A new method for power generation planning and problems of uncertainty

This paper presents a new method for power system planning which considers uncertainties of various system parameters as a part of the criterion function. The criterion function involves both the expected values of total system costs and the cost variances of different power plants. Previously, the authors proposed that these cost variances be evaluated from the existing power system plans in the form of the solutions to the inverse optimization problem, and then applied this procedure to the planning of power systems in Japan. However, because of the correlation between fuel and facility costs, the cost variances could not be determined as specific values. Here, to overcome this difficulty, the authors now introduce hydropower plants into the model as the so-called nonrisky elements and focus on the expenditure shares rather than the plant capacity shares. As a result, more reasonable results are obtained. This improved methodology is applied and the derived cost variances are shown. Further, using the resultant cost variances, the authors make Pareto-optimal power generation plans and demonstrate also the robustness of the method to price fluctuations.