Integrating Storage to Power System Management

TL;DR

This paper explores the integration of energy storage in power system management with wind energy, using approximate dynamic programming to handle large networks efficiently and compare with classical methods.

Contribution

It introduces an approximate dynamic programming approach for managing wind and conventional power units with storage, addressing scalability issues of traditional methods.

Findings

The proposed algorithm balances solution time and accuracy effectively.

It performs well on networks of various sizes.

Comparison shows advantages over classical dynamic programming on small networks.

Abstract

Wind integration in power grids is very difficult, essentially because of the uncertain nature of wind speed. Forecasting errors on output from wind turbines may have costly consequences. For instance, power might be bought at highest price to meet the load. On the other hand, in case of surplus, power may be wasted. Energy storage facility may provide some recourse against the uncertainty on wind generation. Because of the sequential nature of power scheduling problems, stochastic dynamic programming is often used as solution method. However, this scheme is limited to very small networks by the so-called curse of dimensionality. To face such limitations, several approximate approaches have been proposed. We analyze the management of a network composed of conventional power units as well as wind turbines through approximate dynamic programming. We consider a general power network model with ramping constraints on the conventional generators. We use generalized linear programming techniques to linearize the problems. We test the algorithm on several networks of different sizes and report results about the computation time. We also carry out comparisons with classical dynamic programming on a small network. The results show the algorithm seems to offer a fair trade-off between solution time and accuracy.