ADVANCES IN ADHESIVE JOINING OF DISSIMILAR MATERIALS WITH SPECIAL REFERENCE TO STEELS AND FRP COMPOSITES
L.C. Hollaway School of Engineering – Civil Engineering, University of Surrey, Guildford, Surrey, UK
Email: L.Hollaway@surrey.ac.uk ABSTRACT Generally, it is relatively easy to obtain a high joint strength with most modern bonding systems if these are cured under ideal conditions. However, civil engineering construction joints generally are cured on site and are required for the lifetime of the structural member. This implies that the joint might not be as well constructed and cured as in a factory and, in addition, will be under harsh site environmental exposure. Consequently, the joint should be post cured on site and the free surfaces protected from the environmental influences. The paper discusses the advantages and limitations of a proposed partially cured carbon fibre composite prepreg and a compatible film adhesive both developed specifically for the construction industry. The system has been developed from one used in the aircraft industry. Two types of carbon fibre will be considered: an ultra high stiffness and a high stiffness carbon fibre. The technique is compared to that of a more conventional bonding two part system used in civil engineering. Two test techniques will be used to characterise the proposed structural material namely, the double shear butt test and a flexural test using the composite material as rehabilitation for an artificially degraded structural member. In the latter test, two geometric shapes have been used to investigate any advantage that may be gained in shaping the upgrading composite. The paper has shown that a partially cured prepreg manufactured in the factory, shaped geometrically on site, incorporating a film adhesive and applying a low elevated temperature cure under pressure to the prepreg and film adhesive has an advantage over other adhesives particularly when using ultra high carbon fibre composites but the strain to failure of these fibres is relatively low and must be carefully considered. Moreover, the thinness of the film adhesive will reduce to a minimum the ingress of any harsh environments onto its free surface; this will improve the durability of the adhesive over that of the two part adhesive. KEYWORDS Adhesives, advanced polymer composites, durability, retrofitting steel beams, double strap butt joints. INTRODUCTION Most structural adhesives used in construction harden by chemical reaction. Curing takes place within the bulk of the adhesive and adhesion occurs at the interface, Van der Waals forces contribute to adhesion as these are the normal attractions between atoms and molecules and chemical bonding, mechanical interlocking, diffusion, electrostatic and weak boundary layer all play a part in forming the joining of two adherends together. Epoxies are the best known and most widely used structural adhesives. There are only a few commercial epoxy resins but a wide range of curing agents, including amines and acid anhydrides, which can polymerise these resins. There are no volatiles forming on hardening and shrinkage is very low. Structural adhesives can be both cold cured and hot cure and all are cross-linked, this renders the polymer insoluble and infusible and these characteristics greatly reduce creep of the adhesive. The cold cured polymer will cure at room temperature and should be post cured at a temperature of 500C. To avoid brittle behaviour, using additives toughens most modern epoxy adhesives. All amorphous polymers have a glass-transition temperature (Tg), below this temperature they are relatively hard and inflexible and are described as glassy and above it they are soft and flexible and are then described as rubbery. It is unacceptable for adhesives to pass from one state to another during service. Most cold cured epoxy polymer resins will have a Tg of between 500C and 600C and therefore will soften at this temperature when exposed, for instance, to the sun’s rays.
Adhesive bonding relies upon the establishment of intermolecular forces between a substrate and the polymeric adhesive itself. Consequently, it is necessary to pre-treat the substrate of the adherends to enable the required surface properties to be achieved. This treatment will be different for different adherends, the FRP composites are highly polar and hence very receptive to adhesive bonding, whereas the metals and aluminium adherends will range from a physical to a chemical method. The former includes solvent degreasing, abrasion and grit blasting and the latter pre-treatment includes etching and anodising procedures and thus by definition causes chemical modification to the surfaces involved. Sometimes a primer solution is painted on to the surface of the substrate. Furthermore, concrete adherends will require their particular pre-treatment. It will be clear that in all cases the performance of the adhesive joint is directly related to the successful application of the pre-treatment and this in turn depends upon the quality of the surface characteristics of the substrate in terms of topography and chemistry. Furthermore, water is the substance which generally gives rise to the greatest problems in bonded joint durability; consequently, a thin adhesive is essential. It is relatively easy to achieve high initial joint strengths but a satisfactory performance in service over a number of years will require a careful selection of polymers, fabrication methods and a pre-treatment of the adherends. For the long term service of civil engineering bridge structural steels the main environments which cause serious loss of joint strength in civil engineering service are moisture and salt spray and therefore a pre-treatment must be considered in these cases. To achieve greater durability of the joint grit blasting in conjunction with a silane is often used. Silanes have been shown to enhance the durability of bonded steel structures, but compatibility between the adherend and the silane must be achieved. Gettings and Kinlock (1972), found that premixing with γ-glycidoxpropyltrimethoxysilane (γ-GPS) considerably improved the durability of grit blasted steel joints, whereas two other silanes did not. Primers, usually in the form of thin organic coatings, are sometimes used as an addition to pre-treatment of steels. Primers can provide several advantages. They usually (i) have much lower viscosities than the adhesives and can therefore achieve greater contact with the substrate, (ii) have greater interaction with the substrate and adhesive, (iii) can contain corrosion inhibitors, (iv) can protect the adherend’s surface until the bonding process is undertaken. This paper will concentrate upon advances in the bonding of composites to metallic structural members but with special reference to steels. It will be assumed that the steel adherend has received the relevant pre-treatment. Two examples where this technique is used are (i) the rehabilitation of steel bridges and (ii) the upgrading of steel structural buildings to enable present day traffic requirements on bridges or an increase in loading requirements of structural building to be realised. The general technique of upgrading steel uses advanced polymer composite in the form of pre-cast plates, such as pultrusion or preimpregnation (prepreg) systems, and these then require to be bonded, under site conditions, on