VFD Flow Control

[IASMike]

We need to use a vfd and pump to control flow for a pump down application, where the pump is at the bottom of the tank. The problem we are having is that when the static head in the tank changes the pump charcteristics change. This causes the PID to oscillate. Can anyone give us some pointers or where we can find info on setting this application up?

 

[nbucska]

It depends on the specs.

1.) bang-bang control doesn't oscillate ( besides the _-_-)
2.) You may change the gain coefficients vs the tank level
This needs some hight sensor.
3.) use MFA (model free adaptive ) control
4.) You me redesign the rest of the system to reduce
the requirements for the level control


<nbucska@pcperipherals DOT com> subj: eng-tips
read FAQ240-1032

 

[steviefive]

I don't know if I get your problem right ...
When the tank is full and pump is on the bottom, the water level will "help" the machine pumping. Hence the pump will go to or go beyond max. flow conditions. The speed of the VFD should be principally reduced. If the controller fulfils this principle than you should lower the gain.

 

[laundry]

hi IASMike
look in threads programmable logic controllers what is PID I think this will help you.

 

[Marke]

Hello IASMike

What are you using as a feedback signal for the PID loop?

Best regards,

Mark Empson

http://www.lmphotonics.com

 

[IASMike]

We have a Mag Flow Meter as feedback for flow.

 

[Marke]

Hello IASMike

OK so you are looking for constant flow. I would increase the integral time and reduce the derivative gain on the PID. If the derivative gain is too high and you have a system lag, you will get instability. If you have a sudden change in load conditions, then the derivative gain can be used to anticipate the reaction and counteract its effect. I do not believe that this is the case here.

Best regards,

Mark Empson

http://www.lmphotonics.com

 

[itsmoked]

Yes Marke, I would say that the loop should just be PI controlled not PID.

 

[IASMike]

These do not work since when the pump has 21 foot of head the pump starts pumping at 100 rpm. But at 7.5 foot of head the pump will not start pumping until 433rpm. The PID works fine with 21 feet of head, but at 7.5 feet of head it oscillates due to the bottom end having do response. I believe what we have to do is change the scaling as the head comes down. But I have been unable to locate any info on how to determine what to change the scaling to.

 

[itsmoked]

Can you try bang-bang control?

 

[jraef]

IASMike,
No offense, but you appear to be "in over your head" (pun intended). nbucska's first response included the information you need, Marke's response was right on yet in simpler terms, but you still don't seem to accept it. The system should work regardless of the changing pump operating head conditions, that is always the case with pumps in variable level applications. It may be that your pump was not designed to operate at the lower level, but that is not a VFD / PI control problem per se.

IMHO you need to get someone involved who really understands dynamic flow operations using PI control in VFDs like this. It is done all the time with great success using the methods already described here. This forum is for tips, it isn't meant to be a classroom where we tell you every detailed step to take.

"Our virtues and our failings are inseparable, like force and matter. When they separate, man is no more."
Nikola Tesla

 

[katmar]

IASMike,

All the advice you have received so far is from Electrical/Instrument engineers. I don't believe you have an instrument problem. Let me give you a chemical engineer's take on the problem. If the only tool you have is a hammer, then every problem looks like a nail. Let me show you my hammer - you've seen all the electrical hammers!.

In order to understand your problem you need to get hold of the pump curve. The standard pump curve has a separate curve for various different impeller diameters. This is NOT what you need. You need the chart where there are separate curves for different rotational speeds of the same size impeller. If you ask the pump supplier he should be able to generate it for you, or you could do it yourself using the affinity laws (this is tedious work - hopefully the supplier has software to do it).

Once you have the curve, draw a vertical line at the required flow rate. You will see that where it cuts the curves, each curve has a different slope. What this means is that your system does not have a fixed characteristic. The poor controller is trying to use a fixed set of rules to solve a varying problem. You may overcome it with a "clever" controller, or with exactly the right tuning, but I doubt it.

To solve the problem you need to move your operating regime to an area where the curves are more parallel. Generally this would mean moving to the left or upwards on the chart. It may mean that you need a bigger pump running less efficiently. You may be able to use the same pump by putting an restriction orifice on the pump discharge to move the operating point up the chart. Either way, I suspect you are going to be using more power, which negates the use of the VFD. In this case a fixed speed pump with a classical arrangement of a control valve on the discharge might actually be simpler and better.

regards
Katmar

 

[jraef]

Well stated katmar, and that is what I was alluding to in the last sentence of the first paragraph in my response. It may in fact be a pump design issue, and that is why I thought he needs the help of someone well versed in this who can observe the situation first hand and come up with real suggestions, rather that keep fretting about how to do it with free advice from the keyboards of people with other jobs.

That said, I have made PI control loops with VFDs work in situations such as this with regular success, as have many other of the poster and readers of this forum. This is a common problem that we have all faced at one time or another, and in my opinion, the solution is available to him with the equipment he already has. I leave open the possibility that I am wrong and he needs to replace the pump, but I put that as a last resort.

"Our virtues and our failings are inseparable, like force and matter. When they separate, man is no more."
Nikola Tesla

 

[katmar]

Hi jraef,

For sure, the last thing you want to do is change the pump. Apart from the pump cost, chances are he would need new foundations, new piping and valves, and a new higher power VFD. But I believe it is worth doing the analysis on a pump curve to see if there is a problem or not.

I must confess (with my process engineer's hat on) that I have seen you instrument guys work miracles with control that was totally impossible just a few years ago. But sometimes I think it is a bit like putting 10 computers into a flying brick and calling it a Tomcat. Give me a simple old Cessna that can glide home when the engine stops!

regards
Katmar