Nuclear energy technology R&D portfolio selection under scenario uncertainty: distributionally robust ordinal priority approach

A Selecting appropriate nuclear energy technology (NET) R&D portfolios is essential for shaping the national nuclear energy landscape, supporting global carbon reduction efforts, and advancing the UN Sustainable Development Goal for affordable and clean energy. However, research on NET R&D portfolio selection (NET-R&D-PS) remains limited and fails to adequately address the scenario uncertainty. Thus, this study proposes a distributionally robust ordinal priority approach (OPA-DR) for NET-R&D-PS under scenario uncertainty that affects the importance of evaluation attributes. Although the alternative rankings under possible scenarios and their corresponding nominal distributions would be provided, the high uncertainty of future R&D scenarios renders the nominal distributions unreliable. To address this, this study introduces an ambiguity set based on Kullback–Leibler (KL) divergence for OPA-DR, with ambiguity set sizes designed for large- and small-sample problems, characterizing all possible attribute ranking distributions derived from the nominal distribution. This study develops an efficient exact solving algorithm for OPA-DR, requiring only the solution of a one-dimensional equation and the calculation of the optimal solution in closed form with polynomial time complexity, making it suitable for large-scale problems. This study analyzes the OPA-DR sensitivity under varying utility functions and constraint perturbations. The effectiveness of OPA-DR is validated by the NET-R&D-PS for China 2030 Vision Plan, providing insights for scenario analysis, attribute selection, and portfolio selection.