Computing stuff tied to the physical world

Gas flow measurement artifacts

In Hardware on Mar 14, 2011 at 00:01

The raw gas consumption graph always looks a bit odd:

Screen Shot 2011 03 13 at 12.59.45

Green is gas flow, red is electricity consumption.

This is the same oddity I described two years ago on this weblog. The values are calculated from the last rotation time of the meter, but with gas flow things are a bit different: gas consumption is not continuous but a discrete on/off process, with some “modulation” once on.

So instead of each of those “tapering off” effects, what really happens is that the flow stops completely for a while and then resumes again later on.

This is an example of where the measurements need to be “fixed” (!), by injecting fake data points. It also illustrates how you can’t really know anything about consumption between disk rotation pulses.

The vertical grid lines in the above graph are spaced 500 l/h (resp. 500 W) apart, btw.

  1. Thanks JCW, love reading you’re blog.

  2. I’m just waiting for my partner to free some funds for me to buy some nodes off you and an appropriate relay plug :D

  3. This is a misleading graph indeed. You really can’t connect data points like that, since it does not integrate correctly. A better plot would be to connect every data point with a horizontal line to the previous. In that case, you join blocks of equal area, which actually correctly represents your total gas consumption.

  4. I see someone else needs to get inside the meter and paint a load more dots on the magic disk!

    Or maybe invent a non invasive gas flow measuring clamp device! Actually I seem to remember something about something like that involving ultrasonics or something… Unfortunately it’s a vague dusty memory from the back of my cranium, so no idea if it even works, let alone if it’s available!

  5. @TankSlappa: Allready invented! Have a look at this link (German):

  6. Interesting technique, but it sounds like the price tag is going to be, ehm, “very industrial”.

    • Just that big shiny metal housing looks expensive, let alone the contents!

      Example pictures of it bolted to something resembling an oil refinery don’t bode well for cheapness either :-(

  7. The gas meter and pipes in my metering cabinet make noise when gas is consumed. The amount of noise appears to be proportional to usage. The sound or associated vibrations can be measured of course. Problem is that it’s different in every installation (or maybe not?) and the noise level may depend on pressure, and the exact gas mixture, configuration of piping etc. I can’t really see this work.

  8. It should be possible to detect the rotation of the disk with the mirror as opposed to actual passage of the mirror

    But that would mean some more elaborate video processing…

Comments are closed.