stiffness of the joint parallel to grain is crucial for the buckling load and the shear modulus. Nail spacing used in industry does only make use of about 2% of the ultimate vertical load of an entirely glued element. The shear modulus of nailed elements is very low, so that bracing is needed.

7 Finally, three models were examined that take into consideration the distribution and stiffness of the joints. The beam grid model using stiffness data from small-scale tests fitted well the experimental results regardless the joint pattern. Even though, an analytical model based on orthotropic plate theory provided reasonable results. REFERENCES Gabriel, S., 1996, Statische Untersuchungen an Brettstapelwänden unter horizontaler und vertikaler Beanspruchung, diploma thesis, Institut für Baukonstruktionen und Holzbau, Technische Universität Dresden Giese, B., 1995, Untersuchungen der Lastverteilung von Brettstapeldecken, thesis, Institut für Baukonstruktionen und Holzbau, Technische Universität Dresden and IBOIS, Ecole Polytechnique Fédérale de Lausanne Haller, P. (ed.), 1999, Brettstapelbauweise und ökologische Dämmstoffe, Tagungsband, Institut für Baukonstruktionen und Holzbau, Technische Universität Dresden Haller, P., Pannke, K., 1997, Statische und bauphysikalische Untersuchungen an Brettstapelelementen, 2. Fachkolloquium des Landesinnovationskollegs Wärmedämmsysteme, Institut für Textil- und Bekleidungstechnik, Technische Universität Dresden Haller, P., Pannke, K., August 1998, Structural and Physical Behaviour of Nailed Laminated Timber Elements, Proceedings, World Conference of Timber Engineering, Montreux-Lausanne, Switzerland, vol. 2., pp Langner, F. J., 1999, Untersuchung des Einflusses der Stoßverteilung auf das Tragverhalten von Brettstapeldecken, diploma thesis, Institut für Baukonstruktionen und Holzbau, Technische Universität Dresden Leroux, J.-N., 1997, Les parois en planches clouées, diploma thesis, Institut für Baukonstruktionen und Holzbau, Technische Universität Dresden Natterer J., Herzog, T., Volz, M., Holzbauatlas II, Architekturdokumentation Naumann, T., Gabriel, S., 1996, Experimentelle Untersuchungen an Brettstapelelementen, thesis, Institut für Baukonstruktionen und Holzbau, Technische Universität Dresden Neumann, T., 1997, Statische und dynamische Untersuchungen an Holz-Beton-Verbunddecken, diploma thesis, Institut für Baukonstruktionen und Holzbau, Technische Universität Dresden Pannke, K., 1994, Statische und bauphysikalische Untersuchungen an Brettstapelelementen, diploma thesis, Institut für Baukonstruktionen und Holzbau, Technische Universität Dresden Sattler, S., Haller, P., Neue, J., 1997, Feuchte- und Wärmeverhalten von Wandkonstruktionen mit ökologischen Dämmstoffen, Tagungsband, 2. Fachkolloquium des Landesinnovationskollegs Wärmedämmsysteme, Institut für Textilund Bekleidungstechnik, Technische Universität Dresden