Life cycle inventory uncertainty in resource-based industries : a focus on coal-based power generation

Notten, Philippa Josephine

Life cycle inventory uncertainty in resource-based industries : a focus on coal-based power generation

Doctoral Thesis

2001

Publisher

University of Cape Town

Department

Department of Chemical Engineering

Faculty

Faculty of Engineering and the Built Environment

Abstract

The aim of this thesis is to develop an approach to support prospective environmental decision-making in resource-based industries. The specific focus is on coal-based power generation. The objectives of the approach are that it be able to adequately reflect the environmental burdens arising from primary industries, and to make explicit the trade-offs often encountered in environmental decisions. In addition, it needs to take into account that the context in which the assessment takes place affects data availability and quality significantly, and consequently the certainty with which systems can be evaluated. Resource-based processes typically involve large-scale disruption of the local and regional environments, with imprecise processes and diffuse emissions. The modelling of the environmental performance of such processes therefore raises significant challenges, where many disparate sources of data, available at different levels of aggregation, and over various time intervals, have to be brought together into a coherent assessment. An "uncertain" definition of the system is therefore much more meaningful, in which variables are defined over ranges of values to cover inconsistencies and imbalances in the system. The inherently high variability of mining and minerals processes further supports their modelling as ranges of potential performance rather than "typical" operations, where the relevant process of interest must be identified and the variability within the particular process incorporated into the assessment Life cycle assessment (LeA) has received increasing attention for its role in environmental decision making processes, where it supports the process of defining the contribution of human activities to (at least the environmental dimension of) sustainable development. It is therefore the structured approach to environmental decision-making investigated in this thesis to organise the large data sets of varying quality and completeness available around resource-based industries into useful information, able to provide the environmental objective in a decision-making process. LeA is an inherently uncertain procedure in that it combines data sources of varying accuracy and representativeness, and employs subjective judgement in applying this data to future operating systems. Subjective judgements are also present in the definition of the systems, and in the modelling choices determining the accuracy and complexity of the inventory and impact models used. Nonetheless, LeA results are most often presented as single values, which in a comparative analysis, gives the often incorrect impression that one system is always better or worse than another system. A framework has been developed in this thesis to include all relevant sources of uncertainty encountered in LCA models explicitly, where empirical parameter uncertainty, model parameter uncertainty, and uncertainty in model form are investigated in a looped fashion. The innermost loop assesses empirical uncertainty in an iterative probabilistic analysis, using Latin Hypercube sampling of the uncertain input distributions to propagate the data uncertainty to the output, and rank-order correlation analyses to determine the relative uncertainty importance of the parameters input into the model. Model parameter uncertainty is assessed next, by a parametric analysis, or by a combination of sensitivity analyses and a parametric analysis, if a large number of model parameters require consideration. The top-most layer is an assessment of model form, in which alternative model forms are investigated in a sensitivity analysis.

Keywords

Chemical Engineering

Reference:

Collections

PhD / Doctoral

Full item page