EXPERIMENTAL INVESTIGATIONS ON FLEXURAL STRENGTHENING OF TIMBER STRUCTURES WITH CFRP

Transcription

1 Proceedings of the International Symposium on Bond Behaviour of FRP in Structures (BBFS 25) Chen and Teng (eds) 25 International Institute for FRP in Construction EXPERIMENTAL INVESTIGATIONS ON FLEXURAL STRENGTHENING OF TIMBER STRUCTURES WITH CFRP K.U. Schober and K. Rautenstrauch Department of Timber and Masonry Engineering Bauhaus-University of Weimar, Marienstr. 13A, Weimar, Germany ABSTRACT This paper presents a study of reinforcement techniques for restoration and strengthening of existing timber floors under bending loads, based on the use of carbon fibre reinforced plastics on the building site, whereas the removal of the overhanging part of the structure as well as the inserted ceiling is not necessary. Commercial adhesives, as well as products modified for this particular application, were tested. The tests showed, the arrangement of the reinforcement, the bond surface quality and the stiffness of the load transmitting materials were of decisive influence for the overall strength of the specimen. The presented paper provides a summary of recent test series and investigations on structural upgrading and debonding failure of timber floors using CFRP s. KEYWORDS FRP, timber floors, strengthening, structural upgrading. INTRODUCTION Particularly old timber structures in residential houses have a couple of damages and need specific interventions to restore these damages for renovation. They are designed for lower live loads as specified in performance and design standards and need new technologies to increase the load-carrying capacity of the members for state-ofthe-art housing conditions, including reinforcement or repair due to previous overloading, insect and fungal attack. Frequently, neither decision makers nor building contractors have the necessary knowledge on suitable reinforcement techniques and damaging interventions are often made. Therefore, a study of reinforcement techniques for restoration and strengthening of existing timber floors under bending loads has been carried out at the Bauhaus-University of Weimar. The experimental study based on the use of carbon fibre reinforced plastics on the building site, whereby the removal of the overhanging part of the structure as well as the inserted ceiling is not necessary. Reinforcement techniques for structural timber elements, based on the use of adhesives on site, have been applied for some decades as an extension of procedures that became very common for the repair or the upgrading of other structures. Such techniques are very promising as they minimize disturbance of the building and to its occupants during the intervention. However, some problems have prevented the wider use of adhesives, particularly in historical timber structures, where sufficient reliability cannot yet be guaranteed. One reason is that a life-long service life has not yet been fully proven for synthetic adhesives, since the oldest bonded joints are around sixty years and greater ages cannot be simulated by existing accelerated ageing tests. Fibre composites were investigated to reinforce wood as early as 1964 when E. J. Biblis tested very small beams that were faced with fibre glass. Theakston (1965) performed a feasibility study for reinforcing timber beams with fibreglass; Spaun (1981) undertook a study to investigate the potential of fibreglass reinforcement to improve the tensile strength and bending stiffness of wood members. Plevris and Triantafillou (1992) investigated wood beams that were reinforced with non-prestressed carbon sheets and Triantafillou and Deskivic (1992) investigated wood beams that were reinforced with prestressed carbon sheets. Dorey and Cheng (1996) investigated glulam beams strengthened with glass fibre reinforced polymer. The bond strength between fibre composites and wood was investigated by Abdel Magid et al. (1996). The purpose was mainly for determining the bond strength between the laminations of glue laminated beams that are reinforced with fibre composites. Although epoxy based adhesives have been used in most cases for on-site repair jobs, most formulations were developed for other materials. These adhesives are generally too rigid for bonding timber and there is no chemical bonding or suitable mechanical anchorage in wood. The bond surface is prone to fail because of 457

