Plotting for an energetic future
IBM Research – Haifa joins European partners to develop smart grid technology
High-performance computing (HPC) and communication technologies are being efficiently applied to solve large-scale computational challenges in nearly every field from genomics to e-commerce.
Researchers at IBM Research – Haifa, longstanding leaders in HPC and high speed messaging, wondered how such platforms could be put to use to help the world's system for supply and demand of electrical power. Energy demand has grown dramatically over several decades, and now electricity distribution technologies need to evolve to meet this demand.
The issue has the attention of world leaders across the globe. In fact, the U.S. alone, from both private and public funds, has dedicated more money to this cause than to any other single energy overhaul in history. The solution everyone seeks lies in the creation of more flexible and smarter power grids.
High performance solution
IBM got on board by joining an impressive international consortium of partners—Brunel University, Electricite De France, the University of Oxford, UK Power Networks, Union Fenosa, Indra, GTD, Korona, Elektro Gorenjska, and Fraunhofer IWES. The HiPerDNO project, part of the European Union's 7th framework programme, is being coordinated by Brunel University in London.
The project aims to develop a new generation of distribution network management systems that exploit HPC and communication solutions for smart electricity grid operation and management.
Gidon Gershinsky, who leads the Haifa team, explains the problems with the current system. "Electric companies have to build and maintain an infrastructure that can support the highest peak of demand. But when demand falls below that level, which is most of the time, we are still paying for what we're not using—in time, in effort, and even in the resource itself." Another challenge is management of the so-called renewable energy generation. Consider, for example, a solar farm. Production goes up or down according to the day's weather. Yet demand may be up when it's cloudy and go down as the sun comes out. Wind farms face similar challenges. And while these are both important sources of clean energy, their difficulties illustrate the need for electrical distribution systems to become more flexible.
"A smart grid," Gershinsky continues, "would match production to needs, conserving energy when demand is down, saving resources for when they are really needed. It could also enable better demand management at peak times."
Smart meters and applications
Developing smart grid technology includes the deployment of smart meters, much talked about in the media. These work by measuring consumption in real time and potentially offering control of some electric appliances by using a system of two-way communication with utility data and control centers. These command stations can use the information they receive to balance the energy distribution. Consumers could, for example, allow the grid to automatically run certain appliances (like a clothes dryers, which could work at arbitrary hours) when demand and cost are lower.
The second step in the solution involves integrating applications into the infrastructures that manage overall demand and response, in an attempt to shift and reduce consumption. This might involve, for example, financial incentives to consumers, such as discounted electricity during non-peak hours.
IBM's role in the project is two-fold. Gershinsky's team is focusing on the communication aspects for the grid. Given the many millions of European households, each one containing multiple electric appliances, the deployment of smart meters will be massive. The team will create a high-speed messaging layer to be incorporated into the smart grid for the acquisition of the vast amounts of information produced by the smart meters. This technology will avoid congestion by monitoring the network and ensuring that data are dynamically analyzed and digested as they travel to the utility companies' electricity network control centers.
The second component of IBM's role, performed by the machine learning team led by Yaacov Fernandess, involves data mining and analysis. Fernandess and team will focus on consumption planning, using global data to provide insights that will help predict and avoid problems. "The challenge of integrating machine learning into a smart grid is that the system, after a mass deployment of network equipment sensors, will be generating vast amounts of data round-the-clock. To be of any use, data mining and analysis will need to achieve nearly real-time results," explains Fernandess.
A brighter future
The consortium continues to meet periodically for updates and releases of new versions of the developing technology. Gershinsky sums up the importance of their work: "Smart grid technology will not only transform the electricity distribution system as we know it, but it will promote energy-efficient options for consumers, paving the way for the growth of renewable resources that have until now been on the back burner."