Bonding of Carbon Fibre-reinforced Plastics with Wood

Transcription

1 Bonding of Carbon Fibre-reinforced Plastics with Wood René STEIGER Dr. sc. techn. / civil engineer Swiss Federal Laboratories for Materials Testing and Research EMPA Wood Laboratory Dübendorf, Switzerland René Steiger, born Civil engineering degree 1985, PhD 1996 at the Swiss Federal Institute of Technology ETH Zurich structural engineering. Since 1999 researcher at the Wood Laboratory of the Swiss Federal Institute for Materials Testing and Research EMPA. Aim and summary of the presentation The popularity of timber structures has increased amazingly over the last years. Several (partly even spectacular) timber structures represent real milestones. This trend is based on different causes. One of the most important factors is the spreading use of laminated wood, especially of glued laminated timber, which helped to break several natural limits of massive timber, such as dimensions and homogeneity. Connections have always played a major and often limiting role in timber structures. Investigations and developments in this sector therefore have always had and will still take an enormous influence on timber structural work. For example in the second half of the 19 th century, engineered timber construction could only establish itself because of the use of steel elements to realize connections of high performance (high strength and stiffness). Some years ago different researchers focused their work on the use of high strength fibres (HSF) for timber structures. The fibre s high strength and stiffness, their corrosion resistance, their light weight and their low thermal conductibility made these products appear very promising to apply in timber structures. In 1994 the strengthening of glued laminated beams by HSF was patented in the United States of America. From the beginning the Swiss Federal Laboratories for Materials Testing and Research EMPA have played an important role in the research on the use of HSF to strengthen and retrofit mainly concrete and masonry but also timber structural elements. In 1990 EMPA s Wood Lab started a broad research program on the use of HSF laminates for joints in timber structures after having studied possible applications and developments. It was at that time decided to focus on connections not primarily because research on strengthening and retrofitting was already in progress at other research institutes but mainly because of the results of the application study which showed a high potential of the HSF products to be used in connections. The basic idea was to joint tensile elements by spliced plates of HSF laminates. A preliminary series of small, clear tensile test specimens was carried out as a kind of prototype testing. The tests were encompassed on different parameters like type of glue, thickness of glue line, length and edge distance of the HSF laminates, quality of the finish of the surfaces before gluing, etc. This first series, which showed the feasibility of jointing wood members by means of HSF laminates, was followed by a second one, focused on bigger sized structural elements. The study consisted of tensile and bending tests. The influence of climate (temperature and relative humidity) and of long-term loading (creep) was evaluated as well for a reduced sample size. The main results of this second part of the research were: The strength of the joint tends to be higher with increased lap length, but the average shear stress at failure is due to the characteristics of the shear stress distribution along the joint clearly smaller with the long laps. The stress concentration at the lap s ends, which reduces the strength significantly, may be decreased by a stiffer laminate (which thereby increases the joint strength) and by enlarging the width of the groove in the wood near the joint. Also the compensation of the transverse tensile forces by beech plywood reinforcement increased the joint strength markedly. Number and spacing of the laminates have a significant effect on the strength and failure mechanism. A close spacing increases the joint strength but the tensile strength may then become the limiting factor. Splice plates close to the surface of the specimen efficiently reduce the peeling off effect. The reduction of strength of nine specimens, which were exposed more than one year to high humidity and changing climatic influences proved to be minor.

2 The strength of the bending specimens was very satisfactory (MOR between 30 and 34 N/mm 2, referring to the gross section) and the increase of deflection due to the splices was merely 6 to 12%. A specimen subjected to a long-term bending test with a bending stress of 22.5 N/mm 2 increased its deflection within one year for about 20%, showing a gradual stabilization of the creep movement. The recent test series aimed to improve the gluing of the HSF laminates on wood and to obtain more information on temperature resistance and creep. Commercial (mainly epoxy-type) adhesives, as well as products modified for this particular application were tested. A first evaluation of low viscosity adhesives was performed with torsion oscillation tests. The primary parameter of interest was the glass transition temperature. The tests confirmed the specifications of the adhesive producers, indicating a glass transition temperature between 40 and 55 C for epoxy type resins cured at 20 C. This value can be increased by thermal activation (post-curing at 80 C for four hours), which would allow a higher permissible service temperature. The adhesives selected on the basis of the torsion tests were then subjected to tensile tests. Both the torsion oscillation tests and the tensile tests have shown that at a temperature above 50 C the shear modulus of adhesives containing a filler drops sharply. In this higher range of temperatures adhesives without filler have a higher shear modulus, and accordingly they also show no temperature related drop of strength. Above 70 C none of the glues tested demonstrated sufficient strength and stiffness. Therefore, the bestadapted glue for a specific application must be selected in function of the expected temperature range in service. The duration of the long-term tests was limited to max. 24h. The results still gave first indications of the application range of the tested adhesives. Adhesion problems occurring at 50 C between the wood and some glues, could not satisfactory be resolved by applying a resorcinol primer.