2 dimensional changes in the wood induced by moisture content variations, even under Service Class 2 applications (moisture content of timber up to 18 %). Moreover, on-site application of adhesives is somewhat difficult and the consequent quality of adhesive bond is not easy to evaluate. Since properties of reinforced elements very much depend on the care put in the work, such difficulties have to be overcome and first procedures for applying and controlling were established (CEN TC 193/SC1/WG11 23). MATERIALS AND METHODS Adhesive and Reinforcement Two basic approaches are discussed here: the use of reinforcement materials embedded internal in the wood specimen, the use of external reinforcement resulting in a system of composite type. The old solid wood beams were reinforced with a continuous carbon fibre lamella S&P 15/2 with intermediate modulus fibres from S&P Reinforcement Ltd. and 3.15 m length within the clear span for external reinforcement, otherwise over the full length of 3.5 m embedded in the wooden beams. The CFRP layer with a cross-section of 1.4 x 5 mm was glued / embedded by means of the commercially available epoxy resin StoPox SK 41 from StoCretec Ltd, which has a technical approval for gluing together with the carbon fibre lamellas. Because of the tests and the fracture modes observed, the resin will be modified for this special application on timber members. The epoxy resin consisted of two parts of component A and B mixed in a ratio of 4:1 by volume, as required in the technical description. The mean mechanical properties are shown in Table 1. Table 1 Mean material properties of used epoxy resin and CFRP Property Unit Epoxy Matrix CFRP Tensile Strength MPa Tensile Modulus of Elasticity GPa Ultimate elongation % Density g/cm³ Three different reinforcement schemes were evaluated in the testing program, series 1 with external and series 2 and series 3 with internal reinforcement. Table 2 Mean dimensions of used timber beams and reinforcement schemes Series Height [cm] Width [cm] Type Description Vh x 1.4 x 5 mm bonded centrally to the tension zone, horizontal on bottom Vs x 1.4 x 25 mm bonded laterally to the tension zone 3 cm from bottom in slot Vv x 1.4 x 5 mm bonded centrally to the tension zone, vertical on bottom The gluing of the lamellas was done under practice-related conditions. For reinforcement type Vh, the wood surface must be even, unweathered and clean by planning the surface until sound wood is reached to ensure a continuous bound. The CFRP lamellas were cleaned by using acetone. The timber surface was primed using StoJet HIS, a two-part epoxy to saturate the wood prior gluing to avoid desiccation of the resin and to ensure a full compound between resin and wood surface. 458

3 Beam specimens For practice related investigations, a carefully selection of the test specimen was required. In this case a couple of preloaded spruce rafters and ceiling joists with an age of more than 1 years where removed from an old residential house in Bavaria/Germany. The influencing factors like pre-ageing and weathering, moisture content when bonding, wood species, density, cross-section, existing cracks, knots and damages were analyzed. For determination of the strengthening effect after bonding and the Young s modulus of the specimen, the real cross-sectional data including cracks and damages A, I y and the averaged properties A mean, I y,mean where taken into account. A typical cross section of the test specimens and the outline for calculating the real cross section data are shown in Figures 1-2. Due to the usual existing cracks in historic timber beams, a reduction of the moment of inertia and the bending stiffness of about 8-23% (mean value 17 %) could be observed. Based on the reviewed test specimen cross section, a section modulus modification factor, k a of.8 may be justified generally for old timber beams that have similar properties and history as the beams used in this study. The detailed values are shown in Table 3. Test procedure Figure 1-2 Typical cross-section of test specimen The main task of the test series performed was gathering and generating qualitative and quantitative knowledge on CFRP suitable for on site repair of timber structures. We have studied three practical solutions of bonding on the building site in order to investigate the different performance of the composite structure and to attain the highest possible quality of the bond surface. Questions exist concerning the proportion of the quality, influenced by the properties of the wood surface, and how adhesion and the forming of the interface between adhesive and the wood surface really work. For investigating the strengthening effect of the carbon fibre reinforcement in different, practice-related positions the flexural behaviour of all specimen were tested first within the elastic range with and without reinforcement. The tests of the un-reinforced beams were accomplished to determine the bending stiffness and the Modulus of Elasticity of the specimen within the elastic range. After unloading, the same specimen was reinforced described above and the tests were executed on each beam again within the elastic range for Modulus of Elasticity and bending stiffness examination of the reinforced specimen. The applied loads were increased until failure to evaluate the fracture behaviour and ultimate load. There were four beams in each series with a length of the specimen of 3.5 m, span 3.25 m. All beams were surveyed for geometric dimensions and wood damages. The humidity ratio was measured on different locations. The average value of the humidity ratio was % with a COV of 6 %. FLEXURAL-BOND TEST The tests were executed as a four-point bending test according to EN 48. The load span was equal to 6 h m (Figure 5). The beams were loaded with a 2 kn actuator and a spreader beam. The spreader beam, cantered about the mid-span, created a zone with constant moment and zero shears. The vertical displacements of the beams were measured using six inductive transducers in the span and 2 inductive transducers placed on the support brackets and attached to the beam mid-depth, additionally using a 2D close range photogrammetry approach by analyzing the deflection field in the midspan of the specimen (Rautenstrauch et al. 23; Rautenstrauch et al. 24). For measuring the horizontal deflection in the midspan area 1 inductive transducers were used. The longitudinal deflection between the specimen and the CFRP lamella were measured using 459

