Joint supply, demand, and energy storage management towards microgrid cost minimization

The problem of real-time power balancing in a grid-connected microgrid is studied. We consider a microgrid powered by a conventional generator (CG) and multiple renewable generators (RGs) each co-located with one distributed storage (DS) unit. An aggregator operates the microgrid and aims to minimize the long-term system cost, including all RGs' cost, the CG's cost, and the cost for exploiting external energy markets. We jointly manage the supply side, demand side, and DS units, taking into account the randomness of the system, and incorporating the ramping constraint of the CG. A real-time algorithm is proposed, which does not require any statistics of the system. We analytically characterize the gap between the system cost under our algorithm and the minimum cost, demonstrating that our algorithm is asymptotically optimal as the DS energy capacity increases and the CG ramping constraint loosens. In simulation, we compare the proposed algorithm with a greedy algorithm as well as a lower bound on the optimum. Simulation shows that our algorithm outperforms the greedy one and its performance can be close to the optimum even with small DS units.