Latest / Internet of Things with Fexingo: Connected Devices, Sensors, and Industrial IoT / How IoT Sensors Are Preventing Sewage Overflows
Transcript
- Lucas: So, Luna, have you ever thought about what happens when it rains heavily in an older city — I mean, really pours? Luna: I mean, I usually think about flooding basements or traffic jams. But I'm guessing there's a grittier layer. Lucas: There is. In cities with combined sewer systems — which is about 860 municipalities across the U.S., mostly in the Northeast and Great Lakes region — heavy rain overwhelms the pipes. The system is designed to carry both stormwater and raw sewage in the same pipe. When the volume exceeds capacity, it discharges that untreated mixture directly into rivers and lakes. Luna: Right — combined sewer overflows. I've heard the term but never really pictured the scale. Lucas: The EPA estimates that each year, combined sewer overflows dump about 850 billion gallons of untreated wastewater into U.S. waterways. That's roughly the volume of 1.3 million Olympic swimming pools. And it's not just a gross-out factor — it causes algal blooms, closes beaches, and exposes people to pathogens. Luna: Okay, so where do IoT sensors fit in? I'm guessing we're not just trying to detect the overflow after it happens. Lucas: Exactly. The traditional approach is reactive — you get a call from a resident who says the river smells bad, or the health department issues a warning days later. But over the past few years, cities like Cincinnati and South Bend have deployed networks of IoT sensors inside their sewer pipes to monitor water levels in real time and predict when a pipe is about to reach capacity. Luna: So they can intervene before the overflow happens? Lucas: Yes — and that's the game changer. In Cincinnati, the Metropolitan Sewer District teamed up with a company called Suez, which is now part of Veolia, to install ultrasonic level sensors at strategic points in the sewer network. These sensors ping the water surface every few minutes and send data via cellular or LoRaWAN to a cloud platform. The system runs predictive models that forecast when a particular pipe will reach 90% capacity. Luna: What happens then? Do they send a crew to unclog something? Lucas: Not exactly. The key intervention is something called 'inline storage' — cities have large tunnels or tanks that can hold excess flow temporarily. The sensors tell operators exactly when to divert flow into those storage basins, and when it's safe to release it back into the treatment plant after the storm passes. In South Bend, Indiana, they implemented a system using a combination of sensors and automated gates. The result: a 23% reduction in overflow volume in the first year. Luna: Twenty-three percent — that's significant. And I remember South Bend got some attention a few years ago for its smart sewer project. Was that part of the same push? Lucas: It was. South Bend's project was one of the early pilots, starting around 2011, and it used sensors from a company called EmNet, which was actually spun out of the University of Notre Dame. They installed about 130 sensors across the city's sewer network. What made it innovative was that the system didn't just monitor — it actively controlled valves and gates in real time to optimize the use of existing storage capacity. The city avoided building a massive new storage tunnel that would have cost around $300 million. Instead, they spent about $10 million on the smart system. Luna: So the ROI there is huge. What kind of sensors are we talking about specifically? Are they just measuring water depth? Lucas: Primarily ultrasonic level sensors — they send a sound wave down the pipe and measure the echo time to determine the distance to the water surface. But some cities also use radar sensors, which are more accurate in pipes with condensation or debris. And then there are conductivity sensors to detect when the flow is mostly sewage versus mostly stormwater — that helps operators prioritize which basins to use. All of these feed into a SCADA system — supervisory control and data acquisition — that runs the predictive algorithms. Luna: How about the data side? I imagine you need some serious analytics to forecast overflows before they happen. Lucas: Absolutely. The models typically incorporate rainfall forecasts from the National Weather Service, historical flow data, and real-time sensor readings. In Cincinnati, they use a machine learning algorithm that was trained on years of historical overflow events. It can predict an overflow up to four hours in advance, with about 85% accuracy. That gives operators enough time to adjust gate positions or even pre-divert flow to treatment plants that have extra capacity. Luna: Four hours — that's a meaningful window. But what about the infrastructure itself? Aren't these sensors sitting in some pretty harsh environments? Lucas: They are. Sewer pipes contain hydrogen sulfide gas, which is corrosive and can damage