4 inductive transducers at the face of the specimen (Figure 3). In test series 1 with external reinforcement additional 25 strain gauges were placed on the external bonded CFRP lamella to obtain the elongation of the reinforcement itself (Figure 4). The results have been used to calculate the bonding stresses in the resin and for an examination of the load-slip rule of the groove. The data were recorded during the tests using an automatic data acquisition and recording system. In test series 2 and 3 no additional strain gauges were placed by now due to the more difficult test set up for measuring the elongation of the embedded lamellas. Figure 3-4 Deflection measurement on face and bottom of the specimen The Young s modulus of the specimen was measured by recording the load deflection behaviour of all beams between two load levels according to EN 48 with a maximum load value of 12 kn. Loads for the measurement of the flexural capacity of the wood beams were applied monotonically to failure. The test for the evaluation of the bending stiffness achieves failure in 1 to 15 minutes. The results for the un-reinforced beams are reported solely for quantitatively evaluating the effectiveness of the interventions through a comparison with the results for strengthened beams (Figure 6). Load [kn] 16 Figure 5 Load arrangement 12 R 2 = R 2 = S 9/1 unreinforced S 9/1 reinforced Displacement [mm] Figure 6 Load-deflection curve within the elastic range (regression line of 16 data points) 46

5 RESULTS AND DISCUSSION Vertical deflection and fracture behaviour Each piece of wood differs in the amount of stiffness-reducing defects such as knots, splits, and checks and therefore, it is hard to say at what stress level the other reinforced beams would have failed if they had not been reinforced. Such is the nature of wood. Several experimental tests (Triantafillou 1997; Borri et al. 23) showed that the most frequent fracture mechanism is caused by the failure of the traction zone without the complete plasticization of the compression region, depending on the quality of the wood. However, under particular conditions it is possible to note the other failure mechanism, which is theoretically preferable for several reasons. First, the section shows a more ductile behaviour, while the stresses in the FRP material with reinforcement are highly increased and therefore, the composite material is more involved. Initially the load defection is shown to be linear elastic up to local failures induced by the presence of defects e.g. knots and cracks. Wood yielding produced a non-linear response terminated by a sudden drop of the load because of CFRP rupture. This was immediately followed by wood fracture in the tension zone, resulting in the collapse of the beams. The loaddeflection behaviour of the different reinforcement schemes is shown in Figure 7. Failure occurred at a load level between 35 kn and 87 kn (mean kn) due to the different fracture behaviour and sections. The tests have shown that the most frequent failure mechanism is the one in which traction failure and shear failure occurs with or without partial plasticization of the compressed zone. The differences of failure modes for different strengthening methods are described in Schober and Rautenstrauch (25). Load [kn] Displacement [mm] S 3/1 - Vh S 4/1 - Vh S 5/1 - Vh S 7/1 - Vh S 2/2 - Vs S 3/2 - Vs S 5/2 - Vs S 8/1 - Vs S 2/1 - Vv S 4/2 - Vv S 9/1 - Vv S 11/1 - Vv Figure 7 Load-deflection curves for different reinforcement types Generally, for all tests on reinforced specimens a linear phase could be observed until reaching the maximum Service Load F s with low creak and first small cracks in the specimen. Depending on the wood quality and reinforcement scheme, a linear or non-linear phase until the Ultimate Load F u where the first good visible cracks occurred and a further increasing of the deflection until reaching the Maximum Load F m with brittle failure of the specimen follows the linear phase. The different load levels are shown in Table 3. It could be ascertained that on all tests in series 1 and 3 the Ultimate Load is reached within the linear range and on all tests of series 2 the specimen could not support further load increments after the first break. The determination of the flexural strength of the composite structure has been done in the elastic range by recording of the load deflection data of 461

