Mechanical behavior of fiberglass reinforced timber joints Chen, Chi-Jen 1 ABSTRACT The objective of this research is to investigate the mechanical performance of dowel-type timber joints reinforced by fiberglass fabrics as reinforcements. Some critical characteristics such as the anisotropy of wood and splitting failure in structures and in joints demand more skill and limit the engineering design owing to the complexity of the design. According to the results of experiments, the fiberglass reinforcements leads to a higher performance and provides a good security factor to the timber joints. Experimental work, numerical analysis and the optical observation to evaluate the reinforcing technique are reported. 1. INTRODUCTION Joint, is one of the weakest points in a timber structure. The loss of perfect continuity in the structure caused by the presence of joints results in a decrease in the global strength of the structure which in turn implies an increase in the dimension of the assembled elements. The singularity of joints is not only due to a combination of different materials, for example, timber and steel, but is also due to the anisotropy of timber itself. Therefore, we cannot design a timber joint without taking into consideration these parameters. In material point of view, the tensile strength perpendicular to the wood grains represents a weakness of timber structure which affects the global performance of the whole structure. The complexity of timber joints is caused mainly not only by the anisotropy characteristics of wood, but also by the non linearity of fasteners. The bearing capacity is influenced directly by various parameters like embedding strength, grain directions. Moreover, the brittle fracture behavior of timber splitting must be avoided by regarding the minimum space and the end-distances between the connectors. 2. OBJECTIVES For a typical bolted or doweled joint, smaller diameter bolts signify, on the one hand higher ductility due to their ability to undergo large deformations but, on the other hand, show weakness in their bending capacity. The stiffness can be improved easily by increasing the diameter of the bolt. However, the increase in stiffness cannot be realized without increasing the dimension of the joint, otherwise, the longitudinal splitting in timber may cause a premature failure. Fiberglass as local reinforcement is proposed to improve in one hand, the enhanced embedding strength and enhanced ductility, in another hand, to prevent the splitting failure of timber joints. Such a joint is achieved by an optimization of the embedding strength and by the prevention of a premature failure of the timber [1]. 3. EXPERIMENTS The reinforcement materials used are the Vetrotex roving fiberglass bi-directional fabrics woven in 600g/m2, and epoxy resin of WEST SYSTEM. The fiberglass reinforcements are glued laterally on the timber strips. See Figure 1. The investigation of reinforced timber joints consists of three parts. The first part relates to the fundamental properties of reinforced timber, notably the tensile strength perpendicular to wood grains and the shear strength. The embedding strength test is carried out as second part of experiments, which defines the bearing capacity of timber joints in compression. The third part concerns a series of tests on reinforced doweled joints loaded in tension involving various geometric parameters. Qualitative results related to the strength, the stiffness and the failure modes are reported. 1 Ph.D., Civil Engineer, IBOIS-Laboratory for Timber Structure, Department of Civil Engineering, Swiss Federal Institute of Technology, Lausanne Switzerland, IBOIS-DGC-EPFL, 1015 Lausanne, Switzerland
