Rolling-horizon dispatch of advanced adiabatic compressed air energy storage based energy hub via data-driven stochastic dynamic programming

Faced with environmental pollution and energy crisis, energy hub yields an improvement on efficiency and flexibility of multi-energy supply. Advanced adiabatic compressed air energy storage (AA-CAES) is a promising large-scale energy storage technology and is attracting increasing attention due to its heat-electricity co-storage potentials. This paper investigates the external characteristics of advanced adiabatic compressed air energy storage and exploits its ability to implement an energy hub. First, a dual state-of-charge (SoC) model of advanced adiabatic compressed air energy storage is presented, taking into account the system off-design features and the impact of ambient temperature. The state-of-charge of the air storage tank depends on the mass of stored air, whose mass flow rate affects the charging and discharging electric power. The state-of-charge of the high-temperature thermal energy storage depends on the mass of heat transfer oil, whose mass flow rate determines the reserving and releasing heating power. Adjusting the mass flow rates of air and oil offers flexible control on the power and thermal outputs. An energy hub is built based upon the advanced adiabatic compressed air energy storage. To address the daily self-dispatch of the energy hub facing the uncertainties of load and ambient temperature, a data-driven stochastic dynamic programming model is proposed which allows a rolling horizon implementation. The Kernel regression is employed to estimate the conditional probability distribution of uncertainties. The cost-to-go functions in the Bellman equation are approximated via sampling and interpolation. Case studies validate the effectiveness of the proposed approach. The results indicate that: 1) The proposed dynamic programming method outperforms model predictive control in computational efficiency. 2) Neglecting the temperature effect on compressed air energy storage operation leads to 4.5%, 7.8%, and 9.2% regulation errors of charging, discharging and heating power, respectively.