Design of dimensionally stable composites by evolutionary optimization

The problem of designing stable composite plates is considered. In a first step, in-plane design of composite plates for high stiffness, low thermal and moisture expansion is performed. It is shown that ([+/- theta(1))(p)/(+/-(2))(q)](s) stacking sequences are sufficient for in-plane design. An expert enumerative algorithm is coupled with an evolutionary search to obtain optimal plates for the in-plane problem. In a second step, a Monte Carlo laminate analysis is introduced, i.e. material and geometrical properties are distributed around their nominal value. Implications of property distributions on the optimal design are discussed. Under the effect of property variations, the coefficients of thermal and hygral bending of optimal in-plane designs can take on substantial values. Variations in coefficients of thermal and hygral bending are minimized by changing the laminate stacking sequence. This paper offers a unified approach to reliably designing stable composites. Design applications are given for organic and metal matrix composites. (C) 1998 Published by Elsevier Science Ltd. All rights reserved.