Can an Ultrasonic wave penetrate a LPG cylinder

[darklord27]

Hey guys, First time poster here.
Presently I'm doing a project to measure the levels of LPG remaining in an LPG cylinder. I'm currently trying out ultrasonic sensor that primarily works as a distance sensor(HCSR04). As far as I'm aware, the lower the frequency higher the penetration of the wave hence I'm using the 40KHz sensor. But I'm not getting any results and I'm not technically adept enough to measure the second echo either. It would be helpful if you guys could help me out with the necessary details regarding my problem. Alternate suggestions to my problem are also welcome

 

[NascentOxygen]

Hi darklord27.

"not getting any results" is a rather sweeping statement. Can you elaborate, and describe what you have been attempting?

 

[darklord27]

Okay, The sensor I'm using is an HCSR04 ultrasonic distance sensor that uses a 40KHz wave. I've been trying to get it penetrate the Cylinder by placing it on its surface first. The echo returns back in no time. I've been thinking of using an impedance matching agent such as a fine gel but there is a good possible chance of damaging the sensor. I'm having doubts on using the sensor as such since it is made to measure distance i.e detects only the first echo. The things I need to know are.

1. Is it possible to penetrate the Cylinder using ultrasound sensors?
2. If so, what frequency and bandwidth should I use?
3. Brief Idea about the complexity of the circuitry required.
4.Any alternate methods that i could refer to for my purpose if this isn't the way to go forward.

The questions are many and my knowledge about what I'm into is low. Please bare with me .

Regards.

 

[davenn]

maybe some info on this page may help you ...

http://www.ni.com/white-paper/3368/en/

 

[davenn]

This isn't my field of expertise,( I come to it via my activities in the past with seismic surveying using sound)

I suspect that the majority of the sound energy will be reflected back to the sensor from the inner metal surface as shown in the first 2 images on that page and any energy that does actually penetrate into the LPG gas/air space possibly won't have have enough energy to re-penetrate the gas metal boundary and make it out bto the sensor ?

definitely worthy of some experimentation

just my initial thoughts

Dave

 

[anorlunda]

Have you considered measuring the change of sonic reflectivity of the tank wall above the liquid level compared to below liquid level?

I tap the side of my LPG tank to make it ring like a bell. The quality of the bell sound changes when I tap below the liquid level.

 

[rbelli1]

Or you could try from the bottom of the tank. You would still have to detect multiple reflections as you get a reflection from each surface the wave hits but you don't have to deal with a gas.

BoB

 

[anorlunda]

Or you could weigh the tank.

 

[Alec Dacyczyn]

I've tinkered with those modules...

I think it's unlikely that you'll be successful using that sensor. It's strictly for non-contact applications in air. You've correctly surmised that that sharp impedance change between the air and steel is reflecting effectively all of the ultrasonic energy and preventing the sensor to "see" the liquid level inside.

But it might be fun to try to modify it. I actually have one of those (about 10ft away from where I'm typing this). I don't see any obstacles to careful cutting away the protective cowling and sticking the transducer faces directly onto the tank.

However, the speed of sound through steel is so much faster than it would be through liquid or gaseous propane that the sound wave might not exit the steel into the inside of the cavity or re-enter the steel sidewall after bouncing around. For that matter, the return signal may be dominated by the original wave propagating around the circumference of the tank and back to the sensor.

I did a google search for "ultrasonic tank level measurement" and found that all of the commercial offerings (on the first page, I admit I didn't look hard). are meant to be installed on the insides of tanks. There is probably a good reason why this is.

 

[darklord27]

Or you could weigh the tank.

Weight is the easiest method I agree, but I was looking for a non invasive tech that doesn't use weight as the principal measuring meter.

 

[darklord27]

Have you considered measuring the change of sonic reflectivity of the tank wall above the liquid level compared to below liquid level?

I tap the side of my LPG tank to make it ring like a bell. The quality of the bell sound changes when I tap below the liquid level.

I will try that out. Thanks. Do you know about any sensor or application that could help measure sonic reflectivity?

 

[Nidum]

Use a transmission method rather than a reflection method .

Introduce ultrasound on one end of a diametral line and sense signal at other end .

Signal detected after passing through liquid will be stronger than signal after passing through gas .May also be detectable different phase shifts compared to energising signal .

Several ways of making a practical version . For a portable device possibly a caliper with source and detector at ends of arms . Used by manually applying device at incremental heights from bottom of cylinder .

 

[Baluncore]

You need to measure acoustic reflectivity from inner surface of the cylinder wall. That will require a contact sensor that you can slide up the wall.

