Probabilistic FEM vibroacoustic analysis of fibre-reinforced composite having spatial random parameters

Fibre-reinforced composites (FRC) exhibit significant uncertainties which arise from the variable characteristics of the individual layers, their interconnection and the manufacturing process. The FEM modeling of such composites requires discretization the random space to be compatible to discretized spatial space by FEM. In this paper, uncertain and spatially varying parameters are approximated as stochastic field by means of the Karhunen-Loève (KL) expansion. Stochastic FEM structural vibroacoustic analysis is performed in which random vibroacoustic responses are represented using generalized polynomial chaos expansions with arbitrary random orthogonal basis. Knowing the KL expansions of the random parameters, the non-intrusive stochastic technique is employed on a set of random collocation points where the deterministic FE model is executed to estimate the unknown coefficients of the polynomial chaos expansions. This strategy helps us to decouple the FEM modeling and stochastic calculations where the deterministic FE code treats as a black box. The major advantage is the using any commercial FEM software for modeling of the problem. The method application is practiced to stochastic sound transmission througth a FRC plate. A fluid-solid FEM model is employed to calculate random acoustic transmission loss. The results show the impact of parameter uncertainties on the level of the loss factor. The proposed method is also proven to yield closer predictions to the numerical simulation from sampling methods, e.g. Monte Carlo method, however, with low cost in terms of computation time.