You are here:
Advancing and Accelerating Release Estimations for Chemical Processes: Opportunities for Unit Operations, Data Mining, and Machine Learning
Citation:
Smith, Raymond l., David E. Meyer, Gerardo J. Ruiz-Mercado, Michael A. Gonzalez, William M. Barrett, AND John P. Abraham. Advancing and Accelerating Release Estimations for Chemical Processes: Opportunities for Unit Operations, Data Mining, and Machine Learning. Foundations of Computer-Aided Process Design, Copper Mountain, CO, July 14 - 18, 2019.
Impact/Purpose:
This extended abstract reviews work done in the Chemical Safety for Sustainability Research Program on rapidly estimating releases for use in exposure and risk assessments. While release information is often unavailable, it is needed to determine exposures and is combined with hazard information in risk assessments. This work emphasizes releases from chemical manufacturing, focusing on previous work in data mining, simulation, and machine learning. Future work will to expand rapid release estimations to other parts of the life cycle of a chemical.
Description:
Computer-aided process designs, risk assessments, and life cycle assessments can incorporate environmental impacts, but rapid estimation methods are needed to approximate the releases used in these assessments first. Through methods of simulation, data mining, and machine learning the releases from processes can be estimated. Simulation offers a unit-operation or bottom-up perspective, while data mining uses established data bases. Newer efforts have focused on machine learning through the use of classification and regression trees and demonstrate the ability to predict emissions within the same range of the other methods. Future work will apply these methods for their use in exposure and risk assessments.