6 the specimen with and without reinforcement. For further investigations, the Modulus of Rupture on failure was calculated. The increasing of the bending stiffness by applying CFRP reinforcement can been done by defining a fictitious Modulus of Elasticity, calculated from the cross-section data of the un-reinforced specimen: 2 a l1 (F2 F) 1 Efict = (1) 16 I y,mean (w 2 w 1) where F 2 -F 1 denote the load increase in the elastic range, w 2 -w 1 the equivalent deflection values difference, a the distance from the support to the loading point an l 1 the span of specimen. The reinforcing scheme increased the load bearing capacity by mean 5.86 % in comparison to the values measured for the un-reinforced wooden beams. The strength increase k r,b was defined as the bending stress of the reinforced specimen at the deflection in linear range before failure divided by the bending stress of the un-reinforced specimen at the same deflection value and shown for each series in Table 3. The measured data correlate with the experimental results on glulam beams with horizontal reinforcement over the full with by Blaß and Romani (2, 21). There, the bending stiffness is calculated in the linear-elastic range and plastic deformations are not considered, since the stiffness is used for serviceability limit states and the stiffness increase for horizontal reinforced beams with a reinforcement over the full with. Table 3 Comparison of the different load levels and the load-carrying capacity Specimen F s F u F m MOR MOE MOE fict E fict / E k a k r,b [kn] [kn] [kn] [MPa] [MPa] [MPa] [%] --- (series) S3/1-Vh ,622 14, S4/1-Vh ,9 15, S5/1-Vh ,229 14, S7/1-Vh ,744 12, S2/2-Vs ,675 17, S3/2-Vs ,954 14, S5/2-Vs ,373 16, S8/1-Vs ,156 11, S2/1-Vv ,837 16, S4/2-Vv ,747 17, S9/1-Vv ,837 11, S11/1-Vv ,77 18, Mean E fict / E = EI, reinf / EI COV The wood beams reinforced with CFRP lamellas revealed more ductile behaviour with respect to un-reinforced beams. The presence of CFRP reinforcement arrest crack opening, confines local rupture and bridges local defects in the timber. Therefore, the specimen can support higher loads before failure. Moreover, as can be observed from the load-deflection curves of the reinforced and un-reinforced specimen and the calculation of the fictitious Modulus of Elasticity E fict, there is an increase of the load-carrying capacity because of the quoted crack opening arrest. Especially in reinforcement type Vh and Vs, with in wood embedded lamellas, an improvement of the load-carrying behaviour can be observed. The failure of the structure in this cases occurred by shear due to ripping of the section in longitudinal direction between cracks (Figure 8-9). To avoid this additional shear strengthening with connectors usable under practice-related conditions should be applied in further test series. Figure 8-9 Section ripping due to horizontal cracks 462

7 Horizontal deflection In the midspan the horizontal deflection on the timber faces was measured with inductive transducers and close range photogrammetry. The results obtained by using inductive transducers are shown in Figures 1 and 11. They have been used to compare the measurement data with different stress-strain laws given e.g. by Plevris and Triantafillou. A linear strain distribution over the specimen height could be observed in the elastic range for all test series. Where plastic behaviour in the compression zone was reached a non-linear strain distribution nearby the state of failure could be observed height [mm] 9 6 height [mm] ,3 -,2 -,1,1,2 strain,3 [%],4,3,2,1 -,1 -,2 -,3 strain -,4 [%] Design Applications S 4/2-Vv, linear range (F = 33,3 kn) S 4/2-Vv, plastic range (F = 44,2 kn) S 5/1-Vh, linear range (F = 24, kn) S 5/1-Vh, plastic range (F = 37,2 kn) Figure 1-11 Typical strain distribution for reinforcement. type Vv and Vh Predicting the stiffness and strength increase that occurs when a timber beam is reinforced with carbon fibre is a complex problem and difficult to establish because of the natural defects that occur in wood and drastically reduce the stiffness. However, design of wood structures has been accomplished for decades by applying stress modification factors to the allowable design stress values according to national codes for a given size and grade of timber. Accordingly, allowable stress modification factors have been conceptually developed based on the experimental results from this study. With further research and more comprehensive testing, these modification factors could be used by engineers to determine the safe load carrying capacity of old timber beams reinforced with carbon fibre. A review of the load versus deflection curves shows the beams in this study behaved mostly in a linear manner up until failure. In addition to providing confinement, the carbon fabric adds tensile strength and allows the beams to yield more in compression before they fail in tension. Based on test results showing increases in bending stiffness of about 6% and flexural strength up to 25%, a bending stress modification factor, k R,b of 1.7 to 1.12, depending on the type of reinforcement, may be justified for beams reinforced with carbon fibre that have similar mechanical properties as the fibre used in this study. Further testing has to be done with a larger number of specimens to confirm these modification factors. CONCLUSIONS The use of CFRP as a strengthening technique can be applied without necessitating the removal of the overhanging part of the structure. This is very promising in many cases of reinforcement of old, historical structural wood parts. Upgrading traditional timber structures of old solid wood with carbon fibre reinforcement on different locations are described and discussed here based on experimental investigations. For practice related investigations effective cross-sectional data including existing cracks, knots and damages were used with a reduction of the initial bending stiffness and the moment of inertia respectively of about 17 % compared to the cross-sections without defects. The results of the experiments have highlighted the limitations of the composite structure as well as the advantages of the various reinforcement positions and present numerous interesting aspects. The various theories of bonding developed so far are not able to explain comprehensively the observed effects. Chemical bonding always has been seen as the optimal form of combining two surfaces with each other, but its contribution to the overall bonding mechanism is still unclear. The properties of the glue line can be described among other methods by the analysis of the microstructure of the bond surface. This includes the adhesive penetration into the wood surface, the effect of ageing of a glue bond as well as the description of the cohesive strength of the glue line in terms of an optimization of the brittle and elastic ratio of the glue line. 463

