(579b) Bittorrent-Inspired Distributed Energy Resource Management: A P2P Framework for Utility 2.0

Authors: 
Titus, M. C., Rensselaer Polytechnic Institute
Bequette, B. W., Rensselaer Polytechnic Institute

This work examines a peer-to-peer (P2P) approach to management and control of distributed energy resources (DER) on an electric power distribution grid based on a modified version of the BitTorrent filesharing protocol1. It is proposed that customers with DER - defined as small (<50 kW) generation, storage, and/or flexible loads - be treated as active peers in a power distribution swarm coordinated via the Internet. The sheer number of possible future IP-enabled devices of this category (the so-called ‘Internet of things’) motivates the need for a decentralized management strategy, as centralized approaches are not well suited to this scale and complexity of nodes.

The proposed management strategy is inspired by the ability of the BitTorrent filesharing protocol to efficiently distribute large files over bandwidth-constrained networks, while minimizing network capacity requirements2. Our modified protocol applies the principles of BitTorrent file distribution to the problem of electric power grid ancillary services provision by making an analogy between data and energy flows, in which DER can upload or download energy to or from grid peers at a certain power, respectively. In this framework, simple distributed client controllers autonomously dispatch local DER as needed, while attempting to satisfy both local and swarm objectives using common advanced control and optimization techniques such as model predictive control. This approach can both reduce transmission congestion by increasing local resource utilization and decreasing peak demand as well enable resilient and efficient operation of microgrids post-emergency (e.g. Hurricane Sandy).

To test the modified protocol, an open source python BitTorrent client is modified to manage a small swarm of window air conditioning units via infrared using a Raspberry Pi Model B single-board linux machine3 and a custom IR transceiver circuit. The A/C units function as physical devices participating in an otherwise virtual simulation. Real time power use of each A/C unit is collected every 2 seconds by Tweet-a-watt4 modified Kill-a-watt power meters using ZigBee communication. This small-scale swarm demonstrates the potential of the BitTorrent-inspired framework to reduce peak demand and enhance efficiency through integrating local execution of real-time models and optimization with P2P dispatch of network DER.

References:

1)  B. Cohen, Proc. First Workshop on Economics of Peer-to-Peer Systems, (2003)

2)  A. R. Bharambe, C. Herley, V. N. Padmanabhan, SIGMETRICS’05, June 6–10, (2005)

3)  Pi, Raspberry. "An ARM GNU/Linux box for $25." Take a byte (2012)

4)  Flahiff, Daniel. TWEET-A-WATT! A power monitor that twitters your energy use. Inhabitat. Retrieved on March 05, 2009 from: http://www.inhabitat.com/2009/02/04/tweet-a-watt-by-ladyada/ (2009)