3 Abstract In 1990 EMPA s Wood Lab started a broad research program on the use of high-strength fibre (HSF) laminates for joints in timber structures after having studied possible applications and developments. The basic idea was to joint tensile elements by spliced plates of HSF laminates. The tests focused on spruce, since this species is most used in central Europe and especially in Switzerland. A preliminary series of small, clear tensile test specimens was carried out as a kind of prototype testing. The tests were encompassed on different parameters like type of glue, thickness of glue line, length and edge distance of the HSF laminates, quality of the finish of the surfaces before gluing, etc. This first series, which showed the feasibility of jointing wood members by means of HSF laminates, was followed by a second one, focused on bigger sized structural elements. The study consisted of tensile and bending tests. The influence of climate (temperature and relative humidity) and of long-term loading (creep) was evaluated as well for a reduced sample size. The recent test series aimed to improve the gluing of the HSF laminates on wood and to obtain more information on temperature resistance and creep. Commercial (mainly epoxy-type) adhesives as well as products modified for this particular application were tested. A first evaluation of low viscosity adhesives, was performed with torsional vibration tests. The primary parameter of interest was the glass transition temperature. The tests confirmed the specifications of the adhesive producers, indicating a glass transition temperature between 40 and 55 C for epoxy type resins cured at 20 C. This value could be increased by thermal activation (post-curing at 80 C for four hours), which would allow a higher permissible service temperature. The adhesives selected on the basis of the torsional vibration tests were then subjected to tensile tests. Both the torsional vibration tests and the tensile tests have shown that at a temperature above 50 C the shear modulus of adhesives containing a filler drops sharply. In this higher range of temperatures adhesives without filler have a higher shear modulus, and accordingly they also show no temperature related drop of strength. Above 70 C none of the glues tested demonstrated sufficient strength and stiffness. Therefore, the bestadapted glue for a specific application must be selected in function of the expected temperature range in service. The duration of the long-term tests was limited to max. 24h. The results still gave first indications of the application range of the tested adhesives. Adhesion problems occurring at 50 C between the wood and some glues, could not satisfactory be overcome by applying a resorcinol primer. Keywords: Bonding, epoxy, carbon fibre reinforced plastics, connection, joint, strengthening, reinforcement, bending, tension, high strength fibre laminates 1003

4 1. Introduction Economic design of large span and/or heavy loaded structures demands the use of high-performance materials, which means not only high absolute strength and stiffness but also a high ratio of strength and stiffness to self weight (fig. 1). Carbon fibres Glass fibres Aluminium Timber C40 Timber C24 Prestressing cables Prestressing bars Steel S460 Steel S Virtual specimen length for tension failure due to self weight [m] Fig. 1: Specific material strength (ratio of tension strength to density [m]) New materials based on high-strength fibres (HSF) of glass, carbon, boron, aramide etc. have been developed. Due to the high costs, the use of HSF-laminates first was mainly focused on retrofitting and strengthening of existing timber constructions. The continuously decreasing price of these high-tech materials made the new technology more economical and more interesting and opened new perspectives. The Swiss Federal Laboratories for Materials Testing and Research EMPA have played a major role in developing structural applications of HSF and especially of carbon fibres (CF) for many years [1, 2]. One of the first applications was the reinforcement of existing concrete and masonry structures by carbon fibre-reinforced plastics (CFRP). Since the combination of fibre-reinforced plastics (FRP) with wood is of high interest as well, in 1992 the EMPA Wood Laboratory launched a program for applying FRP in timber structures. The present paper provides an overview on the many activities of this program with special focus on the recent test results. 1.1 Mechanical properties of FRP Regarding the possibilities of applying and combining different materials, it is useful to compare their most important characteristics. Figure 2 [3] shows the orders of tensile strength and modulus of elasticity of some materials often used for building tasks. It is to note, that the comparison is rough, since timber exhibits considerable variation in mechanical properties and especially shear and tension strength perpendicular to the grain are very low. Since CFRP provides better performance compared to GFRP and since EMPA at that time had already successfully applied CFRP to strengthen concrete and masonry structural elements, it was decided to mainly concentrate the research on the use of CFRP in timber structures Tensile strength [N/mm 2 ] Spruce II to the grain PVC GFRP CFRP Concrete Steel S235 Modulus of elasticity [kn/mm 2 ] Spruce II to the grain PVC GFRP CFRP Concrete Steel S235 Fig. 2: Mechanical properties (tensile strength and modulus of elasticity) of different materials

