On a Novel Approach for Optimizing Composite Material Panels Using Surrogate Models

This paper describes an optimization procedure to design thermoplastic composite panels under axial compressive load conditions. Minimum weight is the goal. The panel design is subject to buckling constraints. The presence of the bending-twisting coupling and of particular boundary conditions does not allow an analytical solution for the critical buckling load. Surrogate models are used to approximate the buckling response of the plate in a fast and reliable way. Therefore, two surrogate models are compared to study their effectiveness in composite optimization. The first one is a linear approximation based on the buckling constitutive equation. The second consists in the application of the Kriging surrogate. Constraints given from practical blending rules are also introduced in the optimization. Discrete values of ply thicknesses is a requirement. An ad-hoc discrete optimization strategy is developed, which enables to handle discrete variables.