(401b) Towards Systematic Design on Life Cycle Assessment Models By Accounting for Uncertainty and Network Complexity | AIChE

(401b) Towards Systematic Design on Life Cycle Assessment Models By Accounting for Uncertainty and Network Complexity


Ghosh, T. - Presenter, The Ohio State University
Bakshi, B., Ohio State University

Life cycle assessment(LCA) methods are used for studying environmental impacts of products and manufacturing processes from their "cradle to grave". LCA is plagued with several serious drawbacks such as subjective system boundary selection, uncertainty in data sources and aggregation. Neglecting activities from the inventory can result in exclusion of significant environmental impacts as negligibility of these processes is difficult to guarantee. Hybrid LCA methods were introduced for filling boundary 'gaps' not accounted for by traditional process system LCA using national input-output LCA data. Although hybrid approaches try to address the system boundary problem, location of boundaries between the different scales of hybrid LCA were arbitrary. From a broad literature review on LCA studies, it is found that the boundary selection as well as delineation of boundaries between different scales in multiscale LCA models have an impact on the validity of the final results and there is a compelling need to address this problem through a mathematically rigorous method directed towards removing subjectivity in these decisions. Very few studies have addressed the problem of boundary selection and delineation between multiple scales in LCA models without determining the uncertainty in result estimation.


This presentation describes a methodology using Structural Path Analysis(SPA) for exploration of the life cycle network of an activity and prioritize the LCA inventory according to the amount of environmental impact. An algorithm is developed in this study that generalizes the construction of multiscale life cycle models and determine model quality based on uncertainty analysis using analytical error propagation. It can be applied to single scale models, hybrid models as well as the Process to Planet framework (1). This work also combines two distinct modules to accomplish the model generation goal - network theory and uncertainty analysis and extends analytical error propagation approach to multiscale LCA models for performing uncertainty analysis to determine model quality. Another important addition of this work is the introduction of parameters for determining the complexity of life cycle models using price based or data aggregation granularity information. Overall, this algorithm reduces the subjective-ness in determining system boundary in LCA models and introduces parameters that guide the model generation process.


The algorithm is applied for the design of a biodiesel refinery life cycle network. First using SPA, the life cycle network information is extracted from economic models and then used to create a life cycle model using this algorithm. The important questions such as selection of activities to be included in the system boundary and the determination of confidence in the results are all obtained through this algorithm. Significant challenges still exist in terms of availability of uncertainty and complexity data at different scales of a life cycle model. The algorithm is built into a software with a graphical user interface for ease of use for LCA practitioners who without going into the intricacies of the algorithm’s functions can use it for building LCA models. We intend to provide a small demo of the software along with the presentation.


  1. Hanes, Rebecca J., and Bhavik R. Bakshi. "Sustainable process design by the process to planet framework." AIChE Journal 61.10 (2015): 3320-3331.