TR2016-095

Accelerating Convergence to Competitive Equilibrium in Electricity Markets



We present a single time-period decentralized market clearing model based on the DC power flow model. The electricity market we study consists of a set of Generation Companies (GenCos) and a set of Distribution System Operators (DSOs). We model the DSOs as a single node having deferrable loads. The Independent System Operator (ISO) determines the market clearing generation and demand levels by coordinating with the market participants (GenCos and DSOs). It is assumed that each market participant shares limited information with the ISO. We exploit the problem structure to obtain a decomposition of the market-clearing problem where the GenCos and DSOs are decoupled. We propose a novel semismooth Newton algorithm to compute the competitive equilibrium. Numerical experiments demonstrate that the algorithm can obtain several orders of magnitude speedup over a typical subgradient algorithm.