EXPERIMENTAL AND NUMERICAL EVALUATION OF THE STRUCTURAL PERFORMANCE OF GLUED SOLID TIMBER BEAMS MADE OF FRENCH BEECH Van-Dang Tran, Marc Oudjene, Pierre-Jean Méausoone Université de Lorraine, Laboratoire d Etudes et de Recherche sur le Matériau Bois (LERMAB EA 4370), 27 Rue Philippe Séguin, 88051 Epinal, France van-dang.tran2@univ-lorraine.fr Abstract: This paper aims to present a first series of experimental and numerical studies that have been planned in the framework of a research project on the structural performance glued solid timber, made of french beech, currently on-going at the LERMAB. Glued solid timber beams have been manufactured, where the finger joints and the layer interfaces were bonded with structural Melamine Urea Formaldehyde (MUF) adhesive that fulfil current approval criteria for the use in load-bearing timber components. The fracture of glue-lines and the finger joints has been simulated using the cohesive surfaces of ABAQUS finite element software. The simulation results were compared to experimental ones showing good agreements. INTRODUCTION Engineered wood products, such as manufactured glued solid timber (GST) or glued laminated timber beams and columns, have been widely used in Europe and their share is likely to grow, due to the present times characterized by specific circumstances in the climate change and fuel cost uncertainties. These products find extensive applications in timber construction of both new buildings and renovation of older buildings, due to their flexibility and the advantages they offer in terms of different shapes and forms in final assembled systems. Also, where possible, their use is often more favoured, because timber is recognized as being more environmentally friendly, and often more aesthetic, when compared to traditional materials such as steel and concrete [1]. Due to the fact that buildings represent one of the largest energy consumers and greenhouse gas emitters, energy saving strategies related to buildings is strongly recommended [2]. In France, there is a strong demand in the market of GST beams and significant increases are expected in the future. Thus, around 80% of GST beams used in the French market is produced in Germany and Austria. Moreover, the review of the life cycle inventory of glulam undertaken by the FCBA (Technological Institute of wood and wood construction) has demonstrated that transportation was a major source of greenhouse gas with 90% of undesirable emissions. Also, the life cycle inventory conducted by Puettman and Wilson [3] has emphasized transportation as a major source of gas emissions. Considering the growing importance of imported engineered wood products, minimizing transportation would allow for higher energy efficiency and, consequently, for reduced greenhouse gas emissions. This implies local production by using local resources and skills. In this context, a research project on the evaluation and classification of French beech for the use in load-bearing timber components is currently on-going at the LERMAB. In France, minimising transportation is gaining ever more support, since there are vast forest resources, with a large variety of species. The current work presents an exploratory research project on the suitability of producing structural GST beams and columns from local resources. The specie of GST in this study is French beech (Fagus sylvatica), which is one of the most important renewable species cultivated in the eastregion, traditionally used in furniture and rarely used in construction as wood frame. This exploratory research project aims to demonstrate the ability to produce GST 553
structural elements, made of French beech, with satisfactory strength and stiffness requirements. This means that the product mechanical performances must satisfy the European Standard requirements ([15, 16, 17]). Short and long GST beams with different number of layers, ranged between 45 mm and 85 mm (Fig. 1), where manufactured with or without finger joints (Fig. 2). The finger joints and the layer interfaces were bonded with structural MUF adhesive that fulfil current approval criteria for the use in load-bearing timber components. Figure 1: Example for glued solid timber made of two and five laminations, t l : lamination thickness, 45 t 85 [ 15]. l Figure 2: Typical profile of a finger-joint: l j : finger length, p: pitch, l t : tip gap, b t : tip width, : finger angle, l t : tip gap [15]. However, to study experimentally all possible design variations and combinations would be much cost and very time consuming. Therefore, a computational model, using the finite element method, incorporating delamination and fracture of the bondlines at the layer interfaces, is developed to investigate more effectively the flexural behaviour of GST beams. In fact, once numerical models are validated against experimental results then they can be used to thoroughly undertake a parametric study. Although comprehensive studies have been conducted on the structural behaviour of glulam, with or without reinforcements, there is very little research on the structural behaviour of GST including delamination and fracture of bond-lines. EXPERIMENTAL WOR In order to simulate accurately the global structural behavior of GST beams, using the finite element method, the following characteristics are required: - The mechanical properties of the individual lamellas, - The delamination and fracture of the bond-lines at the layer interfaces, - The behavior of finger joints. 554
