Fibre reinforced polymer (FRP) composite sandwich structures are increasingly used in the construction of civil engineering applications because of their outstandingstrength and light weight properties. However, the use of FRP products has some design difficulties as a result of the composition of the fibre and matrix. The designvariables usually are fibre and matrix properties, fibre direction, laminate composition, and core thickness. The combination of the design variables leads to acomplex design problem, and the optimisation of fibre composite sandwich structures is rarely straightforward. This is due to the complicated behaviour, and themultiple design variables and objectives required to be considered. This research deals with the presentation of a glass fibre reinforced polymer (GFRP) sandwichstructure analysis and design. Based on the literature review, a design optimisation methodology was proposed for the FRP composite structures. The methodologycontains three parts; experimental investigation, Finite Element Analysis (FEA) with modelling verification, and design optimisation of the GFRP sandwich structures.Several experimental static and free vibration tests were made on the GFRP sandwich beams and slabs. The experimental investigation provided good information about understanding the behaviour of the GFRP sandwich structures. A user subroutine UMAT was written to model the GFRP sandwich skins in three dimensions (3D) FEA. The FEA model was verified with the structural experimental behaviour in static and free vibration tests. The FEA analysis helped in-depth understanding of the GFRP sandwich structure behaviour, and provided an acceptable model for design optimisation.The design optimisation considered the Adaptive Range Multi-Objective Genetic Algorithm (ARMOGA) as an optimisation method. ARMOGA has robustness, ability in dealing with both continuous and discrete variables, and it has excellent searching for a global optimum. A design optimisation was done with the multi-objective cost and mass minimisation. The design optimisation was done on GFRP slab designs in one-way and two-way spaning. In addition, the optimisation of the single and glue laminated GFRP sandwich beam was also investigated.Single and glue laminated GFRP sandwich beams behaviour was investigated.Static four point tests were conducted for the beam investigation. The investigation showed that shear span to depth ratio (a/d) is the main factor controlling thebehaviour of the GFRP sandwich beam under combined shear and moment. Single sandwich beams showed higher shear and bending strength than glue laminated beams. The static experimental results indicated that there are three types of failure that can be seen in the GFRP sandwich beam; core crushing, core shear, and top skin compression failure. The GFRP sandwich beam did not show debonding as a failuremode because the skin-core interaction strength is close to the tensile and shear strengths of the core. The prediction shear equation showed acceptable results for beams with an a/d less than 2, and the bending equation showed good results for the beams of a/d greater than 4.5.One-way and two-way GFRP sandwich slabs were tested under static point load. GFRP sandwich slab tests showed that the core to skin ratio and the total slab thickness have a big effect on the GFRP sandwich slab load capacity. Slabs with 18 mm thickness and with a 3 mm skin thickness showed double load capacity compared to 15 mm slab thickness with a 1.8 mm skin thickness. In addition, the support system has an effect on the slab behaviour and it represents an important aspect in the design. The two-way supported slab has approximately double loading capacity compared to the one-way supported slab. Square slabs with ±45 degrees fibreorientation have a lower deformation and higher stiffness than 0 degrees/90 degrees orientation two-way square slabs. The effect of screw boundary restraint has a small influence on the behaviour of GFRP sandwich slabs. The effect of the slab width to length ratio is small at service load levels while it has more impact on the ultimate failure load level. The ultimate failure load decreases as the slab width to length ratio is increasing.One-way and two-way slabs were tested for free vibration behaviour in single and continuous support systems. The free vibration tests showed that the span lengthof the slab had an impact on the natural frequency with an increase in span length reducing the natural frequency of the slab. Two-way slabs have a higher natural frequency than one-way slabs. Three boundary restraint types were investigated. Moreover, glue restraints have a larger frequency than screw restraint slabs. The 0 degrees/90 degrees and ±45 degress skin fibre orientations were also studied. GFRP one-way sandwich slabs with ±45 degrees fibre orientation had a lower frequency than slabs with 0 degrees/90 degrees fibre orientation, while, the GFRP two-way sandwich slab with ±45 degrees fibre orientation had a higher frequency than slabs with 0 degrees/90 degrees fibre orientation.Non-linear FEA revealed that the material models for the skin and phenolic core give an acceptable behaviour. The comparison of the FEA results was done with different experimental tests for the slabs and beams. The FEA model using the CRUSHABLE FOAM model and Hashin model gave a good prediction for the GFRP sandwich structure’s behaviour. The core part did not reach the hardening behaviour when the structure failed due to core shear and top skin compression. The same FEA model was used to predict the free vibration of the GFRP sandwich slabs. The FEA model developed in this work provided a good prediction of the free vibration behaviour of GFRP sandwich beams and slabs. This model can be used for design optimisation with confidence.Multi-objective optimisation revealed that slab thickness is affected by the slab span. The required slab skin thickness and core thickness have an approximatelylinear relationship with the slab span length. The slab and beam designs are controlled by mid-span deflection limits. The strength constraints showed no contribution to the design optimisation. The design showed that the optimum core to skin thickness ratio of the beam is 11.0. The glue laminated beam optimisation indicated that the single sandwich beam has an optimum depth design less than theglue laminated beam. The depth of the glue laminated beam increases with the increase of sandwich layers.From this study, it was concluded that experimental investigations gave a better understanding of the behaviour of novel GFRP sandwich structure. In addition, theFEA modelling added more knowledge to understanding the behaviour of such structures. The optimisation design presented the design variables of the GFRPsandwich beams and slabs. Screw restraint slabs have a higher frequency than the simple restraint slabs.
展开▼