to the structure with a cold setting adhesive. With this site fabrication method post curing of the adhesive polymer is not generally undertaken. This implies that the (Tg) of the adhesive is of the order of 500-550C; this is not a high value for temperate climates. However, some adhesives will post cure if the environmental temperature rises above their cold cure temperature but there is a maximum temperature rise (the environmental temperature in this case) above which no further post cure will take place. This post cure will raise the Tg to a few degrees above that post cure temperature value. It is recommended here that all adhesives should be post-cured. Another method that is used for the rehabilitation of structural members, such as RC columns, arched skeletal metallic structures and some structural beam members when certain areas of the member are difficult, physically, to reach with pre-cast sections is the wet lay-up system. In this method the matrix material also acts as the adhesive. The aerospace industry has been using for some time a factory made partially cured prepreg for patch repair. This system can be modified for use in construction where the curing cycle requires an elevated temperature which is acceptable to the civil engineering industry; this system has been used successfully to upgrade a curved-in-plan I section beam on the Boots Building Nottingham UK, Figure 1. The advantage of this material and the curing system is that an adhesive film is used in conjunction with the prepreg, both can be fabricated around the structural member of any geometrical shape and then compacted and cured in one operation; the system that will be discussed in this paper has a compaction vacuum of 1 atm. and cure temperature of 650C. The site compaction (for minimum voids), the film adhesive system (curing simultaneously with the matrix material of the composite for greater interpenetration contact) and site fabrication method (to provide the upgrading material with an ideal fit to the structure) implies that a high quality composite/bonding system has been employed. The Advanced Composite Group (ACG) Derbyshire, UK, has developed a prepreg carbon fibre composite for the aerospace industry, which has undergone satisfactory trials in that industry. ACG is now introducing this system into the construction industry and has developed, specifically for this industry, two matrices and a compatible film adhesive to be used in conjunction with these two matrices to form high modulus (H-M) and ultra-high modulus (UH-M) carbon fibre composites. This material is now one of the potential new bonding techniques for construction and could have
advantages over the two part adhesive for on site bonding procedures; these advantages and limitations are discussed in Section 3.
The aims of this paper are to introduce a possible new bonding system into the construction industry, it will focus upon comparing this system with a more conventional one; the paper will concentrate upon: (a) Metallic/FRP joint considerations using a FRP composite adherend and a steel adherend. (b) The effectiveness, the advantages and the limitations of two alternative composite/steel/adhesive bonding
systems. (i) Partially cured factory made FRP prepregs bonded on site to structural members with adhesive films under a low elevated temperature and pressure. (ii) Factory made pre-cast plates bonded to steel adherends with a conventional two part adhesive resin.
(c) Geometric shapes of FRP upgrading components, when utilising ultra high modulus and high modulus carbon fibre (UHMCFRP) and (HMCFRP) prepreg composites bonded to steel. The methods which are used to illustrate the two joining techniques [defined in item (b)] are: (i) the tensile strengths of bonded double strap butt joints where the cover straps are the FRP composites, (ii) the FRP bonded geometric shapes to rehabilitate flexural steel beams using adhesive film [the investigation of item (i) showed the film adhesive to be the most suitable].
METALLIC JOINT CONSIDERATIONS WITH PARTICULAR REFERENCE TO STEELS To date advanced polymer composite materials have been used to upgrade a small number of steel structures compared with those of reinforced concrete. In support of such steel structural applications, only a limited amount of research appears to have been conducted, for example, (Mosallam and Chakrabarti 1997, Lui et al, 2001, Tavakkolizadeh and Saadatmanesh 2003 and Photiou et al, 2006). Nevertheless, the high tensile strength and modulus of elasticity of carbon fibre polymer (CFRP) composites make them ideal candidates for upgrading steel structures, but it is necessary to appreciate the possible limitations of their mechanical properties, their interaction with the steel substrate and of their long-term behaviour in harsh environments. When analysing/designing a joining system there are a number of considerations that must be addressed. It has already been stated that it is relatively easy to achieve a high initial joint strength and this statement could be verified by undertaking a simple double lap joint as laid down in various experimental specifications. However, joints in civil engineering construction also require to maintain this high strength over long periods of time when various loading regimes are applied to it and when the joint is exposed to aggressive environments. Some of these considerations will be mentioned here to justify the configurations used in the joining systems. Thermal expansion A further problem when rehabilitating steel members with FRP material is the differential thermal expansion between the two elements. The FRP composite can lead to high adhesive stresses; these must also be considered
Figure 1 Final placement of the carbon fibre prepreg around flanges and web of beam at Boots Building Nottingham, UK. (By kind permission of Taywood Engineering London, UK. and ACG
in the adhesive joint. Adhesives with high curing temperatures may be unsuitable for some uses below room temperature because of large thermal stresses, which develop as the joint cools below the fabrication temperature. Effects of Bond Defects Defects in adhesive joints, which are of concern, include surface preparation deficiencies, voids and porosity, and thickness variations in the bond layer. Of the various defects which are of interest, surface preparation deficiencies are probably the greatest concern. These are particularly troublesome because there are no current non-destructive evaluation techniques, which can detect low interfacial strength between the bond and the adherends. Bond thickness variations usually take place in the form of thinning due to excess resin bleed at the joint edges. This leads to overstressing of the adhesive in the vicinity of the edges. Common aerospace practice involves the use of film adhesives containing scrim cloth, some forms of which help to maintain bond thicknesses. It is also common aerospace practice to use mat carriers of chopped fibres to prevent a direct path for access by moisture to the interior of the bond. This process is probably not relevant to civil engineering but the use if film adhesive is as will be shown later in the paper. Durability of the Adhesive Joints Some general observations on joint durability are summarised in Table 1, which give the main factors which have an influence on joint durability.
