Modular multi-domain AI framework for sustainable construction material optimisation

Unofficial AI-generated summary based on the public title and abstract. Not an official translation.

🔬研究者:Provides a novel integration of AI and LCA for multi-criteria material optimization, offering a methodological template for further research.

🏢実務担当者:Offers a practical tool for early-stage design to evaluate material trade-offs, aiding sustainability reporting and procurement decisions.

🏛政策担当者:Demonstrates how AI can enhance LCA-based building regulations and support embodied carbon reduction policies.

The construction sector is a major contributor to global environmental impacts, largely driven by the embodied emissions of structural materials. Early-stage material selection therefore represents a critical opportunity for impact reduction; however, conventional life-cycle assessment (LCA) approaches remain limited in systematically evaluating large numbers of feasible material configurations under multiple and often conflicting sustainability criteria, constraining effective decision-making during design. To address this limitation, this study develops an artificial intelligence – driven decision-support framework for the systematic exploration and optimisation of construction material configurations within a fixed building design. The framework integrates LCA modelling, scenario-based material substitution, predictive modelling using mixture-of-experts backpropagation neural networks, and sustainability-oriented optimisation within a unified workflow. A reinforced-concrete office building in the United Kingdom is used as a case study to generate 1500 technically feasible material scenarios by varying concrete compositions (including GGBS and fly ash substitution), reinforcement steel production routes, cement formulations, and end-of-life pathways. Scenario outputs from OneClick LCA are used to train the predictive models, enabling rapid estimation of multiple sustainability indicators, which are subsequently coupled with a shuffled frog leaping optimisation algorithm using a composite sustainability index. The results demonstrate that the framework efficiently identifies material configurations that balance environmental, circularity, and cost-related performance while expanding the range of feasible design solutions beyond conventional scenario-based evaluation. The proposed framework provides a scalable and flexible decision-support tool for early-stage design, enabling systematic assessment of trade-offs and supporting informed material selection under practical design and feasibility constraints.

• semanticscholar https://doi.org/10.1016/j.eiar.2026.108582first seen 2026-08-15 05:30:43 · last seen 2026-08-16 05:42:48