For a network that powers the country, the United States electric grid is increasingly fragile. Millions of people, households and industries rely on the grid’s ability to balance the supply and demand for energy but extreme weather events and challenges predicting renewable energy generation levels have put significant pressure on it. Grid failures like the one that happened in Texas in 2021 can affect millions of people.
“The only way to prevent cascading outages is to better predict them,” said electrical engineering and computer science Professor Sara Eftekharnejad. “If we can predict the most probable grid failures then mitigative actions could be taken to prevent those failures .”
Eftekharnejad received an NSF CAREER Award to research the impacts of the uncertainties within the electric power grid and develop enhanced methods to predict disruptions. Her research will focus on two main areas –
- Statistical modeling for power grid failures that are often caused by severe weather and interconnectivity issues.
- Modeling the generation uncertainties, particularly as more power is generated by renewables that depend on weather conditions like wind and solar.
The two are separate issues but can also come together to cause significant problems.
“Generation and outages are interdependent. If there is an unforeseen shortage of renewable power generation, that could potentially lead to outages. Similarly, severe outages could disconnect the distributed renewable generation resources from the grid” said Eftekharnejad. “If the grid operators are aware of the impending failures considering these uncertainties, they can take actions that prevent large-scale blackouts.”
To better predict power generation uncertainties and outages across the more than 7,000 power plants and 2.7 million miles of power lines that make up the United States power grid, Eftekharnejad and her research team will develop statistical predictive models.
“When outages cascade, it can affect millions of people. We are trying to develop better methods to estimate the probabilities of outages considering the uncertainties of the available generation resources,” said Eftekharnejad. “We are going to find a way to quantify the uncertainties using large-scale data and machine learning methods.”
Using historical or synthetic data, they will develop statistical models for outage predictions over hundreds of power lines.
“We are looking for a dynamic model that can adjust in real time. The model would learn from the system measurements and adjust itself to better capture the existing uncertainties,” said Eftekharnejad. “Once we know how to model outages and predict them – now we have a way to quickly predict outages in seconds.”
Eftekharnejad and her team also want to develop better forecast models for wind and solar power generation.
“If we can better predict the day-ahead generation uncertainties, we can better plan for those uncertainties and ensure adequate reserves are available,” said Eftekharnejad.
In addition to preventing large-scale blackouts, Eftekharnejad says better modeling of the grid uncertainties could also have significant economic benefits.
“Reducing disruptions is better for the electric utilities and customers. More reliable power could reduce costs for both,” said Eftekharnejad.
“Receiving an NSF CAREER award is an important accomplishment and recognition for new faculty. The awards support pre-tenure early-career assistant professors,” said Jae Oh, the David G. Edelstein Professor for Broadening Participation and chair of the Department of Electrical Engineering and Computer Science. “The EECS department has been regularly producing CAREER awardees in recent years, and we expect this trend to continue in many future years.”
“Dr. Eftekharnejad’s research reveals the power of algorithms in modern society. We obviously cannot afford to rely on nonrenewable resources exclusively for power, nor can we always imagine what power demands and failure events will happen in the future,” said J. Cole Smith, Dean of the College of Engineering and Computer Science. “Her research will ultimately serve to make our power grid more effective in normal operations, more reliable in times of disruptions, and more efficient in using renewable energy sources. This kind of research is just so vital to addressing modern challenges to our national security and quality of life, and the College is excited to see what she and her team of students will produce with this prestigious award.”