Ultrasonic Flowmeter in Water Dosing

[asg0127]

Is it good idea to use ultrasonic clamp type of flowmeter for water dosing?
I already using the flowmeter pulse output by setting to 10L per pulse. Installation set-up was followed and signals seems working fine, flowrate reading seems on target at 2.9 liters per second (approximate equal during bucket test) This pulse output is fed to PLC digital input and have a program of Total Volume = Total Volume + 10.

The problem is, it is always off target. Actual volume either over or under the target.

I am starting to doubt if ultrasonic flowmeter can be used for dosing of liquid. Did anyone used the same type of flowmeter?

By the way, the conductivity of pure water that was used is less than 2 microsiemens that is why electromagnetic was not considered before.

 

[parky]

Many questions come to mind:
What is the frequency of the pulses i.e. 10ms, 100ms
If the pulses are quite fast are you using high speed inputs or interrupts
Also why totalflow = totalflow + 10 why not use actual flow (count) X 10
Are you using a oneshot to do the addition it could be that the plc scan time is faster than the length of the pulse i.e. it adds 10 more than once for example the pulse goes high it adds 10 to the total & does it on next scan again on the same pulse.
Are you getting air turbulance (bubbles) in the water (can affect the count)
Although I never actually implemented a clamp on FM I did trial one on a CIP system & it was pretty accurate.

 

[asg0127]

What is the frequency of the pulses i.e. 10ms, 100ms
>Every 10 liters (approx. every 3 or 4 second if flowrate is constant at 2.8 liters/sec)

If the pulses are quite fast are you using high speed inputs or interrupts
>Used only Standard digital input since the pulse rate is bit slow.

Also why totalflow = totalflow + 10 why not use actual flow (count) X 10
>That's what came into my mind xD. Kidding aside, it is the usual formula discussed in this forum. I am not sure what will be the difference between this two formula, mathematically speaking.

Are you using a oneshot to do the addition it could be that the plc scan time is faster than the length of the pulse i.e. it adds 10 more than once for example the pulse goes high it adds 10 to the total & does it on next scan again on the same pulse.
> yes, I used one shot.

Are you getting air turbulance (bubbles) in the water (can affect the count).
> I am thinking about this right now and will try to find out. But with regards to the installation, I just followed the 10D upstream and 5D downstream (actually, it is positioned more than 10D and 5D) and it is installed in upward flow to ensure pipe is always full. By the way, the pipe is a stainless and as what I saw during installation, all fittings are done neatly, inside and out.

Although I never actually implemented a clamp on FM I did trial one on a CIP system & it was pretty accurate.
> yes, it is said accurate as what the documents and datasheet is telling and the reason I decided to chose Ultrasonic flowmeter.

 

[cardosocea]

Is the quantity to be dosed into the process the same that you're dosing into the bucket? 5L seems very small to catch deviation errors if your dose is to be 100 or 1000L...



What's your setup? Does the pipe have a gooseneck to reduce the amount drained? Are the valves automated? When you start the process, does some liquid drain into the bucket as soon as the valve opens? All of these interfere with error of the measurement, usually in a consistent way provided that you repeat the same procedure all the time.



Then, assuming your bucket is 5L, you fill it in less than two seconds with a 2.9L/s flow rate. That seems too fast and too little quantity to be able to get a good consistent reading since the biggest source of error is the beginning and end of the dose. I think we need a bit more information to give you more tips.

 

[asg0127]

Is the quantity to be dosed into the process the same that you're dosing into the bucket? 5L seems very small to catch deviation errors if your dose is to be 100 or 1000L...
> I did bucket test just to verify the flowrate displayed in the indicator. I tried to check how many seconds to fill a 50 liter bucket and seems equal (bucket filling starts when the flowrate is stable)

Quantity to be dosed in production start at 350 liters to 1500 liters.

 

[parky]

My suggestion you make it say 2 litres a pulse at that rate most PLC's could easily see it alternatively just make it 5 litres just to see if it improves it what PLC are you using ?.
Most PLC's can have a scan time at less than 100ms, that means you should be able to see pulses of 5 per second allowing for a scan between rising & falling edges.

 

[Chris_M]

How far off target are your doses?
What variance is the flow meter rated for?
When I was researching some clamp on meters (admittedly a few years ago) the variance was higher than our other options.

 

[cardosocea]

What's your pipework configuration?

 

[asg0127]

How far off target are your doses?
>around 20%, sometimes its less than 20%. Generally, random error is observed.

What variance is the flow meter rated for?
>1% according to datasheet

When I was researching some clamp on meters (admittedly a few years ago) the variance was higher than our other options.

 

[asg0127]

How far off target are your doses?
>around 20%, sometimes its less than 20%. Generally, random error is observed.

What variance is the flow meter rated for?
>1% according to datasheet

When I was researching some clamp on meters (admittedly a few years ago) the variance was higher than our other options.

 

[asg0127]

What's your pipework configuration?
> sensors are mounted on upward flow at the highest point of piping layout where I believe pipe is full of water during flow.

 

[cardosocea]

And the cut off valve is where? just after the flowmeter? You seem to have the flowmeter in the right spot, but the valve can also add some imprecision unless you give it time to always drain down.



Have you tried accumulating the analog value? If this is in a Rockwell PLC there's a dedicated instruction to do this for you.

 

[asg0127]

Yes, after the flowmeter.

I tried totalizing the analog value but result is bad. So I tried pulse output (10liter per pulse) but still bad.
I guess, the problem is with the instrument. Flow reading is stable, but seems having problem with the repeatability as same setpoint was inputted but different results.

I am using a Delta PLC

 

[asg0127]

https://drive.google.com/file/d/1-Qaj72wQUpv7BCw8zemRYz6QKz6WP0-e/view?usp=drivesdk

Here is the picture of sensor

 

[Ken Moore]

What is the scan time of the plc, and update speed of I/O card, and how long is the pulse. You may be missing pulses, if the length is shorter than your update time/scan time.