J’ai abandonné mes interrupteurs à flotteur pour des capteurs de niveau ultrasonique — voici la franchise après deux ans

You know what finally pushed me over the edge? A float switch jamming at 2 matin. on a caustic soda tank, and nobody realizing the level was creeping up until morning. After that mess I started looking seriously at non-contact level sensors. Here’s the honest version after actually installing and living with ultrasonic units across a couple of sites.

Premier, the basics, because the sales sheets love to overcomplicate it. An ultrasonic sensor just shouts a high-pitched sound down at the liquid and listens for the echo. A chip times that round trip, does the math against the speed of sound, and tells you the distance to the surface. No wetted parts, nothing touches the gunk. For nasty chemicals, that part is genuinely the safest play you’ve got — sulfuric acid, caustic, solvants agressifs, whatever; the sensor sits up top and the liquid never sees it.

But — and this is the part nobody prints on the brochure — it’s not magic, and I learned that the hard way. Foam will wreck you. On a wastewater pit with a frothy surface, the sensor started reading the top of the foam instead of the water and our numbers drifted by nearly a foot. Steam and heavy vapor do the same thing: they slow the sound down, so the “distance” comes back wrong. Temperature swings matter too. The units do carry a built-in thermistor to compensate, and it works most of the time, but in a tank that sees a 40°C swing between a hot afternoon and a cold night shift, I’ve seen steady 2–3% drift. And there’s the blind zone — get the liquid up into that dead band right under the transducer and it just spits out garbage. You have to mount it high enough that your max level never reaches it. So no, it’s not “pinpoint accurate” in every situation. Know your tank before you trust it.

People always ask me: why not just use radar? Fair question. Radar fires microwaves instead of sound, so it laughs at steam, écume, poussière, et la température. If you’re measuring through a blanket of vapor or inside a hot, pressurized vessel, radar is the right tool. The catch is the price. A decent radar transmitter — say an Endress+Hauser Micropilot or a Siemens Sitrans LR — runs $1,500 À $4,000+ per point. Ultrasonique? For the same water, huile, or mild-chemical job you’re looking at a third to half of that. Pour le 90% of tanks that just hold water, Mazout, or non-aggressive stuff, ultrasonic is the sensible pick, and I’m not embarrassed to say it.

Concrete numbers from what I’ve actually spec’d and bought:

Flowline EchoSonic / EchoPod — the budget workhorse, $200–$500 a unit. Dead simple, 4–20mA out. Great for a basic water or chemical tank.
Senix ToughSonic — $250–$600, tough little things, easy to hang on a narrow pipe.
Pepperl+Fuchs VEGASON — $600–$1,800, more industrial, better beam control.
Siemens Sitrans LU150 / LU240 — $700–$1,500, solid, plays nice with their PLC ecosystem.
Endress+Hauser Prosonic FMU30 / FMU90 — $800–$2,500, the “buy once” option if you want the name and the support.
Installation isn’t rocket science, but it’s not zero either. Physically: cut a hole in the lid, point the sensor straight down, keep it clear of agitators and ladder rungs (those throw false echoes). For a clean retrofit I budget about an hour per tank plus the wiring. The 4–20mA run to the PLC is straightforward if you’ve already got conduit; if not, an electrician tagging in costs me roughly $200–$400 a drop depending on run length. Then calibration: you tell the software the empty “zero” distance and the full “span” distance, and that’s about it. Most units have echo-mapping that learns the tank’s fixed obstructions so it stops false-tripping on a beam or a stirrer. Honnêtement, the software took me longer to trust than to configure.

One thing that surprised me: our liability insurer actually cared. After we documented continuous non-contact monitoring on the acid and caustic tanks — no more guessing, no more a guy climbing up with a dip tape — they knocked a chunk off the environmental/spill rider at renewal. Not a huge number, but combined with not risking a person on a tank, it paid for the sensors fast. If you’re storing anything hazardous, call your broker before you buy; the monitoring requirement is sometimes already buried in your policy, and getting ahead of it saves grief.

Where it’s been great: a municipal wastewater lift station (I used an IP68 unit — it survived constant splashing and humidity), an agricultural irrigation reservoir (a cheap Flowline, I just needed to know “is the pond low”), and a food-grade syrup tank where not touching the product mattered for hygiene. Where I’d go radar instead: a heated asphalt tank, a vessel with heavy surface foam, or anything throwing off a lot of hydrocarbon vapor — ultrasonic just isn’t happy there.

So here’s my honest take. If your budget is tight and your liquid isn’t doing something weird — no foam, no steam, no crazy heat — an ultrasonic sensor is more than enough, and you’ll wonder why you put up with floats for so long. The real protection isn’t the sensor catching a problem after it happens; it’s knowing the level is wrong while you’re not even on site, and being able to do something about it. Spend the money you saved versus radar on a cellular gateway so the alert actually reaches your phone — that’s the part that keeps you out of trouble.