DISTRIBUTING DESIGN DOMAINS FOR TOPOLOGY OPTIMIZATION IN SYSTEMS DESIGN

When objectives, constraints, boundary conditions, and loading conditions are clear, topology optimization is useful to generate the optimal shape for an individual component. However, in systems consisting of multiple components, component interfaces and their locations may have a significant influence on the component, and, thus, the global system performance. Before topology optimizations can be performed, design domains and associated loading conditions must be allocated to the individual components when designing the system. Unfortunately, to optimally define interfaces in lightweight design, the mechanical performance of each component must be known. This paper proposes a simple and robust approach to specifying the settings for component optimization in a system context. In the first step, requirements are elicited and categorized in system- and component-level. Next, a topology optimization formulation at the component-level is derived from the requirements. Third, different interface definitions are generated, and optimal topologies are computed using sampling. This database is evaluated using a Parallel Coordinates tool, determining the optimal distribution of design domains and incorporating formal and informal constraints. Lastly, final geometries are computed using topology optimization with the setup resulting from system optimization. By informing system-level design with information from the detail design, design domains can be optimally distributed. In a use case, the approach is applied to the control system of a glider airplane, where a common design domain is allocated to three brackets. Optimization at system- and component-levels lead to a reduction in mass by ~11 % compared to an intuitive distribution of design domains. With the proposed approach, individual topology optimizations of single components will contribute to the global system performance in an optimal way.