Latest / Internet of Things with Fexingo: Connected Devices, Sensors, and Industrial IoT / How IoT Sensors Are Preventing Elevator Breakdowns in Real Time
Transcript
- Lucas: So last month, a 38-story office tower in midtown Manhattan avoided what could have been a very bad Tuesday morning. One of its eight passenger elevators started sending a specific alert — motor vibration outside the normal range. Not a breakdown yet. Just a sensor saying hey, something's off. Luna: And that alert came from an IoT sensor attached to the motor housing. This is Schindler's predictive maintenance system, right? Lucas: Exactly. Schindler, the Swiss elevator company, has now deployed over a million connected sensors on its equipment globally. The one in that Manhattan building detected a bearing that was starting to wear — a couple degrees hotter than normal, a slightly different vibration signature. The system flagged it 72 hours before it would have failed. Luna: And the old way would have been waiting for the car to stop between floors with people inside. Then a technician shows up, diagnoses on the spot, orders parts — you're looking at hours of downtime, maybe a full day. Lucas: Right. Emergency repairs on an elevator in a commercial building run about four to five thousand dollars on average, plus the intangible cost of 40 people stuck in a metal box. In this case, the building's maintenance team scheduled the bearing replacement for Sunday at 2 a.m. — off-peak, no disruption. Luna: That's the kind of story that makes the business case for IoT really concrete. But before we go deeper — you know, this is exactly the kind of tech conversation where listener support makes a real difference. If you've gotten something useful out of episodes like this, a couple of dollars a month is genuinely what keeps these going — buy me a coffee dot com slash fexingo. Lucas: It's true. Small contributions from people who actually listen are what allow us to stay ad-free and dig into the specifics. No agenda, no sponsor talking points — just the data and the use case. Luna: So back to those sensors. What exactly are they measuring besides vibration? Lucas: There are typically four or five sensor types on a modern connected elevator. Motor temperature is the big one — a 2-degree Celsius rise above baseline is often the earliest signal. Door cycling count is another: every time the doors open and close, the microcontroller logs it. After about 500,000 cycles, you know the door operator mechanism is approaching end of life. Luna: And cable tension is measured continuously, right? That's one I hadn't thought about. Lucas: Yeah, load cells on the hoist cables measure tension in real time. If one cable starts taking more load than the others — even a 5 percent imbalance — the system flags it. That used to be something a technician checked manually every six months. Now it's every second. Luna: So the data stream is constant. How much data are these sensors generating per elevator per day? Lucas: Rough estimates I've seen from Schindler's deployments suggest about 10 to 15 megabytes per day per elevator. That's not huge — a single smartphone generates more. But when you multiply by a million sensors across thousands of buildings, you need serious edge processing. A lot of the anomaly detection happens right on the sensor module itself, not in the cloud. Luna: Edge computing makes sense because you can't afford latency. If you're waiting for a round trip to the cloud to decide whether that vibration is dangerous, the car could be stuck by the time you get an answer. Lucas: Exactly. The sensor module — about the size of a deck of cards — runs a lightweight machine learning model that's been trained on thousands of historical failure cases. It knows what a healthy bearing sounds like, what a worn bearing sounds like, and what a catastrophic failure sounds like. It only sends an alert upstream if the pattern crosses a threshold. Luna: Let's talk about the economics for the building owner. You mentioned 72-hour advance warning. What's the actual ROI on installing these sensors? Lucas: Schindler has published some data on this. Buildings that adopt predictive maintenance see a 30 to 40 percent reduction in unplanned downtime. The cost of a sensor retrofit is roughly $2,000 per elevator — that includes the hardware, installation, and the first year of data service. A single avoided emergency repair in a Class A office building pays for sensors on all eight elevators. Luna: And the sensor itself is probably very cheap at scale. I've seen BOM estimates for these multi-sensor modules at around $50 to $80 in volume. Lucas: That sounds right. The real cost is the software and the analytics platform. Schindler actually built its own IoT platform called Schindler Ahead. It's a subscription model — the building owner pays a monthly fee per elevator, and that covers the sensors, the connectivity, the analytics, and the guaranteed response time if something gets flagged. Luna: So it's not just selling hardware. It's selling uptime as a service. Lucas: Exactly. And this is happening across the whole elevator industry. Otis has its ONE IoT platform. Kone has 24/7 Connected Services. Thyssenkrupp has MAX. They're all racing to retrofit existing installations because the installed base is enormous — there are roughly 18 million elevators worldwide, and most of them are dumb. Luna: What about the data privacy angle? These sensors are in common areas, but they're constantly monitoring motion, door cycles — could you infer occupancy patterns? Lucas: It's a valid concern. The sensor data is typically aggregated and anonymized before it leaves the building. Schindler says they don't track individual floor destinations or specific people — just mechanical