Discrete convolution in generating capacity reliability evaluation-LOLE calculations and uncertainty aspects

The authors present a practical and general method based on discrete convolution techniques for evaluating the expectations and variance of the loss of load indices for a single-area power system by treating forced outage rates and peak load forecasts as random variables with specified means and variances. The proposed method easily accounts for partial forced outage rates associated with unit derated states, without any restrictions. The proposed algorithm is applied to typical generating systems, including the IEEE Reliability Test System and the Brazilian South/Southeast System planned for completion in 1991