Generalizing Demand Response Through Reward Bidding

Abstract

Demand-side response (DR) is emerging as a crucial technology to assure stability of modern power grids. The uncertainty about the cost agents face for reducing consumption imposes challenges in achieving reliable, coordinated response. In recent work, [Ma et al. 2016] introduce DR as a mechanism design problem and solve it for a setting where an agent has a binary preparation decision and where, contingent on preparation, the probability an agent will be able to reduce demand and the cost to do so are fixed. We generalize this model to allow uncertainty in agents' costs of responding, and also multiple levels of effort agents can exert in preparing. For both cases, the design of contingent payments now affects the probability of response. We design a new, truthful and reliable mechanism that uses a "reward-bidding" approach rather than the "penalty-bidding" approach. It has good performance when compared to natural benchmarks. The mechanism also extends to handle multiple units of demand response from each agent.