Hedging large-scale distribution system investments against uncertainty

The paper addresses the problem of large-scale distribution network planning under uncertainty. We develop a hedging algorithm to deal with scenario representation of uncertainty in large-scale distribution systems. Specific algorithms are developed to meet the large-scale combinatorial problem, and resolve the conflicts on scenario first-stage decisions. The problem is decomposed into a sequence of scenario deterministic problems, solved approximately by a network specific evolutionary procedure: convergence constraints are relaxed. The hedging process is taken to resolve decision conflicts by successively proposing better objective pseudo-decisions. Those are to be progressively approached by each scenario. The marginal costs of each resolved conflict provide valuable first-order information on the decision robustness. The process robustness is discussed and illustration is provided for a large-scale real network planning problem.