Abstract In the early stages of product design, considering environmental impacts throughout a product’s life cycle is essential to support sustainable industrial development. Traditional Life Cycle Assessment (LCA) methods, while comprehensive and systematic, are often limited in the conceptual design phase due to the lack of detailed information and the time required for modeling. Parametric LCA approaches attempt to bridge this gap but still demand extensive data and expertise. To address these challenges, this study proposes a methodology based on Machine Learning (ML) techniques, enabling the implementation of a Surrogate LCA for preliminary environmental evaluation during early design. ML models can learn from existing LCA datasets to predict environmental impacts using limited design parameters, providing rapid and informed feedback to designers. This approach transforms LCA into a proactive design aid, capable of handling data uncertainty and dynamically adapting to design variations. The study demonstrates the methodology through a case study, showing how product geometry influences prediction accuracy and, consequently, the uncertainty of environmental impact estimations. Results highlight the potential of ML-driven LCA tools to enhance early-stage design decisions, supporting the transition toward circular economy principles and environmentally responsible product development. Graphical abstract
Manuguerra et al. (Fri,) studied this question.