From Reaction Stoichiometry to Life Cycle Assessment: Decision Tree-Based Estimation Tool

Decision-making during the early stages of research and development (R&D) should be informed by both economic and ecological perspectives. While early stage cost assessments are well established, life cycle assessment (LCA) is still largely descriptive but should expand to a more prospective tool for early assessing the ecological effects of future processes. Chemical processes should be first assessed as early as when only the reaction equation is known. Our previous comparison of estimation methods based on the reaction equation identified three requirements to foster early stage LCA: (1) estimate inventories rather than final impacts to ensure flexibility, (2) distinguish between processes, as single values cannot reflect the variety of chemical processes, (3) provide a measure of uncertainty. In this publication, we propose regression trees to estimate key inputs for industry-scale life-cycle inventories of chemical processes directly from the underlying reaction equation. In detail, the regression trees yield the raw materials' impact, the direct greenhouse gas (GHG) emissions in CO₂eq, and the demands for electricity, steam, natural gas, cooling water, and process water. The regression trees outperform the current best available proxy values and provide inventory information that is as accurate as cost estimates. Thus, our work enables decision-makers to consider environmental aspects with the same level of accuracy as costs projections.