Process Engineer transitioning to Machine Learning & IIoT- Muby

Hi @muby,

  Others may know more, but I would guess ultrasound waves generated and detected from outside of the cylinder will be highly influenced by the metal cylinder, itself. Of course it may be possible to 'interpret' such interactions, as they will be affected by the presence of the liquid on the inside, but I doubt if it will be as simple as the usual transmitter/receiver pair available in the Arduino 'set of experiments.

On the other hand, if you can mount the sensors inside the cylinder, the situation would be very difficult. Of course, that will present a host of new problems, particularly mechanical, electrical connection, sensor tolerance of the lpg chemically and pressure-wise, and safety.

The obvious method is to measure the weight of the cylinder, but I guess that isn't practical in your situation. Alternately, a sliding float and position sensor, inside the cylinder, but again there will be many mechanical, etc. issues.

Your laser suggestion seems a bit far fetched unless your cylinder has transparent windows?  .. lasers produce light beams which do not travel through metals very well! Or maybe, you were thinking of putting it inside? ..

Good luck!

@muby

The sound projected would have to traverse the metal twice, once going in and once coming out.

Would running a thermal sensor down the outside of the cylinder work to determine where the gas is cooling the bottom part of the cylinder ? The "full" section should be a slightly lower temperature than the "empty" section.

@davee 

Thanks for the feedback Dave. 

Yes, an important objective of the solution is to be non invasive, thus mounting sensors inside the cylinder would not be feasible. The idea is to attach the sensor to the base of the cylinder, externally. I have read on other (old) forums that ultrasound can be used for such an application. However, i don't know the precise type of ultrasound sensor (or configuration) that would be required.  

For the laser, i was thinking of cylinders that are made up of composite material. But my focus is really on the metal cylinders for now. 

@will 

Thanks for the feedback Will. 

Yes, you are correct. It will have to traverse the metal twice, thus it will need a sufficient amount of energy to get back to the receiver. 

Thermal sensor is an interesting idea.  I have come across a similar idea, where a sticker was used to determine the temperature profile along the height of the cylinder. The sticker turned out to be unreliable 

 The principle is correct. Heat is transferred from the cylinder to the liquid, thus cooling the area. 

However, we are looking for a method that can tell you precisely how much lpg you have left and track your rate of consumption. Not just, whether it is empty or not. But in theory, this should work. I will consider how i can develop this further. 

Thanks 

Hi @muby and @will,

  The thermal sensing idea is indeed creative. I guess this would depend upon the temperature difference between the liquid and the gas. If the gas is being used at a 'high' rate, then the evaporative cooling effect of converting the liquid to gas will also be 'relatively' high, so it might be possible to monitor it with a number of external sensors, although metals generally conduct heat fairly well, so this will reduce the magnitude of any gradient across the cylinder. By contrast if no gas is being used, and the temperature of the environment surrounding the cylinder is also stable, then the temperature of the whole system should reach equilbrium.

As for composite materials and lasers, 'composite' is an extremely general term, so I don't know what you have in mind. For a 'time of flight' laser system, I guess the material would need to be optically transparent (and not disperse the beam shape too much) to the wavelength of the laser. And of couse the liquid would need to reasonably efficiently reflect that wavelength.

@davee and @muby

Yes, it's a really interesting problem, trying to peek inside a closed, pressurized, metal (for now, at least) object to detect the level of a clear, colourless fluid.

I was reasoning this way (and I'm no chemist/physicist, so I could be way off base) ...

Assuming he ambient temperature is well above 40 degrees, the bottom part of the tank containing the liquid will always be absorbing heat through the cylinder wall (since the ambient temperature is always higher than the LPG's boiling point). This, in turn will cause a slight increase in pressure and temperature in the top (gaseous) area of the cylinder, resulting in the surrounding air to be warmed.

Since the tank is in equilibrium, we would expect that the "leakage" of heat into the bottom of the tank would equal the "leakage" of heat out of the top of the tank. This should create a small difference in the heat of the wall of the cylinder.

Since one requirement of the measuring process specified by @muby is that the system be "non-invasive" i was thinking that if the cylinder was stored vertically (since any other orientation would be even harder to measure) then a small sensor (such as a TMP006 module) could be mounted on a slide just outside the back of the cylinder and be driven up and down a slider a few millimetres away from the tank wall.

This would allow a contact-less heat sensor to be positioned anywhere along the height of the cylinder and sense the temperature by I/R alone. Contact would never be needed (nor allowed). The resulting temps can be recorded by a micro-controller and analysis of the temperature at each height might yield enough difference to determine the liquid level to a sufficient accuracy. Accumulated results saved with times could then be used to determine usage rate(s) and expected lifetime for the cylinder's contents.

Volume of the remaining LPG would be the volume of the hemisphere at the bottom of the cylinder plus pi times the cylinder radius squared times the height of the liquid phase.

This suggestion was intended as an example of an alternative to sonic and laser sensors, a test rig should be very easy and inexpensive to build and would not waste many valuable resources (except time, of course) in construction and operation of a prototype.

Hope this helps explain my reasoning.

Hi @will and @muby,

 If the cylinder is actually supplying gas to a 'system' outside of the cylinder, at an appreciable rate, then (Will), your suggestion maybe plausible, as this flow of gas will both reduce the pressure inside the cylinder and also remove the higher energy molecules from inside the cylinder, causing the average temperature of the contents of the cylinder contents to reduce.

However, if the gas flow is stopped, then the contents of the cylinder will transition back towards an equilibrium state. Over a range of pressures and temperatures, the boiling temperature of a liquid depends upon the pressure. A cylinder of (say) butane that has been kept in a (vibration-free) room, at constant 15 degC/59 degF with the valve closed, for at least a few hours to reach thermal equilibrium, does not contain a 'boiling' liquid. If you could tag and watch individual molecules of butane within that cylinder, then you would observe molecules moving around, including some of the molecules near the surface of the liquid 'breaking free' and becaming gaseous, but this would be balanced by gaseous molecules rejoining the liquid. A 'normal resolution' video camera inside the cylinder would only observe a pool of still liquid.

(The butane would only start to 'boil' when the valve is opened, releasing the presssure, and reducing the effective boiling point temperature of the butane within the cylinder to below the present liquid temperature.)

So, my hypothesis is that a temperature based scheme requires the gas to be flowing out of the cylinder (or some equivalent energy flow mechanism), to create a non-(thermal)equilibrium situation. And in practical terms, this gas flow must be sufficient to produce significant localised cooling, given the masking effects of thermal conduction by the cylinder, etc.  Depending upon the gas demand profile for your application, this might or might not exist.