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

Abstract

This article introduces a novel constraining approach for structural optimization, which 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 analyzing the mesh information. This article reviews approaches from the literature focusing on CAD-environments, sampling methods, data analytics and optimization techniques for design and sizing optimization with FE-models. The evaluated approaches are integrated into a Python-based optimization environment. Accordingly, the introduced methodology enables the environment to handle geometric infeasible designs. The presentation of the first results focuses on the feasibility of structural assemblies and the results demonstrate the viability of the NSGA-II for optimization tasks. The example considers the design of a generic b-pillar structure under crashsafety requirements. Using this approach, the NSGA-II algorithm avoids geometric infeasible areas and comparably increases structural performance.

Methodology for Neural Network-Based Material Card Calibration Using LS-DYNA MAT_187_SAMP-1 Considering Failure with GISSMO

“…For both MPI methods used, a systematic experimental scheme was created with varying settings. The experimental design of the optimization-based method including the results is shown for the 1 Step Strategy in Appendix A, Table A1 and for the 3 Step Strategy in Table A2. As already mentioned in Sections 2.5.1 and 2.5.2, the dimensionless dynamic time warping distance was used as evaluation criteria in addition to the qualitative SOC curve shapes.…”

“…In recent years, the application of FE analyses has become an indispensable tool in the product development process, especially when it comes to the design and dimensioning of structural components with complex materials such as polymers or fiber-reinforced plastics [1][2][3]. Due to the increasing computational capacity and the progress in the software development of FE programs, the level of detail in calculations has increased in recent years and the generated data have become more accurate [4].…”

Test and Validation of the Surrogate-Based, Multi-Objective GOMORS Algorithm against the NSGA-II Algorithm in Structural Shape Optimization

“…In particular, SFE CONCEPT offers the ability to generate structures quickly and pre-process Finite Element (FE) decks easily to follow Finite Element Analysis (FEA). Hence, the SFE CONCEPT tool takes over the design and pre-processing to optimize structural components [18,24,25]. A vector of design variables (DVs) x describes the geometry variably.…”

“…As this article deals with implementing a new optimization algorithm into an existing optimization framework, the following paragraph briefly introduces the present implementation. The basic framework and approach are described in [18]. In addition, an extension of the method and more detailed optimizations are performed in [20].…”

“…Engineering optimization typically fails to handle these geometry-based constraints during optimization. For this reason, a new approach for the handling of geometrical constraints in package and design optimization was implemented in a Python-based framework called IKOS [18] by the authors. The proof of concept for the IKOS configuration used the well-established NSGA-II algorithm, following [19].…”