Table 1 Main factors affecting joint durability
Fatigue Resistance and Creep Characteristic of an Adhesive Joint Adhesive joints are more susceptible to fatigue failure than are the FRP strengthening material but they generally have superior fatigue characteristics to those of an equivalent riveted joint. A limited number of fatigue tests on FRP strengthened structures have been performed. For instance, Mertz et al (2001) and Miller (2000) used small scale fatigue tests and then continued with full size girders which had a high degree of initial corrosion; these structures showed no signs of fatigue damage or loss of stiffness after 106 cycles loading. All polymers will creep under load; FRP composites will creep less than un-reinforced polymers as the fibres in the former material have a stabilising effect on the polymer. Under sustained loading, continued creep will eventually lead to creep rupture of the adhesive and this will place a limit on the adhesive. For the majority of steel strengthening schemes the permanent load carried by the strengthening is low, however, if the composite FRP strengthening system is prestressed then the permanent load carried across the adjacent joint can be significant. In the latter system the FRP plate would be anchored at each of its free ends and consequently a portion of the permanent and superimposed load would be carried by the anchorage. In addition, as the rigidity of the composite in the principal direction is governed by the fibres, a composite with a high volume fraction of fibres will have relatively low creep as the fibres have almost zero creep. Stress Considerations of an adhesive joint It is generally assumed that the FRP rehabilitated steel beams will fail by either the adhesive strength being exceeded or by an inter-laminar shear strength failure of the FRP as this will generally be lower in value than the shear/inter-laminar shear strength of the metallic substrate, unless the latter has been degraded by corrosion. Two approaches can be used to predict the failure of adhesive joints, (i) an elastic stress based analysis and (ii) a fracture mechanics approach. The latter one examines the energy required for unstable crack propagation along the joint but the limitations of this approach currently is the geometry of the plate and temperature effects;
Factors Effects on Joints
Water and salt solution If high pH and soluble aggressive ions are absorbed by the adhesive they may plasticize and toughen it. Temperature (i) Elevated temperatures will increase the rate of (i) degradation, (ii) creep. (ii) Elevated temperatures may post-cure and may plasticize and toughen cured adhesives. Oxygen Can contribute to metallic corrosion and polymer degradation. The adhesive Poor interfacial contact implies poor preparation of metallic substrate, and induced air voids.Composition of adhesive Chemical types affect cured structure, bulk properties, interfacial composition and stability.Adherend - Metals Surface stability Adherend - FRP comps. Moisture content, permeability. Nature of primer (if applicable). Bonding conditions.
