Thursday, October 18 , 2018, 9:43 am | Fair 62º

 
 
 
 

UCSB Engineer Finds Way to Predict Cascading Power Outages

A method of assessing the stability of large-scale power grids in real time could bring the world closer to its goal of producing and utilizing a smart grid. The algorithmic approach, developed by UC Santa Barbara professor Igor Mezic along with Yoshihiko Susuki from Kyoto University, can predict future massive instabilities in the power grid and make power outages a thing of the past.

“If we can get these instabilities under control, then people won’t have to worry about losing power,” said Mezic, who teaches in UCSB’s Department of Mechanical Engineering. “And we can put in more fluctuating sources, like solar and wind.”

While development of more energy efficient machines and devices and the emergence of alternative forms of energy give us reason to be optimistic for a greener future, the promise of sustainable, reliable energy is only as good as the infrastructure that delivers it. Conventional power grids, the system that still distributes most of our electricity today, were built for the demands of almost a century ago. As the demand for energy steadily rises, not only will the supply become inadequate under today’s technology, its distribution will become inefficient and wasteful.

“Each individual component does not know what the collective state of affairs is,” Mezic said.

Current methods rely on a steady, abundant supply, producing enough energy to flow through the grid at all times, regardless of demand, he explained. However, should part of a grid already operating at capacity fail — say in times of disaster, attack or malfunction — widespread blackouts all over the system can occur.

“Everybody shuts down,” Mezic said.

The big surges of power left unregulated by the malfunctioning component can either overload and burn out other parts of the grid, or cause them to shut down to avoid damage, he explained. The result is a massive power outage and subsequent economic and physical damage. The Northeast Blackout of 2003 was one such event, affecting several U.S. states and part of Canada, crippling transportation, communication and industry.

One alternative to solve the situation could be to build more power plants to produce the steady supply to feed the grid and have the capacity to handle unpredictable failures, fluctuations and shutdowns. It’s a solution that’s costly both for the environment and for the checkbook.

However, the method developed by Mezic and partners promises to prevent the cascade of blackouts and their subsequent effects by monitoring the entire grid for early signs of failure, in real time. Called the Koopman Mode Analysis (KMA), it is a dynamical approach based on a concept related to chaos theory, and is capable of monitoring seemingly innocuous fluctuations in measured physical power flow.

Using data from existing monitoring methods, like Supervisory Control And Data Acquisition (SCADA) and Phasor Measurement Units (PMUs) KMA can track power fluctuations against the greater landscape of the grid and predict emerging events. The result is the ability to prevent and control large-scale blackouts and the damage they can cause.

Additionally, this approach can also lead to wider development of, demand for and use of renewable sources of energy, Mezic said. Because energy from systems like wind, water and sun are weather-dependent, they tend to fluctuate naturally, and this ability to respond to fluctuations can dispel what reservations utilities may have about relying on them to a greater degree.

Mezic’s research is published in the Institute of Electrical and Electronics Engineers journal Transactions of Power Systems. Other collaborators in Koopman Mode Analysis research include researchers from Princeton University, Tsinghua University in China and the Royal Institute of Technology in Sweden.

Support Noozhawk Today

You are an important ally in our mission to deliver clear, objective, high-quality professional news reporting for Santa Barbara, Goleta and the rest of Santa Barbara County. Join the Hawks Club today to help keep Noozhawk soaring.

We offer four membership levels: $5 a month, $10 a month, $25 a month or $1 a week. Payments can be made using a credit card, Apple Pay or Google Pay, or click here for information on recurring credit-card payments and a mailing address for checks.

Thank you for your vital support.

Become a Noozhawk Supporter

First name
Last name
Email
Select your monthly membership
Or choose an annual membership
×

Payment Information

Membership Subscription

You are enrolling in . Thank you for joining the Hawks Club.

Payment Method

Pay by Credit Card:

Mastercard, Visa, American Express, Discover
One click only, please!

Pay with Apple Pay or Google Pay:

Noozhawk partners with Stripe to provide secure invoicing and payments processing.
You may cancel your membership at any time by sending an email to .(JavaScript must be enabled to view this email address).

  • Ask
  • Vote
  • Investigate
  • Answer

Noozhawk Asks: What’s Your Question?

Welcome to Noozhawk Asks, a new feature in which you ask the questions, you help decide what Noozhawk investigates, and you work with us to find the answers.

Here’s how it works: You share your questions with us in the nearby box. In some cases, we may work with you to find the answers. In others, we may ask you to vote on your top choices to help us narrow the scope. And we’ll be regularly asking you for your feedback on a specific issue or topic.

We also expect to work together with the reader who asked the winning questions to find the answer together. Noozhawk’s objective is to come at questions from a place of curiosity and openness, and we believe a transparent collaboration is the key to achieve it.

The results of our investigation will be published here in this Noozhawk Asks section. Once or twice a month, we plan to do a review of what was asked and answered.

Thanks for asking!

Click Here to Get Started >

Reader Comments

Noozhawk is no longer accepting reader comments on our articles. Click here for the announcement. Readers are instead invited to submit letters to the editor by emailing them to [email protected]. Please provide your full name and community, as well as contact information for verification purposes only.