A binary‐regularization‐based model predictive control applied to generation scheduling in concentrating solar power plants

Abstract

SummaryThis paper proposes the use of model predictive control (MPC) with binary‐regularization to manage the electric power generation problem in concentrating solar power plants with thermal energy storage. The main advantage of the of MPC with binary‐regularization formulation is the inclusion of a power block protection method based on a binary‐regularization term that penalizes power generation variation (also called generation cycling) differently according to the power block situation, ie, normal operation, startup, or shutdown. This distinction simplifies the choice of schedules with reduced variation and high energy sale profits. The interest in this reduction is the achievement of a higher lifetime of the power block elements, lower maintenance costs, and easier plant operability. A benefit of the generation scheduling based on MPC is the capacity of rescheduling the power generation at regular periods, taking advantage of the most recent energy prices and weather forecast, and of the plant's current state. An interesting question is if the proposed protection mechanism affects the economic results of the MPC black strategy. In this regard, an economic study based on a realistic simulation of a 50 MW parabolic trough collector‐based concentrating solar power plant with thermal energy storage, under the assumption of participation in the Spanish day‐ahead energy market scenario, is included. Realistic values for actual and forecasted solar resource and for energy price are used, and for penalties for deviation from the committed generation schedule. The economic study shows that the proposed scheduling method provides an important reduction of the generation cycling without decreasing energy sales profits. Another advantage of the proposed method is the possibility of estimating the highest level of power block protection, which maintains the profits by means of historical data, which favors its practical implementation.