Fiberglass Fabric Timber Figure 1 Configuration of Reinforced Joints An additional series of embedded tests is implemented using the Electronic Speckle Pattern Interferometry. The optical real time observation offers directly the information on the state of in-plane deformation fields on the specimen surface. The results show a considerable improved deformation behavior of the reinforced over the non-reinforced specimens. 3.1 Fundamental Properties of Reinforced Timber Two principal mechanical properties are studied by means of standardized tests: shear strength and tensile strength perpendicular to grains. The tension stresses perpendicular to grains are obtained form applied tensile force divided by the surface of screened zone on specimens supposing the deformation of dowel is neglected. The results are summarized in Table 1. Unreinforced Timber Reinforced by 1 layer Reinforced by 2 layers Numbers 16 16 16 Theoretical Strength (N/mm 2 ) 10.97 19.48 Experimental Strength (N/mm 2 ) 1.50 10.51 19.10 Increase (%) 600 1173 Table 1 - Results of Tensile Tests For shear strength, specimens with area of shearing plane is 20 mm x 20 mm are undertaken. Two types of specimens considering two shear planes (TL and RL) are tested. The strength of unreinforced joints is varied from 5.53 N/mm 2 for RL plane and 8.22 N/mm 2 for TL plane. The reinforced specimens (shear in RL plane) with one and two layers of fiberglass are tested by the same testing configuration, the improved results are shown in Table 2. Reinforced by 1 layer (RL) Reinforced by 2 layers (RL) Specimens Numbers 20 20 Average Density (kg/m 3 ) 429 430 Average Shear Strength (N/mm 2 ) 7.10 8.67 Increase (%) + 28.4 + 56.8 Table 2 Results of Reinforced Shear Specimens 3.2 Embedding Strength in Compression Compression tests are executed in directions both parallel and perpendicular to the wood grains. The specimens are prepared according to Eurocode 5 [2], regarding especially to the end distance and the slenderness of fasteners. The fiberglass reinforcements are glued laterally on the timber strips. Following parameters are taken into account to the investigation:
- wood density - grain direction of loaded specimens (parallel and perpendicular to the grain) - quantity of reinforcements (number of layers) - proportion of reinforcement by varying the thickness of specimens Table 3 describes the improvement of embedding strength by reinforcing fiberglass on timber joints. Figure 4 presents the relationship of wood density and bearing strength of reinforced timber. The strengths increased proportionally when wood density increases. Theoretical strength is calculated by using equation of embedding strength in Eurocode 5: σ f 0 = whereσ 0.082 (1-0.001d) ρ f 0 : embedding strength parallel to wood grains d :diameter of dowel ρ :density of wood Unreinforced 1 layer reinforced 2 layers reinforced Mean density (kg/m 3 ) 420 430 400 Theoretical Strength EC5 (N/mm 2 ) 28.93 29.62 27.55 Experimental Strength (N/mm 2 ) 32.39 38.02 39.7 Theoretical Increase 1.12 % 28.4 % 44.1 % Table 3 Comparison of Results from Embedded Tests Different directions of wood grains are investigated by taking the specimens reinforced by one layer of reinforcement. As the bearing strengths of different directions of fiberglass fabrics are defined experimentally. The estimated embedding strength are proposed by integrating the embedding strength of fiberglass and the embedding strength of wood. Figure 4 shows good collation between experimental and estimated embedded strength. Embedding Strength (N/mm 2 ) Embedding Strength VS. Grain Direction 35 Experimental (Reinforced) Estimated (Reinforced) 30 EC 5 (Unreinforced) 25 20 15 0 15 30 45 60 75 90 Grain Direction ( ) Figure 4 - Comparison of Bearing Strength between Experimental Results and Estimation 3.3 Bearing Strength in Tension The specimens with various end distances, with the same geometry and the same reinforcement (RT600) are executed in tensile tests.
