Flow rate instantaneous value to Totalised value

can any one help with the math for converting an instantaneous value to a totalised value for Gas Consumption usage every hour.

in the plant we have an existing gas flow meter installed for main gas supply line the value is usually around 2000m3/hr.

the citect SCADA system "samples" this instantaneous value every 20 seconds and logs this value to a trend file. i need to totalise these values for Gas usage consumption every hour.

for this exercise lets assume all instrumentation is calibrated and reading correctly in the I/O device (Bailey Infi 90) and correctly displayed in the SCADA

so for 1 Hour i will have 180 samples (3 / min x 60min) do i simply sum all these values and divide by a value (180?) to get a usage value or have i missed something here ?

That is 11.11111111111111 per 20 seconds so if you are going to accumulate every 20 seconds you need to take the actual value divide it by 180 then add that to your totalizer. In saying that it will not give you very accurate results as it may vary during the 20 seconds you are not reading it but should give you a rough total.

The easiest way to remember how to do this is to imagine a trend of flow versus time. The totalised value is the area of that graph above zero. So if you break the area down in PLC samples, you can approximate each sample to a square and get the area by multiplying X and Y and adding it to the total variable.

Does it make sense? This pops up fairly regularly here and it's not that hard if you can picture it this way.

If possible, I like to do 1.0 second sampling for totalizer functions.
Much better accuracy.
You have to assume the flow is constant between samples, 20 seconds is a long time for such an assumption.

Ken Moore said:

If possible, I like to do 1.0 second sampling for totalizer functions.
Much better accuracy.
You have to assume the flow is constant between samples, 20 seconds is a long time for such an assumption.

Click to expand...

Bear in mind that on average the rising and falling signals will equalize, and if the totalization is over a long period relative to the sampling rate the errors will balance out. This may not be good enough for custody transfer applications but is probably more than sufficient for process control unless the flow rate fluctuations are severe.

I suggest taking the average of the instantaneous flow at the start and end of the sample period, multiply by the time elapsed, and add to the previous total. Make sure you have consistent units.

60 min/hr x 60 sec/min / 20 sec/cycle = cycle/hr

NewTotal = OldTotal + (cu_m_per_hour1+cu_m_per_hour2) / (2 x 180)

Many thanks to all responders, very much appreciated.


thanks for the extra calculation info Tom, this is most helpful