5 1.2 Possible applications of FRP in timber structures The first step in research consisted of the evaluation of possible applications and investigations concerning the combination of FRP and other high strength materials with timber. The study [4] aimed to find and characterize problems in timber construction, where the use of FRP might be promising basically study the (dis)advantages of the combination of different materials. It was found, that there is a wide range of application of FRP in timber structures (fig. 3) and that further studies should concentrate on the evaluation of suitable synthetic resins and fibres tension tests on small specimens with FRP joints the fabrication of 1:1 prototypes to get an idea on the feasibility the behaviour of the FRP joints under long-term loads and in different temperature ranges the sensitivity of the FRP application with regard to adhesive quality (precision of resin / hardener mixture) and workmanship (laboratory and practical conditions!). beam or truss elements plates connections steel squared timber eng. wood prod. glulam timber + material FRP AFRP GFRP CFRP conventional composite section reinforcement prestressed (re)strengthening AFRP: CFRP: FRP: GFRP: Aramide Fibre Reinforced Plastic Carbon Fibre Reinforced Plastic Fibre Reinforced Plastic Glass Fibre Reinforced Plastic local global shear compression perp. to the grain tension perp. to the grain connection bending Fig. 3: Possibilities of making use of FRP in timber engineering Application 1: Reinforcements [5] Upgrading structures for higher loads or restoring original strength has always been an important engineering task for structures of any material. Before FRP were available, steel was mostly used for such purposes. The bonding of steel plates onto concrete was developed very successfully at EMPA in the seventies. In the early eighties the steel plates were substituted by CFRP. Today the technique is well-established and has been used successfully on approximately 400 structures world-wide. The continuously decreasing prices of HSFlaminates made the technology more economical and more interesting in the meantime. The main advantages of using CFRP-laminates rather than steel plates are their light weight and their corrosion resistance, as well as their flexibility, which allows convenient and easy transport on rolls to the application or building site. It was very tempting to use this material on timber structures as well. A considerable number of timber structures (made of solid timber and of glulam) have already been successfully reinforced with FRP, mostly in North America, but in Europe too. Figures 4 and 5 show examples realised in Switzerland. The FRP-reinforcement consists of stripes/plates/rods of HSF embedded in a polymer matrix and of a bonding agent (glue, mortar or casting compound). The FRP lamellas can be installed either inside the timber member, which presents a number of difficulties for the realisation, but has important advantages (invisibility, protection against fire) or outside, which is the most common way when reinforcing/retrofitting existing structures. 1005

6 The bonding of the laminates has to be prepared very carefully: The wood surface must be flat, clean and not weathered, to insure proper adhesion. The wood surface must be dry at the time of gluing. The gluing interface of the FRP-lamellas must be degreased and clean. The application of the lamellas must take place at a reasonable temperature (> 10ºC) and there must not be any stresses on the glue lines during the curing of the bond. An extended curing time may be necessary at low temperatures and can be very cumbersome when used on important traffic ways. To overcome this problem an accelerated heat curing method has been developed and used with good success. By means of a special generator direct current is applied between the ends of the lamellas, in order to obtain a permanent curing temperature of approximately 70ºC. Thus the curing time can be reduced to about 2 hours. Fig. 4: Retrofitted bottom chords of the Aare river bridge at Murgenthal (Switzerland) Fig. 5: Pillar in a historical building, strengthened perpendicular to the grain Application 2: Connections For timber structures the design of the joints plays a decisive role and is a challenge for innovative trials especially regarding the transfer of tensile forces. The use of FRP-laminates appears to be a very good possibility to face problems like local stress peaks and multiaxial stress situation. Compact joints of high stiffness can be realised. The thermal conductivity of HSF-laminates is much smaller than that of steel. When the HSF-laminates are protected by timber, the fire resistance of the connection is good. Whereas joints subjected to compressive loads are rather simple to realise, tensile joints have always to be optimised regarding strength, stiffness and geometrical dimensions. It was therefore decided to study a possible use of FRP-laminates based on the testing of specimens with joints subjected to tension forces. Gathered knowledge when studying various parameters can be easily transformed to the application of FRPlaminates as reinforcing elements as well. All tests reported in the subsequent sections were carried out on Spruce specimens!

