A validation process for probabilistic production costing programs

Validation of power production cost forecasts is addressed. Such programs cannot be validated by simply comparing predictions to outcomes. EPRI's chronological simulation program, BENCHMARK, can be used in a Monte Carlo mode to produce approximate production cost distributions. An exhaustive enumeration experiment can be conducted to measure their accuracy. An example is given for a system with seven thermal units. Three price scenarios were included on the cost of replacement power for generation shortfalls. Chronological simulation programs such as BENCHMARK can also be calibrated when used in the deterministic mode to match system operating policy. The Monte Carlo procedure can then be used to validate production costing results for large systems with some assurance. Cumulative probability distribution curves produced in this process also present useful information for decision-making. >