120 80 40 Figure 5 - Three Variants of Specimens with various end distance The shear strength is varied from 5.53 N/mm 2 for RL plane and 8.22 N/mm 2 for TL plane Different failure modes of unreinforced specimens took place due to the variations of end-distance. By referring to Figure 5, The left specimen shows a sufficient end-distance (L=7.5d) leads to an significant bearing failure at first. The splitting failure occurs while the end-distance is decrease by the slip of dowel. The splitting failure due to the decreased end-distance (L=5d) are shown in the middle specimen. The exceeded tensile strength perpendicular to wood grains causes the splitting failure along two extremities of end-distance. The right specimen shows the shear-out failure due to a insufficient end-distance (L=2.5d), where the failure takes place by exceeded shear strength. For the same geometry but reinforced joints, the failure modes are replaced by a significant bearing failure. The dowel shanks into timber with a slip along the direction of load. Not only the bearing capacity is increase explicitly, but a good ductility of joints appear during the tests. The brittle failure (splitting crack) is well avoided by reinforcing fiberglass fabrics. See Figures 6 and 7. Table 8 resumes the comparative results. Figure 6 Failures of Unreinforced Specimens Figure 7 Failures of Reinforced Specimens End Unreinforced 1 Layer 2 Layers Increase Increase distance by 1 Layer (%) by 2 Layers (%) I = 7.5d ( 120 mm ) 24.4 28.39 32.79 16% 34% I = 5d ( 80 mm ) 21.05 26.89 30.89 27% 46% I = 2.5d ( 40 mm ) 15.66 18.4 30.33 18% 93% Table 8 - Comparison of Bearing Strength in Tension (N/mm 2 ) 3.4 Holographic Observations A series of tests were carried out to study the evolution of the in-plane displacement fields in a reinforced timber specimen when subjected to compressive loading. Observation is carried out in the region surrounding the loaded bolt. The interferogram showing the in-plane displacement distributions in horizontal direction are captured by video camera
and saved as image files on the computer. Observations are compared to the results of a previous study [3] with unreinforced specimens. Figure 9 shows the in-plane displacement fields at different levels of loading. A noticeable concentration of strains is observed around the bolt for the unreinforced timber joints. Due to the stiffness of fiberglass fabrics, moderate deformations are observed in the reinforced joints. Even under a higher increment of load, the reinforced wood specimen shows a quasi non-deformed map around the bolt The onset of crack propagation on the specimen is shown at the end of the figure of unreinforced joints. The crack initiated at the point of highest stress concentration on wood. The displacement map in last figure of reinforced joints, demonstrates the ability of the reinforcement to prevent successfully the occurrence of crack at the same load level on the reinforced specimen. Unreinforced joints Low Load High Reinforced joints Figure 9 Holographic maps of unreinforced joints (upper row) and reinforced joints (lower row) 4. NUMERICAL MODELING The numerical model is developed based on Finite Element Method, which provides the flexibility to study analytically the dowel-type timber joint. By varying the material characteristics as well as geometric parameters, the evaluation of mechanical behaviors of joint is thus recognized. The design criteria can consequently be defined. [5] Due to the high orthotropic characteristics of timber, the splitting and shearing failures are observed for most of unreinforced specimens. A strength criterion is thus proposed taking two critical material strengths: shear strength and tensile strength perpendicular to grains, into consideration.(4) 2 æ σ xy ö æ σ TT ö 1 + çç ç è Y ø è S ø where σ TT : Tensile stress perpendicu lar to wood grains 2 σ xy : Shear stress parallel to wood grains Y S : Tensile strength perpendicu lar to wood grains : Shear strength parallel to wood grains The shear strength is varied from 5.53 N/mm2 for RL plane and 8.22 N/mm2 for TL plane in accordance of experimental results. The tensile strength perpendicular to grains of 5 N/mm2 is taken from reference [4]. The re-built failure criterion
is applied to define the critical loads of unreinforced joints in tension. The embedding strengths are calculated by dividing the fund critical loads by the embedded surface A (= diameter of dowel X thickness of wood). For reinforced joints, the failure criterion is modified considering the failure mode. In previous experiments, the reinforced joints failed in bearing and in shearing out. The high transversal tensile strength avoids successfully the splitting failure. The failure criterion is thus modified taking compressive strength and the shear strength into consideration. The criterion can be written as following: 2 2 σcp σ xy + 1 X S where σcp : Compressiv e stress parallel to the wood grains σ xy : Shear stress parallel to the wood grains S X : Reinforced shear strength perpendicu lar to the wood grains : Compressiv e strength parallel to the wood grains ( 50MPa) In general, the applied strengths in failure criterion are defined by means of experiments. They are nevertheless, changeable depending on many parameters such as: species of wood, studied plans and mechanical testing methods. The onset of failure of doweled joint is predicted by estimating a critical load. Figures 10 presents the feasibility of models comparing to the experimental results. The mechanical properties of timber joints are introduced in numerical models in accordance with the experimental results. The error from unreinforced numerical model is about 11.1%. The tendency of load evolution behaves a good collation between experimental and numerical models. For reinforced numerical model, the error is of 6.67 %. Bearing Strength (N/mm2) 30 25 20 15 10 Comparison of Results (Unreinforced Joints) Numerical average results 5 Experimental results 0 20 40 60 80 100 120 140 Edge distance from bolt (mm) Bearing Strength (N/mm2) Comparison of Results (Reinforced Joints) 40 35 30 25 20 15 Numerical average results 10 5 Experimental results 0 20 40 60 80 100 120 140 Edge distance from bolt (mm) Figure 10 - Comparison of Numerical and Experimental Results 5. APPLICATION EXAMPLE In practical applications, the fiberglass reinforcement can be applied easily on the inserted element in a composed section (sandwiched system). The possible reinforced joints of timber structure are given in Figure 11. In timber structures such as truss joint, axial loaded joints in tension or in compression. Figure 12 shows A pedestrian footbridge made in timber truss using fiberglass-reinforced joints is presented [6]. Trussform bridge crossing over the river has total span of 24.5 m and with deck width of 2 m. All diagonals are built in solid wood, the upper and bottom chords are built in laminated wood.
