Uncertainty in life cycle impact assessment of toxic releases: Practical experiences - Arguments for a reductionalistic approach? Impact assessment of toxic releases in a substance flow analysis for PVC in Sweden

This paper describes the experience with impact assessment of toxic releases in a Substance Flow Analysis (SFA) for PVC in Sweden. For this system, all emissions related to the PVC-chain were inventoried. They have been evaluated making use of the Life Cycle Impact Assessment (LCIA) step from the CML-guide, including the new toxicity equivalency factors calculated with the Uniform System for Evaluation of Substances (USES). The application of this method led to the conclusion that LCA Impact Assessment of toxic releases is still a major bottleneck: the USES-equivalency factors are not to be trusted due to outdated data, inappropriate defaults, etc. in the USES' substance properties database. Therefore, a second USES-set of factors was calculated that differed up to factors of 1,000 or more from the old ones. Even these factors probably suffer from unacceptable high structural, in practice not reducible uncertainties. In conclusion, we warn the LCA community not to overestimate the possibility of LCA Impact Assessment to obtain a meaningful priority setting with regard to toxicity problems. Instead, we propose developing indicator systems for LCIA of toxic releases that genuinely deal with all relevant types of uncertainty: data uncertainty, modelling uncertainty and particularly paradigmatic uncertainty.