therefore, it has yet to be successfully applied to infrastructure strengthening applications, Cadei, et al (2004). The former analysis is generally used to evaluate the distribution of stress along the adhesive joint and is based upon an elastic characterisation of the adhesive layer. It is able to analyse beams with varying cross-section or FRP plates that taper along the beam; the analysis requires the results of lap shear tests. Several closed-form stress analysis are available that predict the distribution of bond stresses along a plate bonded to a beam, (Albat and Romilly 1999; Denton, 2001; Frost et al 2003; Miller 2000; Smith and Teng 2003). These publications assume that the adhesive is linear-elastic and involve a number of simplifying assumptions. The low-order solutions Deng et al 2004 are simpler in form but assume that the stress does not vary through the adhesive thickness and consequently does not satisfy the requirement that the adhesive stress at the free end of the adhesive is stress free. High-order solutions Yang et al 2004 model the variation in stress through the thickness of the adhesive layer, but are complex. Stratford and Cadei 2006 introduce a method for designing an adhesive joint (originally for strengthening metallic structures) between the strengthening plate and the beam. The method predicts stress distributions along the adhesive joint and can be used up to failure of the adhesive or the substrate in the immediate vicinity of the joint; these are failure modes typically associated with adhesive joints involving metallic or substrates. Adhesive joints are generally characterised by high stress concentrations in the adhesive layer, which originate in the case of peel stresses due to unequal straining of the adherends or because of the eccentricity in the load path. Considerable ductility is associated with shear response of typical adhesives, which is beneficial in minimising the effect of shear stress joint strength. Response to peel stresses tends to be much more brittle than that to shear stresses and a reduction of peel stresses is desirable for achieving good joint performance. Wherever possible the joint should be designed to ensure that the adherend fails before the bond layer. This poses a significant challenge to metallic/FRP composite joints in civil engineering, as the normal adhesives are inherently much weaker than the FRP composite or metallic elements being joined; one of the adherends, if not both, will be thick particularly when upgrading steel structural units, consequently, bond failure will invariably occur. STRUCTURAL SYSTEMS The structural systems discussed here investigate one unique method for upgrading steel structural members and compares the results of this method with those of a more conventional method. The results presented here have been derived from studies undertaken at the University of Surrey over a number of years. They consist of two areas: (i) The first being small scale tests to obtain the characteristics of bonding of two dissimilar materials using two techniques. (ii) The second being tests on scaled rectangular hollow cross section steel beams upgraded using the two techniques of item (i) and in addition two different geometrical shapes of the composite material. The method discussed in item 1 was first introduced in Photiou et al (to be published in ASCE Journal of Materials in Civil Engineering) and the method used in item (ii) was first introduced in Photiou et al (2006). THE DOUBLE STRAP BUTT JOINT. The two adhesive fabrication systems which will be utilised for the FRP rehabilitation of steel structural members will be. (i) The standard civil engineering thixotropic epoxy resin adhesive Sikadur 31, which is a two part adhesive material, (a product of Sika, Ltd. Welwyn Garden City, Herts), bonding a rigid FRP composite plate to the adherend. The mechanical properties of the adhesive are given in Table 2. (ii) The partially cured factory made epoxy film adhesive and prepreg composite material, developed by Advanced Composites Group Ltd. Derbyshire, (developed from an aerospace prepreg material). The upgrading system is to be fabricated and finally cured on site at a low elevated temperature of 650C under a pressure of 1 atm. The material properties are given in Table 3, and Figure 2 shows the test arrangement of the double strap butt joint. The two types of composite materials used are the HMCFRP and UHMCFRP composite prepregs, both fibre systems being uni-directionally aligned are impregnated with a low elevated temperature cure epoxy resin system (cure temperature 650C). In addition, a further adherend system consisting of one laminate of + 450 glass fibre/epoxy (GFRP) composite prepreg fabricated on to both sides of the two laminates of carbon fibre UHMCFRP/GFRP prepreg; one of the glass fibre surfaces eventually being bonded on to the steel surface adherend. The reason for laminating the GFRP composite prepreg onto the free surface of the composite system is to maintain a symmetric laminate and the reason for using the GFRP laminate adjacent to the adhesive is to allow for a more uniform stress transfer between the adherends and, in addition, to overcome any galvanic action that might take place. Table 3 gives tensile stress~strain characteristics for the GFRP prepregs. For the two-part adhesive, two thicknesses were analysed, namely 0.1mm, and 0.5mm; for this discussion only 0.5 mm thick adhesives will be considered. For site work it is not advisable to go lower than 0.5 mm for a two
part adhesive due to the difficulty of making a satisfactory joint. One layer of adhesive film had a thickness of 0.1 mm; the application of one or two layers was investigated (Photiou, et al 2003) but it was concluded that there is no benefit from having more than one film layer. The thin steel plates used in the strap joints had thicknesses of 3.0mm and 6.0 mm, with typical 0.2% proof stress of 700 N/mm2 and a limit of proportionality at approximately 550 N/mm2. Table 2 Properties of Sikadur 31 (Abstracted from the manufacturer’s data sheets
Figure 2 Line diagram for double strap butt joint
Preparation of the double strap butt joints tensile coupon samples The composite prepreg is stored at –200C immediately after its factory manufacture and is maintained at this temperature until required for use. Before cutting the partially cured prepreg samples to size, the material is allowed to thaw for approximately 2 hours. The designed layers of prepreg are placed onto the relevant mould, and a halar film and breather blanket are placed to covers the mould and composite and the whole unit is placed in a vacuum bag. The vacuum bag, under a vacuum assisted pressure of 1 atm., is transferred to an oven (at 650C) for 16 hours. After curing, the temperature of the oven and sample are reduced to room temperature at a steady rate. The flat plates are then cut to the required size in preparation for forming the joint using the two part adhesive. For the joints which use a film adhesive with the precast plates, these latter are placed on either side of the butt-jointed steel plates with the adhesive film between the composite plates and the steel. The joints are exposed to the same pressure and cure temperature as discussed above for the formation of the double lap butt joints. For those specimens which use a two part cold setting adhesive, this later is applied to both surfaces of the steel adherend and the CFRP/GFRP composites are placed in position, a special jig is used for accurate positioning.
Table 3 Stress~Strain Characteristics for CFRP and GFRP laminates. Material
0.37 For the joints which are made and bonded with adhesive film in one operation, the method for cutting and placing the partially cured prepreg is similar to the above but now the adhesive film is also cut to size and placed in position on the top and bottom of the steel adherends and the whole is cured in one operation as described above.
