Latest / Internet of Things with Fexingo: Connected Devices, Sensors, and Industrial IoT / How IoT Sensors Track Bridge Fatigue in Real Time
Transcript
- Lucas: So, Luna, you know how sometimes you're driving over a bridge and you see those little signs that say 'weight limit 15 tons' and you kind of just hope the engineers really did their math? Luna: All the time. Especially on older bridges. It's like a leap of faith every crossing. Lucas: Right. Well, the leap of faith is shrinking, thanks to IoT. I'm talking about structural health monitoring — real-time sensors that track fatigue in bridges. And I want to zero in on one specific technology that's taken off since the I-95 bridge collapse in Philadelphia back in 2023. Luna: The one where a tanker truck fire caused a section to collapse? That was a wake-up call. Lucas: Exactly. That was a fire, not gradual fatigue, but it exposed how fragile our bridge monitoring really was. After that, the Federal Highway Administration fast-tracked funding for IoT-based monitoring systems. And the technology that's been getting the most attention is acoustic emission sensing. Luna: I've heard of acoustic emission in like, pressure vessels and pipelines. How does it work on a bridge? Lucas: Same principle. You attach piezoelectric sensors — small ceramic discs — to the steel girders or concrete. When a micro-crack forms or grows, it releases a burst of elastic energy, a tiny acoustic wave. The sensor picks that up, and the system locates the source by triangulating the arrival times at multiple sensors. Luna: So you're literally listening to the metal groan. Lucas: Exactly. And the key advance in the last few years is machine learning filtering. Because otherwise, you'd be swamped by noise — trucks rumbling, wind, thermal expansion. The model learns what's a real crack signal versus just traffic. One study from Purdue showed they reduced false positives by 94 percent using a convolutional neural network. Luna: 94 percent — that's huge. So the data becomes actionable rather than just a firehose of alerts. Lucas: And that's exactly what state DOTs need. They have over 46,000 structurally deficient bridges in the US. You can't inspect them all visually every year. But with acoustic emission sensors, you can monitor continuously and prioritize repairs based on actual crack growth rates, not just age. Luna: I read about a bridge in Minnesota — the I-35W St. Anthony Falls Bridge replacement — they installed over 500 sensors. But that was a new build. What about retrofitting older bridges? Lucas: That's actually where most of the action is. Retrofits. Since 2024, about 1,200 bridges have been retrofitted with IoT monitoring systems. The sensors are relatively cheap — a few hundred dollars each — and you can install them during routine maintenance closures. No need to shut down the bridge for weeks. Luna: And power? Because a lot of these bridges are in remote areas, no grid access. Lucas: That's the clever part. Many of these sensors are energy-harvesting — they use piezoelectric materials that generate electricity from the vibrations of traffic. So they're essentially self-powered. Or they use small solar panels with supercapacitors. Some systems even use the temperature differential between the bridge surface and the air below. The point is, they're designed to run for years without battery changes. Luna: So the infrastructure itself powers the monitoring. That's elegant. What about data transmission? Cellular? LoRaWAN? Lucas: Depends on location. Urban bridges often have cellular, but rural ones use LoRaWAN or even satellite IoT. The data rate is tiny — just bursts of timestamps and amplitudes. A bridge might send a few kilobytes a day. So it's cheap. The real expense is the analytics backend and the engineering time to interpret the data. Luna: And I imagine that's where the bottleneck is. You can have all the sensors in the world, but if no one knows how to read them, you're just collecting data. Lucas: That's the next frontier. But there are companies like Strainstall, which is part of James Fisher, and also some startups like Brimrose that provide turnkey monitoring as a service. They handle the sensor installation, the data pipeline, and the monthly reports to the DOT. It's becoming a subscription model. Luna: That makes sense. DOTs don't want to hire data scientists. They want a red flag that says 'inspect this girder before next Tuesday.' Lucas: Exactly. And here's a concrete example: In 2025, the New York State DOT used acoustic emission monitoring on the Tappan Zee Bridge replacement — the Mario Cuomo Bridge — and detected abnormal crack growth in a weld on the main span. They were able to schedule a repair during low traffic hours, avoiding a full closure. Estimated savings: $2 million in avoided lane rental penalties alone. Luna: Wow. So not just safety, but economics. That's the kind of story that gets funding approved. Lucas: And that's really the theme of this episode. IoT isn't just about smart toasters. It's about making critical infrastructure smarter, safer, and more cost-effective. We're talking about a $40 billion annual maintenance gap in US infrastructure. Any technology that helps prioritize spending is a game-changer. Luna: It's one of those things where if it saves one life, it's worth it. I mean, if today's conversation gave you something usable — maybe a new perspective on why that bridge you cross every day has a sensor glued to it — I think that's worth a coffee to support the show. Lucas: Yeah, honestly, if today was worth a coffee to you, that's the link — buy me a coffee dot com slash fexingo. It keeps the show ad-free and helps us dig into topics like this. Luna: And we really appreciate that. So, back to bridges — you mentioned the cost savings. But what about the challenges? Like calibration, or false negatives? Lucas: Great question. False negatives are the silent risk. If a crack grows slowly and quietly, the sensor might not catch it until it's larger. That's why you still need periodic visual inspections. The sensors don't replace engineers; they augment them. Also, calibration is tricky because every bridge has a different acoustic signature. You need a baseline — and that often means intentionally creating small cracks in lab samples to train the model. Luna: So it's not plug and play. There's a learning curve per structure. Lucas: Exactly. And that's why adoption has been slower than some predicted. But the trend is clear. In the next five years, I think we'll see acoustic emission sensors become standard on all major river crossings and long-span bridges. The technology is proven. It's the deployment logistics and the interpretation skills that need to scale. Luna: And maybe some regulatory push. After the I-95 collapse, there was talk of mandating real-time monitoring for certain bridge classes. Lucas: There was. The Bridge Infrastructure Monitoring Act, introduced in 2024, didn't pass, but it's been reintroduced this year. It would require continuous monitoring for all bridges longer than 500 feet in urban areas. If that passes, you'll see a massive spike in sensor deployment. Luna: So we might be at an inflection point. IoT sensors moving from early adopter to mainstream safety equipment. Lucas: That's my take. We're going from 'we should monitor this bridge' to 'which monitoring system gives us the best data per dollar?' And that's a good place to be. Luna: All right, I'm never going to look at a bridge the same way again. Thanks, Lucas. Lucas: You're welcome. And if you're driving over one, just know there might be a tiny ceramic disc listening for cracks, keeping you safe.