Position paper: Characterising performance of environmental models
Environmental Modelling & Software
| Hi-index | 0.00 |
Life Cycle Assessments (LCAs) of the same products often result in different, sometimes even contradictory outcomes. Reasons for these differences include using different data sets and deviating modelling choices. This paper purposely used different data sets and modelling choices to identify how these differences propagated in LCA results. Vehicle for this methodological exploration was an LCA case study of a typical polystyrene (PS) disposable cup. An initial LCA of PS cups was made using only one data set per process. Contribution and sensitivity analysis identified those processes with influential contribution to the overall environmental impact. Next additional data sets were acquired for all influential processes. The spread in impact results for each life cycle process was calculated after impact assessment for each individual inventory data set as to preserve the correlation between inventory data within each individual data set. The spread in impact results reflects uncertainty existing between different data sets for the same process and due to modelling choices. The influence on overall LCA results was quantified by systematically applying all combinations of data sets and modelling choices. Results from the different data sets and modelling choices systematically point to the same processes as main contributors to all impact categories (PS production, cup manufacturing, PS incineration and PS recycling). The spread in toxicity indicators exceeds the energy-related impact categories. Causes of spread are resources and energy used (type, amount, date and origin), reported emissions, and applied allocation procedures. Average LCA results show slight preference for recycling PS compared to incineration in most impact categories. Overlapping spread in results of the two waste treatments, however, does not support the preference for recycling. The approach in this paper showed how variation in data sets and modelling choices propagates in LCA outcomes. This is especially useful for generic LCAs as systematic use of multiple data sets and multiple modelling choices increases the insight in relative contributions of processes to, and uncertainty in the overall LCA. These results might be less easy to perceive, but they provide decision makers with more robust information.