Results of double strap butt joints Tables 4 and 5 summarise the geometric properties, the failure loads and the displacements of the two types of adhesive materials used. The tables also list the first peel load, displacements and the failure modes of the specimens. Theoretically the adhesive layer should not be the weak link in the joint and wherever possible the joint should be designed to ensure that the adherend fails before the bond layer. However, in a steel/CFRP composite joint the adhesive is much weaker than the FRP composite or steel elements being joined as the thickness of the adherends are thick compared to the adhesive and therefore the bond stresses become relatively large until the bond failure occurs at a lower load than that for which the adherends fail. Consequently, in a well-bonded steel/FRP composite joint, failure should occur within the adhesive (cohesive failure) or within the adherend (FRP inter-laminar failure). Failure at the adherend-adhesive interface (interfacial or bond-line failure) generally indicates that a stronger bond should be made. In the tests under discussion, four failure modes were observed and are referred to as mode A (cohesive), mode B (interlaminar) and mode C (interfacial) and mode D (ultimate tensile failure of FRP). The categorisation is based on a visual examination of the failed specimens; an estimate was made of the percentage of each of the above criteria over the relevant surfaces with the lap lengths corresponding to 100%.
Steel thickness 3.0 mm A = Cohesive failure B = Interlaminar failure C = Interfacial failure D = Ult. Failure of FRP Four parameters for the double-strap joint will be discussed, these are (i) HM-CFRP bonded with Sikadur 31 epoxy adhesive, (ii) HM-CFRP bonded with adhesive film, (iii) UHM-CFRP bonded with adhesive film and (iv) UHM-CFRP/GFRP bonded with adhesive film. Tables 4 and 5 give the test results for the parameters in items (i) and (ii) above and it can be seen that joints with Sikadur 31 epoxy adhesive have lower average peel and failure loads, with values between 30.3 kN and 28.2kN, respectively, than joints with the adhesive film layer, which varied between 33.1 kN and 33.4 kN, respectively. The failure modes for the Sikadur 31 varied between cohesive, interlaminar and interfacial, whereas for the adhesive film, all specimens failed by the interfacial mode on both sides of the steel surface. Turning to the test results, in Tables 6 and 7, for the parameters in items (iii) and (iv) above, it will be seen that the thickness of the steel adherend had to be increased to 6 mm due to yielding of the 3 mm thick steel. With the 6 mm thick adherend steel it will be noticed that the failure mode in both cases is by the ultimate strength of the UHM-CFRP cover plate being exceeded. However, the composite cover plate which has the UHM-CFRP/GFRP laminates (the GFRP laminate being adjacent to the adhesive film), is able to carry 26% greater load than that carried by the UHM-CFRP laminate only. As the adhesive film is clearly able to support higher loads than the two-part adhesive the two latter tests were not carried out using Sikadur 31 adhesive. Further tests were undertaken, but not reported here, where deformations were measured over the whole tensile sample. These deformations showed that the maximum strains reached a value of about 3400 µε, which is a value comparable to the tensile failure strain of the UHM-CFRP composite. As both UHM-CFRP/GFRP and UHM-CFRP composites reached their ultimate strengths, it would indicated that the GFRP laminate provides a more gradual transfer of load between the two high moduli materials. In the design analysis of adhesively bonded joints a key parameter is the adhesive thickness. Analytically, it can be shown that the thicker the adhesive the better is the load transfer under shear, although overall joint stiffness is decreased. However, as the adhesive thickness increases, so does the likely occurrence of bond-line porosity; this decreases the shear and peel strengths markedly over the life span of the joint. In an earlier investigation two thicknesses (viz. 0.5 mm and 0.1 mm) of Sikadur 31 were used and it was observed that there was little difference between the performances of the two joints. Forming an adhesive bond line thickness, using a two
Specimen Ref. No
CFRP thickness each side
First Peel Load (kN)
Displace- ment – (mm)
Final Failure Load (kN)
Failure Mode 1st Side
Failure Mode 2nd Side
DSS 01 0.6 0.5 32.3 2.34 29.5 2.91 80% A- 20% B 80% A- 20% B
100% A 80% A- 20%B
DSS 02 0.6 0.5 29.5 1.71 28.9 2.50 80% A- 20 % B 90% A - 10 %B
80% A- 20% B25% A-50% B-
DSS 03 0.6 0.5 29.1 1.71 28.3 2.44 100% A 100% A
80% A- 20% B80% A- 20 %B
part adhesive, of 0.1 mm on site is difficult, consequently all such joints, in this paper, utilised a 0.5 mm joint thickness. Furthermore, the double strap butt joint in which HM-CFRP plate was bonded to the steel substrate using the film adhesive failed by interfacial failure which might suggest a poor surface preparation. However, great care was taken over all surface preparations and it must be concluded that as the surface preparation cannot be improved this is the true failure criteria for this type of construction material for the double lap joint.