Lower technology is usually better. I pour a litre of boiling water over the side of the cylinder. As the water dries, the internal liquid level becomes visible as a distinct line. That is due to the difference in heat transfer at the liquid level inside the cylinder.

 

[Jeff Rosenbury]

What about a thermal sensor?

Apply a little heat. The temperature of the steel should change differently depending on what's behind it. You might use a thermal camera?

 

[darklord27]

Power constrains are low. I did think about a thermal principles and tried to apply the hot water through the sides in an engineering perspective. Thermal sensors and heating etc would require good amounts of power and it seems to be impractical at this point.

 

[anorlunda]

Are we talking about a little tank, 10 kg? or a big one 1000 kg?

If it is big, then I return to the simple idea of weight. You could install a strain gauge (also called load cell) on one of the supporting legs of the tank, To calibrate, you only need to know the sensor reading tank full, and again tank empty, then do a linear interpolation.

I'm a big fan of KISS in engineering. As Baluncore said, low tech is better.

 

[SteamKing]

Okay, The sensor I'm using is an HCSR04 ultrasonic distance sensor that uses a 40KHz wave. I've been trying to get it penetrate the Cylinder by placing it on its surface first. The echo returns back in no time. I've been thinking of using an impedance matching agent such as a fine gel but there is a good possible chance of damaging the sensor. I'm having doubts on using the sensor as such since it is made to measure distance i.e detects only the first echo.

It sounds like (sorry for the pun) all you've been doing is measuring the wall thickness of the gas cylinder with your sensor. The sound travels from the transducer into the metal and is reflected immediately back to the sensor from the interface (back wall) between the cylinder wall and the interior of the cylinder, which is why you get the instantaneous return.

Ultrasonic thickness gaging is an NDT way of measuring metal thickness for a number of different applications. I've used it, for example, to gauge the plating thickness on floating vessels, to tell how thick the plating is. It's used all the time in the marine industry to determine which plates need to be replaced because their thickness has been reduced below safe levels by corrosion. I'm pretty sure this method is also used for shore-side applications.

It's possible your particular sensor is calibrated to pick up the echo from the back wall and to ignore any other sound reflections. You might have to hook up the receiver to an oscilloscope to see if you are obtaining multiple reflection returns from inside the cylinder which would be spaced out and return at different times.

In any event, you'll need some advice from a true expert, someone who knows NDT and ultrasonics.

 

[Jeff Rosenbury]

Power constrains are low. I did think about a thermal principles and tried to apply the hot water through the sides in an engineering perspective. Thermal sensors and heating etc would require good amounts of power and it seems to be impractical at this point.

I was thinking more of a 5W strip up the side, then viewing the way the heat dispersed. You shouldn't need a lot of power, but you will need an expensive camera.

Applying a lot of heat all over (like a water bath) would make it difficult to watch the heat flow.

 

[Jeff Rosenbury]

If you fed a single sync pulse from the bottom you could range gate away the first transition reflection. That signal would continue to reverberate (reflect back and forth off both sides of the steel) with the strength falling off with time. This might swamp the signal from the liquid/gas interface, but it might not. I've used this technique successfully with UWB radar, but that doesn't work well through metal.

 

[sophiecentaur]

Going in from the floor of the tank would allow better matching and coupling into the liquid than into gas through the steel. I don't know how directional the beam that gets into the liquid would be and you could expect significant echos from the sides of the tank unless the liquid level is less than the radius of the tank (i.e. near empty`)
Perhaps a system using transmission rather than reflection would work. The shortest delay would correspond to the direct path from bottom to top and the delay value would be least with a full tank and most with an empty tank (relative speeds in gas and liquid).

 

[NascentOxygen]

I vaguely recall there used to be a liquid-crystal strip that you could buy to glue down the side of your gas cylinder----as propane was drawn off latent heat loss cooled the liquid's surface, and the strip (basically a liquid crystal thermometer) would show a colour change at the boundary. I'm not sure how it worked to accommodate wide-ranging ambient temperatures; perhaps it comprised multiple thermometer strips in parallel?

 

[Baluncore]

Maybe step the cylinder sideways quickly. That impulse will induce a 'U' shaped sloshing in the liquid. That internal oscillation will have a low frequency with a magnitude determined by the liquid mass in the tank. You might be able to measure the oscillation of the liquid mass with the same transducer that applied the impulse.

 

[darklord27]

Are we talking about a little tank, 10 kg? or a big one 1000 kg?

If it is big, then I return to the simple idea of weight. You could install a strain gauge (also called load cell) on one of the supporting legs of the tank, To calibrate, you only need to know the sensor reading tank full, and again tank empty, then do a linear interpolation.

