Toward Corrective Methods for Sustainable Power Grid Control

February 8, 2017

Enrique Mallada, assistant professor in the Department of Electrical and Computer Engineering, designs new methods to coordinate network-wide control that allow a more efficient integration of renewable wind and solar energy into power grids. Mallada is currently working on a project funded by a seed grant from the Environment, Energy, Sustainability and Health (E2SHI) to equip power grids with fast acting controllers that can quickly react in the event of failure to avoid blackouts, while operating the system more efficiently between these failure events.

Mallada joined Johns Hopkins University in January 2016. He is part of both the LCSR and E2SHI and has a secondary appointment in Mechanical Engineering. Mallada’s research focuses on predicting the behavior of dynamic power systems and making power grids more reliable and sustainable.

This semester, Mallada is teaching “Networked Dynamical Systems,” a course he created. The idea is predicated on the idea that everything—from transportation grids and power systems to global economies—is interconnected and has dynamic behavior. The course explores how individuals’ interactions affect the overall behavior of complex systems.

Mallada’s most recent project uses machine learning tools to identify the causes of voltage fluctuations (from the loss of generators to cyber-attacks) with the goal of eventually moving the power grid from a preventive system to a more sustainable corrective system.
According to Mallada, the power grid behaves in such a very specific and interconnected way that a singular event can change its entire makeup. Mallada uses this characteristic to discern what happens during the failure, so that operators can make better, more informed decisions on how to mitigate the impact of those events. Mallada analyzes different properties of events in the grid to identify whether these measurements are caused by an event which has its own dynamic behavior, whether it was just an error in measurement, or whether it was a malicious attack.

Today’s power grid operation is slow and preventive: the generators lack the ability to collectively react fast to problems that arise in the grid, Mallada says.

Because of the rare probability of component failure, the grid operates daily in a way that is more expensive than it could be, just to ensure that such failures do not generate a blackout. Over a year, transmission lines, generators, or distribution feeders are disconnected at such an infrequent rate that there are many days in which nothing happens, and yet the grid loses the possibility of generating cheaper energy.

Developing faster, pro-active technology is often viewed as too expensive, time-consuming, and risky, he explains.

“It is often easier to act preventatively than to create technology that would eventually be more efficient and reliable,” says Mallada. This idea of “overprovisioning,” he explains, often occurs in infrastructure systems such as telecommunication networks and power grids; people tend to do things in a preventive but easier way that is more expensive in the long run.

According to Mallada, two major problems cause the power grid to act preventatively: line failure and the uncertainty of a reliable sustainable power source.

Power flows from generators to demand through the transmission lines, with each line capable of transmitting only a limited amount of power. Thus, the grid must be set up so that power does not overload the lines. Problems thus can result from the generator’s inability to efficiently respond to a change in the grid, Mallada says.

For instance, if one line is lost, power will be quickly rerouted through those remaining. This “rerouting” of power can overload other lines, causing them to fail too and producing a “cascading failure” that leads to a blackout.

However, if operators can quickly detect the line overflows and change the generation before the line heats up, failure can be prevented without the need of constant use of expensive generators, Mallada says.

The other major challenge in the current movement toward sustainable energy is the uncertainty of the source. Conventional methods such as nuclear and combustion generators create emissions, but with more environmentally friendly forms of sustainable energy, supply can be unpredictable. It is difficult to accurately predict the accessibility of the supply of sun and wind at a given point, and in a power grid, it is critical that supply and demand are balanced at all time, according to Mallada.

He says that one solution for this uncertainty is “smart demand,” in which demand is not a set point, but a range that can fluctuate over time depending on the imbalance. “Smart demand” allows for some variance in power consumption, depending on the availability of the sustainable energy (wind, sun) at that moment.

“If suddenly you have no more wind, then maybe instead of using an expensive generator to increase production, or a generator that creates a lot of emissions, you can use ‘smart demand’ to reduce consumption for some time until the renewable energy supply returns. In this way we can compensate for uncertainty,” Mallada says.

“Smart demand” as a strategy also allows the power grid to proactively react to a change in power supply. If a power line is down, “smart demand” can avoid failure by compensating for fluctuations in the power supply, he explains. If the grid operates faster in a more corrective approach, it is actually more efficient, more flexible, and cheaper.

By compensating for uncertainties in the grid and preventing “cascading failures,” a corrective method in the grid is more reliable.

“It is also more efficient, because it can use less expensive resources and more, cheaper energy,” he concludes.


Johns Hopkins University

Johns Hopkins University, Whiting School of Engineering

3400 North Charles Street, Baltimore, MD 21218-2608

Laboratory for Computational Sensing + Robotics