Positive displacement pump control

[Buzzen]

Hello All,
I am using a acqoustic type flow meter made be flexim. I am controlling my pump with a PID loop. Has anyone been successful trying to control a flow with this type of setup when the flow is as low as 200ml/m or roughly 3 feet per minute, using a positive displacement pump?

 

[RussB]

Sorry but a positive displacement pump and 3 feet per minute, does not compute. Also very slow update devices are difficult with PID. You may be better off with logic control, even with a rather long scan.
To control a PDP, monitoring RPM is key. Do you have an encoder or other feedback device on the pump?
More information may be helpful in getting a better answer from someone.

 

[Buzzen]

The feedback is clamp-on transducers that feed info into a flowmeter that gives off a 4-20ma signal. One of the problems that I am having is the pulsing action of the pump is hard to control at such a slow rate and the flowmeter is bouncing 20% below the setpoint and 20% above. I have tried using loginc to minimize the bouncing effect by using an average that goes into pid. But I am still getting the bouncing and I am afraid that the pump will eventually burn-up. Now the error value in the pid is hitting the zero mark so I know that I have tuned it correctly its just that darn pulsation screwing everything up. I have been told that I may get more control using a smaller diameter flow pipe is that true.

 

[RussB]

Maybe, but one can only guess, like the designer did it looks like. Not knowing any details of the pump, fluid, line diameter or length.
Sounds like a classic example of a mechanical engineer installing a system without input from the controls guy.

I would toss the flow meter and install an encoder on the motor then use it and a little math to control it plus act as a flow meter.

 

[widelto]

I´ve tried it the other day and it´s a PITA due to pulsations. I change my pump and use a dose pump with known flow per minute.

 

[mellis]

You haven't mentioned what kind of positive displacement pump you have. It makes a big difference. If you must live with a pump that pulsates, and you must use a PID loop to control flow (that's two big IFs), then filtering the flow signal so the PID doesn't chase the pulsations is required.

Also, you state a line velocity of 3 feet per minute. That sounds way low for an ultrasonic to pick up.

 

[Peter Nachtwey]

So what do you want to control? Position or Flow?

Hello All,
I am using a acqoustic type flow meter made be flexim. I am controlling my pump with a PID loop. Has anyone been successful trying to control a flow with this type of setup when the flow is as low as 200ml/m or roughly 3 feet per minute, using a positive displacement pump?

It makes a big difference as to how to approach the problem.

 

[BryanG]

It depends on how quickly you need to control the flow and how quickly your set point changes? All the flows that I control/monitor are from reciprocating pumps and so I have some experience. My set points don't change so I just need to keep the pump output constant'ish depending on how the back pressure from the system affect the pump output.

The solutions we use are:
Move the flow sensor as far away from the pump as possible, pipe work has a nice averaging effect.
As long as you can take having a slightly sluggish system then add more averaging to the flow sensor input.
Increase the PID sample time, again leads to a more sluggish system.
You can even add an accumulator which is a mechanical device that absorbs pressure/flow spikes/drops. In a low pressure system you can do much the same thing with a short vertical section of pipe with air trapped inside.

So, yes it can be controlled, depending on the accuracy and speed you need.

 

[Buzzen]

I wanted to share what I have learned about controlling the poitive displacement pump with a pid loop via flowmeter. The problem of bouncing was straightened out a little with adding damper factor in the menu of the flowmeter. Then I used an average of five points then at first I divided by five, when that didnt quite do the trick I started experimenting by dividing by different numbers. For example 4.88 worked. Then I added a timer and a counter so that would slow the flowmeter down and with some patience I finally tweaked it to the point that it worked! Thanks everyone for the Help.

 

[leitmotif]

Here we go again pulling answers or hens teeth?

To adequately answer your question we need specific system parameters
Pump type - piston gear screw diaphragn etc etc
Volume per Revolution peristalic or stroke
Liquid pumped
It would be nice to have volumetric flow rate vs RPM or stroke curve.
System pressure

OR
you can send a crystal ball

Dan Bentler

 

[Peter Nachtwey]

I wanted to share what I have learned about controlling the poitive displacement pump with a pid loop via flowmeter. The problem of bouncing was straightened out a little with adding damper factor in the menu of the flowmeter.

But you aren't controlling flow you are only filtering it!!!! If you want to know the instantaneous flow that will not do.

Then I used an average of five points then at first I divided by five, when that didnt quite do the trick I started experimenting by dividing by different numbers. For example 4.88 worked. Then I added a timer and a counter so that would slow the flowmeter down and with some patience I finally tweaked it to the point that it worked! Thanks everyone for the Help.

OK. If you are happy then I guess we are done.

 

[raveon]

For lobe style pumps I have used closed loop control systems. I use a pickup (example prox) for a lobe in the pump to calculate a rev of the pump. In turn this gives me ML's per lobe. I then run this data through a totaliser function block in a PLC. I have Temp sensor and pressure regulators as well. I use the temp sensor to watch for changes in viscosity and i regulate pressure to optimise the fluid low. Generally speaking its an adaptive fluid curve in which the pump and regulator work together to achieve efficient fluid transfer without shearing the living heck out of the product. PIDE runs both the regulator and the pump. it took about a dozen iterations to get it to work, but i use all the time now.. not enough O's in smooth. It's a pricey control package though.