IMPROVEMENT ON STATE FORMULATION, STRATIFICATION AND ESTIMATION IN MONTE-CARLO PRODUCTION COST SIMULATION

A computer algorithm designed for Monte Carlo production cost simulati on is described. The design seeks to enhance the precision of producti on cost estimation at a reduced computation time compared with the exi sting approach. The algorithm has three major computer steps: (i) to f ormulate a state population according to the uptime/downtime of the ge nerating unit, hence to simulate generation availability; (ii) to stra tify this population by a mathematical rule and draw sample states out of each stratum; and (iii) to estimate the population mean of unit co mmitment production costs by the rank statistics of the sample. The st ratification rule aims to remove any judgemental input and to render t he stratification process entirely mechanistic. The estimator, given b y rank statistics of the sample, can avoid identification of the regre ssion model and thus save computation time. The algorithm has been eva luated and compared numerically with the existing approach. Hence, the effectiveness on precision improvement is demonstrated.