For this end, the three above requirements have been characterized using appropriate experimental procedures. The characteristics of the MUF adhesive were assessed using three modes of failure (Fig. 3), namely opening mode, shearing and tearing modes. Figure 3: Three different modes of fracture. To describe the opening mode (mode I), the wedge-splitting specimen similar to that studied in [7] has been tested (Figs. 4a and 5a). Specimen used to study mode III is shown in Figs. 4b and 5b. Specimens shown in Fig. 5a are characterized by a trapezoidal shape groove aimed to force the crack growth inside the glue-line zone. 30 unbonded Adhesive ( MUF) 43 9 50 250 12 A A - A 15 unbonded Adhesive ( MUF) 15 25 50 250 50 A Figure 4: Geometry of specimens: for mode I, for mode III. Figure 5: Experimental setup: in mode I, in mode III. 555
After that, the individual lamellas, the finger joints as well as the manufactured GST beams were tested in flexural destructive tests according to the EN 408:2010 requirements Figs. 6-7 [16]. From the experimental results, various parameters were obtained: namely the global modulus of elasticity (Eq. 1) and the flexural strength (Eq. 2). Other properties were obtained according to EN 338 [17]. a=6h±1.5h 6h a=6h±1.5h F (N) h F2=0.4Fmax w F1=0.1Fmax L=18h±3h Figure 6: Plan of test for measure the global modulus of elasticity in flexion; h, b, L: height, width and length of beam, respectively (mm); G: shear modulus (N/mm 2 ) [16]. 2 3 3aL 4a 3Fa E m, g (1); f m 2 (2) 3 w w 6a bh 2 1 2bh 2 F2 F1 5Gbh w1 w2 w (mm) Figure 7: Force/deflexion curve on the range of elasticity displacement; F1; F2: increase of force along the line of regression (N); w1, w2: increase of deflexion corresponding to F1, F2 (mm); Fmax: maximal force of measurement (N) [16]. The material parameters of the individual lamellas have been obtained from both experimental tests and by mean the relations suggested by EN 338 [17]. The average test results are summarized in Table 1. Table 1: Elastic properties of French beech E L E R E T RT RL G TL RT G RL G TL 11900 748 500 0.558 0.038 0.015 40 700 700 NUMERICAL MODELLING The ABAQUS finite element code was used in the present study. 2D and 3D finite element models were assumed. Orthotropic anisotropic elasto-plastic material model [4, 5, 14] has been assumed for the timber behavior. The crack propagation at the bond-lines as well as in the finger joints have been described by cohesive surfaces based on the traction-separation law (Fig. 6) supported in ABAQUS. t t t max max max n s, t t G TC max n max max s, t Figure 8: Traction-separation law [13] 556 f n, f s f t
In that behaviour law, the stress vector, t, is depending on the stiffness matrix [] such that [13]: t n t t s t t nn sn tn ns ss ts nt st tt n s t Where: n, s and t represent the separations in the normal, shear and tangential directions, respectively. During softening, t is defined on the basis of a damage parameter matrix D ( i ) as follows: I D ( ) t; i n, s t (3) t i, (4) For a better reading on the traction-separation low, the reader is referred to the ABAQUS manual [13]. FINITE ELEMENT MODELS Before simulating both the behaviour of finger joints and the two-layer GST beams, all the cohesive parameters entering the damage model (crack initiation and its growth) of the glue-lines were first identified using an appropriate parametrical study. 1. Simulation of crack propagation in modes I and III 3D finite element model and eight-node hexahedral elements have been used for the discretization of the timber (Fig. 7). Only one half of the model has been considered since the geometry admits one plane of symmetry. Figure 9: Finite element model for mode I, Finite element model for mode III. Several nonlinear analyses were conducted using ABAQUS software [6-12] for different combinations of cohesive behaviour parameters in order to obtain the optimal values that fit well the experimental results. Tables 2 and 3 summarize the optimal cohesive parameters used for fracture propagation in mode I and mode III, respectively. Table 2: Optimal parameters for the fracture mode I n (MPa/mm) T n G n (mj/mm 2 ) 3 0.75 0.11 557