7 2. Preliminary tests with small tensile specimens [6] As a first step of experimental research and development a number of tensile tests on small clear specimens jointed with FRP splices (fig. 6) were carried out as a kind of prototype application to prove the feasibility and to identify possible difficulties. At this stage mainly carbon fibres were used as a reinforcing material. The tests encompassed a variety of parameters: type of glue (epoxy resins) thickness of glue line ( mm) length of FRP lamination (15 90 mm) stiffness (thickness) of FRP lamination ( mm) edge distance ( mm) quality of the finish of the surfaces of the wood and the FRP laminates before gluing plywood reinforcement glued to the sides of the specimens to counteract secondary transversal stresses. The strength of the connections proved to be quite high for most of the specimens, i. e. with most of the test parameters. In some cases the shear strength of the timber (more than 10 N/mm 2 ) was reached, which means that higher strengths could only be reached with a larger shear area. It was shown that the shear strength depends on the length of the bond line (fig. 7). With increasing lap length the shear strength of the joints decreased due to the uneven distribution of shear stresses along the bond line showing peaks at both ends (fig. 8). This phenomenon is well known from similar situation in lapped joints. Due to the low number of specimens compared to the high number of test parameters, the results were statistically not relevant but rather showed trends of influences. Fig. 6: Preliminary tests with small and clear tensile specimens (Example VZ 1.3.3: b = 25 mm, h = 30 mm) Shear strength [N/mm 2 ] Lap length [mm] τ max τ max τ mean τ mean Fig. 7: Shear strength in the bond line for different lap lengths Fig. 8: Distribution of shear stresses for different lap lengths and equal stiffness E A of jointed members l 1 l 2 > l

8 Finally the stress distribution was analysed by means of a finite element model [7]. This model was also used to perform thermal calculations of specimens, which were tested in tension at different temperatures (fig. 9) Deformation [mm] Ultimate load Strength [kn] Deformation at ultimate load Temperature [ C] Fig. 9: Influence of temperature on the tensile strength of the joints The gathering of experiences concerning the fabrication of the joints, respectively the test specimens was of equal importance. The fabrication proved to be quite delicate and a laborious work: A number of factors affect the strength and the quality of the glue line: Accuracy of the mixing ratio of the resin/hardener components, finish of the surfaces of the wood and the FRP-laminates before gluing, curing conditions etc. Of considerable importance is the viscosity of the (fresh) glue: Low viscosity resins show compared to higher viscosity resins a much better penetration into the joints and the voids, which leads therefore to a better quality glue line. The shrinking of the glue during the curing is on the other hand higher with low viscosity resins and their strength is lower. Since gluing deficiencies are very difficult to detect, the danger of a low quality glue line of a high viscosity resin should not be underestimated. The results of the investigations demonstrated the feasibility of the FRP-spliced joints for timber structures. It was at that time however, indispensable to deepen the understanding of the behaviour of this type of joint and to clarify the influence of some more parameters like differential expansion between wood and FRP (due to shrinking, swelling or temperature). 3. Tensile and bending tests with glued-laminated specimens [3, 8] 3.1 Tensile tests The subsequent test series consisted of over one hundred tensile tests on bigger size glued laminated spruce specimens having cross sections up to 110 x 110 mm (fig. 10). Since the strength of the laminates was actually much higher than the transmittable shear force, GFRP- instead of the more expensive CFRP laminates were used. This change however required another optimisation of the embedding resin. An epoxytype adhesive proved to be suited best. Similar test parameters as in the preliminary series were studied: type of resin, length of connecting plate, stiffness (E A) of plates, reinforcement of the cross section (to compensate for transverse tension), shape of grooves for the plates, climate exposure, etc. In addition the number and the spacing of the plates was a new parameter. The main aim of this series was the geometrical optimisation of the joint, together with the reduction of secondary tensile stresses perpendicular to the grain. Three to five plates with lap lengths between 50 and 120 mm and with different spaces and arrangements were used (fig. 11). The performance of the joint generally increased with increasing number of plates. A distance of 20 mm between the laminates and an anchorage length of 100 to 150 mm proved to be optimal. With such geometry loads up to almost 250 kn could be transmitted in the joint, representing a tensile stress of 22.7 N/mm 2 with respect to the gross cross section. With respect to the net cross section, tensile strengths up to 40 N/mm 2 were measured, representing a stress level, which for most spruce timber elements in practice is sufficient. An optimally designed joint is characterized by shear strength of the adhesive to be higher than that of the wood in the contact zone and by exhausting both the shear and the tensile strength of the timber.

