I have a riddle for you

[Buzzen]

Thanks Tom,
very helpful I will apply these factors and re-post how it all worked! It was a real pain trying to have a control loop via flow meter. The PID was always floating.

 

[Narlin]

The flow is really low on some applications in the area of 3-4 gallons per hour, with about 5 psi. I am using a flow meter that will respond fairly accurately when the tank is full. As the tank empties the psi is around 1-2 which will throw off flowmeter.

Most flow meters that work in the 3-4 gal/hr range also report accurately at 1-2 gal/hr? Although you may want a booster pump to keep the flow up, I would guess that your meter is accurate and that the "positive displacement" pump is pushing strokes that aren't completely filled with fluid due to the lower pressure. Consequently, calculating flow from pump strokes whenever there is known variable pressure can cause an error.

narlin

 

[Buzzen]

Well I did a little research and came up with a pump report which nobody that I work with ever told me existed. Ever have that happen? Here are the numbers.

Average flow (LPH)= 98.8936@60hz . Would I be better off trying to figure out Liters per Hz, then control the flow that way? OR do I still need to figure out pump speed? Is the RPM's of the motor the pump speed? Or is there something else I need?

 

[Lancie1]

OR do I still need to figure out pump speed?

Yes.

Is the RPM's of the motor the pump speed? Or is there something else I need?

If it is a direct drive pump (no gears) then the motor speed is also the pump speed. Now you need to tell us exactly what pump and motor that you have. You can buy a small Variable Speed Drive to control the speed. With a drive with a known speed, you can then calculate the expected pump flow rate.

Another possibility is to add some type of sensor (maybe a tachometer) to measure the motor or pump RPM.

Or you can buy a metering pump that has settable flow rates.

 

[mellis]

Chemical metering pumps come in many flavors, such as peristaltic, piston pumps (fixed and variable stroke), gear pumps, and so on. Most are "positive displacement" types - that is, for every rotation of the pump a specific volume of liquid is moved from the inlet to the discharge. For most pumps this volume remains fixed, and at a constant speed the flow rate is fairly constant across a wide range of of inlet and discharge pressures. The volume moved on each revolution is the pump displacement, and is usually expressed in liters per revolution, or cubic inches per revolution, or such. If you know the pump speed you essentially know the flow rate:

Q = d x rpm
Q is flow rate
d is displacement

If you have one of these pumps you don't need a flow meter - multiply the speed by the displacement and you have the flow rate.

Now some pumps have variable displacement - they can adjust the effective stroke of the piston by moving a swash plate or by some other mechanism:

http://www.miltonroy.com/corp/details/0,10294,CLI1_DIV39_ETI3654,00.html


In this case, at 50% adjustment the displacement is 50% of maximum, and again you can calculate the flow from the pump speed. The pump manual should explain all of this, and give you the accuracy of the adjustment and so on. The stroke adjustment can be manual, or respond to a 4-20 mA signal or whatever.

If you don't need high accuracy and there is no expensive or dangerous consequence to not delivering the correct flow, I agree that you can get away without a flowmeter.

However, having done many installations using flowmeters and metering pumps together I can say there are a lot of failure modes and operational problems with metering pumps that flowmeters detect quite easily.

My personal experience is that the internal relief valve popping open can be difficult to troubleshoot without a flowmeter. Most of the metering pumps I've worked with require a backpressure regulator on the discharge for proper operation. Getting these set properly can be a problem, especially at extremely low flow rates.

While it is true that for every rotation of the pump a specific volume is supposed to be transferred, that is the ideal case. In actual practice, metering pumps do have slip and it varies with material characteristics. In general, the higher the viscosity, the lower the slip.

Mechanical failures like motor coupling (normal or mag) can also be an issue.

 

[Tom Jenkins]

You can use a little short cut if you use a VFD. Simply take the percent of pump full speed (Actual Hz / 60) x Q max = Q Actual. At 30 Hz your flow will be 49.45 lph

 

[Tom Jenkins]

I can say there are a lot of failure modes and operational problems with metering pumps that flowmeters detect quite easily.

I agree - that's why I recommended keeping the flow meter for alarming.

Slip is usually figured into the displacement. Again, the pump manual should identify the impact. The accuracy of the control can be improved by using slip to determine the zero flow speed, often called the slip rpm. Then, instead of Q = rpm x d you would use the formula Q = (rpm - slip)x d.

An even greater refinement would be to use a bucket and a watch to measure actual performance, making sure to maintain the same backpressure on the pump. This will probably provide accuracy equal to that of a flow meter.

 

[mellis]

An even greater refinement would be to use a bucket and a watch to measure actual performance, making sure to maintain the same backpressure on the pump. This will probably provide accuracy equal to that of a flow meter.

I generally do a bucket check even on flowmeters, it's a good idea to independently verify the performance.

What we're talking about here is the difference between accuracy and repeatability. An open-loop system without a flowmeter can produce good repeatable results.

 

[Buzzen]

THANKS GUYS FOR YOUR GREAT RECOMENDATIONS, IT IS VERY HELPFUL! We still have calibration columns that I was going to use, but based on what you are saying a simple bucket would work even better beacuse, I would be working with the back pressure valve in place. Yes, it is controlled with ABB ACS35O VFD. The pump is a Milton Roy with no variable speed control built into it the model is a CS. The motor is a regular 3-phase 480 volt induction motor with external fan@1765 RPM.