Table 5: Double-Strap Joints HMCFRP bonded with Adhesive Film
Specimen Ref. No
CFRP thickness each side
per Side No
First peel load (kN)
Final failure load
DSFO 01 0.6 1 - - 36.4 2.24 100% C 100% C
100% C 100% C
DSFO 02 0.6 1 33.1 1.83 27.2 2.53 100% C 100% C
100% C 100% C
DSFO 03 0.6 1 - - 36.7 1.94 100% C 100% C
100% C 100% C
Steel thickness 3.0 mm A = Cohesive failure B = Interlaminar failure C = Interfacial failure D = Ult. Failure of FRP
Table 6: Double-strap Joints UHMCFRP bonded with adhesive film
Specimen Ref. No
CFRP thicknesseach side
per side No
First peel load (kN)
Final Failure Load (kN)
1st Side 2nd Side
DSFN01 0.6 1 - - 25.4 1.63 Mode D DSFN02 0.6 1 - - 29.5 2.34 Mode D DSFN03 0.6 1 - - 31.8 3.09 Mode D
Steel thickness 6.0 mm A = Cohesive failure B = Interlaminar failure C = Interfacial failure D = Ult. Failure of FRP
Table 7: Double-strap Joints UHMCFRP/GFRP bonded with adhesive film
Steel thickness 0.6 mm D = Ultimate failure of FRP THE REHABILITATED FLEXURAL BEAMS The aim of this part of the program was to restore an artificially degraded beam to its full strength/stiffness by upgrading it with UHM-CFRP/GFRP and HM-CFRP/GFRP by using the most advantageous bonding method. The failure criteria of the double strap butt joints given above showed that the best performing adhesive technique for a steel/composite join was by utilising a compatible adhesive film with respect to the polymer of the partially cured FRP prepregs. Two artificially degraded rectangular beams of dimensions 120 mm x 80 mm were rehabilitated by means of a U-shaped prepreg unit and another two beams by a prepreg flat plate unit. One of each of the geometric types was manufactured from an UHM-CFRP and from a HM-CFRP. In all cases the laminates were bonded to the tensile flange of their respective steel beams by a compatible film adhesive. Both the U-shaped and the flat plate unit had identical laminate lay-up material. The carbon fibres and the glass fibres used in the CFRP/GFRP prepreg laminates are aligned directionally and at +450 to the longitudinal direction of the beams, respectively. As the glass fibre composites have only 0.20 stiffness values of those of the CFRP composite and because of the glass fibre position on the soffit of the beam, they added little to the overall stiffness of the upgrading. The steel beam was characterised by a modulus of elasticity of 205 GPa and a 0.2% proof stress of 375 MPa. with no distinct yield stress or yield plateau. The beam was artificially degraded by machining 2.5 mm thickness of material from the soffit flange and all surfaces were grit blasted to the Swedish Code SA 2½ Grade 3 Dirk grit. Figure 3 a is a line diagram sketch of the U-shaped upgrading composite bonded to the rectangular beam and the upgrading flat plate, respectively. Preparation of composites for the rehabilitated beam The fabrication of the prepreg composites on to all steel beams commenced with a single GFRP prepreg ply placed onto the soffit of the beams and this was followed by one double ply CFRP laminate and by a further single ply GFRP prepreg. This was followed by another double ply CFRP laminate and a final GFRP prepreg. Only the GFRP prepregs of the U-shaped CFRP/GFRP composites were taken up the vertical sides of the beams
Specimen Ref. No
CFRP Thickness Each Side
Adhesive Layers Per
First Peel Load (kN)
Final Failure Load (kN)
1st Side 2nd Side
DSFNG1 0.6 1 - - 39.4 2.02 Mode D DSFNG2 0.6 1 - - 35.4 2.06 Mode D DSFNG3 0.6 1 - - 34.4 1.92 Mode D
to the height of the modified neutral axis; this vertical component was a continuation of the +450 GFRP composite. The UHM-CFRP and the HM-CFRP ply thicknesses were 0.3 and 0.6 mm, respectively. At the end of the laying-up procedure the laminates were de-baulked and following this operation a halar film and breather blanket was placed over the beam and the whole unit was placed in a vacuum bag. The bag, under vacuum assisted pressure of 1atm., was transferred to an oven at 650C. The curing operation followed the same procedure as for the tensile coupon specimens discussed above. It was anticipated that the U-shaped GFRP composite unit would prevent peel failure at the free end of the reinforcement or a peel failure following a strain failure of the carbon fibre composite. Figure 3(a) Line diagram sketch of U-shaped upgraded steel beam Figure 3(b) Line diagram sketch of test beam The test rig span for each beam was 1700 mm and a four point loading system was used with the two external loads being positioned at 200 mm on either side of the centre line of the beam; a line diagram is shown in Figure 3b. Strain gauges were positioned at strategic positions on the beam and two displacement transducers were placed on either side of the beam at mid span to measure the central deflection and to observe any torsional effects that might occur. Results for the flexural beams A summary of the results for the artificially degraded beams tested to failure is given in Table 8. The most significant result is that associated with the UHM-CFRP/GFRP beams. The steel reached 0.2% proof stress at about 23 kN per jack and thereafter deformed non-linearly up to failure of the composite. The two geometric types of UHM-CFRP/GFRP upgrading (viz. U-shaped and the flat plate composite sections) failed in the location of the pure moment region when the strain in the CFRP of 0.4% was reached. However, after the CFRP/GFRP composite reached its ultimate strain the U-shaped section composite had no apparent bond failure on either side of the failed composite and therefore was still able to provide a degree of stiffening. The ultimate load reached by the strengthened beam exceeded the plastic collapse load of the full steel beam and when the beam was unloaded it had a permanent deformation of 12 mm. On reloading the beam it was able to support a load in excess of the capacity of the steel beam alone indicating that the FRP system was still contributing to the strength of the beam and that the adhesive bond was still active. The load deflection path for beam 3 (upgraded with UHM-CFRP/GFRP composite flat plate) followed a similar pattern to that of beam 1 and reached an almost identical load to that beam, at which point failure occurred as a result of carbon breakage at 0.4% strain, but in this case the fibre breakage triggered a near complete debonding of the composite system from the steel flange. Unlike beam 1 no residual composite action remained between the FRP and the steel beam. Beam 2, which contained the U-shaped HM-CFRP/GFRP upgrading system, was loaded to 50kN/jack. The steel reached 0.2% proof stress at a load of about 23 kN/jack; beyond this load the beam continued to deform non-linearly. At 50 kN/jack the test was stopped due to an excessive deflection of 42 mm. The ultimate stresses of the CFRP or GFRP composites had not been reached and no failure of the composites occurred. The maximum load was in excess of the plastic collapse load of the original steel beam. Beam 4 containing the flat plate HM-CFRP/GFRP upgrading system behaved in a similar manner to that of beam 2.
