Streamlined Life Cycle Assessment under Uncertainty Integrating a Network of the Petrochemical Industry and Optimization Techniques: Ecoinvent vs Mathematical Modeling

Abstract

Environmental databases have recently become an essential instrument in the sustainability evaluation of products. Unfortunately, these repositories still contain a limited number of chemicals. Furthermore, they are based on an attributional life cycle assessment (LCA) approach that considers fixed mass flows reflecting static industrial settings that are in practice dynamic, which might lead to errors. Building on some recent developments, we explore here an alternative approach to quantify the LCA impact of chemicals based on a network representation of the petrochemical industry coupled with linear programming, stochastic modeling and allocation methods. This method was applied to a network comprising 178 processes and 144 products, generating for most of the chemicals results that are consistent with those available in Ecoinvent for widely used impact categories such as GWP or those included in the ReCiPe 2008 methodology. The network model provides estimates of the life cycle impact embodied in chemicals under varying yields and demands, even for chemicals missing in standard repositories. Overall, we advocate for the use of network models of the petrochemical industry capable of carrying out consequential LCA under uncertainty as a complement to existing databases. This would allow to enlarge the capabilities of LCA repositories, thereby promoting the wider use of LCA in the chemical industry by improving the transparency and flexibility of the LCIA phase.