Distributed secondary frequency control algorithm considering storage efficiency

Storage technologies, such as batteries, have higher ramping capability than conventional generators, and so they could lead to better dynamic tracking performance when following a frequency control signal. However, storage units are energy constrained, so the energy content of the signal they follow should preferably be low. Our first contribution is a method for decomposing the target signal into a different signal for each generator and each storage unit, depending on their characteristics and on the storage energy levels. Faster units receive signals with higher ramping rates. Storage units with low energy-to-power ratios ( τ_i ) receive signals which are energy-neutral (efficiency-corrected) in shorter time frames than the signals for storage units with larger τ_i. Our method is particularly suitable when the τ_i vary by orders of magnitude over the storage population. It dispatches the units by solving a modified economic dispatch problem. The energy content of the storage unit signals is managed through a term penalizing energy level deviation in the cost function. The second contribution is a distributed implementation of our method, which approaches the dispatch of its centralized counterpart. Our distributed implementation innovates compared to other distributed algorithms in that it considers non-unity storage efficiency. Using a population of 1000 units we show that the distributed implementation is: 1) up to 700 times faster than the centralized implementation, therefore allowing for real-time implementation with a large number of units; and 2) robust to loss of communication links.