Renewable sources are increasingly penetrating the electric energy systems. However, renewable production is fluctuating, and the forecast of future production involves uncertainty. Thus stand-by energy production and/or demand response is required to maintain power sufficiency. As battery costs are reducing, battery storage technologies emerge as a credible alternative to mitigate the uncertainties of the current grid, especially as the number of Electric Vehicles increases.
As investment costs for Battery Energy Storage Systems (BESSs) represent a large portion of the total cost, and the BESSs can be used only for a limited number of cycles, it is important to optimize the BESS operation over its lifetime, which is, however, operation dependent. To this end, we formulated a novel stochastic optimal control problem consisting of the ratio of two long-time average cost criteria, namely the ratio of long-time average profits over long-time average degradation. Then, we developed efficient Dynamic-Programming-based algorithms that exploit the problem structure for this kind of criterion. This work is currently being extended to include a realistic, physics-based battery degradation model.
Submitted Paper:
I. Kordonis, A.C. Charalampidis, P. Haessig 'Approximately Optimal Control of MDPs over a
Long Operation-Dependent Time Horizon and
Application to Battery Energy Storage Systems'
For the Battery Energy Storage Systems research, I cooperate with Professors Alexandros Charalampidis and Pierre Haessig, from CentraleSupélec, Rennes.