Multi-Scale Control of Battery Systems

Electrochemical storage provides dynamic flexibility to modulate electrical loads at various frequencies and aid the control of distribution and transmission networks.  This flexibility is key to manage large amounts of intermittent and non-dispatchable renewable power and ensure energy quality.  Generating revenue for batteries in electricity markets, however, is challenging because: i) there is significant uncertainty on electricity prices and on revenue streams that can be generated from different markets (day-ahead, real-time, regulation) and ii) there is significant uncertainty on the effect of charge/discharge profiles and capacity limits on the lifetime of the battery.  As a result, it is not clear how to optimally determine storage and sell/buy allocations given local electricity market conditions. 

In this project we are developing stochastic optimal control formulations that capture revenue streams at multiple time scales from day-ahead, real-time, and frequency regulation markets and that seek to determine optimal allocation portfolios and buy/sell policies that maximize revenue. We perform exhaustive year long and minute-by-minute closed-loop simulations using real data from ISOs at different pricing hubs. We are also using historical market data to determine probability distributions for different market prices and measure risk of different revenue streams. To deal with the associated complexity of the computations involved we use cutting-edge optimization tools capable of running of multi-core computing systems. Our models are implemented in PLASMO and solved with DSP/PIPS. 

 Li-Ion Battery System