AI in Mechanical Design for Sustainability: AI-Assisted Optimization of Product Sustainability in Design and Manufacturing Preparation

Abstract

This thesis investigates how AI-driven generative design and topology optimization methods combine with circular economy (CE) principles and life-cycle assessment (LCA) to enhance the sustainability of mechanical components. The existing AI methods enable significant reduction of weight and component count, but environmental and circularity factors are considered as secondary design elements, which create a disconnect between local geometric optimization and global sustainability targets.

The study adopts a practical research framework together with an inductive approach that relies on existing literature to conduct case study investigations. A detailed investigation of eight industrial cases was conducted, which demonstrated the use of AI for designing and manufacturing mechanical components. The study evaluated AI-driven topology optimization and generative design as well as AI-enabled DFMA/DFD and ML-based LCA approaches. The study extracted comparable data from each case study to analyze design context and AI methods and improvements achieved through weight reduction, part consolidation, environmental indicators, CE elements, and DFD components. Overall, the case studies show that the main opportunities for sustainability improvement are reducing material use and component count through AI-driven topology and generative design and improving circularity through DFMA/DFD and ML-enabled LCA integration.

The empirical findings show that AI-driven topology and generative methods deliver substantial lightweighting and assembly simplification results, yet their optimization process lacks LCA and CE and end-of-life integration, which only occurs during post-hoc evaluations. Machine learning-based LCA surrogates show potential to provide environmental feedback during design processes, yet they encounter difficulties with training data and model validation and software integration. The thesis builds upon these findings to develop a conceptual Lifecycle Aware AI Design Framework (LAAIDF), which establishes connections between life cycle stages and artificial intelligence module categories that include topology/generative design and DFMA/DFD optimization and ML-enabled LCA surrogates and AI for circular scenarios. The framework provides a structured response to the research question and offers theoretical, managerial, and policy-level implications for organizing AI initiatives that support lifecycle sustainability and circular economy objectives in mechanical design organizations.