The diagonals in simple sections that are inserted and are connected into the double sections. The fiberglass fabrics are applied discretely on the inserted sections where the dowels connect the chords and diagonals, shown in Figure 13. The connections on reinforced zones are thus designed with the higher security as well as the enhanced load capacity. Truss Joint of Sandwiched System Tensile Joint Moment-Resisting Joint in Solid Wood Moment Resisting Joint in Laminated Wood Figure 11 - Possibility of Reinforced Timber Joints Figure 12 View of footbridge Figure 13 View of reinforced detail 6. CONCLUSION The improvements on mechanical performance of reinforced timber joints reflect not only on bearing capacity and on several strengths of joints, but also behave a higher ductility and failure resistance which provide the wider possibility in practical designs. The premature failure especially the splitting failure parallel to wood grains was avoided by replacing
the bearing failure which provides the higher loading capacity of timber joints. The improved tensile strength in perpendicular direction of wood grains transforms the brittle failure modes through the less fragile failure modes. Several modifications of geometry on reinforced joints propose the designing possibility for reinforced timber joints. The use of the electronic speckle pattern interferometry to determine the in-plane displacement fields in the neighborhood of bolt-loaded holes in a timber specimen. A significant improvement of the mechanical strength is observed for a reinforced bolted wood joint. As the mixed failure mode is observed and assumed to represent the main failure mode for doweled joint loaded in tension, the proposed failure criterion is thus effective for predicting the strength of joints. The reinforcing technique can be evaluated by analyzing the interaction between stresses with the variations of geometry by the numerical models. Moreover, the critical end-distance as well as edge distance can be optimized while the mechanical properties of reinforcements are well known. 6. REFERENCES [1] P.Haller, C.J.Chen, J.Natterer (1996), "Experimental Study on Glass-fibre Reinforced and Densified Timber Joints", Proc. of International Wood Engineering Conference, October 28-31, 1996, New Orleans, Louisiana, USA. [2] pren 383. Structures en bois - Méthodes d'essais - Détermination de caractéristiques de fondation de la portance locale d'éléments d'assemblage de type broche. Bruxelles : Comité Européen de Normalisation, 1993. [3] P. Rastogi, C. J. Chen, "Mechanical Behavior of Bolted Wood Joint using Electronic Speckle Pattern Interferometry", International Conference on Experimental Mechanics: Advances and Applications, Singapore, pp. 404-407, December 4-6, 1996. [4] L. Daudeville, L. Davenne, M. Yasumura (1996), "Experimental and Numerical Analysis of Failure in Bolted Joints", Proceedings of International Wood Engineering Conference, October 28-31, 1996, New Orleans, Louisiana, USA. [5] C. J. Chen, Mechanical Behavior of Fiberglass Reinforced Timber Joints, Ph.D. Thesis, No. 1940, EPFL, March, 1999, Lausanne Switzerland. [6] Gärtl, K. Holzbrüke mit Rautenfachwerksträgern, Schweitzer Holzbau, N0. 12, 1998, pp. 8-11