Implementation of uncertainty analysis and moment‐independent global sensitivity analysis for full‐scale life cycle assessment models

TLDR

Life cycle assessment models and databases have grown in size, resolution, and complexity, increasing the number of uncertain inputs analysts must manage. The study aims to develop systematic methods for assessing uncertainty and sensitivity in LCA model outputs. The authors provide a theoretical framework and implement it in the open‑source Activity‑Browser LCA software, combining uncertainty analysis with moment‑independent global sensitivity analysis, and demonstrate it on a crystalline silicon photovoltaic case. The approach was successfully demonstrated on a crystalline silicon photovoltaic case study.

Abstract

Abstract Life cycle assessment (LCA) models and databases have increased in size, resolution, and complexity, requiring analysts to rely on an ever‐increasing number of uncertain model inputs. Such increased complexity calls for systematic approaches to assessing the uncertainty of the output results of LCA models and the sensitivity of LCA model outputs to the model's uncertain inputs. In this contribution, we provide a theoretical basis and present a practical software implementation that combines uncertainty analysis and moment‐independent global sensitivity analysis, which can be readily applied to full‐scale LCA models. We implemented our approach in the Activity‐Browser open source LCA software and it is made available for use in LCA studies. We demonstrate the approach and software implementation with a case study of crystalline silicon photovoltaics.