A stochastic discrete choice dynamic programming model of power plant operations and retirement

We present a methodology to estimate fixed cost parameters relevant to the decision to operate, mothballor retire an open-cycle gas turbine (OCGT) using a dynamic discrete choice model, based on fuel andelectricity prices, as well as technical data and the operational status of OCGTs in the PJM market area.With operational and mothballed OCGTs, we find for both, age of the power plant and plant vintagestatistically significant positive correlations with the fixed operation and maintenance (O&M) costs. Wealso show a statistically significant negative relationship between the installed capacity and the fixed O&Mcosts, confirming that an increase in scale results in lower specific costs. The estimated fixed O&M costparameters for an operational OCGT vary from 15.3 USD/kW/yr for new, large, high-efficiency units, to 50.8USD/kW/yr for older, small, low-efficiency units. Mothballing a plant reduces these costs by 75% to 95%,depending on plant vintage and size. Decommissioning an OCGT was found to be cash flow negative, whichmeans that the associated cost exceeds any scrap value the equipment may have on secondary markets. Ourestimated cost parameters depend on operational status, capacity, vintage, and age of a generation unit. Thisdifferentiation is valuable for a better understanding of costs in the context of competition policy. It wouldalso allow for a more realistic parameterisation of power market models. Using the estimates and marketdata, we also compute the probabilities of operating, mothballing or retiring an OCGT. Sensitivity analysesregarding changes in prices of capacity, electricity, and natural gas reveal that the operating decisions forOCGTs are significantly affected by the profitability potential, most notably by electricity prices.