Multidisciplinary design optimization of a generic b-pillar under package and design constraints

Yannis Werner,Thomas Vietor,Matthias Weinert,Thomas Erber
DOI: https://doi.org/10.1080/0305215X.2020.1837791
IF: 2.5
2020-11-09
Engineering Optimization
Abstract:This article introduces a novel constraining approach for structural optimization that aims to support the conceptual engineer during the early embodiment phase for structural lightweight design. It reduces the time spent on structural engineering studies by enabling optimization algorithms to detect geometric intersections by analysing the mesh information. This article reviews approaches from the literature focusing on computer aided design environments, sampling methods, data analytics and optimization techniques for design and sizing optimization with finite-element models. The evaluated approaches are integrated into a Python-based optimization environment. Accordingly, the introduced methodology enables the environment to handle geometrically infeasible designs. The presentation of the first results focuses on the feasibility of structural assemblies and the results demonstrate the viability of NSGA-II for optimization tasks. The example considers the design of a generic b-pillar structure under crash-safety requirements. Using this approach, the NSGA-II algorithm avoids geometrically infeasible areas and increases comparative structural performance.
engineering, multidisciplinary,operations research & management science
What problem does this paper attempt to address?