Early-Stage Environmental Impact Forecasting of Chemicals with Machine Learning and Data Analytics Tools

Life Cycle Assessment (LCA) is a method used to evaluate the environmental impacts of materials and processes throughout their entire life cycle, from production to end-of-life. However, performing LCA at the early design stage of a chemical process is often challenging because Life Cycle Inventory (LCI) data for new or emerging chemicals are not readily available. To enable impact assessment under these data-limited conditions, this study employs Machine Learning (ML) and Scaling Index Regression models to estimate environmental impacts across all life cycle stages. Artificial Neural Network (ANN) and eXtreme Gradient Boosting (XGBoost) are employed to develop models to predict Life Cycle Impact Assessment (LCIA) endpoint metrics such as Human Health Impact (HHI), Ecosystem Quality Impact (EQI), Global Warming Potential (GWP), and Resource Utilization Impact (RUI) during the production phase of a chemical based on thermodynamic and molecular descriptor properties of the chemicals. Regression models are then applied to estimate the impact of technologies used during the use phase and end-of life phase by determining emission factors for the different technologies involved in the process. To demonstrate the accuracy of the proposed framework a case study is presented to validate the model’s performance.