Affinely Adjustable Robust ADMM for Residential der Coordination in Distribution Networks

It is becoming increasingly important for network operators to coordinate the use of prosumer-owned DER so that their full value can be harnessed without violating the network's technical limits. Unfortunately, this coordination is complicated by the highly volatile uncertainties of PV and prosumer load, as well as the distributed nature of the problem. To address this challenge, we present an affinely adjustable robust extension of the distributed ADMM algorithm that is resilient to forecast deviations. As with ADMM, every prosumer negotiates with the grid over a receding horizon to obtain locational marginal prices and their coordinated 'here-and-now' decisions prior to the uncertainty realizations. However, our new AARO-ADMM approach robustly coordinates the first time step of each negotiation to enable prosumers to take local 'wait-and-see' recourse decisions that compensate deviations from the forecast in real-time. Our experiments on a modified 69-bus network show a significant reduction in the frequency of negotiation and communication needed by ADMM to maintain grid security, with just a small cost increase over an idealized but unachievable baseline.