L. Canning et al. describe an interesting study on the development and flexure testing of a novel advanced composite/concrete duplex beam. The structural concept shown in Fig. 1 optimises the specific stiffness and specific strength of a simply supported beam by having concrete in compression and a fibre reinforced plastic (FRP) profile to support the tensile and shear forces. A lightweight unit is achieved by having a box-shaped FRP profile with a central void. Early in the paper the authors convey that an area that must be addressed with all composite constructions is the shear component required between the two dissimilar materials, to enable full composite action to be developed. This is the main aspect that will be discussed in what follows.
Canning et al. used continuous fibre lamenae and a vacuum prepreg processing technology to manufacture, at best, one beam per day. Such a long production time for a unit of dimensions 70 × 175 × 1500 mm suggests that it is suitable for low production numbers and small size. Fabricating the FRP profile by a labour-intensive process has enabled Canning et al. to construct a very lightweight unit. Deskovic,19 who first studied duplex beams, and the author20 recommend that, on balance of structural performance and economical considerations, the FRP profile be manufactured by pultrusion. Recently, the American pultruder Strongwell21 processed an optimised simply supported beam with external dimensions of 914 × 439 mm. The application for their hybrid carbon/glass fibre reinforced profile is as girders in road bridges. Such a large profile size can be pultruded at metres per hour.
It is not clear from the paper why the prepreg profile has both glass and carbon fibre reinforcement. In the original concept by Deskovic (see Fig. 1), the reason for the carbon reinforcement, with a lower strain to failure than the glass, was to develop a pseudo-ductile moment–deflection characteristic. It is noted that the load/mid-span deflection of the prepreg beam is virtually linear elastic to ultimate failure (se Fig. 12).
To develop the optimum duplex unit, all concrete must be in compression at the ultimate limit state. The FRP/concrete interface will therefore be close to the neutral axis, and under optimum conditions the longitudinal shear component at the interface will be high. Both Canning et al. and Deskovic19 used an FRP profile with a flat top surface. Deskovic found the bond area was insufficient to transfer shear, and steel bolts (shear studs) were necessary to attain the calculated ultimate moment. Practical units should not have metallic shear studs if long-term durability is not to be compromised. To improve the bond, Canning et al. have FRP side walls to increase bond area, roughened FRP surfaces (after removing a sacrificial peel ply), and indents in the side walls to enhance mechanical bonding. In the event, the initial failure mode, from the single beam specimen tested, was bond failure at half the ultimate moment.
Canning et al. give a list of techniques that could provide effective full composite action. None of these techniques includes that favoured by Hall and Mottram,20 which follows from the strengthening of reinforced concrete using thin FRP sheets.22 Their technique is to cast concrete onto an uncured layer of epoxy mortar (e.g. Sikadur) applied to the FRP. Hall and Mottram used similar four-point bending tests to show that the concrete–adhesive–FRP bond did not govern resistance. Although their non-standard pultruded panel was not optimised for duplex beams, it possesses the geometric feature of T-upstands that can be cast into the concrete. Such upstands provided additional bond area and mechanical interlocking. The combination of a higher relative bond area, lower relative shear force and an epoxy mortar bond was shown in Reference 2 to overcome the shear transfer weakness found by Canning et al. This was despite having cracked concrete, as part of it was in tension. Furthermore, the mortar bond provides the major advantage22 that it should be tolerant of alkalinity and will seal the glass FRP from the concrete.
It is not straightforward to compare the relative stiffness of the two FRP–concrete beam units. Rough calculations of specific stiffness show that the initial value (to ± 10%) is in the ratio 1·0:1·4:2·7, using beams 7 and 8 by Hall and Mottram20 as unity, followed by a conventional reinforced concrete beam and the Canning et al. duplex beam. For the pultruded beam the stiffness is constant to a mid-span deflection of span/250. Bond failure in the prepreg beam occurred when the deflection was lower, at span/333. Serviceability limit state for building application would require a deflection of span/250
Canning et al. have demonstrated that, by optimising the duplex unit's geometry and materials distribution, a high specific stiffness can be achieved. Pultruded profiles could be developed with T-upstands that would give a specific stiffness higher than 1·4 and be used for beam, panels and permanent formwork. It is the author's opinion that there are other optimisation criteria, such as application cost, bond performance, and buildability, which need to be equally considered when developing practical solutions.
Authors' reply
We thank Dr Mottram for his interest in the paper and for presenting his discussion on the paper.
