Development of a multiattribute value function for long-range electrical generation expansion

Long-range expansion planning of generation facilities for electric utilities is an extremely complex problem for several reasons. One of the major reasons for complexity is the multiple objective nature of the problem. The two most prevalent objectives in generation planning involve: 1) minimizing the discounted capital and operating costs of the system over some planning horizon and 2) maximizing some measure of reliability over the planning horizon. Commonly, only the first objective is considered explicitly while the second objective is treated implicitly, by means of a constraint placed on the measure of reliability (e.g., loss of load probability and reserve generation capacity). In this paper, we employ and demonstrate a multiattribute value (MAV) measurement technique due to Keeney and Raiffa [11] to identify a MAV function for generation expansion planning, containing attributes of: 1) the loss of peak load probability for each time period (e.g., one year) of the planning horizon for a proposed generation system and 2) the total discounted capital and operating costs over a planning horizon for a proposed generation system. This MAV function is developed through interaction with planning department personnel (decision maker) of a midwestern electric utility company. Indications are given as to how this value function can be interfaced with optimization models for generation expansion planning and to the feasibility of extending the function to consider attributes concerned with, for example, environmental degradation and other social factors.