Sensing Liquid Level/Presence through a container wall

I'm trying to come up with a way to sense the presence of liquid on the other side of a container wall. The wall is less than 1 mm thick and is made of stainless steel.

This is basically a liquid level determination. I just need to know if there is liquid on the other side of the wall at the point the sensor is placed. I think some sort of ultrasonic density sensor would work.

The trick is that I cannot place anything in the container or modify the container in any way. So, the usual trick of using an ultrasonic distance sensor and sensing the level from the top of the container will not work. Also, the container is sealed.

What I am trying to do is determine the liquid level on a 5 gallon Cornelius keg. The sensor doesn't need to be permanently placed. I would just move the sensor up and down the side of the keg and watch the readings to determine where the liquid line is.

I assume the typical ultrasonic sensors won't help here as they would just read zero distance on contact with the keg wall.... or would they use the wall as a transmitter and still return a different reading depending on whether there is liquid or gas directly adjacent to the sensor?

I know something like this is possible:

I bought one of those hoping it would work, but the propane tanks are thicker and there is no way to recalibrate the sensor.

Thanks for any ideas.

An old trick to estimate the liquid level in metal tank containing cool liquid is to pour a small amount of hot water down the outside of the tank. Then you quickly run your finger up and down the tank, sensing the tank temperature. The liquid level is indicated by the boundary between the warm and cold regions.

This works because the thin metal tank walls heat up quickly where liquid is not present. Some variation of the technique might work, for example you could use a self-heating thermistor as the probe.

Edit: By the way, are you sure that the ultrasonic probe you linked would need to be recalibrated? What it detects is the change in reflection coefficient at the metal/liquid versus metal/gas interface.

make two sensorhousings in each a resistor of about 1000 ohms and a ntc
the resistors going to an output
the ntc to a analog input.
when all touching the metal wall the one under liquid keeps colder due to the higher conductivity of the liquid

an ultrasonic will work too as the mass is higher

when all touching the metal wall the one under liquid keeps colder due to the higher conductivity of the liquid

The problem I see with that is that stainless steel is not a very good conductor of heat.

Just thinking about it a slightly different way, could you possibly use a servo with something metallic to "knock" on the container and "listen" for the sound you hear back? Basically apply the same principle that's used to make water musical instruments?
Is should sound different when full vs empty.

I was once involved with a project to look at the level of molten iron in a blast furness. We used a radioactive source and giger counter to act as a beam that the iron broak. You could do the same thing with back scatter.

Thanks for all the input so far!

Grumpy_Mike, do they make a backscatter sensor that will work with an Arduino (and is cheap?)

Psychephylax, a thumper is a little more complex than I’d like. Plus, I suspect that it would have to be calibrated for each keg as I would guess each keg is different enough that their acoustics aren’t very similar. If I just wanted to know if it was empty or full it would probably work fine, but I want to know where the liquid line level is.

Shooter, these kegs are stored in a refrigerator long term, so they’re pretty much the same temp as the liquid inside them. I’d have to be able to heat the exterior somehow (hot water, torch, etc.) and still get a temp sensor array on there fast enough to detect the temp differences between the areas under liquid and gas. It would work, but I think it might be difficult to use in practice.

Shooter, any suggestions on how to use an ultrasonic sensor? All the sensors I can find seem to be for measuring distance and in my case the distance from the sensor to the keg would be zero. I need one that can use the keg wall as its interface. All it would need to do is “see” through the <1mm stainless wall enough to tell if there is something dense behind it (liquid) or something much less dense (gas).

JRemington, I’ve purchased one of those GasCheck devices and they don’t work on the kegs. They work fairly well on a propane tank though and they’re advertised as ultrasonic, so I know the principle is valid.

Grumpy_Mike, do they make a backscatter sensor that will work with an Arduino (and is cheap?)

There is no differance between a back scatter sensor and a normal sensor. The cheapest work from a giger counter so that is not very cheap

Are you allowed to move/reposition the keg? finding the tilting point is an indication of the fill level.

Is the weight constant?

Measuring the weight of the keg should work, after initial calibration.

I support the weighing idea. Almost foolproof.

If that doesn't work, then the temperature idea. You can try to heat the surface and see how warm it gets. Regardless of the fact that the stainless steel is not as good a conductor of heat as some other metals are, it will still stay cooler where there is liquid behind it, than if there isn't. In fact, the low thermal conductivity may work to your advantage, as it will reduce the heat transfer away from the point where you are applying the heat, in the plane of the container surface. If the wall is only 1 mm thick, most of the heat is going to go to the liquid inside and not sideways through the metal.

The self-heating thermistor sounds like a good approach to me. It could initially run at low power so as to get a measurement with no heating applied. Then the power could be increased, and another measurement taken after a short while, to determine the amount of heating.

To use ultrasonics, I would glue a piezoelectric cylinder to one end of a stainless steel block, and glue a mass to the other end of the cylinder. The other end of the block would be machined to match the profile of the tank. When the piezo cylinder is given a burst of signal at its resonant frequency, the intensity of the reflected burst will depend on whether there is liquid in the tank. This will only work if the block is in good contact with the tank.

Another approach would be to use a thin piezoelectric cylinder mounted inside a housing that is pressed against the tank. One end of the cylinder will make contact directly with the tank. The other end is connected to a heavy mass and then a spring, so that it is held securely against the tank. Then the resonant frequencies and damping factor of the piezo cylinder will depend on whether or not there is liquid in the tank.

Well, after reading everyone's responses and thinking about it further, I think measuring the weight of the keg may be the best approach.

These things sit in my kegerator (beer fridge) until they're empty, so I'd rather not move them around. Plus, the 3rd keg is behind the front two, so it is hard to access anyway.

So.... a new question. These kegs are about 9 inches in diameter and weigh about 8 to 9 pounds empty and about 50 pounds full.

What would be the best weight sensor (pressure transducer?) to use? I would need three of them and would need to leave the kegs sitting on them full time in a 35 degree F environment. There is also some moisture present due to condensation when I open the fridge, but it is generally dry.

Thanks again

What would be the best weight sensor (pressure transducer?) to use?

Search for "load cell".