Dynamic programming-based energy storage siting and sizing: Application to enhance flexibility of large-scale power grid

Abstract

To address the issues of limited Energy Storage System (ESS) locations and the flexibility unevenly distributed in the large-scale power grid planning, this paper introduces the Dynamic Programming (DP) theory into flexibility planning, and proposes a DP-based ESS siting and sizing method. This method reduces the computational complexity of siting and sizing to ensure a sufficient number of ESSs allocated. It provides each partitioning area with a certain degree of flexible ramping capability so that the flexibility is evenly distributed in the large-scale grid.

The proposed method starts with a high-voltage pruning partition algorithm to hierarchically partition the large-scale grid, with the partitioning outcomes serving to divide the various DP stages. Then a state transition equation is established with the ESS rated power as the state variable, considering all nodes which satisfy the voltage level requirements as potential ESS sites to ensure a sufficient number of locations. Following this, a DP basic equation is formulated with the ESS capacity as the decision variable, setting flexibility constraints for all partitioning areas to achieve an even distribution of grid flexibility. By combining the state transition equation and the DP basic equation, the proposed method culminates in the energy storage allocation dynamic programming model, which determines the optimal locations, capacities, and rated powers of ESSs, along with the construction cost.

This paper further explores the development of the Flexible Resource Allocation Intelligent Decision Software (FRAIDS) building upon the proposed method. Case analysis in an actual grid verifies that the calculations from FRAIDS significantly enhance the entire grid flexibility. Additionally, day-ahead dispatching results indicate that, following ESS allocation, net load fluctuates between 15,295.5 MW and 17,794.9 MW with the conventional method, compared to a more stable range of 16,309.8 MW to 17,417.4 MW with the proposed method. This shows that the proposed method effectively reduces net load fluctuations, thereby significantly alleviating flexible ramping pressure on thermal units.