Consumer-Aware Distributed Demand-Side Contingency Service in the Power Grid

We consider the problem of flexible loads providing contingency reserves to the electric grid. We provide a distributed gradient projection (DGP) algorithm to minimize loads' disutility while providing contingency services. Loads use locally obtained grid-frequency measurements and inter-load communication to coordinate their actions. Privacy is preserved: disutility or consumption information is not exchanged. We provide a proof of convergence of the algorithm and compare its performance through simulations to that of a "dual algorithm" previously proposed in the literature that solved the dual problem. The DGP algorithm solves the primal problem. Its main advantage over the dual algorithm is that it is applicable to convex-but not necessarily strictly convex-consumer disutility functions, whereas the dual algorithm is not. A disutility function that is not strictly convex can better model consumer behavior than one that is strictly convex, since it is insensitive to small changes in consumption. Simulations show that the DGP algorithm is effective in arresting grid-frequency deviations, and performs better or similarly to the dual algorithm in cases where the two can be compared.