A stochastic study on the robustness of a stiffened composite structure

Abstract

COCOMAT is an ongoing four year European Commission project aimed at exploiting the large reserve of strength in composite structures through more accurate prediction of collapse. As part of the research program, curved stiffened composite panels of various designs have been manufactured and tested in compression. In the current pool of experiments it is possible to observe the scatter in results caused by manufacturing defects and material variations. In order to gain more insight into the results, a stochastic framework has been developed where variations are introduced into the finite element analyses and the effects measured. An issue arises as to what the stochastic boundaries should be, and the how this affects the final result. In a bid to understand how the stochastic method can be useful in future analyses a design criterion, robustness, has been used. A qualitative measure of robustness in the form of a Robust Index has been derived so that comparisons between the results achieved can be made.