Bringing Science and Pragmatism together in a Tiered Approach for Modelling Toxicological Impacts in LCA

Artikel i vetenskaplig tidskrift, 2004

Goal, Scope and Background. The EU 5 th framework project OMNIITOX will develop models calculating characterisation factors for assessing the potential toxic impacts of chemicals within the framework of LCA. These models will become accessible through a web-based information system. The key objective of the OMNIITOX project is to increase the coverage of substances by such models. In order to reach this objective, simpler models which need less but available data, will have to be developed while maintaining scientific quality. Methods. Experience within the OMNIITOX project has taught that data availability and quality are crucial issues for calculating characterisation factors. Data availability determines whether calculating characterisation factors is possible at all, whereas data quality determines to what extent the resulting characterisation factors are reliable. Today, there is insufficient knowledge and/or resources to have high data availability as well as high data quality and high model quality at the same time. Results. The OMNIITOX project is developing two inter-related models in order to be able to provide LCA impact assessment characterisation factors for toxic releases for as broad a range of chemicals as possible: 1) A base model representing a state-of-the-art multimedia model and 2) a simple model derived from the base model using statistical tools. Discussion. A preliminary decision tree for using the OMNIITOX information system (IS) is presented. The decision tree aims to illustrate how the OMNIITOXIS can assist an LCA practitioner in finding or deriving characterisation factors for use in life cycle impact assessment of toxic releases. Conclusions and Outlook. Data availability and quality are crucial issues when calculating characterisation factors for the toxicity impact categories. The OMNIITOX project is developing a tiered model approach for this. It is foreseen that a first version of the base model will be ready in late summer of 2004, whereas a first version of the simple base model is expected a few months later.