Scaling-up topology optimization with target stress states via gradient-based algorithms

Research output: Contribution to journalResearch articleContributedpeer-review

Contributors

Abstract

Benchmark artifacts serve as an appropriate mean to represent quality measures in additively manufactured components. Especially witness specimens, which represent structural properties as a subtype of benchmark artifacts, are supposed to reproduce target stress states as they are critical for component failure. This study aims at proposing an approach to effectively scale the results of topology optimized witness specimens with failure-critical target stress states using gradient-based methods. Therefore, possible formulations with analytical sensitivities are derived and implemented in a Matlab framework in order to contextualize the actual formulation within classical approaches consistently. It has been confirmed that gradient-based topology optimization with analytical sensitivities is not appropriate for a direct formulation of failure-critical target stress states. Thus, an indirect formulation based on compliant mechanisms is presented in this paper. It is shown that plausible, functional structures can be produced. Target stresses are achieved in scaled-up results in two and three dimensions with an accuracy up to a relative error of 3.7 %.

Details

Original languageEnglish
Article number107766
JournalComputers & structures
Volume314
Publication statusPublished - Jul 2025
Peer-reviewedYes

External IDs

Scopus 105002556506
ORCID /0000-0003-1185-0046/work/182727857
ORCID /0000-0001-6182-7341/work/182728449

Keywords