Preliminaries in Data-Driven Prioritization of Sustainability Indicators for Stationary Energy Storage: An Interpretable Machine Learning Approach

The study proposes a novel method for identifying sustainability indicators for stationary energy storage devices in their early design stages. In this preliminary assessment, literature-based data were collected and established in a database, then analyzed using the Apriori algorithm for association rule mining implemented in Python. Scope-and dimension-based evaluations revealed a well-balanced co-existence of environmental, social, economic, and ethical indicators under the scope of “optimization of integrated sustainability outcome.” The algorithm generated more than 100 rules, and by applying a support threshold of 0.25, a final set of indicators covering all four dimensions was derived. Some of the selected indicators include: GHG emissions, Renewable energy penetration, Energy efficiency, Levelized Cost of Storage, Operational/maintenance costs, Payback period, Resource consumption, Investment cost, Installation cost per kW, System lifetime, Supply chain transparency and traceability, Use-phase emissions, Accessibility of clean energy, Combined environmental impact, and Fairness in demand flexibility distribution. The method is non-biased, data-driven, and replicable, offering a robust foundation for sustainability evaluations.