parameters. But as these systems get more sophisticated, the line between mechanical monitoring and behavioral monitoring could blur. Regulators in Europe are already looking at this under GDPR. Luna: Right, because if you know that floor 12 gets heavy traffic between 8 and 9 a.m., you know that floor 12 has a busy office. That's commercially valuable information. Lucas: It is. And the industry is going to have to be transparent about what data is collected, how it's used, and whether building owners can opt out of certain data streams. Right now, the mechanical data is fairly benign, but the infrastructure is there to collect much more. Luna: Let me ask about a specific failure mode that sensors can catch. The one that scares me most is the overspeed governor — the mechanism that engages if the elevator starts free-falling. Can IoT monitor that? Lucas: It can, and it does. The governor is a centrifugal brake — it spins, and if it exceeds a set speed, it triggers mechanical brakes. Sensors measure the governor's rotational speed continuously. If there's any deviation from the expected acceleration profile — say the governor is a bit sluggish because of corrosion — the system flags it. That's a rare failure, but when it happens, it's catastrophic. Luna: Schindler actually had a well-known incident in 2018 at a building in Singapore where a counterweight fell. Did that accelerate their IoT investment? Lucas: It did. That incident involved a broken guide rail bracket — something that's very hard to inspect visually. After that, Schindler ramped up its sensor development specifically for guide rail condition monitoring. They now have sensors that measure rail vibration and alignment continuously. If a bracket starts to loosen, the vibration signature changes subtly over days or weeks. Luna: One more number I want to pin down: what's the global market for elevator IoT right now, and where is it headed? Lucas: A report from MarketsandMarkets published this spring pegged the global elevator IoT market at about $7.5 billion in 2025, growing to $12 billion by 2030 — that's a 14 percent compound annual growth rate. The biggest drivers are the installed base aging — the average elevator in North America is 25 years old — and building owners wanting to differentiate their properties with smart building features. Luna: And the sensor retrofit itself is a fraction of the cost of a full modernization. A full elevator replacement can be $150,000 per car. A sensor kit at $2,000 is a no-brainer in comparison. Lucas: Right. And the payback period is typically under 18 months just from reduced emergency repairs and lower insurance premiums. Some insurers now offer discounts for buildings with predictive maintenance, because the risk of a liability claim drops significantly. Luna: Is there a comparison you like to make between elevator IoT and something more familiar, like a car's check engine light? Lucas: That's exactly the analogy I use. A modern car has 100-plus sensors and an onboard diagnostic system that tells you, 'Hey, your oxygen sensor is drifting, you've got 500 miles before the check engine light comes on.' Elevator IoT is the same idea — except the stakes are higher because failure means people trapped in a vertical box. And the sensor data is now precise enough that you can schedule maintenance literally weeks in advance. Luna: So the next time I step into an elevator and see a Schindler or Otis logo, I'll assume there's a tiny computer watching the bearing temperature. Lucas: If it was installed in the last three years, almost certainly. And in five years, the dumb elevator will be the exception. The technology is cheap enough and the ROI is clear enough that it's hard to justify not doing it. Luna: One thing we haven't touched on — what happens when the sensor itself fails? Do you get false positives? False negatives? Lucas: That's a real concern. These sensors are industrial-grade, but they still have a failure rate. Schindler's system has built-in self-diagnostics — each sensor checks its own health every 10 minutes. If a sensor goes silent, the system knows it's a sensor problem, not an elevator problem. False positives are more common in the first month after installation, because the algorithm is learning the baseline for that particular elevator. After about 30 days, false alerts drop significantly. Luna: And false negatives — missing a real problem — that's the nightmare scenario. How do they guard against that? Lucas: By having multiple sensors with overlapping coverage. The motor temperature sensor, the vibration sensor, and the current draw sensor all correlate. If one sensor misses a hot bearing, the vibration sensor will catch the change in signature. The algorithms are trained on thousands of real failure sequences, so they're pretty good at pattern recognition. But no system is perfect — that's why regular manual inspections still happen on a reduced schedule. Luna: So the technician's role shifts from 'go look at everything' to 'investigate the specific things the sensors flagged.' That's a big productivity gain. Lucas: Massive. Technicians can cover more elevators per day, and they show up with the right parts already in the truck. Before IoT, a typical service call was diagnose, order part, come back next week. Now it's 'the sensor says the door operator motor coupling is worn, I'll bring a new coupling and replace it in 20 minutes.' Luna: And that's the real-world impact of IoT — not a futuristic fantasy, but a bearing replacement on a Sunday at 2 a.m. in a Manhattan skyscraper. Lucas: Exactly. This is industrial IoT doing exactly what it promises: preventing failure before it happens, saving money, and keeping people safe. And it's happening right now, in millions of elevators every day.