Laminate optimization of blended composite structures using a modified Shepard's method and stacking sequence tables

This article presents an optimization tool for the stacking sequence design of blended composite structures. Enforcing blending ensures the manufacturability of the optimized laminate. A novel optimization strategy is proposed combining a genetic algorithm (GA) for stacking sequence tables with a multi-point structural approximation using a modified Shepard's interpolation in stiffness-space. A successive approximation approach is used where the set of design points used to create the structural approximations is successively enriched using the elite of the previous step. Additional improvement in the generality and efficiency of the algorithm is obtained by using load approximations thus enabling the implementation of a wide range of stress-based design criteria. A multi-panel, blended composite problem is used as an application to demonstrate the performance of the developed tool. The optimization is performed with mass as the objective to be minimized, subjected to strength and buckling constraints. The results presented show that completely blended and feasible stacking sequence designs can be obtained, having their structural performance close to the theoretical continuous optimum itself. Additionally, the multi-point Shepard's approximation shows a considerable saving in computational costs, while the limitations of inexpensive stiffness-matching optimizations are observed.