8 The wood beams reinforced with CFRP lamellas revealed more ductile behaviour with compared to unreinforced beams. The presence of CFRP reinforcement arrests crack opening, confines local rupture and bridges local defects in the timber especially for reinforcement types other than on the bottom face. The failure of the structure in this cases occurred by shear due to separating of the section in longitudinal direction between cracks. To avoid this, an additional shear strengthening with connectors usable under practice-related conditions should be applied in future test series. These investigations are currently going on and will be supplemented by using further practical boundary conditions. REFERENCES Abdel-Magid, B., Scholsky, K., Shaler, S., Dagher, H. and Kimball, T. (1996). Interfacial bonding between phenolic matrix composites and wood, Fiber composites in infrastructure, H. Saadatmaneshand and M. R. Ehsani, eds, Biblis, E. J. (1965). Analysis of wood-fiberglass composite beams within and beyond the elastic region Forest Product Journal, 15(2), Blaß, H.J. and Romani, M. (2). Load-bearing capacity and deformation behaviour of FRP reinforced glulam composite beams, Research report: Schlussbericht AiF-Vorhaben 1147/N. Karlsruhe. Blaß, H.J. and Romani, M. (21). Design model for FRP reinforced glulam beams, International Council for Research and Innovation in Building and Construction, Working Commission W18-Timber Structures, CIB- W18/ , Venice. Borri, A et al. (23). FRP reinforcement of wood elements under bending loads, Proceedings, Structural Faults and Repair, London. CEN TC 193/SC1/WG11 (23). Adhesives for on-site assembling or restoration of timber structures. On-site acceptance testing : Part 1: Sampling and measurement of the adhesives cure schedule. Doc. N2. Part 2: Verification of the shear strength of an adhesive joint. Doc. N21. Part 3: Verification of the adhesive bond strength using tensile proof-loading. Doc. N22. Dorey, A. and Cleng, J. J. (1996). The behavior of GFRP glued laminated timber beams Advanced Composite Materials in Bridges and Structures, M. M. El-Badry, Ed., Canadian Society for Civil Engineering, Montreal, EN 48: (1995), Timber structures - Structural timber and glued laminated timber. Determination of some physical and mechanical properties, CEN European Committee for Standardization. Plevris, N. and Triantafillou, T.C. (1992). FRP Reinforced wood as structural material, Journal of Materials in Civil Engineering, 4 (3), Rautenstrauch, K. et al. (23). Strain analysis of solid wood and glued laminated timber constructions by close range photogrammetry, International Symposium: Non-Destructive Testing in Civil Engineering, Berlin. Rautenstrauch, K. et al. (24). Strain analysis of solid wood and glued laminated timber constructions by close range photogrammetry, Bauingenieur, Düsseldorf, Springer VDI-Verlag, Schober, K.U. and Rautenstrauch, K. (25) Strengthening of timber structures in-situ with an application of fiber-reinforced polymers, FRP Composites in Civil Engineering CICE 24, Seracino (ed). Taylor & Francis Group, London, ISBN , Spaun, F. D. (1981). Reinforcement of wood with fiberglass, Forest Products Journal, 31(4), Theakston, F. H. (1965). A feasibility study for strengthening timber beams with fiberglass, Canadian Society of Agriculture Engineering. Triantafillou, T.C. (1997). Shear reinforcement of wood using FRP materials, Journal for Materials in Civil Engineering, ASCE, 9(2), Triantafillou, T. and Deskivic, N. (1992). Prestressed FRP sheets as external reinforcement of wood members, Journal of Structural Engineering, 118(5),