9 L 1 L 2 B H L1 = 1300 mm L2 = 1000 mm B = 110 mm H = 100 mm Fig. 10: Tensile tests on glued-laminated GFRP-spliced specimens in structural sizes Type Lap length l 1 50 mm 2 70 mm 3 90 mm mm Type 5 Type 6 Type 7 Type 8 Type 9 Type 10 Type 11 Type 12 Type 13 Type 14 Type 15 Type 16 Type 17 Type 18 Type 19 Geometrically optimised specimen Tested configurations/parameters: Lap lengths: l: 50, 75 and 115 mm Lap distances a: 16, 18, 20, 24, 26, 30 mm Lamellas: GFRP, t = 2.8 mm E A = 5.8 to 6.2 MN GFRP, t = 3.7 mm E A = 9.8 MN GFRP, t = 4.7mm E A = 16.4 MN Steel S 235, t = 4.0 mm E A = 84 MN Thickness of bond line: 0.70, 1.00, 1.10, 1.15, 2.10 mm Fig. 11: Geometry of the tensile test specimens 1009

10 In accordance to the tensile tests on small specimens (see 2.) the joint s strengths tended to be higher with increased lap length, but the average ultimate shear stress is due to the characteristics of the shear stress distribution along the joint clearly smaller with the long laps. This shear stress distribution, together with transverse tensile forces and the shear stress concentration at the joint govern the failure mechanism. The stress concentration could be decreased by using stiffer laminates (E A) which thereby increased the joint strength. The stress concentration could also be reduced by enlarging the width of the groove in the wood near the joint. A chamfer of 15 mm length and 1 mm width (fig. 10) increased the strength of the joint by as much as 25%! Increased strength was obtained also by reinforcing the specimens with beech plywood to absorb the transverse tensile forces (fig. 11, type 6). Another efficient method to reduce the transverse tensile forces and thereby increase the strength of the joint is to augment the number of the splice plates. A close spacing increases the strength of the joint but the tensile strength may become the limiting factor. Splice plates close to the surface of the specimen efficiently reduce a peeling off effect. In order to study possible weakening of the joints caused by secondary stresses due to differential shrinking and swelling provoked by changes in MC, long term tests (1 year) were carried out under several climatic situations: wet climate: 20 C and 95% relative humidity of air changing climate: 6 weeks intervals of 20 C and 95% / 35% relative humidity of air outdoor, shelter: natural cycles of temperature and humidity of air. Compared to the specimens tested with a moisture content of 12% and not subjected to any ageing, the shear strength was about 15% lower. Particularly the severe test with changing climate did not provoke a significant loss in shear strength. 3.2 Bending tests Timber structures in practice often make use of beam elements. Joints of beams have to be stiff in order to prevent deflection. Hence short and long term 4-point bending tests on beams with cross sections of 120 x 180 mm and lengths of 2.80 m spliced at the centre (fig. 12) were carried out. Three different lap lengths (fig. 13) of the splice plates were tested. The geometrical design and the fabrication of the joints were based on the knowledge gathered from the tensile tests. Fig. 12: Bending test configuration Width of slot: 6 mm Lap lengths: l1 to l3 = 75 mm l4 = 50 mm l5 = 115 mm Type 1 Type 2 Type 3 Type 4 Type 5 Fig. 13: Geometry of the bending test specimens