I'm a big fan of KISS in engineering. As Baluncore said, low tech is better.

It is a standard 15.6 Kg tank. Weight sensor's I've already calibrated and got the readings too. Then i started searching for other methods that would be possible to implement.

 

[meBigGuy]

The most common way is to measure the temperature of the outside wall after releasing some gas.
http://www.amazon.com/dp/B00LW2VZQS/?tag=pfamazon01-20

 

[Nidum]

I understood that you wanted a practical quick reading device which could be used on any cylinder and without modifying the cylinder .

I have to admit that for the cylinders in my workshop I use some of the rough and ready methods mentioned by others above .

The company where I did my training and first worked had the capability to x-ray cylinders but I think that is going a bit far !

 

[Nidum]

PS: For transmission device I envisaged that caliper would be used metal detector style with output just as a sound which changed frequency as the liquid/gas interface was passed .

 

[sophiecentaur]

Maybe step the cylinder sideways quickly.

. . . . or oscillate it from side to side with an actuator, to find the resonant frequency of the 'sloshing'? At least the Low Frequencies involved would not be impeded by the steel bottle.

 

[anorlunda]

After all these ideas, please post a follow-up after your final decision. Let us know what you chose and how it worked.

 

[darklord27]

Hello Guys,
Thanks for all the suggestions and Sorry for the late reply. I did go through the concepts of all your replies and found the transmission method most suitable to use. I tried the circuit and kept the transmitter and receiver at diametrical ends and again there was no echo to be received. I guess the power is too low or the formation of standing waves within the cylinder might have impeded the strength.Or the wave might have reflected within the steel walls itself. I'll keep on trying modifications to this and while I'm on it, temperature seems to be a very good option. If there is any way you guys know to practically engineer a way to get the temperatures at different points of the cylinder ie. sensors that detect minute differences in temperature( i don't know how much the temperature would vary when gas is being used), if you could redirect me towards that way, it would be very helpful. The actuator technique sounds promising and I'll surely look into that too. Thanks for all the replies and I'll keep posting on developments. Please feel free to redirect me to sensors that would help me engineer the required tech.
Regards

 

[Nidum]

The HCSR04 transmitter/receiver device is low power and only works for non contact detection in air .

Try to find devices with separate transmitter and receiver and which are designed for direct contact applications .

 

[sophiecentaur]

The HCSR04 transmitter/receiver device is low power and only works for non contact detection in air .

Try to find devices with separate transmitter and receiver and which are designed for direct contact applications .

Matching the power in and out is very important and it's important to use the appropriate devices.

Temperature sensing shouldn't be too much of a problem because it's pretty easy to use your hands to establish where the level is by the temperature variations over the surface. A number of fixed sensors (say 10) would give you an accuracy of 10% in level. You could perhaps do better than that with some clever interpolation between readings.

 

[NascentOxygen]

I vaguely recall there used to be a liquid-crystal strip that you could buy to glue down the side of your gas cylinder----as propane was drawn off latent heat loss cooled the liquid's surface, and the strip (basically a liquid crystal thermometer) would show a colour change at the boundary. I'm not sure how it worked to accommodate wide-ranging ambient temperatures; perhaps it comprised multiple thermometer strips in parallel?

Amazon will enlighten you: here are two such gauges. https://www.amazon.com/dp/B0010P3OU2/?tag=pfamazon01-20

The customer comments teach you how each can be used successfully.

 

[sophiecentaur]

Amazon will enlighten you: here are two such gauges. https://www.amazon.com/dp/B0010P3OU2/?tag=pfamazon01-20

The customer comments teach you how each can be used successfully.

That device is very convenient to use when all you need is a visual inspection (plus you have to actually get to the site of the bottle). It could be the basis of a system, perhaps based on the change in reflectivity of the whole strip as the coloured portion changes length. That could produce an electrical signal etc. etc.

 

[meBigGuy]

Post 24 was a link to an LCD strip.

I wonder how the electrical characteristics of the LCD might change when the color changes. Putting them at intervals and measuring XXX (whatever it might be) would provide what you are looking for. Anyone have the materials science background to comment of the electrical characteristics of LCD strips with temperature?

 

[sophiecentaur]

If you want to go electronic then there's no reason why you couldn't use an array of thermistors (ten, twenty or whatever) in series, stuck to the side and measure the total resistance. The scale would be a bit non-linear but it would at least be monotonic and the signal out could be read and recorded easily. An identical control chain, mounted near (or perhaps around the top of the bottle) could compensate for overall temperature changes.