Restructuring Infrastructure: Building Smarter Systems
America relies on its infrastructure: the countless networks of roads, bridges, and waterways that make our modern way of life possible. “The United States’ infrastructure is one of the most valuable assets in the country,” says George Lederman (PhD '16). “So obviously its important for the government to monitor it carefully—particularly as it ages.”
The problem, according to Lederman, is that much of that infrastructure was built in the 1950s and 60s, and was designed for a 50 year life. “Now that it’s coming to the end of its design life, we have to figure out ways to address the problems that come with aging,” he says.
In particular, Lederman is interested in the lifecycle of bridges. For the past two years, he has been monitoring the vibrations caused by Pittsburgh’s light rail system passing over bridges. By monitoring the vibrations before and after repairs to the track, he can assess the condition of the structures underneath the railroad. But ultimately, he says he’d like to crowd source this data.
“When you go over a bridge, there’s an interaction: the bridge vibrates, which causes the vehicle to vibrate. And one of the things I’m interested in is, if you had a cellphone in the vehicle, could you tell the condition of the bridge based on the vibrations?”
Like Lederman, Milad Memarzadeh (PhD '15) is developing smarter, more interactive ways of monitoring and controling our infrastructure systems—in particular, integrated systems, such as networks of pipelines, roads, and wind turbines, made up of individual parts. He says that currently, the only way ensure that these networks are operating smoothly is to assess the condition of their individual components, which makes maintaining these networks not only costly, but time-consuming.
To that end, Memarzadeh has been developing a framework that would allow infrastructure managers to monitor and control systems more efficiently, by incorporating data about individual components of these systems in a way that will speak to the whole, and that will respond to changes in the data that is collected.
“You might have a wind farm with 100 turbines, but you can only afford to collect data for maybe 20 of them,” he says. “What I’m interested in seeing is, once you’ve collected data on some of them, can you use that data to make generalizations about the system as a whole?”
Not only will Memarzadeh’s research allow for better, more efficient monitoring, it will encourage smarter planning by allowing data to dictate changes in the model. “It’s adaptive,” he explains. “As you receive more and more data, you learn more, plan, and move forward.”
Lederman and Memarzadeh were finalists in this year’s 3 Minute Thesis championship (3MT), which challenged students to explain the significance of their current research in just three minutes, using language that is compelling and accessible to a non-specialist audience.
Both Memarzadeh and Lederman were grateful for the opportunity to share their work with the public. “The 3 Minute Thesis is really a cool competition,” says Lederman. “It forces you as a PhD student to step back and think, what is my research ultimately about, and then condense that down so you can help people see why this is a challenge, why it’s important, and what I’m doing about it.”
Milad Memarzadeh - Probabilistic learning and planning framework for optimal management of systems under uncertain environments
George Lederman - Good vibrations: How to protect our infrastructure.