Computing Sciences masthead Berkeley Lab Computing Sciences Berkeley Lab logo
Share/Bookmark

Exascale for Energy


Optimization and Control of Electric Power Systems

Mathematicians from Berkeley Lab’s Computational Research Division (CRD) are working to increase the reliability of the electrical grid and improve the nation’s ability to respond to energy disruptions. By advancing the technologies needed to implement a smart grid, Berkeley researchers will play an important role in avoiding costly, cascading blackouts like the August 2003 blackout that affected eight northeastern U.S states and Canada.

Although improvements to the nation’s power grid have since been put in place, further research is needed to expand the grid’s ability to simultaneously respond to multiple power outages and automatically re-route electricity to avoid broader blackouts. The three-year project, in which Berkeley Lab researchers are collaborating with mathematicians and power engineers from Cornell University and the University of Wisconsin, will develop optimization algorithms that detect vulnerabilities in the power grid, analyze cascading outages, and perform resource allocation across multiple locations and times. They will then combine these algorithms into an integrated optimization framework.

“The North American power grids combine to make one of the largest interconnected systems on earth,” says Juan Meza, head of the High Performance Computing Research Department and Lead Principal Investigator for the project. “It has a complex transmission network, containing more than 9,200 electric generating units connected to over 200,000 miles of transmission lines.” Electricity is distributed across North America on multifaceted grids, which are controlled by a variety of players. These large interconnects are divided among a number of regional utility companies which are responsible for maintaining their portion of the grid and delivering power to consumers. 

Decision-making timescales
Figure 19. Representative decision-making timescales in electric power systems. Behaviors at very fast time scales (e.g., requirements for grid resilience against cascading failures) impose constraints on longer time scale decisions, such as maintenance scheduling and grid expansion.
Source: J. Meza, LBNL

Because electricity cannot be stored, the electric power system must constantly be adjusted to ensure that the generation of power matches the demand. This task falls on hundreds of Control Area Operators across the continent using computerized control centers. The regional utility companies are responsible for coordinating the area operators working on their portion of the grid.

“Although interconnectedness ensures that the system is reliable most of the time, it also means that a few critical line failures could potentially cause a massive blackout,” says Meza.

The grid delivers electricity from power plants to consumers via two primary systems, the transmission system and the distribution system. First, the transmission system transports electricity from power plants to substations near populated areas; then the distribution system delivers electricity from substations to consumers’ homes and businesses. If each power plant, substation, and consumer is represented as a point on the electric power grid, then multiple redundant lines connect each point to ensure that power can travel a variety of routes, from any power plant to any substation and consumer. This means that a line failure between any of these points will not cause a power outage for consumers because power can be rerouted.

The problem becomes even more complex because not all transmission lines on the grid are created equal. Some lines and routes are created to carry more power than others, and options for how to reroute power could affect the severity of an outage—will it affect a neighborhood or an entire city?

Meza notes that one current approach for determining the optimal detour routes in the event of a transmission line failure is to build a large-scale model of the grid and test different scenarios to determine which will cause a blackout. Although this approach has worked so far, experts warn that it will soon be impractical as the demand for power grows and the grid becomes ever more complex.

Instead of building a model of the entire electric power grid, Meza and his collaborators will use combinatorial techniques and graph algorithms to identify all the critical line failure scenarios that could lead to a blackout. He notes that by taking advantage of these mathematical methods, the computational cost is drastically reduced, and the methods can easily be modified to detect vulnerabilities in the grid as it becomes more complex.

Their mathematical methods will also identify optimal detours for rerouting power in the event of multiple line failures, and will rank the solutions from those that will cause minor outages to those leading to severe blackouts. This will help facilitate decision making between the many players involved in both the short- and long-term process of delivering electricity.

“Our combinatorial techniques can analyze vulnerabilities of large complex systems in a fraction of the time needed by previous methods,” says Meza. “This vulnerability analysis will be an important component in decision support and policy making, thereby providing a more reliable and efficient power grid.

”With electricity demand expected to increase by 35% in the next 20 years, maintaining and planning for system reliability is a recognized problem in the national interest. Long-term planning needs to account for actions that take place over a wide range of time scales, from fractions of a second to the 15 years it takes to site and construct transmission lines (Figure 19). Advanced computational tools can take much of the guesswork out of this planning by providing algorithms and software to support optimum power flow across multiple temporal and spatial scales.

 


<< Previous page