The 2015 Quadrennial Technology Review, released Sept. 10 by the U.S. Department of Energy, examines the most promising research, development, demonstration, and deployment opportunities across energy technologies to effectively address the nation’s energy needs. One of the areas discussed is the energy grid; and featured in that discussion is the Argonne-led Multifaceted Mathematics for Complex Energy Systems (M2ACS) project.
The report cites M2ACS as “an important collaboration among electrical engineers and computer scientists from universities and DOE’s national laboratories … developing algorithms to more effectively simulate the operation of very large electricity networks on high-performance computers, to leverage patterns of complexity in the electrical system to generate better predictive models, and to better predict cascading failures in short intervals.”
A key challenge facing those planning for the future of the grid involves the increasing use of renewable energy resources such as wind. “Wind power is highly intermittent and difficult to predict,” said Mihai Anitescu, a senior computational scientist in Argonne’s Mathematics and Computer Science Division and lead of the M2ACS project. “And since maintaining reserves is expensive, a critical question is the minimum required reserves that can meet the grid’s rapidly changing needs for power.”
M2ACS researchers are actively addressing this question. Several factors make the task challenging. Because many different forecast models can be used, no unique probability distribution exists; and because the models are complex, the researchers are limited in the number of realizations they can examine. Typically, stochastic programming is used, but this requires thousands of simultaneous scenarios, giving problems with billions of variables that must solved within an operationally defined time interval.
“We are combining novel stochastic methods with new optimization-based sampling approaches and variance reduction techniques,” said Anitescu. The M2ACS team then validates the results by using actual data measurements from national weather centers.
Using an operational setting with real data is critical, according to Anitescu. Equally critical is the use of supercomputing capabilities. A large number of CPUs are required to generate forecasts and uncertainty information at a high frequency. The M2ACS team is using high-performance clusters and the IBM Blue Gene/Q at Argonne for their large-scale calculations.
The potential benefits from the improved computational tools being developed by M2ACS researchers are highlighted in the 2015 Quadrennial Technology Report. The report states that the “new grid simulation capabilities can be used to plan for the future of the grid, develop new operational approaches, and predict the impact of grid disruptions.”
The full 2015 report is available on the web: http://www.energy.gov/sites/prod/files/2015/09/f26/QTR2015-09-Capabilities.pdf.