3 No. 1 laminates of GFRP
1 No. double laminate of CFRP
120 mm x 80 mm artifically degraded beam.
2.5 mm thick degraded soffit
5 mm thick
W W 1700 mm
650 mm 650 mm400 mm
5 mm thick
Figure 8 Failure criteria and remarks for the CFRP/GFRP composite using film adhesive plated beams
Beam 1 UHM-CFRP/GFRP
U-shaped 45 kN/jack
Beam reached its 0.2% proof stress at 23 kN/jack with a defln. of 7.4 mm. A non-linear defln. up to 45 kN/jack, at which point carbon fibre strain failure of 0.4% but adhesive bond on either side of fibre failure remained in tact. On reloading (after 15 mins), beam sustained load of 40 kN/jack, which was in excess of steel beam alone.
U-shaped 50 kN/jack
Beam reached its 0.2% proof stress at 23 kN/jack with a defln. of 8 mm. A non-linear defln. up to 50 kN/jack, test stopped due to excessive defln. of 42 mm. No actual failure of FRP
Beam 3 UHM-CFRP/GFRP
Flat plate 45 kN/jack
Loading behaviour almost identical to beam 1. Failure occurred due to carbon fibre strain failure at 0.4% strain – complete debonding of FRP plate from steel beam - no residual composite action.
Beam 4 HM-CFRP/GFRP
Flat plate 50 kN/jack
An almost identical beam result to that of beam 2. No actual failure of FRP at 50 kN/jack – test stopped due to excessive deflection.
CONCLUDING REMARKS The paper has been concerned with the bonding systems which are available to joint FRP composites to steel structural member and then transferring this knowledge to ascertain how the more positive system could be incorporated to strengthen/stiffen a structural steel member used in construction. Attention was given to two specific systems, a two-part epoxy adhesive cold cured or at a low elevated temperature and an adhesive film which was cured under pressure at a low elevated temperature. The procedure was illustrated by an experimental analysis of a double butt joint using HM-CFRP and UHM-CFRP composites joining steel adherends. It was mentioned that to obtain an 0.1 mm thickness of adhesive joint on site with a two part epoxy adhesive is not practically possible, consequently, values quoted have been derived for joint thicknesses of 0.5 mm and 0.1 mm for the two part adhesive and the film adhesive, respectively. The joints with the adhesive film performed well when compared to the two part epoxy adhesive, with higher failure loads being achieved in the former joints. The failure load joint values for the HM-CFRP composite plates bonded with the film adhesive were higher than those of the two part adhesive. Similar higher failure load values occurred in the case for the UHM-CFRP composite plate/steel joint but it should be noticed that the comparison for the film adhesive joint is made with a 6 mm thick steel substrate compared with the 3mm thick adherend for the two part adhesive. It was necessary to increase the substrate thickness to 6 mm as the 3 mm steel substrate yielded under the tensile load. As the joint thicknesses are different for the film and the two part adhesives a linear numerical analysis was undertaken in another study (Photiou et al, 2006) and one of the parameters that this analysis yielded was the longitudinal shear stresses at the interface between the steel and the FRP composite at its free end and at the discontinuity of the steel substrate. The peek stresses at the free end of a 0.1 mm thick joint was 57% greater in value compared with the 0.5 mm thick joint. Likewise, at the position of the discontinuity of the steel adherend the corresponding value was 21%. The higher values as the adhesive thickness decreases have been noticed by other research workers. In the current F.E. analysis no distinction was made between the two different bonding methods. On a positive side, the thinness of the joint and its intermixing with the polymer matrix will be a great advantage when considering durability of the adhesive; this point has been discussed in the paper. An advantage of using an adhesive film is that the compaction and curing operation of the composite prepreg plate is undertaken in one operation, it also allows an intermixing of the adhesive and the matrix of the composite to take place during curing thus producing a high degree of molecular interlocking. Furthermore, as the fabrication of the combined prepreg and adhesive film is performed under a pressure of 1 atm. the compaction of the two component parts is good; the compaction of the composite has been measured as leading to about 2% voids. By introducing a low modulus GFRP composite prepreg between the steel and composite it has been shown that a greater load can be supported and from previous work it has been shown that a more gradual transfer of shear stress between the two adherends can be achieved. A further gain is the protection that the GFRP composite provides against any possible galvanic action between the steel and the CFRP composite. Furthermore, the thinness of the film adhesive, compared with the thicker ones, reduces to a minimum the ingress of harsh environments, such as liquid or salt solutions, through its free surface and this improve the overall durability of the joint. A disadvantage of the UHM-CFRP is its low strain to failure.