11 The achieved strengths were satisfactory for all three splice lengths. The MOR was between 30 N/mm 2 and 34 N/mm 2 (referring to the gross cross section), which is an adequate level when designing beam elements made of glulam. The increase of deflection of the beams due to the splice was merely 6% to 12% compared to specimens without any joint, i. e. the stiffness of the splice proved to be remarkable. A first exploratory long-term bending test on a specimen with joint configuration 2 according to figure 13 was carried out by increasing the bending stress up to 27.5 N/mm 2 over time intervals of 28 days (first step) and 7 days (second step). At a stress level of 25 N/mm 2 the deflections increased significantly. In a second long-term creep test a beam having a maximum bending stress of 22.5 N/mm 2, which is about 2x the allowable stress, increased its deflection within 26 months for about 20%, showing a gradual stabilization of the creep movement. After a time span of 22 days without any load, the beam was subjected to an increasing short term bending load. A stress level of 33.4 N/mm 2 was reached, showing no reduction in bending strength compared to the specimens tested only under short term loads. The level of the long term load was quite high but there was no change in climatic conditions. Based on a rough extrapolation of the creep curve (fig. 14) creep factors of 1.0 for a time span of 100 to 200 years have to be taken into account. 20 Span: 2.50 m elastic deformation: mm Deformation w 1 [mm] Reference length: 1.0 m elastic deformation: 2.69 mm Power failure Deformation w 2 [mm] Days Fig. 14: Creep deformation under long-term load (22.5 N/mm 2 ) during 800 days (w 1 and w 2 refer to fig. 12!) 4. Evaluation and optimisation of different adhesives [9] The next step aimed to optimize the bonding of CFRP on wood and to evaluate the temperature resistance and creep characteristics. Commercial adhesives as well as products modified for this particular application (table 1) were evaluated with special regard to costs, process ability (working life, viscosity, toxicity, curing temp.), strength and stiffness of the joint, creep characteristics and influence of temperature (up to 80 C). Table 1: Evaluated original products and modifications Product Type Components Filler (original) Modification A Epoxy 2 no B Epoxy 2 no C Epoxy 2 no D Polyester 2 no E Epoxy 2 no F Epoxy 2 no G Epoxy 1 yes H Epoxy 1 yes I Epoxy 1 yes None C-Mod1 Epoxy 2 no Quartz powder: 0.78 (GResin + GHardener) C-Mod2 Epoxy 2 no Quartz powder: 1.04 (GResin + GHardener) D-Mod Polyester 2 no Quartz powder: 0.78 (GResin + GHardener) E-Mod Epoxy 2 no Quartz powder: 0.78 (GResin + GHardener) 1011

12 4.1 Torsional vibration tests A first evaluation of low viscosity adhesives was performed on base of torsional vibration tests according to DIN The primary parameter of interest was the glass transition temperature, i.e. the temperature at which the glue transits from the hard-elastic to the soft-elastic state, as well as the temperature at which the shear modulus (storage modulus G ) starts to drop off (fig. 15) Dropping of G' at: 90 C Shear modulus G': 335 N/mm 2 Shear modulus (G') [N/mm 2 ] Dropping of G' at: 43 C Shear modulus G': 784 N/mm 2 Shear modulus of epoxy hardened at 23 C Shear modulus of epoxy hardened at 80 C Temperature [ C] Fig. 15: Shear modulus and glass transition temperature of an epoxy-type adhesive derived by a torsional vibration test The tests confirmed the specifications of the adhesive producers, indicating a glass transition temperature between 40 and 55 C for epoxy type resins cured at 20 C. This value could be increased by thermal activation (post-curing at 80 C for four hours) (fig. 16). The glass transition temperature of not filled products was higher than that of filled ones. The storage modulus G of not filled products was lower than that of the filled ones (fig. 17). Some products/modifications exhibited higher storage modules G when post-cured at elevated temperature, most however did not. For temperatures below T = 50 C, the storage modulus G of filled products/modifications was significantly higher than that of not filled ones. G drops sharply above T = 50 to 60 C. Post-cured at T = 80 C during 4h, four products/modifications could be found with higher glass transition temperature (+41 to +55%) and only ignorable loss of G -modulus (-15 to -19%) Cured 7d at RT Cured 1d at RT, 4h at 80 C Cured 7d at RT Cured 1d at RT, 4h at 80 C [ C] G'-Modulus [MPa] A B C D E F G H I C-Mod1 C-Mod2 D-Mod E-Mod filled products / modifications Fig. 16: Glass transition temperature of evaluated products/modifications derived by torsional vibration tests 0 A B C D E F G H I C-Mod1 C-Mod2 D-Mod E-Mod filled products / modifications Fig. 17: Storage modulus G of evaluated products/ modifications derived by torsional vibration tests

13 4.2 Tensile tests Subsequently tensile tests on small clear specimens (fig. 18) were carried out, in order to evaluate the performance of adhesives/modifications selected on base of the torsional vibration tests. The shape of the specimens was optimised in order to force failure to the joint [10]. The tensile test series aimed to get information on the aptitude and process ability of adhesives evaluated by torsional vibration tests strength/deformation at different temperature levels (20 C, 50 C, 60 C, 70 C, 80 C) effect of loading time (short time (2 min.), ramp load and permanent load during 0.5h, 16h, 21h) effect of curing (cured 7d at RT, cured 1d at RT and post-cured at 80 C during 4h) effect of primers to improve adhesion influence of lamella s stiffness. l l 1 Specimen: Glue line: Lamellas: l = 600 mm b = 30 mm t = 25 mm l 1 = 30 mm b = 25 mm t = 1 mm t1 = 1 mm t2 = 2.8 mm Groove to centre the lamella CFRP lamella T700 E = 152 GPa ft = 2300 MPa εu = 1.51% CFRP lamella T300 E = 109 GPa ft = 1115 MPa εu = 1.05% Reinforced clamping zone to prevent specimens from being crushed perpendicular to the grain Fig. 18: Tensile tests on small clear specimens to optimise materials stiffness and evaluate best adhesives 4.3 Results Influence of lamellas stiffness The tests performed at ambient temperature showed that the stiffness of the materials transmitting the loads, i.e. wood, CFRP and the bonding agent, are of decisive influence for the overall strength of the joint. The stiffness ratio timber/glue/lamella needs to be balanced. Adhesives and reinforcements of greater stiffness provided better test results. Considerably higher strength of joints on base of T700-lamellas where found due to better adhesion glue/lamella and optimised stress distribution at contact zones (fig. 19) Effect of fillers All over the tested temperature range filled adhesives had higher strengths than such without any filler (fig. 20). Joints with filled adhesives exclusively exhibited timber failure. 1013

