Optimal Solution Analysis and Decentralized Mechanisms for Peer-to-Peer Energy Markets

TL;DR

This paper analyzes optimal clearing in P2P energy markets, showing conditions for market equivalence with pool markets, and introduces a decentralized ADMM-based trading mechanism with convergence guarantees and learning schemes.

Contribution

It provides a novel decentralized optimization method for P2P energy trading using ADMM, with analytical insights and parameter tuning schemes.

Findings

P2P market outcomes match pool markets under connected communication structures.

Unconnected communication structures lead to clustered P2P markets.

The proposed ADMM-based method converges quickly and effectively facilitates decentralized trading.

Abstract

This paper studies the optimal clearing problem for prosumers in peer-to-peer (P2P) energy markets. It is proved that if no trade weights are enforced and the communication structure between successfully traded peers is connected, then the optimal clearing price and total traded powers in P2P market are the same with that in the pool-based market. However, if such communication structure is unconnected, then the P2P market is clustered into smaller P2P markets. If the trade weights are imposed, then the derived P2P market solutions can be significantly changed. Next, a novel decentralized optimization approach is proposed to derive a trading mechanism for P2P markets, based on the alternating direction method of multipliers (ADMM) which naturally fits into the bidirectional trading in P2P energy systems and converges reasonably fast. Analytical formulas of variable updates reveal insightful relations for each pair of prosumers on their individually traded prices and powers with their total traded powers. Further, based on those formulas, decentralized learning schemes for tuning parameters of prosumers cost functions are proposed to attain successful trading with total traded power amount as desired. Case studies on a synthetic system and the IEEE European Low Voltage Test Feeder are then carried out to verify the proposed approaches.