A graph-based framework for casting-aware connectivity refinement in free-size optimized shell structures

Integrating casting-oriented manufacturing considerations into structural optimization remains challenging, particularly for thin-walled components optimized using free-size formulations. Such designs often exhibit spatially disconnected regions of increased wall thickness that may impair feeding accessibility and promote unfavorable solidification behavior during casting. This work proposes a graph-based framework for the connectivity refinement of free-size-optimized shell structures. Regions of increased thickness are identified and clustered, and an A*-based pathfinding algorithm is employed to establish connectivity while balancing geometric distance, required material thickening, and existing structural features. A minimum spanning tree (MST) is subsequently used to generate a globally connected and material-efficient network. The resulting connectivity paths are integrated into the structural model and preserved during a subsequent free-size re-optimization step. The framework is evaluated using benchmark examples, an industrial shock-tower demonstrator, and a complementary casting-validation study. For the industrial demonstrator, the introduced connectivity can be incorporated while maintaining the original compliance target, resulting in a final weight increase of approximately 2–4% after re-optimization. The casting-validation study further indicates that the generated connectivity can improve casting-related behavior while requiring only a limited increase in component volume. The proposed framework provides a computationally efficient strategy for introducing casting-oriented connectivity into shell-based structural optimization workflows and establishes a flexible foundation for future integration of higher-fidelity manufacturing constraints and process simulations.