Real-time self-scheduling of Jintan AA-CAES plant in energy and reactive power markets

Advanced adiabatic compressed air energy storage (AA-CAES) has the advantages of large capacity, long life, and freedom from carbon emissions. The national demonstration project of AA-CAES plant in Jiantan, Jiangsu Province, is a successful application of this technology. Merchant AA-CAES can provide frequency and voltage support through participation in active and reactive power markets. Jiantan AA-CAES plant is charged at maximum power at night and can sell energy to the power grid from 8 a.m. to 12 p.m. This paper proposes a self-scheduling policy for Jintan plant. First, a self-scheduling model for AA-CAES participating in energy and reactive power markets is established. Then, this problem is transformed into a one-way trading problem, and a data -driven weight method is adopted to obtain the online dispatch policy. This method comes down to solving an equivalent linear programming problem and is thus highly efficient. The effectiveness of the proposed method is verified on Jintan AA-CAES plant. Compared to the ideal optimum, which utilizes the exact price sequence the next day, the proposed policy enjoys an optimality gap of 3.92%, significantly better than the classic policy for the one-way trading problem. The proposed strategy will enable Jintan plant to earn an average daily net income of 12,395$ in the energy and reactive power market, which outperforms the model predictive control (MPC) with a 5% prediction error at the prediction time domain of 2 h or 15% at the prediction time domain of 3 h. Compared with the MPC algorithm using ARIMA as the prediction model with a prediction error below 8%, the proposed method has better performance.