If predictions come true that renewable energy sources like solar panels and wind generators are the primary suppliers of tomorrow’s power grids, then the engineers maintaining those grids will need new innovations in fault detection, made possible by researchers like NJIT Associate Professor Joshua Taylor.
The problem is that traditional power grids use fault detection methods designed for what’s called synchronous generation, as with gas power plants, and those methods work poorly for inverter-based generation found in renewable systems.
In synchronous grids, faults such as tree contact are easily detected because they cause symptoms like abrupt changes in voltage. Industrial relays automatically interrupt the power flow, just like circuit breakers in your household.
“Lightning strikes the line and creates an arc between the line and the ground. You open switches on either side, or upstream, and the fault clears because it’s de-energized, and then you close them again and you’re back to normal,” explained Taylor, of NJIT’s electrical and computer engineering department. “When there’s a storm, and you see the lights go off for half a second, that’s the switches opening and reclosing to clear faults.”
But with inverted-based renewable energy sources, fault currents can be quite small, making detection much harder. And there’s a second problem: traditional grids rely on the physics of synchronous machines to detect faults, while inverter systems from different manufacturers can all behave differently, due to the programming.
“At a certain point, if you have enough of the grid being fed by inverters, then suddenly the fault currents start to look different. There are switches that are supposed to open, but the sensors and the logic that’s supposed to make that decision, it might make a bad call,” he noted.
Taylor and his peer, Alejandro Dominguez-Garcia of the University of Illinois Urbana-Champaign, are principal investigators and received $275,000 each from the National Science Foundation to fund graduate research into solving this challenge.
Their best idea is to add what engineers call a perturbation onto the line, such as some form of asymmetry, serving as a mathematically provable signal for the circuit to recognize. “The mathematical formalization of this problem will constitute a streamlined, optimization-based procedure for designing new detection schemes,” Taylor and Dominquez-Garcia wrote in their proposal.
“I mean, it’s not what a mathematician would consider math. But it has a nice theoretical element. Where there are existing theoretical tools for fault detection in general systems — you could be talking about automobiles or chemical plants and trying to figure out when a system is not operating — the way it’s supposed to be is a classical problem. [Now], some of the specific features of this problem, we expect to lead to new insights in fault detection,” Taylor said.
“You have a power network, and a fault can occur anywhere in that network. You have relays all around the network, and they don’t have communication with other parts of the grid. Typically they have to look at where they are and make the call — fault or not? And so part of the idea with the inverter is injecting a perturbation. Well, the inverter doesn’t have a communication channel to the relay. So it’s kind of using the actual physics of the grid to send information to the relay.”
A doctoral student at NJIT will formulate the design problem, while a peer in Illinois would work on the software modeling for three-phase power grids. Illinois also had a testing laboratory that simulates the grid. Taylor said he is hopeful that both graduate students will begin in the spring 2025 semester.
