• Automated LCA workflow developed for building refurbishment projects • NLP and ML extract material inventories directly from Bills of Quantities • Embodied impacts for A1-A3 calculated automatically and reproducibly • Workflow extended to refurbishment stages A4-B5 using standardised tools • Real refurbishment case completed within minutes and validated against benchmarks The construction sector is a major contributor to global greenhouse gas (GHG) emissions, with a substantial share arising from the embodied impacts associated with the production and use of building materials. Refurbishment can reduce these impacts by extending building life and limiting new material demand. Life Cycle Assessment (LCA) quantifies these benefits, but refurbishment LCAs are often slowed by inconsistent project documentation and manual workflows. This study presents a data-driven workflow for automated LCA in building refurbishment. Natural language processing and machine learning are used to extract material inventories from Bills of Quantities (BoQ), link them to environmental databases, and calculate embodied impacts for the A1–A3 modules (production stage). The workflow is further extended to additional refurbishment-relevant life cycle stages, including A4 (transport to site), A5 (construction and installation), and B2–B5 (maintenance, repair, and refurbishment) by linking material inventories automatically extracted from BoQ to refurbishment stage calculations through transparent, stage-specific assumptions. The main contribution of this study is the integration of BoQ-based material extraction, machine-learning database matching, and life cycle impact calculation into a unified automated workflow, together with the systematic extension of document-derived material inventories to refurbishment stages, enabling the explicit calculation of refurbishment impacts (B5) within an automated and auditable LCA workflow. Applied to the rehabilitation and expansion of the Town Hall of Fuentes de Andalucía (Seville, Spain), the workflow reports A1–A3 impacts of 300,158.50 kg CO 2 -eq in about 10 minutes. Validation against published refurbishment studies shows consistency with established benchmarks.
Gachkar et al. (Wed,) studied this question.