Developing a Framework for Improving the Accuracy of Process-based LCA for Energy Pathways

  • Author / Creator
    Di Lullo, Giovanni Roberto
  • Life cycle assessment (LCA) is becoming a popular tool to quantify environmental impacts including greenhouse gas (GHG) emissions. Because of the high number of assumptions and low quality of available data, the results of LCA are often viewed with skepticism. It is common for studies to provide deterministic point estimates and a limited sensitivity analysis and not account for uncertainty in the data and assumptions used, which can also lead to a lack of confidence in the results. In order to further increase the usefulness of LCA results for decision makers, a robust methodological framework that can be used to accurately quantify uncertainties and communicate results is needed. Furthermore, obtaining accurate data of certain industrial activities requires complex engineering models that have long computing times, are difficult for non-experts to use, and may contain confidential data. Proxy modeling is investigated here to create an accurate, easy-to-use, black-box model that can be easily shared.A survey of the existing literature was performed to examine how practitioners are currently implementing sensitivity and uncertainty. The survey found sensitivity and uncertainty analyses were inconsistent and basic, and the methods/assumptions lacked proper justification. Multiple sensitivity and uncertainty methods were investigated, leading to the development of the Regression, Uncertainty, and Sensitivity Tool (RUST) and framework. The Morris and Sobol global sensitivity methods used in RUST were examined to determine whether they can accurately identify the key inputs that have the largest effect on overall output variance. RUST was validated using the previously published FUNdamental ENgineering PrinciplEs-based ModeL for Estimation of GreenHouse Gases in Conventional Crude Oils and Oil Sands (FUNNEL-GHG-CCO/OS) and FUNNEL-GHG-Natural Gas Transmission Lines (NGTL) as case studies. After reviewing multiple proxy modeling methods, quadratic and artificial neural network (ANN) regression proxy models were investigated to create an accurate, easy-to-use black-box model that can be easily shared. Generating target values needed for training from the engineering software can be time consuming; hence, adaptive sampling methods were examined (random, spread, high error, and 50/50 random/high error). It was found that while both the Morris and Sobol methods can identify the key parameters, the Morris method requires fewer than 1/100th as many model evaluations as Sobol. RUST and the corresponding framework can be used to improve the quality of the LCA and reduce the time required by the practitioner. Quadratic proxy modeling works well for models that exhibit nearly linear behavior, but the ANN proxy models are superior for iterative non-linear models. The results found that ANN proxy models are more accurate than quadratic regression, and the high error sampling method reduced the maximum error but increased the average error. Because of uncertainty in LCA input values, reducing average error is less valuable than reducing extreme errors. The regression model can be easily published, it does not require a large effort to make a user-friendly version of the model, and it conceals confidential data if necessary. The simplified model makes it easy for policy makers to investigate how changes in critical parameters affect LCA results without having to learn how to use the full complex model.

  • Subjects / Keywords
  • Graduation date
    Fall 2022
  • Type of Item
  • Degree
    Doctor of Philosophy
  • DOI
  • License
    This thesis is made available by the University of Alberta Library with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.