Auctions for congestion management in distribution grids

Abstract

Large controllable loads, such as electric vehicles, are increasingly penetrating electricity distribution feeders. To avoid local congestion, their consumption behaviour must be steered, for which a real-time price propagated down from the transmission system does not suffice, as it does not reflect local grid conditions. To efficiently steer the charging of EVs by multiple self-interested parties, we propose an auction framework which accounts for local grid conditions, the limited flexibility of EVs, and the uncertainty inherent to small-scale networks. We formulate the EV charging problem as a job scheduling problem for self-interested aggregators, and auction network capacity
for discrete time slots using sequentially-cleared auctions, which run in parallel. We simulate this auction on a local network using realistic data for EV driving behaviour and network capacity, showing this method leads to feasible allocations which are fairer in case one party is weaker than the other due to size or information asymmetry.