Total flow from CFH?

[NetNathan]

I am reading CFH (cubic ft per hour) from a flow meter.
I want to calculate total CF used over a period in time...per day....per week....per month.

What is the calculation for this? The flow is not always on.
I assume that if I am flowing 400 CFH for 24 hours then my Total flow is 400x24=9600 CF.
Is this correct?

 

[Bullzi]

Does the meter have a totalizer on it? If so you can read that flow total. If not you will have to totalize it in your PLC.

 

[NetNathan]

4-20ma out for CFH
Pulse out for total flow

The Siemens module for reading pulse input (6ES7550-1AA00-0AB0)
is $551 for 2 channels! I have 6 flows to monitor.
I assume this is the correct module.

 

[Gil47]

What is the make and model number of your flow meter
as pulse output is often more common than a 4 to 20 output
and often both are available.

 

[danw]

A flow meter pulse output represents some volume (configured in the flow meter), for instance, 10 gallons. So to get a total, you count the pulses and multiply by how much each pulse represents. 152 pulses = 1520 gallons

But you have to watch out for terminology/nomenclature.

A frequency output is for flow rate: volume/time, like gallons/minute. An example of a frequency output might be 0-5Kz = 0-100gpm

Frequency outputs consist of pulses, too, which can be confusing.

If you count pulses, you might need a retentive counter to keep the total when the power is off.

>400 CFH for 24 hours then my Total flow is 400x24=9600 CF.
Is this correct?
Ans: yes.

 

[Steve Bailey]

How many cubic feet does each pulse represent? Is that something you can configure in the flow transmitter? If you can keep the pulse frequency low enough so that you can detect each pulse with a standard DC input on the PLC, that's the easiest way to totalize. If you can't handle the pulse frequency with a standard DC input, you'll need the high speed counter modules.

Otherwise, to totalize the flow from the rate, you have to sample the flow rate periodically, multiply the sampled flow rate by the sample period and add that to a running total. Shorter sample times give you better accuracy, but means that you have to able to handle adding small increments to large running totals. That can be a problem if you're dealing with 32-bit floating point numbers. For longer sample times when the flow can change by a large amount between successive samples, you can take the average of the current sample and the previous sample to improve your accuracy.
Search the forum using the keyword "totalizer" and you'll find plenty of discussion on the subject.

 

[NetNathan]

I can handle pulse configuration, Siemens has a nice FB for it.
I just don't want to spend the extra money ($1500+) just to read the pulse input.
I was hoping to use the 4-20ma in for CFH to calculate total flow.

The flow meter is a FOX Thermal FT1. I just noticed the FT2 will talk Profibus-DP, which is one of my Siemens CPU ports, so now I can get tons of info.

 

[Steve Bailey]

Looks like the pulse output is locked at 100 Hz and you can configure what that represents. 100 Hz square wave is 10 mS on/ 10 mS off, so I doubt you'll be able to capture that with a standard DC input module.
An example of the totalizing calculation: Suppose the gas is flowing at 200 CFH and you sample once per second. Each sample period is one 3600th of an hour, so you'll have to divide your sample by 3600 to convert the sample to cubic feet. Your sample of 200 CFH represents 0.055 cubic feet.
If instead, you sample every ten seconds, that is one 360th of an hour and your sample represents 0.555 cubic feet.

 

[NetNathan]

My original intent here was to convert CFH to Total CF over a period of time.....without using pulse input.
CFH is sent as 4-20ma and I wanted to convert this value from CFH to Total Flow over time selected....day, week, month.

 

[Steve Bailey]

Sorry if I didn't make myself clear. The second part of my last post is an example of the totalizing calculation from the analog signal scaled to CFM.

An example of the totalizing calculation: Suppose the gas is flowing at 200 CFH and you sample once per second. Each sample period is one 3600th of an hour, so you'll have to divide your sample by 3600 to convert the sample to cubic feet. Your sample of 200 CFH represents 0.055 cubic feet.
If instead, you sample every ten seconds, that is one 360th of an hour and your sample represents 0.555 cubic feet.

 

[AshleyParr]

i have found with flow rates that vary sampling more often improved the accuracy for me, so once per 500ms or 1s

 

[NetNathan]

Sorry if I didn't make myself clear. The second part of my last post is an example of the totalizing calculation from the analog signal scaled to CFM.

An example of the totalizing calculation: Suppose the gas is flowing at 200 CFH and you sample once per second. Each sample period is one 3600th of an hour, so you'll have to divide your sample by 3600 to convert the sample to cubic feet. Your sample of 200 CFH represents 0.055 cubic feet.
If instead, you sample every ten seconds, that is one 360th of an hour and your sample represents 0.555 cubic feet.

Sorry Steve....you are correct.
Thanx...

I assume I keep adding the flow per second for 24 hours to get total flow for the day....and so on for week and month.

 

[Bullzi]

I assume I keep adding the flow per second for 24 hours to get total flow for the day....and so on for week and month.

I would do that for each day and store the daily total. Then you can add up the daily totals to get a week total and month total. You can then store the month total and get a year total from add the month totals up. You could then store the year total and start over for the next year.

 

[Steve Bailey]

I assume I keep adding the flow per second for 24 hours to get total flow for the day....and so on for week and month.

Correct. And that will be where you may have to balance the desire for accuracy against the limitations of 32-bit floating point numbers.
You get better accuracy by sampling more frequently, but each individual sample is a small value. Totalizing over days, weeks or months produces large values. 32-bit floating point numbers give you seven significant digits. It doesn't matter where the decimal point is, all you get is seven significant digits. In my first example, where each sample represents 0.055 cubic feet, as long as the cumulative total is less than 10000 you can keep adding 0.055 cubic feet at each sample. Once the total gets above 10000 you can only add 0.05 cubic feet per sample (or 0.6 if you round up) and when the total gets to 100,000 cubic feet the smallest increment you can add is 0.1 cubic feet.
There are plenty of strategies to deal with this issue. Bullzi has mentioned one. You just need to be aware of it. And if your PLC supports 64-bit floating point numbers, it is less of a problem.

 

[NetNathan]

Correct. And that will be where you may have to balance the desire for accuracy against the limitations of 32-bit floating point numbers.
You get better accuracy by sampling more frequently, but each individual sample is a small value. Totalizing over days, weeks or months produces large values. 32-bit floating point numbers give you seven significant digits. It doesn't matter where the decimal point is, all you get is seven significant digits. In my first example, where each sample represents 0.055 cubic feet, as long as the cumulative total is less than 10000 you can keep adding 0.055 cubic feet at each sample. Once the total gets above 10000 you can only add 0.05 cubic feet per sample (or 0.6 if you round up) and when the total gets to 100,000 cubic feet the smallest increment you can add is 0.1 cubic feet.
There are plenty of strategies to deal with this issue. Bullzi has mentioned one. You just need to be aware of it. And if your PLC supports 64-bit floating point numbers, it is less of a problem.

I am using a fairly high power Siemens CPU (1516-3). It should support 64 bit.

Yes....my first Siemens job in 15 years (last one was S5). A whole new learning curve.