Flow totalisation in PLC 3

[bmsg]

I have an application where I have to totalise the flow value of coke oven gas (in a steel plant)in a PLC based automation system. For this, I have a flow transmitter in the field which gives a 4-20mA DC output corresponding to the flow range.

Can I use this 4-20mA signal directly in the PLC(through analog input card)for totalisation of this flow? Or, should I convert this 4-20mA DC signal into a pulse signal through an external pulse convertor and then feed the pulse signal to a pulse card of the PLC for totalisation?

Which philosophy will result in better totalisation accuracy?

Or is it better to use an instrument generating a direct pulse signal corresponding to the flow and then feed the pulse signal to a pulse card of the PLC for totalisation?

BMSG

 

[FrancisL]

An external 4-20 to pulse convert is unlikely to improve accuracy, since that still depends on the primary reading.
The best would be a flowmeter with built in integration and a serial connection to the PLC.
If integrating 4-20 from the flow meter beware that when the flow is low accuracy is low. If the flow is zero you may still have a reading from the FT and you could be integrating it leading to a falacious reading. So you may need logic to switch integration off when you know the flow is zero.

 

[Mike6158]

If your PLC supports the HART protocol you could use a multivariable flow transmitter to calculate the flow rate in engineering units and send the data to the PLC. Some venodor's multivariable flow transmitters can be configured to send the unscaled flow value via an analog output. You could then scale the value in the PLC, save it to a REAL (or double word) memory register, do your totalizing calc, and then save the result to a "totalizer memory register". It might help to know which PLC you are using. The capabilites of the various mfg's PLC's varies widely.

 

[doug123]

I use the pulse to a high speed counter where accuracy is required.
To show our accuracy at the export station we typically fill an open 55 gallon drum @ 200 - 250 gpm to the top of the drum without spilling a drop.

 

[FrancisL]

Of course an important question is what type of flow transmitter is being used!

 

[bmsg]

Thanks to everybody for all the replies. But, what I basically wanted to know was whether using a flow transmitter with a 4-20mA output would result in less accurate flow totalisation than a transmitter with a pulse output. The PLC supplier is insisting that there will be large-sacale inaccuracy if I use a 4-20mA DC input. Is it true?

BMSG

 

[FrancisL]

It all depends!
I asked what the flowmeter was because it matters. If it is an orifice plate it is not going to be better that +/-1% at best, and so 4-20mA, even when integrated may be good enough.
If it is a turbine (say) then you should read the pulses from the meter directly, since that is the primary signal.

Ask:
What are the accuracy requirements?
What is the accuracy of the flowmeter?
What is the accuracy of the 4-20mA input card, in fact of all the compoments of the 4-20mA loop?
Given these you can work out whether integrating in the PLC will degrade the accuracy significantly.
If you do decide to integrate in the PLC then make sure that the integration logic is reliable - for example it should run on a time interrupt fast enough to respond to changes in flow rate without losing accuracy.
Your PLC supplier by making dogmatic statement is not impressive!

 

[bmsg]

Thanks FrancisL,

I agree with all you have said.

My flowmeter is an orifice plate with a SMART differential pressure type flow transmitter, which outputs a 4-20mA DC signal proportional to the DP.

My only confusion is that if I integrate the transmitter signal (after square root extraction) in the PLC, will there be any significant error in the process of A/D conversion and integration vis-a-vis direct counting of pulses from a pulse converter?

 

[FrancisL]

I would have thought that a SMART differential pressure type flow transmitter could also do the square root extraction and then send 4-20mA proportial to Flow.
If not then you must linearise it in the PLC then do the integration.
A/D conversion errors depend in the resolution of the PLC'a Analog input. If it is 12 bit then your error will only be about 0.025%, which is far less than the error in the Flow meter, and so it is not likely to degrade overal accuracy significantly.
If it is 8 bits then you have a cheap input card. Even so that would contribute 0.4% which may still be acceptable.
(When adding errors by the way is it normal to use Root Mean Square so adding 1% from the flowmeter to 0.4% from the A-D still only makes 1.07%)
What accuracy are you aiming to achieve?

 

[bmsg]

Thank you again, FrancisL. I would require a nominal accuracy of about +/- 1%. Your reply has been very helpful.

 

[scozans]

I had experienced this kind of problem before, I used a SIEMENS SIMATIC S7-300 with analog input module. The good thing with my project was that errors were not that important. I just evaluated the values read by the CPU from the FT and totalized it using continues addition (since flow is varying, therefore, I used a fast pulse to enable the addition process). Say, try to graph a sample flow reading with respect to time (x, for time and y, for flow) and then plot a sample curve; that curve will be your volume. Can you see the idea now? Segment your curve equally and then measure each segment and add them all, that will be the total! Each segment represents your pulse signal in the program; the signal to add. Now everything is just a piece of cake! Hint: the smaller the segments are the accurate the reading will be. Caution: consider ther size of your program to be able to execute the pulse correctly. If you want some more details just post your interest here.

scozans
Philippines

 

[FrancisL]

With a Siemens PLC I would use a Time based interrupt OB to run the integration code. For example OB32 Cyclic interrupt 2 (default interval: 1 s) This would make it interrupt the main scan every 2 secs so that you do not have to worry so much about the size of your program.

 

[9984]

Don't forget to use the line pressure and temperature in your flow calculation, has a big effect on the accuracy.

 

[Mike6158]

I've written an FB for the S7 that uses flow, pressure, temperature, and analytical data to calculate gas and liquid flow through an orifice. It's a fairly large block but it much more accurate than using a fixed C'