Robust decision making in vehicle LCA, the need for uncertainty propagation

The aim of this manuscript is to enhance the interpretation of the environmental burden of different vehicle technologies, by incorporating uncertainty and variability in the end result. The scope is the Belgium car market of 2012.
One of the main shortcomings in literature on vehicle LCA is the lack of incorporating uncertainties. In most studies, environmental impacts are shown with single values. The complexity, uncertainty and variability of the system are not well approximated with single values. Single values approximate the environmental impact of a vehicle, but fail to provide a wider view on the possible effects. Uncertainties are an inherent part of LCA and should not be avoided but embraced and made explicit in the result. Identifying and integrating uncertainties in the result gives a more robust interpretation.
Many sources of uncertainty exist in LCA, in this paper two distinct classes of uncertainties are investigated, namely parameter uncertainty and modeling uncertainty.
Firstly, parameter uncertainty deals with the variability of the car market, the mismatches between the theoretic values derived from the New European Driving Cycle (NEDC) compared to the real life driving conditions and the unrepresentativity and inaccuracy of the background data (data from general databases). The most spectacular differences between the NEDC and the real-life values concern the emissions of NOx and CO. However, the contribution of the market variability to the overall uncertainty is larger compared with the uncertainties due to differences in driving cycle. The influence of the uncertainties on the result is made explicit by adding error bars in the LCA graphs.
Secondly, modeling uncertainty deals with the uncertainty in life cycle impact assessment due to normalization, weighting and methodology.
The choice of the Life Cycle Impact Assessment method itself introduces an uncertainty. The different vehicle technologies are compared using EDIP, Eco-Indicator 99, EPS and Impact 2002 to calculate the environmental performance.
As the impact methods have different viewpoints on environmental impact a large difference in ranking is observed.