You are here:
Mining Available Data from the United States Environmental Protection Agency to Support Rapid Life Cycle Inventory Modeling of Chemical Manufacturing
Citation:
Cashman, S., D. Meyer, W. Ingwersen, A. Edelen, J. Abraham, W. Barrett, M. Gonzalez, P. Randall, G. Ruiz-Mercado, AND R. Smith. Mining Available Data from the United States Environmental Protection Agency to Support Rapid Life Cycle Inventory Modeling of Chemical Manufacturing. ENVIRONMENTAL SCIENCE & TECHNOLOGY. American Chemical Society, Washington, DC, 50(17):9013-9025, (2016).
Impact/Purpose:
The research presented in this manuscript is a method to rapidly estimate life cycle inventory. The research addresses the need to reduce the time and resources necessary to perform life cycle assessment in EPA. This research will support implementation of the life cycle human exposure models being developed in CSS Project 11.01.
Description:
Demands for quick and accurate life cycle assessments create a need for methods to rapidly generate reliable life cycle inventories (LCI). Data mining is a suitable tool for this purpose, especially given the large amount of available governmental data. These data are typically applied to LCIs on a case-by-case basis. As linked open data becomes more prevalent, it may be possible to automate LCI using data mining by establishing a reproducible approach for identifying, extracting, and processing the data. This work proposes a method for standardizing and eventually automating the discovery and use of publicly available data at the United States Environmental Protection Agency for chemical-manufacturing LCI. The method is developed using a case study of acetic acid. The data quality and gap analyses for the generated inventory found that the selected data sources can provide information with equal or better reliability and representativeness on air, water, hazardous waste, on-site energy usage, and production volumes but with key data gaps including material inputs, water usage, purchased electricity, and transportation requirements. A comparison of the generated LCI with existing data revealed that the data mining inventory is in reasonable agreement with existing data and may provide a more-comprehensive inventory of air emissions and water discharges. The case study highlighted challenges for current data management practices that must be overcome to successfully automate the method using semantic technology. Benefits of the method are that the openly available data can be compiled in a standardized and transparent approach that supports potential automation with flexibility to incorporate new data sources as needed.