It is suggested that the UHM-CFRP/GFRP U-shaped upgrading composite in conjunction with the adhesive film has a structural advantage over the UHM-CFRP/GFRP flat plate upgrade composite and adhesive film in that it has the ability to contain the failure and still provides a degree of stiffening. Referring to the HM-CFRP composite upgrading utilising the adhesive film there seems to be no advantage in using the U-shaped geometric form as opposed to the HM-CFRP flat plate as the beam deforms excessively in both cases without debonding. The strain of the CFRP composite in both cases is still within the limits of its ultimate value of 1.8%. This part of the work has shown that there are many factors which must be considered when designing a jointing strategy, and it is clear that a good understanding of the full load-deflection response of the joining system is essential. REFERENCES Albat, A.M. and Romilly, D.P. (1999), "A direct linear-elastic analysis of double symmetric bonded joints and
reinforcements", Composite science and technology, 59(7), 1127-1137. Cadei, J.M.C., Stratford, T.J, Hollaway, L.C. and Duckett, W.G. (2004). Strengthening metallic structures using
externally-bonded fibre-reinforced-polymers, CIRIA Report C595, London. Deng, J., Lee, M.M.K. and Moy, S.S.J. (2004). "Stress analysis of steel beams reinforced with a bonded CFRP
plate", Composite structures; 65, 205-215. Denton, S.R. (2001). "Analysis of stresses developed in FRP plated beams due to thermal effects", Proceedings
of the international conference on FRP composites in civil engineering (Teng ed), Hong Kong. Frost, S., Lee, R.J. and Thompson, V.K.(2003). "Structural integrity of beams strengthened with FRP plates –
Analysis of the adhesive layer", Proceedings, Structural faults & repair (ed. Ford), London, 2003. Gettings, M. and Kinlock, A. J., (1977). "Surface analysis of polysiloxane/metal oxide interfaces", Journal
Materials Science, 12, 2511-2518. Liu, X., Silva, P.R. and Nanni, A. (2001). “Rehabilitation of steel bridge members with FRP composite
materials.” Proc. CCC 2001, Composites in Construction, Figueiras et al. (eds), Porto, Portugal, 613-617. Mertz, D. R., Gillespie, J.W., Chajes, M.J. and Saboi, S.A. ( 2001)."The rehabilitation of steel bridge girders
using advanced composite materials", IDEA program final report, contract number NCHRP-98-IDO51, Transportation research board, National Research Council, USA.
Miller, T.C. (2000). The rehabilitation of steel bridge girders using advanced composite materials, Master’s Thesis, University of Delaware, Newark, USA.
Mosallam, A.S. and Chakrabarti, P.R. (1997). “Making connection.” Civil Engineering, ASCE, 56-59. Moy, S.S.J. (2001) (ed). FRP composites – Life Extension and Strengthening of Metallic Structures. Institution
of Civil Engineers, 33-35 Photiou, N.K., Hollaway, L.C. and Chryssanthopoulos, M.K. (2003). “Characterisation of adhesively bonded
plates for upgrading structural steelwork”, Proc. Structural Faults and Repair, Forde (ed), (CD ROM). Photiou, N.K. (2005). Rehabilitation of Steel Members Utilising Hybrid FRP Composite Material Systems, PhD
Thesis, University of Surrey. Photiou, N.K., Hollaway, L.C. and Chryssanthopoulos, M.K. "Selection of CFRP systems for steelwork
upgrading", Accepted for publication in ASCE Journal of Materials in Civil Engineering. Photiou, N.K., Hollaway, L.C., and Chryssanthopoulos, M.K. (2006). "Strengthening of an artificially degraded
steel beam utilising a carbon/glass composite system", Jnl. of Construction and Building Materials, 20(1-2), 11-21.
Smith, S.T. and Teng, J.G. (2003). "Interfacial stresses in plated beams", Engineering structures, 23, 857-871. Stratford, T. and Cadei, J. (2006). "Elastic analysis of adhesion stresses for the design of a strengthening plate
bonded to a beam", Jnl. of Construction and Building Materials, 20(1-2), 34-45. Tavakkolizadeh, M. and Saadatmanesh, H. (2003). “Strengthening of steel-concrete composite girders using
carbon fiber reinforced polymers sheets”, Journal of Structural Engineering ASCE, 129 (1), 30-40. Yang, J., Teng, J.G. and Chen, J.F. (2004). "Interfacial stresses in soffit-plated reinforced concrete beams", Proc
inst civil engineers: Structures and Buildings; 157(1): 77-89.