A Multiscale Bayesian Framework for Environmentally Conscious Process Design

Improving the sustainability of chemical processes requires techniques for including life cycle environmental considerations in product and process design. Data for life cycle assessment are usually available at different spatial resolutions such as, the economy, supply chain and equipment. These data vary in their degree of completeness in capturing the life cycle, and their level of uncertainty. For example, equipment scale data are relatively accurate, but narrow in scope, while economy scale data capture the economic network, but are aggregated. This paper presents a new framework for using all kinds of data and models for obtaining the life cycle inventory of a selected product or process. Such information is essential for assessing the life cycle impact of process alternatives and for incorporating environmental aspects in process design. The proposed multiscale framework is represented as a treebased model and can be used in deterministic or stochastic formulations. The stochastic formulation provides the probability distribution for the life cycle state variables via a hierarchical Bayesian data fusion algorithm. Case studies illustrating the framework are in progress.