Table 3: Optimal parameters for the fracture mode III s (MPa/mm) T s G s (mj/mm 2 ) 13 2.5 0.75 Figs. 10 to 13 display the comparison between the experimentally and numerically predicted load-opening displacement curves, for various interface cohesive strength and critical fracture energy values. Figure 10: Effect of the critical interface fracture energy in mode I with the interface cohesive strength T n =0.7MPa. Figure 11: Effect of the interface cohesive strength in mode I with the critical interface fracture energy G n =0.11 MPa/mm. Figure 12: Effect of the critical interface fracture energy in mode III with the interface strength T s =2 MPa. Figure 13: Effect of the interface cohesive strength mode III with the critical interface fracture energy G s =0.75 MPa/mm. 2. Simulation of two-layer GST beam After the parameters for the damage law of glue-lines have been identified in both fracture mode I and mode II, a two-layer GST beam, under four-points bending test (Fig. 14), has been analyzed. 2D finite element model and four-node plane stress elements have been used for the discrimination of one half of the beam. 558
Figure 14: Four-points bending test: experimental; numerical Figure 15: Four-points bending test: experimental; numerical Fig. 15 illustrates the comparison between the experimental and simulation results, where it can be seen a fairly good prediction of both the non-linear behaviour and the post-failure behaviour. 3. Simulation of finger joint In the same way, the computational approach has been applied to simulate the behaviour of finger joint, under four-points bending test. It can be observed from Figs. 16 and 17 that the computational approach is predictive and suitable for the analysis of crack initiation and propagation in glue-lines in the context of glued solid timber components. Figure 16: Four-points bending test: experimental; numerical Figure 17: Comparison between experimental and numerical force-deflection curves CONCLUSIONS AND FUTURE WOR This paper presents the results of a first series of experimental and numerical tests that have been planned in a comprehensive research project on the structural performance of glued solid timber made of of french beech currently on-going at the LERMAB. The 559
obtained results are satisfactory and particularly promissing with regard to the finite element approach, since it can noticeably reduce the expensive experimental procedures. Work is now in progress and future work includes the simulation of fullscale multi-layered beams with finger joints, including geometrical and material parameters optimization. REFERENCES [1] O Loinsigh C., Oudjene M., Ait-Aider H., Fanning P., Pizzi A., Shotton E., Meghlat E-M., Experimental study of timber-to-timber composite beam using welded-through wood dowels, Construction and Building Materials, 36 (2012) 245-250. [2] Premrov M., Dobrila P., Experimental analysis of timber-concrete composite beam strengthened with carbon fibres, Construction and Building Materials, 37 (2012) 499-506. [3] Puettmann M.E., Wilson J.B., Gate-to-gate life-cycle inventory of glued laminated timbers production, Wood and Fiber Science, 37 (2005) 18-29. [4] Oudjene M., helifa M., Finite element modelling of wooden structures at large deformations and brittle failure prediction, Materials and Design 30 (2009) pp. 4081-4087. [5] O Loinsigh C., Oudjene M., Pizzi A., Fanning P., Shotton E., Mechanical behaviour and 3D stress analysis of multi-layered wooden beams made with welded-through wood dowels, Composite Structures, 94 (2012), pp. 313 321. [6] Ted Diehl, Modeling Surface-Bonded Structures with ABAQUS Cohesive Elements: Beam-Type Solutions, ABAQUS User s Conference, 2004. [7] Stefania Fortino, Giuseppe Zagari, Antonio Lorenzo Mendicino and Gerhard Dill-Langer, A simple approach for FEM simultation of Mode I cohesive crack growth in glued laminated timber under shortterm loading, Journal of Structural Mechanics, 45(2012), pp.1-20. [8] Chien-Chung Chen, Daniel G.Linzell, Modeling end notched flexure tests to establish cohesive element Mode II fracture parameters, Engineering Fractures Mechanics, 77(2010) 1338-1347. [9] Pizhong Qiao and Ying Chen, Cohesive fracture simulation and failure modes of FRP-concrete bonded interfaces, Theoritical and Applied Fracture Mechanics, 49(2008) 213-225. [10] N.Dourado, S. Morel, M.F.S.F de Moura, G. Valentin and J. Morais, Comparison of fracture properties of two wood species through cohesive crack simulations, Composites: Part A 39 (2008) 415-427. [11] Giulio Alfano, Silvio de Barros, Laurent Champaney and Nunziante, Comparison between two cohesive-zone models for the analysis of interface debonding, European congress on computational Method in Applied Sciences and Engineering ECCOMAS, 2004. [12] F. rasucki, A. Münch and Y. Ousset, Numerical simulation of debonding of adhesively bonded joint, International Journal of Solids and Structures, 39(2002) 6355-6383. [13] Hibbit arson and Sorensen, ABAQUS, Theory manual, version 6.2, Inc., 2000. [14] M. Oudjene, M. helifa, Elasto-plastic constitutive law for wood behaviour under compressive loadings, Construction and Building Materials, 23(2009) 3359-3366. [15] FprEN 14080, Timber structures- Glued laminated timber and glued solid timber- Requirements, authorized by ARNOR Normalisation, 2012 [16] NF EN 408, Structure en bois, Bois de structure et bois lamellé-collé: Détermination des certaines propriétés physiques et mécaniques, Edité et diffusée par l Association Française de Normalisation (ARNOR), 2010. [17] NF EN 338, Bois de Structure : Classes de résistance, AFNOR, 2010. 560