Main content area

Comparative Evaluation of Chemical Life Cycle Inventory Generation Methods and Implications for Life Cycle Assessment Results

Parvatker, Abhijeet G., Eckelman, Matthew J.
ACS sustainable chemistry & engineering 2018 v.7 no.1 pp. 350-367
case studies, chemical industry, data collection, databases, engineering, global warming, greenhouse gas emissions, greenhouse gases, life cycle impact assessment, life cycle inventory, models, styrene, trade
A life cycle assessment (LCA) practitioner is often faced with the problem of missing chemical life cycle inventory (LCI) data sets, as current databases cover only a small fraction of chemicals used in commerce. Here, we critically review eight different methods used by LCA practitioners to estimate missing chemical LCI data, including process simulation, engineering process calculations, molecular structure-based models, stoichiometric approaches, use of proxies, and omission. Each method is technically described with examples from the literature, description of advantages and disadvantages, and discussion of suitability depending on availability of data and expertise. Methods are then fully demonstrated and compared for accuracy against reported chemical industry data using case studies of styrene and its downstream product, acrylonitrile-butadiene-styrene (ABS). Resulting LCI and life cycle impact assessment (LCIA) values are compared and discussed, with specific attention to method-specific exclusion of particular flows. Out of the four methods with which full LCIs can be generated, the advanced process-based methods give the most accurate life cycle GHG emission results for styrene and ABS, compared to plant data. Stoichiometric calculations, which are the most commonly used approach, underestimate the actual global warming results by 35–50%. Among the 18 impact categories of the ReCiPe LCIA method, results for the estimated LCI data were within 10% of the actual plant results for only 4–5 categories. Based on the critical review and demonstration results, we provide recommendations for appropriate use of LCI estimation methods in various LCA modeling situations.