Modular FRP sandwich structures for building floor construction

Fibre reinforced polymers (FRP) have many advantageous properties compared to traditional steel and concrete, such as lightness in weight and corrosion resistance. FRP is high in strength and, when used as load-bearing members in building structures, can also provide function integration (such as thermal insulation) and environmental benefits (especially when glass fibres are used, i.e. glass fibre reinforced polymers or GFRP). This makes GFRP a promising structural material for building floor construction. However, studies into GFRP building floors have been very limited, particularly for slabs and GFRP-steel composite floor structures. In addition, most pultruded GFRP sections available on the market for floor and deck applications have fixed geometries, which limits design flexibility. Therefore, this thesis presents a modular GFRP sandwich system for building floor construction. This modular system consists of built-up sections consisting of pultruded GFRP web-core profiles (such as box or I-profiles) that are incorporated between two pultruded flat panels to form a web-flange sandwich assembly. This system provides greater flexibility in designing floor systems, and is thus especially advantageous for varying design parameters and load conditions. The objectives of this thesis are to investigate the mechanical performance of modular GFRP sandwich structures either as beams (or one-way spanning slabs), two-way spanning slabs and GFRP-steel composite beams under static loading. Composite action and the corresponding bending stiffness, failure mechanism and the resulting load-carrying capacity were evaluated and analysed. In addition, two different pultrusion configurations were examined in the two-way slabs and GFRP-steel composite beams: flat panels with pultrusion directions either parallel or perpendicular to the web-core profiles, a feature that has not been investigated for such structures. From these tests it was shown that failure modes of the sandwich specimens were dependent on span-to-depth ratios or pultrusion orientation of the flat panels placed parallel or perpendicular to the web-core profiles. Furthermore, all the GFRP-steel composite beams were appropriately designed since they showed ductile load-deflection responses that resulted from yielding of the steel girder, which commenced prior to failure of the GFRP slabs. The type of shear connection (adhesive bonding and novel blind bolts) was also investigated. It was shown that adhesive bonding provided full composite action at all load-levels up to failure in sandwich beams, slabs and GFRP-steel composite beams. However, bolted connections provided only partial composite action in sandwich beams and slab structures. Furthermore, in the case of sandwich beams, bolted connections provided non-linear structural responses because the degree of composite action changed with the load level. In GFRP-steel composite beams, bolted connections with certain spacing provided either full or partial composite action, dependent on the pultrusion configuration of the box-profiles within the GFRP slab and the resulting longitudinal shear force per unit length at the GFRP-steel interface. Adding foam into the core of sandwich beams and slabs significantly enhanced the load-carrying capacity by preventing out-of-plane buckling of upper flat panels in compression, while providing only a small increase in weight. It was found that foam had a negligible effect in improving bending stiffness in the sandwich beams due to its low elastic modulus in comparison to that of the GFRP material. However, improvements in stiffness were significant in the two-way slabs due to improvements in in-plane shear stiffness of the webs when loaded in the transverse direction. Finally, numerical models were developed using ANSYS finite element software to predict bending stiffness and load-carrying capacity of the modular sandwich structures. These models were validated with experimental results, showing good comparisons. The bolted connections were modelled by a simplified approach where nodes were coupled at the corresponding bolt positions in beam, slab and GFRP-steel composite beam specimens. The adhesively bonded connections were modelled by coupling all nodes at adherend interfaces. The simplified approach was highly effective in analysing bending stiffness and both full and partial composite action. Furthermore, analytical models were developed and effectively applied to the sandwich system to predict bending stiffness and failure loads. Parametric analysis was also performed to develop design recommendations.