Evaluation and improvement of variance reduction in Monte Carlo production simulation

A computer algorithm which combines several variance reduction techniques to enhance the precision of Monte Carlo production simulation is designed. The techniques included are stratified and antithetic samplings and linear regression estimation. For stratified sampling, a mathematical rule which can always lead to a near-optimum stratification is presented. The variance reduction by modelling generating units´ outage according to their uptime/downtime distribution in comparison with modeling by forced outage rate is investigated. Numerical test results achieved by applying the algorithm to cost and environment evaluations in an actual Taiwan power system are examined