Thursday, January 24, 2019

- Story shared from Penn State News by Erin Cassidy Hendrick

Not only does every power generation facility — including wind, solar, coal, nuclear, and hydroelectric sources — need to route their energy in a responsive, cost-conserving way, there are also an enormous number of contingencies that arise at a moment's notice. IMAGE: © iStock Photo / metamorworks

Not only does every power generation facility — including wind, solar, coal, nuclear, and hydroelectric sources — need to route their energy in a responsive, cost-conserving way, there are also an enormous number of contingencies that arise at a moment's notice.

IMAGE: © iStock Photo / metamorworks

UNIVERSITY PARK, Pa. — In an effort to modernize and reimagine the United States' power grid, Penn State researchers have qualified for a highly selective, innovative competition sponsored by the U.S. Department of Energy.

The Penn State team of researchers, one of only ten universities chosen for the Grid Optimization Competition's first challenge, is led by Uday V. Shanbhag, the Gary and Sheila Bello Chair and professor in the Harold and Inge Marcus Department of Industrial and Manufacturing Engineering.

"As the United States begins incorporating more renewable energy sources, there are some new and unique challenges that today's infrastructure simply can't handle," Shanbhag said.

Announced by the Advanced Research Projects Agency-Energy (ARPA-E) within the DOE, the competition challenges researchers from universities and national laboratories to solve the fundamental issues facing the electricity infrastructure, while addressing the concerns that widespread renewable energy sources will introduce in the future.

The network

"With a network as large as the U.S. power grid, the optimization problems we need to solve are incredibly large and complex," Shanbhag said.

Not only does every power generation facility — including wind, solar, coal, nuclear, and hydroelectric sources — need to route their energy in a responsive, cost-conserving way, there are also an enormous number of contingencies that arise at a moment's notice.

With the first round of funding, the teams are being challenged to design algorithms that address the next generation of security-constrained optimal power flow (OPF), essentially finding ways to provide electricity more quickly, efficiently, safely and reliably within the current grid. Distinct from past models, the new set of models are complicated by the need to model the flow of electricity, as governed by power flow equations, with much higher fidelity.

Using the mathematical principles of optimization, the software controlling the grid signals that a certain set of generators need to be "dispatched'' to meet current demand. But if one of those generators fails, Shanbhag said, "Can the algorithm controlling the power grid take recourse and keep the lights on?"

"Providing for every contingency possible in a network like this, serving more than 65 million nodes, it's a large and nasty problem," Shanbhag said. "And it is one that has to be solved every ten minutes."

Over the past few decades, the models used in the power grid have been adapted to handle these situations.

"But in their expanded, nonlinear form, it is computationally challenging, so coarse approximations were used," Shanbhag said. "But now, it is essential to consider more accurate models that are complicated by size and uncertainty.''

Between the sheer number of customers, the speed in which contingencies need to be solved, and the fluctuating nature of renewable energies like solar and wind power, the nation is ready for the next generation of power grid technology.

The challenge

Innovating these solutions places a Herculean, but inspiring, challenge ahead of the team.

"There are many reasons why this is a daunting mathematical and computational challenge, and why Dr. Shanbhag's vision for computationally efficient solution methods could be a major game changer," said Hosam Fathy, the Bryant Early Career Professor of Mechanical Engineering. "If the Penn State team wins this competition, it will be an indication that we have made substantial strides in the stochastic grid optimization domain, thereby paving the way towards significant leaps in how the electric power grid is operated both now and in the future."

Their approach will focus on creating a method that is able to both scale appropriately with the size of the underlying optimization problem and address the underlying nonlinearity. The underlying code needs to be able to adapt instantaneously, while also conserving computing resources so the system does not become overburdened.

"With this mindset, the power grid will be able to better deal with the challenges expected to emerge in future power systems," said Mort Webster, professor of energy and mineral engineering in the College of Earth and Mineral Sciences.

Their project will also aim to develop the mathematical tools to enable a trustworthy infrastructure well into the future.

The next phase

In the next phase of the competition, ARPA-E will provide each team with sample data from the power grid to test their algorithms.

"We'll take this actual network information, apply our algorithms, and see how well we do!" Shanbhag said.

Participants that develop scalable schemes for finding minimum-cost solutions to these problems will advance to the next round.

"Penn State's research institute model has been crucial in developing a foundation for this work, as well as facilitating the current research," Shanbhag said. "In fact, the Institute of Energy and the Environment provided crucial seed funding that fueled the conceptualization of these schemes and the Institute of CyberScience has been tremendously helpful as we attempt to develop efficient algorithms for such problems."

The team

Capitalizing on interdisciplinary strengths, a team was assembled from the College of Engineering and the College of Earth and Mineral Science, and also includes of Nilanjan Ray Chaudhuri, assistant professor of electrical engineering and computer science; Chiara Lo Prete, assistant professor of energy economics; and Minghui Zhu, assistant professor of electrical engineering.

This team brings together a diversity of methodological backgrounds in stochastic optimization, nonlinear/nonconvex optimization, and control theory, with domain interests in power systems, electricity market design and operation, and battery modeling and energy storage.

"Penn State has always been a global leader in energy systems research, but in order to maintain this leadership we need to join forces across different disciplines in order to build larger, cohesive teams in the energy area," said Fathy.

Collectively, the group was chosen to include experience with both the theoretical and applied principles surrounding sophisticated power systems, with a particular emphasis on addressing the new questions that are posed by renewable energy.

"The hope was to build a team at Penn State that is not just capable of solving today's energy problems, but also to establish a research infrastructure for the future of power systems and markets," Shanbhag said.

According to Fathy, given the University's pursuit to be at the forefront of a re-imagined energy infrastructure, this competition and team of researchers presents a critical turning point.

"This is an example of what Penn State's Energy University initiative is about. It is not about our individual successes within our individual research silos, but rather about how we come together to do something much bigger," said Fathy.

"By bringing these minds together," Shanbhag said, "we believe that we have a chance to solve this problem."