Treatment of Uncertainties in Life Cycle Assessment

Spurred by European Union directives along with increased interest among US initiatives, life cycle assessment (LCA) is becoming a common tool used in the measurement and evaluation of environmental performance and overall sustainability. Life cycle assessment is an analytical technique for assessing potential environmental, social, and economic burdens and impacts of a given product or process, encompassing all stages of life cycle, from raw material production through end-of-life management. LCA provides metrics that can be used to measure progress toward sustainability. A life cycle assessment includes a number of phases: goal and scope definition, inventory analysis, impact assessment and interpretation. Each of these phases has significant associated uncertainties. Decisions made without regard to these uncertainties may be flawed. This paper describes common sources of uncertainty, methods for quantifying uncertainty, and methods for propagating input and model uncertainties in order to determine the resulting uncertainties in final estimated environmental impacts. An example application is presented to illustrate that even if only the most critical uncertainty sources are identified, it is feasible to obtain useful information about uncertainties in a given assessment. This application also illustrates the need to prioritize the many sources of uncertainty present in a given LCA. There are many practical challenges to consider in this very broad field, but also a variety of areas in which further achievements can be made. This paper serves to illustrate opportunities to further apply the tools of probabilistic modeling to the important and rapidly growing field of life cycle assessment. After describing the importance of this topic and identifying tools for performing relevant analysis, prospects for future progress in this field are considered. There are significant practical challenges to implementing explicit uncertainty treatment, but also many areas in which further achievements can be made. The few examples where uncertainty has been treated explicitly in LCA assessments illustrate the opportunities that exist to further apply probabilistic modeling to this important and rapidly growing field. 2 THE VALUE OF QUANTIFYING LCA UNCERTAINTIES A general motivation for quantifying uncertainties is to increase the transparency of LCA data and results. Uncertainty is undeniably present in many aspects of analysis, and treating it explicitly will aid in several ways, as outlined below. 2.1 Decision support When making comparisons among design alternatives, apparent differences in impacts may be misleading if the uncertainty in impacts is large enough to overwhelm any relative differences between alternatives. Quantification of these uncertainties will support informed decision making (Basson and Petrie 2004; Cowell et al. 2002; Lenzen 2006; 2000). When design alternatives are being evaluated in the presence of significant uncertainties, large uncertainties may make it impossible to determine whether one design alternative is truly superior to another. This type of situation arises frequently in other fields, and the statistics topic of hypothesis testing is a mature field used, for example, to evaluate the true benefit of a new drug that is being considered for approval by a regulatory agency. In the field of LCA, Basson and Petrie (2004) have considered similar problems and termed their approaches “distinguishability analysis.” Whatever the specific name and technique, it is important to recognize that apparent differences in impacts from design alternatives may be small enough relative to uncertainties in estimated impacts as to be insignificant. Consideration of uncertainty is thus critical in this case for sound decision making.