BridgeFinder: Enabling Predictive Optimisation Techniques for High-Level Design of Large Structures in Mechanical Engineering Development
Year: 2025
Type: conference-paper
Abstract: In this study, we present BridgeFinder, a novel high-level design optimisation approach tailored to the demands of large-scale construction engineering. This method facilitates predictive design rather than relying on reverse calculations. BridgeFinder integrates SmilesMe, a newly devised notation and abstraction layer, with an evolutionary algorithm leveraging the specially developed Evolution Shape Function for fitness assessment. Our inspiration to develop BridgeFinder stems from MolFinder, a technique used in chemistry to forecast molecular structures. Current algorithm-based optimisation strategies primarily focus on the later stages of construction design, such as CAD/FEM. Alternatively, some methods lack definitive calculation guidance, leading to overly algorithm-driven outcomes that may not always align with the requirements of high-level design in large constructions (e.g., generative design techniques). BridgeFinder aims to bridge this gap, offering refined calculation guidance, thereby generating construction concepts more suited to the high-level phase of large-scale projects. We anticipate that our findings will illuminate new pathways for advancements in construction design methodologies.
Institutions: Technische Hochschule Ingolstadt, RMIT University
Language: English
Subfield: Architecture
Field: Engineering
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4407319931