On the possibility of reduced variable predictions for the thermoelastic properties of short fibre composites

Computer aided design offers the potential for the rapid development of new advanced structures from short fibre reinforced composites. The advantage of a validated numerical simulation is clear, as it allows a large number of the potential structure's solutions to be studied before proceeding to a manufacturing stage, reducing cost and risk accordingly. In our recent work we have demonstrated that the direct finite-element-based procedure of Gusev could be reliably employed for the prediction of thermoelastic properties of laboratory injection moulded samples, on the basis of Monte Carlo multi-fibre computer models built based on the measured fibre orientation distribution functions. Most injection moulded or extruded structures however, exhibit non-uniform fibre orientation states across the final parts, with a diverging variety of different local fibre orientation states. It Would be impractical to characterize all the possible orientation states by their full distribution functions, and it would be equally impractical to attempt direct numerical property predictions for all the various orientation states. Here we show that real injection moulded and extruded materials show fibre orientation states close to a maximum entropy prediction (i.e. most random) and that local thermomechanical properties can be excellently predicted based on the second order orientation moments and a constant strain assumption between the phases. Our results landmark a practical possibility of computer aided design of advanced structural parts from short fibre reinforced composites. (C) 2003 Elsevier Ltd. All rights reserved.