Cradle-to-gate environmental impact prediction from chemical attributes using mixed-integer programming

Life Cycle Assessment (LCA) has recently gained widespread acceptance in green chemistry as an effective tool for quantifying the environmental impact of chemicals along their life cycle. Unfortunately, LCA studies require large amounts of data that are hard to gather in practice, a limitation that is particularly critical when assessing the complex processes and value chains present in the chemical industry. With the aim at simplifying these calculations and promoting the wider adoption of environmental principles, in this work we develop an approach that predicts the cradle-to-gate life cycle production impact of organic chemicals from attributes based on their molecular structure and theimodynamic properties. The approach presented relies on a mixed integer programming (MIP) optimisation framework that streamlines the LCA calculations by systematically constructing multi-linear short-cut predictive models of cradle-to-gate life cycle impact. These models contain key molecular and theimodynamic attributes that are identified using binary variables. On applying our method to an LCA data set containing 83 chemicals, 17 molecular descriptors and 15 theimodynamic properties, we produced estimates for widely used metrics such as cumulative energy demand (CED), global warming potential (GWP) and Eco-indicator 99 (EI99) with relative errors within acceptable ranges considering the nature of any LCA study. Our optimisation-based streamlined LCA framework ultimately leads to simple linear models that are amenable for implementation in computer aided molecular design software, thereby opening new avenues for the inclusion of sustainability principles in the early stages of the development of new chemicals.