Marcello Balduccini, PhD, is Bettering Our Bridges

The United States is currently undergoing an infrastructure overhaul. The Biden administration’s $1 trillion infrastructure bill aims to provide billions in federal funds to upgrade outdated roads, bridges, transit systems and more. Research being done by Marcello Balduccini, Ph.D., assistant professor of decision and system sciences (DSS) and director of the Haub Innovation Center, is just one example of how those funds are being distributed.

Balduccini received a grant from the Federal Highway Administration to work with students to research methods for leveraging artificial intelligence, machine learning and data visualization that will remotely evaluate the life of a bridge and automatically detect its critical issues. 

There are over 600,000 bridges in the United States and, according to the American Society of Civil Engineers, 42% of them are at least 50 years old. A whopping 178 million trips are taken across structurally deficient bridges every day.

“When a bridge is about to collapse, it gets shut down for at least a few years while engineers try and rebuild it from scratch,” says Balduccini. “And that’s the best case scenario. It’s expensive, inconvenient and dangerous.”

Bridges today are inspected by individuals rigorously trained to gauge their quality and structure. The issue is getting inspectors on site to explore these massive structures for flaws, like corrosion, water penetration and shrinkage.

“Think about how big the Ben Franklin Bridge is,” says Balduccini. “There are a lot of different areas to grapple and climb inside to check. It takes enormous amounts of skill and time. Inspectors can’t possibly do this more than once a year, if even that.”

In-person inspections are also exorbitantly expensive. In this project, Balduccini is working to make the inspection process faster, simpler and — perhaps most importantly — remote.

Balduccini and his students are primarily data crunchers, which means they’re collaborating with civil engineering students and faculty at Drexel University who are physically creating and installing sensors on these huge bridges.

The sensors will measure bridge vibrations and the amounts of force at critical areas on the bridge — two data points that will then go back to Balduccini and his team for analysis.

Engineers from Drexel are also installing cameras on the bridges to collect visual data of traffic conditions on the bridge. All three of these metrics will be taken into consideration to determine if an anomaly is being picked up by the newly installed sensors.

“If there is a large vibration that sets off alarm bells, we’ll need to compare that time stamp with our visual data to make sure it wasn’t caused by a large vehicle,” explains Balduccini. 

Andrew Holmberg ’22, currently a DSS major, is dedicated to the machine learning work on this project under Balduccini — that is, programming the computer to discern what data is of concern and what is not. 

“I’ve been interviewing for a lot of software development internships lately and this project always comes up,” says Holmberg. “It’s really unusual and impressive to employers that large-scale coding is coming into play outside of my computer science classes.

The machine learning system being created by Balduccini and his students will eventually learn how to reconcile all this information and only send up red flags when there is a real issue. 

“Bridge owners will be able to log in to this web-based interface, review any issues that have been flagged, determine how serious they are and move forward with maintenance,” says Balduccini. “That means they’ll catch things before an in-person inspector could and at very little cost.”

Eventually, this could mean fewer unsafe bridges for American motorists.

“Currently, there is very little local funding to fix bridges, which means they usually have to be on the brink of collapse before anyone on the federal level can step in to do anything,” says Balduccini. “Having these readings come in consistently will help reduce the cost of maintenance on bridges, so those smaller local amounts of money can be used on minor patches instead of huge rebuilds.”

Balduccini and his students look forward to implementing this sensor system on bridges across the country in the near future.