The long term analysis of an energy-environment system is fraught with various uncertainties. Classical stochastic programming has been used with large scale LP models but it requires the assessment of scenario probabilities, which may be difficult. We propose a Minimax Regret formulation, which requires only identifying the possible states of nature. The formulation is implemented on the MARKAL model of Québec to analyze a composite scenario with five possible abatement targets, being revealed in the fourth period of the 9 period model. It has been experimentally verified that the minimax regret strategy depends only on the extremal reduction targets and not on the intermediate ones. Key results of minimax regret and minimum expected value strategies are compared.
Paru en mai 1997 , 23 pages