Probabilistic load variation modeling for estimating the variance of annual production cost

Abstract

As the electrical industry becomes more competitive, the demand for more accurate cost simulation models has increased. In addition, there is increasing demand to estimate not only the expected value of cost, but also some measure of the risk associated with the cost estimate. The problem described in this paper arose because engineers in an electrical utility were asked by management to estimate the risk or uncertainty when proposing projects including fuel purchase, power transactions, generation modification and building new generation. This paper presents a method of estimating the risk by estimating the variance of the production cost. Existing chronological production cost models have included the uncertainty of generating unit availability; however they have emphasized the expected cost, rather than the variance of the cost. This paper includes the variance of load as well as the variation in unit availability, and it is shown for a sample system that the load contributes approximately five times as much cost variation as does unit outages. This probabilistic approach models unit availability using Monte Carlo sampling and simulates load variation using three load scenarios based on a stratification technique. The authors apply the proposed approach to illustrate the impact of load uncertainty and emphasize the estimation of cost variance which is induced by load variation and unit outages.