Robust online operation of power systems with advanced adiabatic compressed air energy storage and renewable generation

With the rapid growth of renewable generation in power system, the share of dispatchable power reduces, raising the need for large-scale energy storage to compensate for volatile renewable output. Advanced adiabatic compressed air energy storage (AA-CAES), advantageous in large capacity and emission-free, is promising for grid-level use. This paper studies the operation of power systems with renewable generation and AACAES. The off-design characteristics of AA-CAES are featured by variable efficiencies and limited low-load capacity. A battery-like linear model of AA-CAES is suggested, in which the coefficients can be calibrated from experiments or simulations. A robust online operation scheme is proposed which only requires the currently observable information and makes no reference to renewable forecast. In the day-ahead stage, the unit on- off status as well as lower and upper bounds of unit output and storage levels in all periods are derived from a robust unit commitment model. In the real-time dispatch stage, an optimal power flow problem with pre-allocated operation bounds is solved. Feasibility is guaranteed by the setting of operation bounds obtained from the robust model. Case studies illustrate the precision of the simplified AA-CAES model and validate the performance of the operation method. The results indicate that a linear model is suitable when AA-CAES operates in near constant-pressure conditions, whose error is 0.4% for the charging part and 1.4% for the discharging part. A non-anticipative robust formulation is superior to conventional two-stage robust optimization in feasibility enforcement under uncertainty, the later encounters an infeasible probability up to 42% when the power system lacks limited regulation sources.

Automated summary

This paper investigates the operation of power systems with renewable generation and advanced adiabatic compressed air energy storage (AA-CAES). A battery-like linear model and a robust online operation scheme are proposed to describe and optimize the operation of AA-CAES. Case studies validate the accuracy of the model and the performance of the method.