14 4.3.3 Influence of temperature (fig. 21 and 22) Compared to RT all adhesives exhibited a certain loss of shear strength when tested at T = 50 C. At temperatures up to T = 50 C filled adhesives performed better than not filled ones regarding strength and stiffness. Above T = 70 C none of the tested products and modifications exhibited sufficient strength and stiffness. At temperatures above T = 50 C frequent internal failure of filled adhesives occurred Shear strength [N/mm 2 ] Mean (T700) = 9.69 N/mm % Mean (T300) = 6.45 N/mm 2 Shear strength [N/mm 2 ] Mean (T700, filled) = N/mm 2 Mean (T700) = 9.69 N/mm 2 +15% Mean (T700, not filled) = 8.99 N/mm 2 Mean (T300, filled) = 8.43 N/mm 2 Mean (T300) = 6.45 N/mm 2 Mean (T300, not filled) = 5.66 N/mm 2 +50% A C D E F G H I E-Mod Short time, RT, T300 Short time, RT, T A C D E F G H I E-Mod filled products / modifications Short time, RT, T300 Short time, RT, T700 filled products / modifications Fig. 19: Influence of lamellas stiffness at RT Fig. 20: Effect of fillers at RT Shear strength [N/mm 2 ] A C D E F G H I E-Mod filled products / modifications Mean (T700, 20 C) = 9.69 N/mm 2 Mean (T700, 50 C) = 8.69 N/mm 2 Mean (T300, 20 C) = 6.45 N/mm 2 Mean (T300, 50 C) = 5.61 N/mm 2 Short time, RT, T300 Short time, RT, T700 Short time, 50 C, T300 Short time, 50 C, T700 Fig. 21: Shear strength at RT and T = 50 C -10% -13% Shear strength [N/mm 2 ] C (T700) E (T700) I (T700) Temperature T [ C] Fig. 22: Influence of temperature on shear strength Creep characteristics No significant creep deformations and loss in strength for permanent load applied during max. 16 to 21h at T = 50 C were found (fig. 23). Sometimes even slightly higher strength of joints under constant load at elevated temperature could be observed due to interlacing and reduction of stress concentrations Shear strength [N/mm 2 ] A C D E F G H I E-Mod filled products / modifications Mean (T700, short time) = 8.69 N/mm 2 Mean (T700, long time) = 7.97 N/mm 2 Short time, 50 C, T700 Long time, 50 C, T700 Fig. 23: Shear strength at T = 50 C for different loading times Shear strength [N/mm 2 ] A C D E F G H I E-Mod filled products / modifications Fig. 24: Effect of post-curing (4h at 80 C) Short time, RT, T700, not post-cured Short time, RT, T700, post-cured Short time, 50 C, T700, not post-cured Short time, 50 C, T700, post-cured Long time, 50 C, T700, not post-cured Long time, 50 C, T700, post-cured Effect of post-curing No significant effect on strength and stiffness of the joints was observed for short time and long time loading tests (fig. 24).

