Dempster–Shafer-informed levelized cost of energy: A robust approach for energy portfolio optimization under deep uncertainty

Energy management has become a strategic priority for governments worldwide, especially amid rising uncertainty driven by events such as the COVID-19 pandemic and the Russia-Ukraine war. These disruptions have intensified the volatility of energy markets and challenged traditional planning approaches. To address this, we propose a Robust Energy Portfolio Optimization (REPO) framework that operates under hybrid uncertainty, integrating both probabilistic and epistemic elements to support reliable long-term energy planning. A central contribution of this work is the incorporation of Dempster–Shafer Theory (DST) to model ambiguity in the Levelized Cost of Energy (LCOE)—a critical yet often disputed parameter in energy modeling. Unlike probabilistic methods alone, DST allows for the aggregation of expert opinions without assuming complete knowledge, enabling the REPO model to produce Dempster–Shafer-informed LCOE estimates that reflect real-world uncertainty more accurately. REPO optimizes energy portfolios based on three core objectives: cost-efficiency, environmental impact, and financial risk. We validate the model using both simulated and empirical data across 66 scenario configurations, demonstrating that the proposed portfolios outperform traditional approaches in terms of stability and robustness, while preserving (or slightly improving) their out-of-sample performance. The most environmentally favorable portfolio consists of 0.4% Critical Pulverized Coal Plants (CPCP), 3.7% Wind Power Plants (WPP), 16.7% Photovoltaic (PV), and 79.2% Combined Cycle Gas Turbine (CCGT). The most cost-effective solution comprises 100% CCGT, while the most balanced portfolio—minimizing deviations across all three objectives—includes 56.3% CPCP, 3.7% WPP, 16.7% PV, and 23.3% CCGT. These results underscore the effectiveness of the REPO framework in navigating hybrid uncertainty and aligning portfolio decisions with policy priorities. By explicitly modeling epistemic uncertainty and incorporating expert-informed estimates, REPO offers a flexible, realistic, and actionable tool for energy policymakers facing volatile and complex decision environments. • Proposes a robust energy portfolio optimization (REPO) model for energy management amidst uncertainty. • Introduces Dempster-–Shafer Theory (DST) to account for expert opinions on the Levelized Cost of Energy (LCOE). • Combines robust optimization and Polynomial Goal Programming (PGP) to optimize financial risk, sustainability, and CO 2 emissions. • Demonstrates the model with empirical data for Iran, highlighting the advantages of renewable technologies like wind and solar. • Provides practical insights for energy policymakers, improving decision-making under uncertain global events such as the COVID-19 pandemic and geopolitical crises.