Distributionally Robust Generation Expansion Planning With Unimodality and Risk Constraints

As more renewables are integrated into the power system, capacity expansion planners need advanced long-term decision-making tools to properly model short-term stochastic production uncertainty and explore its effects on decisions. We develop a distributionally robust generation planning model, accounting for family of potential probability distributions wind forecast error uncertainty. Aiming include realistic distributions, we construct informed moment-based ambiguity sets by adding structural information unimodality. operational-stage unit commitment constraints risk operational limit violations in two distinct forms: chance conditional value-at-risk (CVaR) constraints. In both forms, resulting is mixed-integer second-order cone program. Using thorough out-of-sample numerical analysis, conclude: (i) chance-constrained exhibits better performance only if sufficiently accurate about first- moments as well mode location available; (ii) conversely, such unavailable, CVaR-constrained outperforms; (iii) these models have similar when unimodality excluded.