Estimation of production cost variance using chronological simulation.

Abstract

Production costs are usually simulated on the basis that the availability of generation capacity is subject to random failures of system generating units. In order to estimate the variance of cost, both the random forced outages of units and load uncertainty should be modeled in a production cost simulation. In this dissertation, the effects of uncertainties in generation availabilities will be analyzed using a Monte Carlo approach.

In this long-term (annual) load model, emphasis is placed on modeling the variation in chronological load so that a chronological production cost simulator can efficiently produce an estimate of the variance of annual production cost. A stochastic approach, using a conditional weekly sampling scheme, is proposed to model the annual load variation. Then, a probabilistic approach using stratified sampling is proposed to model the load variation on an annual basis. Furthermore, the stochastic and probabilistic approaches are compared in terms of accuracy and effort. The result of this dissertation is to provide a model that can be used to estimate the variance of annual production cost using a chronological production cost simulator. Thus, the variance in production cost can be expressed as a function of load uncertainty and uncertainty in generator availability.

This dissertation describes research on the effects of the uncertainty in annual load variation and uncertainty in generation availability on the variance of production cost in an electrical power system. Two different approaches of load uncertainty modeling are developed. The load uncertainty modeling accounts for uncertainty when reliable weather (e.g. temperature) forecasts are not available. Both approaches can be used to estimate the variance of long-term production cost, typically for an annual study.