Firstly, it must be pointed out that the work presented in the paper was purely a beam study and not an application. While applications were considered, at this stage they were not of primary concern. This was the first of ten beams which have been tested to investigate theoretical, experimental and numerical behaviour. We were interested in producing a structural form and developing a technique to manufacture this lightweight system which had the same ultimate strength as a reinforced beam of the same cross-section. A thin aircraft prepreg advanced material was initially used as, at that time, there was no thick prepreg developed suitable for laying up large, constant cross-section, thicker units. From the manufacturing point of view the technique was time-consuming. The beam in question was laid up in conjunction with the prepreg manufacturer to aircraft specification as a demonstration of how to handle the materials and to manage the manufacturing technique. The beam quality has subsequently been used as a yardstick against which other beams have been compared. Clearly, as Dr Mottram points out, three days is not an economic manufacturing time. We have developed the handling techniques, over ten beams, to adapt manufacture to civil engineering standards such that the material preparation and lay-up is not completed in 4 h prior to the 16 h cure cycle. Currently, we are involved in a Brite-Euram project which has developed a civil engineering prepreg, which is much thicker than the aircraft one we used; this will reduce considerably the manufacturing time of the beams to an estimated 1·5 h.
Dr Mottram suggests that, based upon his experience and the work of Deskovic, a pultruded section would give better structural and economic performance. We disagree with this comment. It is well known that to form a pultrusion section in which the fibres are precisely placed is a difficult operation to undertake whereas the prepregs are positioned precisely by operatives. Any desired stiffness can be designed into a unit. The economics of the two different composites would require a life cycle assessment and a lift cycle costing analysis.
We will now concentrate upon Dr Mottram's specific points.
The reasons for the CFRP/GFRP hybrid flange in our beam were twofold. Firstly, the UD CFRP was to provide the flexural rigidity of the beam at a lower cost and smaller thickness compared with a system wholly made using GFRP material. Secondly, the ±45° GFRP was continuously laid up around the webs and the tensile flange interleaved with the UD CFRP at the latter position. This method of lay-up provided bond length to the ±45° reinforcement and contained the tensile UD carbon. Deskovic provided UD CFRP in the flanges to give an indication of imminent failure of the beam (pseudo-ductility) by the relatively low strain to failure of the carbon fibre; this method has also been used by Meier. In our beam it is recognised that the load-deflection behaviour is essentially linear to failure, although there is some non-linearity near failure due to the ±45° GFRP in the beam; the linearity could present a problem. In reality the solution is self-evident. Once the beam is designed to meet deflection criteria of span/250 to span/750 the factor of safety ranges from 2·5 to 6 respectively, well above that normally required. It is accepted that beams of this type will have deep sections.
It is clear that, to develop a duplex unit, all concrete should be in compression. In our case the beam was designed such that the neutral axis was at the interface between the concrete and composite. As was stated in our paper, a number of systems could be employed to obtain near-composite action between the concrete and composite. The indents in the walls of the permanent shuttering had no assistance from adhesive action. The initial failure of the system appeared to result from separation of the permanent shuttering away from the concrete. However, the beam still functioned and supported loads up to failure at nearly 5 t. One of the methods which we used in a later study, namely an adhesive compatible with fresh concrete (Sikadur 31, which is a filled epoxy not an epoxy mortar), provided a complete composite action between components of the beam as has been reported in our paper.23 Small-scale pull-off tests gave values of shear bond strengths of 3·47 MPa. However, this method of bonding might not be the ideal answer when the engineer is confronted with the expoxy adhesive coating of an 18 m long beam. A more practical solution might be to use studs and accept that a smaller failure load would result, as was mentioned in our paper.
We do not understand the significance of the penultimate paragraph of Dr Mottram's comments. He is suggesting that the specific stiffness of the duplex beam is greater than the Hall and Mottram beam, to which we agree. The comment that the deflection was lower (span/333) at initial mechanical bond failure of the indented beam tested is correct. However, further beam tests where fewer prepreg layers were used and with different methods of sheaer connections, gave ratios of span/200 at 50% of ultimate flexural strength.
The point raised by Dr Mottram that cost, bond performance and buildability should be considered as optimisation criteria is a good one, and one with which we agree. Some of these points are being investigated in the current research.
It should be mentioned that there are always a number of solutions to engineering problems and the one finally chosen will be the best for that particular case. We accept that the pultrusion method for manufacturing structural units is one of the most versatile techniques currently available to the construction industry. It should be realised, however, that other techniques have their place in civil engineering and that our type of beam has specific advantages over the pultrusion unit, notably the lightweight property. A system that meets design deflection criteria but weighs 25% of a reinforced concrete beam will have its place in the market. It is possible to erect the beam without the compressive concrete in place and in this case the weight is 3% of a conventional beam, allowing considerable savings in transportation and placement.
