Role of Distributed Coordination in Resilient and Fine-Grain Control of Power Grids

Conference Chair
Dr. Marija Ilic milic@ece.cmu.edu, Carnegie Mellon University, ECE and EPP

Conference co-chairs:
Dr. Krishna Kant, George Mason University
Dr. Gabriela Hug, Carnegie Mellon University, ECE and EPP
Dr. Orkun Karabasoglu, SYSU - CMU Joint Institute of Engineering
Dr. Zico Kolter, Carnegie Mellon University, CS
Dr. Raj Rajkumar, Carnegie Mellon University, ECE

The NINTH ANNUAL CARNEGIE MELLON CONFERENCE ON THE ELECTRICITY INDUSTRY will be held on February 3-5, 2014 at Carnegie Mellon University. This will be a national workshop, which happens to be held as part of the 9th CMU Annual Electricity Conference as it is an outgrowth of recommendations made by the last year’s NSF sponsored workshop held during the 8th CMU Electricity Conference.

Energy Management Systems are currently an integral part of electric power utility operations. Their main functions have been state estimation based on redundant system-wide measurements using highly centralized Supervisory Control and Data Acquisition (SCADA), the assessment of most critical outages, and generation scheduling for forecasted system demand. The software applications have evolved over time to support reliable bulk power system operations by allocating sufficient standby generation reserve necessary to supply customers without interruptions when any single – or double – equipment outage (contingency) occurs.

However, different software applications are used in isolation from each other. At present there are no IT tools capable of relating specific applications in an on-line manner to advise system operators and/or gradually fully automate decision making as conditions continuously vary. There is a major need for R&D in support of a systematic framework for assessing proximity to reliability problems and for resource allocation to ensure that as different events occur the grid is operated as reliably and efficiently as possible. This need has triggered an initiative by the Federal Energy Regulatory Commission (FERC) with major efforts toward improving centralized software and filling this void. Much U.S. Department of Energy R&D funding is devoted to supporting the work.

Even with the ongoing efforts, it remains necessary to re-think the frameworks for next generation Energy Management Systems. In particular, it is essential to design EMS architectures capable of accommodating interactive information exchange between different layers within the very complex electric power grid for integrating the objectives of the electricity users, producers of unconventional technologies (renewable resources, for example) and the new direct line flow control grid technologies (ranging from HV DC embedded into the existing HV AC grid, low-cost smart wires, and the like). Instead of continuing today’s approach to centralized modeling and decision making over all time scales and all stakeholders, EMS architectures are needed to enable embedding of distributed intelligence into smart users, buildings, resources to manage uncertainties and social preferences, and to minimally coordinatethe system as a whole by the existing EMS ultimately responsible for the system-level reliable and efficient performance. It is also necessary to develop methods which take into consideration the effects of EMS architectures on the ability of grid users to make social choices.

A unique conference will be held with leading engineering, computer science, and policy-making thought leaders to exchange ideas on future energy management systems. The objectives of the new EMS architectures will particularly be targeted to support the premise that IT-enabled distributed loosely coordinated architecture is both efficient and robust during major unplanned natural disasters, such as hurricanes and tornadoes. At present, EMS software does not have applications that enable electricity users to pro-actively respond to system conditions during such major natural disasters. There are no tools for fast bottom-up restoration, either.

The workshop will address possible problem formulations for next generation multi-layered EMS architectures capable of incentivizing the groups of users (one may refer to them as the smart balancing authorities (SBAs) to utilize resources as efficiently as possible during normal conditions, and capable of being sufficiently adaptive and flexible to ensure differentiated reliable service during single/double equipment outages (planned for by the utilities now by keeping conservative standby the worst-case reserves), as well as during natural disasters (for which utilities do not have methods to ensure reliable prioritized service without requiring excessive standby reserve). The workshop will explore how the distributed intelligence and the right IT signals with the right parallel asynchronous algorithms embedded into different layers would enable both social choice and system-level performance.

We point out that what is unique about this workshop is its radical departure from current centralized EMS. Instead, next generation EMS architectures comprising layers of distributed management systems (DMSs) embedded into SBAs, distributed local power grids, and high voltage EMS systems in each control area, and between the control areas for interconnection-wide energy management will be explored. Notably, there is no single entity which supports this thinking at present. The support is biased by the conventional vendors and utilities which do not lend themselves to distributed solutions which enable choice and social benefits. In contrast, it is natural for NSF to support distributed thinking.

Finally, every effort will be made to bring the right balance of engineers and computer science thought leaders. The PI of this proposal has had a long history of pursuing this synergy. Carnegie Mellon University itself is a natural place to host this meeting. We hope to make the connection between folks working on Next Generation Internet (Peter Steekiste) and the leaders of Smart Grid Design (Marija Ilic).

This is a follow-up to the NSF workshop organized by the PI last year on Data-Driven Sustainable energy Services http://www.ece.cmu.edu/~electriconf/2012/. A detailed NSF Final Report on this workshop is available.

Tentative Agenda:

We plan two full days of formal conference sessions. We will require each participant to submit the slides prior to the conference and provide a one-page summary of the main points in the slides. So, instead of spending time on slide presentations, much more time will be allotted to discussing main points/ideas put forward by the presenter. We already have a mechanism for posting and archiving all of the Carnegie Mellon Electricity Conferences since the conference inception and it will be rather straightforward to provide a permanent place to make these documents publicly available to the broader community.

We also envision several keynote papers with state-of-the-art presentations.

Partial support for U.S. faculty and students will be available upon request.