Towards a probabilistic approach in LCA of building retrofit measures

This paper proposes an approach of sensitivity analysis for LCA of building retrofit measures aiming to establish the impact of input data uncertainties on the output variance. The approach includes the quantification of data input uncertainties in terms of their Probability Distribution Functions (PDFs), their sampling and the uncertainty propagation through Monte Carlo (MC) methods. A sensitivity analysis through Variance based decomposition (Sobol' method) techniques are used to point out the key parameters uncertainties that mostly affect the LCA results distributions. The paper presents a building case-study where the MC-based uncertainty and sensitivity analysis method is applied considering different design options (XPS and Cork internal insulation measures) and different scenarios for the assessment of the building energy need (use phase). Results obtained highlight that the differences on the Climate change environmental impact between the two design options is quite limited (about 12%) and this is mainly due to the use phase which is the more relevant input parameter on the overall result. Concerning the Sensitivity Analysis, when the building energy need is considered as a "deterministic" input in the LCA assessment, the unitary impacts of the design options materials uncertainties are the most influential parameters. On the other hands, when the building energy need is represented by a PDF, the quantity of energy carrier consumed and its unitary environmental impact are the most influential parameters on the output variance. .