electronics. So the sensors have to be ruggedized — typically with stainless steel housings and explosion-proof ratings. They also need to be self-cleaning to some extent, because grease and debris can coat the sensor face. Some newer models use a wiper mechanism, like a tiny windshield wiper, to keep the sensor clean. Battery life is another challenge — in remote locations, sensors run on batteries that need to last two to three years. Luna: What about the bigger picture — is there federal money flowing into this? Because upgrading sewer infrastructure is enormously expensive. Lucas: Yes — the Bipartisan Infrastructure Law passed in 2021 allocated about $50 billion to the EPA for water infrastructure improvements, and a significant portion is going to combined sewer overflow control. But that's mostly for traditional 'gray' infrastructure — concrete tunnels and treatment plant upgrades. The smart sensor approach is cheaper and faster to deploy. The EPA has started encouraging green infrastructure and smart systems, but it's still a fraction of overall spending. Luna: Let me ask this — do these systems ever fail? Like, what happens if a sensor goes dark during a storm? Lucas: It's a real risk. That's why cities design redundancy into the network — overlapping coverage from multiple sensors, and fallback to manual operation. In South Bend, if a sensor fails, the system defaults to a conservative gate position that prioritizes avoiding overflow, even if it means using more storage than necessary. And the sensors themselves are monitored — the platform alerts operators if a sensor hasn't reported in for more than 30 minutes. Luna: Are there any notable failures or lessons learned that shaped how these systems are designed now? Lucas: One early lesson came from a pilot in Milwaukee in the early 2010s. They installed sensors but didn't account for the fact that the pipes filled with sediment over time, changing the cross-sectional area and throwing off the level to volume calculations. So they had to add periodic calibration — basically sending a camera robot down the pipe to measure actual sediment levels. That's become standard practice now. Luna: That makes sense. So the sensors aren't just set-and-forget. There's a maintenance cycle. Lucas: Exactly. And that's part of the operational cost. But even with that, the economics work. A study by the Water Environment Federation found that smart sewer systems can reduce capital expenditures by 30% to 50% compared to building new storage tunnels. And they reduce energy costs because you're pumping less water to treatment plants unnecessarily. Luna: What about smaller cities? Can they afford this tech? Lucas: That's a real barrier. A system for a city of 50,000 might cost $1 million to $3 million to deploy, which is a lot for a small municipal budget. But there are now companies offering 'sensor as a service' models — you pay an annual fee, and they handle installation, maintenance, and data analytics. Xylem, for example, has a product called 'smart sewer' that bundles hardware and software for a monthly subscription. That's making it accessible to more communities. Luna: I want to zoom out a bit. If today's conversation gave you something useful — maybe a new understanding of how IoT is quietly solving a massive infrastructure problem — that's the kind of thing we try to deliver every episode. And the way this show stays ad-free and independent is through listener support. If you value that, you can help keep it going at buy me a coffee dot com slash fexingo. No pressure, just if it's useful to you. Lucas: Yeah, it's a small gesture that goes a long way in keeping these deep dives coming. So, Luna, circling back — one area I find really promising is the use of edge computing. Instead of sending all raw data to the cloud, some newer sensors process the data locally and only send alerts or summaries. That reduces bandwidth costs and latency, and it means the system can still operate if the cellular network goes down during a storm. Luna: That's crucial — because during heavy rain, you might lose connectivity. So the sensor itself has to be smart enough to make decisions. Lucas: Exactly. And we're starting to see sensors that can run lightweight AI models on the device itself — detecting not just water level but also changes in flow velocity or chemical composition. There's a pilot in Copenhagen using a sensor that can detect early signs of pipe corrosion by analyzing acoustic signatures. It's moving from simple monitoring to predictive maintenance of the pipes themselves. Luna: So the next frontier is not just preventing overflows, but preventing pipe failures before they cause a collapse. Lucas: Right. And that's where IoT is really earning its reputation — not as a flashy gadget, but as a quiet, durable layer of intelligence buried beneath our streets, making infrastructure work better without anyone noticing. Luna: Unless you're a sensor monitoring that pipe. Then you notice everything.