15 4.3.6 Process ability Concerning the application in practice some very useful hints were found: Preparation of bonding surfaces is important. Strongly exothermic reacting products and such that are very sensitive on failures of mixture resin/hardener might cause problems in practical use. Depending on the application (casting/overhead work) products with adequate viscosity should be chosen. Curing at low temperatures (T < 15 to 20 C) might provoke incomplete hardening. The adhesion problems occurring at 50 C between the CFRP and some glues, could not satisfactory be overcome by applying a resorcinol primer. 5. Conclusions The tests proved, that efficient joints with high stiffness and load bearing capacity can be realised on base of FRP-laminations. In cases where the optical appearance and the corrosion resistance are of importance, the use of FRP-laminations might be a valuable alternative. Technically there do not appear to be major obstacles to the application of FRP-laminates; however applying these products in timber structures requires experience and a higher quality of workmanship than traditional splices and reinforcements. The investigations provided numerous results of quantitative nature, especially strength and stiffness values as well as of quantitative nature, regarding the suitability of materials, which were basically glues and FRPs. Problems of workability, which plays a decisive role for the quality of the splice, were also clarified. There actually are adhesives on the market, which guarantee a reliable, stiff and good bearing bonding of HSF laminates on wood. The products however should be chosen carefully with respect to the temperature range to be faced with in the real situation in practice. The bond between the epoxy resin and the wood, as expected, provided no problems. Satisfactory adhesion to the surface of the FRP-laminations, however, was not easy to achieve. (This problem could not satisfactory be solved by applying a resorcinol primer.) The quality of the filling of the joints depends greatly on the viscosity of the resin which, in turn, influences the volumetric shrinkage and through this the quality and strength of the filling. The anchorage length of the laminations is to be established assuming that the bonding agent is always stronger than the wood and that the maximum possible shear stress which can be transferred is equal to the shear strength of the wood. Thus internal failure of the adhesive must be prevented. Members repaired with FRP-lamellas should be kept dry, as water uptake by the wood will cause swelling and stresses in the bond. The moisture will also be trapped under the laminate and this could lead to decay of the wood. Therefore, additional measures should be applied to shield the repaired surface permanently from rain as well as from direct sunshine. 6. Outlook Further studies should be focused on: creep tests for loading time >> 24 h behaviour of epoxies at low temperatures (down to -5 C) monitored field-tests optimised application techniques. 1015

16 References [1] Meier U., Strengthening of Structures using Carbon Fibre/Epoxy Composites, Construction and Building Material, Vol. 9, No. 6, 1995, pp [2] Meier U., Post Strengthening of Buildings by use of CFRP Lamellas, Documentation D 0128, Schweizer Ingenieur- und Architektenverein SIA, Zurich, Switzerland, 1995, 107 pp. (in German) [3] Meierhofer U. A., Bending and Tension Jointing of Timber by use of High-Strength Fibre Material, Schweizer Ingenieur und Architekt, Vol. 117, No. 43, 1999, pp (in German) [4] Timmermann K., Meierhofer U. A., Fibre-reinforced Plastics in Timber Structural Systems. Investigations and Developments. Part 1: Status and possible Applications and Developments, Research Report No. 115/23 of the EMPA Wood Laboratory, Dübendorf, Switzerland, 1992, 36 pp. (in German) [5] Kropf F. W., Detail Design and Retrofitting / Restrengthening of weather-exposed Timber Structures, Proceedings of the VI. International Workshop on Urban Heritage and Building Maintenance Maintenance and Restrengthening of Wood and wooden Structures, Swiss Federal Institute of Technology ETH, Zurich, Switzerland, 2000, pp [6] Timmermann K., Meierhofer U. A., Fibre-reinforced Plastics in Timber Structural Systems. Investigations and Developments. Part 2: Preliminary tensile Tests with small Specimens with CFRP-spliced joints, Research Report No. 115/26 of the EMPA Wood Laboratory, Dübendorf, Switzerland, 1992, 48 pp. (in German with English summary) [7] Ehrler M., Experimental and numerical investigation of timber tensile joints with splice FRP-laminates. Diploma Thesis, Institute for Building Materials, Swiss Federal Institute of Technology ETH, Zurich, Switzerland, 1995, 60 pp. plus annex (in German) [8] Timmermann K., Meierhofer U. A., Fibre-reinforced Plastics in Timber Structural Systems. Investigations and Developments. Part 3: Tensile and bending Tests with glued laminated Specimens, Research Report No. 115/32 of the EMPA Wood Laboratory, Dübendorf, Switzerland, 1994, 80 pp. (in German with English summary) [9] Steiger R., Fibre reinforced Plastics in Timber Structural Systems. Investigations and Developments. Part 4: Load bearing Capacity and Deformation of a Tension Joint reinforced by CFRP-Lamellas Influence of Materials Stiffness, Duration of Load and Temperature, Research Report No. 115/44 of the EMPA Wood Laboratory, Dübendorf, Switzerland, 2001, 175 pp. (in German with English summary) [10] Abdulahi S., Eggers S., Hauser C., Timber tensile joint on base of FRP-splices. Report, Institute for Building Materials, Swiss Federal Institute of Technology ETH, Zurich, Switzerland, 1995, 63 pp. (in German)