RPI has developed a simulator that can help prevent the catastrophic power grid failure that occurred in Texas
TROY — Rensselaer Polytechnic Institute’s Design Lab has found a way to make electric substations a bit safer and more efficient.
The school, which teamed up with K&A Engineering Consulting, P.C., in White Plains, completed a project where students were tasked with creating a neural network that simulates a substation in the quest to train substation workers on new techniques, emergency situations and testing.
“The idea of the project is if we can simulate equipment going into emergency overload or being damaged, a mathematical simulation such as a neural network can become the digital twin of the substation’s functions and allow us to try different protocols and hopefully avoid a massive catastrophe before the accident ever happens,” Manuel Pimenta, technical program manager at K&A, said.
A neural network, also known as a mathematical simulation, is “a network or circuit of neurons, or in a modern sense, an artificial neural network, composed of artificial neurons or nodes. Thus a neural network is either a biological neural network, made up of real biological neurons, or an artificial neural network, for solving artificial intelligence (AI) problems.”
This type of simulator can provide useful in training for dire situations, like the one presented in Texas in late February. After a fluke snowstorm blanketed the state and affected the state’s unique power grid, a total breakdown of the infrastructure took place and residents were left for days without electricity, running water or heat. The situation escalated quickly and drastically, resulting in rolling blackouts, and by the time the state was able to reinstate power across the board, it was less than five minutes from a total blackout that would have lasted days, if not weeks.
Pimenta added the neural network simulates the substations in real time, meaning workers can simulate the breaks to each station and introduce variables for it to react to like the real station would.
“While an emergency at substations is rare, we want to keep the workforce trained for all types of scenarios so when situations present, there is a well-rehearsed solution to follow,” Pimenta said.
Palmer Feinburg was one of the students in the experiment. During the 16 weeks he was working on the network, the experience provided him valuable insight into his career in engineering. He ultimately secured a spot as an intern with K&A directly.
“The interesting thing about this project was the initiative we had to take,” Feinburg said. “We sent K&A an email every two weeks updating on our progress, but it was unlike many situations where you have a professor walking you through each step of the project. We had very little direction from overhead and had to follow our own theories.”
This type of learning is why K&A and RPI worked so well together for the experiment. As the RPI Design Lab is running dozens of projects at any given time, the real-world experience gives students both the ability and confidence to embark on their careers armed with knowledge.
“The beauty of our design lab is its open, experiential learning space where students can collaborate on projects and have the experience of working in a real lab,” Kathryn Dannemann, director of RPI’s Design Lab. “Our students focus on these particular topics and in time, they become the experts and are able to guide research.”
While the capstone class is mandatory for engineering students, Feinburg said the experience proved valuable, as he learned a lot about neural networks and the benefits of using the technology for training and prevention.
“You don’t need a ton of energy or super-advanced computers to run a neural network,” Feinburg concluded. “They’re faster, more effective and less expensive than a lot of simulations out there. Using this simulation will bring a significant upgrade to substations that utilize it.”
