Give me your opinion!!!

Ok, I'm looking for very general advice about how you would attack this project.

We have 4 pumps that pump water to a section of our building. 3 of the pumps are the same size and the 4th pump is smaller. Currently, only the small pump is on a VFD, the other 3 are on regular contactors. The system has an analog sensor on the discharge side. One of the 3 big pumps runs all the time and as pressure drops, one of the other large pumps comes on, then the 3rd large pump, then if necessary the small pump ramps up and tries to help maintain the setpoint using a PID loop.

We are getting ready to change the system so that every pump has a VFD. My initial thoughts are this: Still have 1 large pump run continuously maintaining pressure with a PID loop. Then as pressure drops start the other large pumps next and finally, if needed, the small pump. Where I'm stuck is this...when 1 pump is running the PID loop will regulate it no problem. When a second pump is needed do I just let that same PID control both pumps?? Same thing if 3 or all 4 pumps are running? One PID loop total, or more than 1?

This is my first adventure with multiple pumps on VFD control with PID. Any other suggestions or advice is of course welcome.

Edit: I'm using AB SLC 5/03 if anyone cares.

Have seen example where two pumps running at half speed are able to supply 100% of load at one qarter the power.
Flow is proportinal to speed ie 2 X .5 = 100
Power is teh cube of speed so
1/2 speed is approx 1/8 power X two pumps = 1/4 power.

Sounds pretty neat BUT
centrifugal pumps do not pump much below 20% rated RPM
= CHECK PUMP CURVE
motors may not have enough cooling at 10% speed or less.
Motor insulation if class B may not be able to handle VFD high voltage and spikes - more so a problem at 480 line than with 240 line.

Am fairly sure am preaching to choir - you can run multiple units on one VFD -- BUT you CANNOT start an additional unit on a VFD with one or more already running.

Is it possible to run smallest pump and be able to supply system under light demand. If so then you could use it as master and rest of units as needed would be slaved. One sensor controls one VFD with others slaved.
Problem here is if you lose sensor or the master VFD. Also I think using PID will you be able to handle different running pumps each of which has a different pump curve UNDER INSTALLED CONDITIONS. How will PLC know when it has excess pumping power and is able to turn one off?

Think I would go with two sensors. Let PLC pick the higher or lower of two and use it. Then if one fails you have automatic backup and an alarm capability for a failed sensor.

Then while you are at it have PLC start and stop different pumps to equalize run hours.

Hmmm lessee how can I make this more complicated?

Dan Bentler

leitmotif said:

BUT you CANNOT start an additional unit on a VFD with one or more already running.

Click to expand...

I'm not sure what you mean by this...? Are you talking about if you have more than 1 motor connected to the same VFD?

leitmotif said:

Then while you are at it have PLC start and stop different pumps to equalize run hours.

Click to expand...

I already does this, I just left that part out to make my original question simpler. I also have things in place to detect a bad sensor. Actually, there are quite a few extra goodies that are on this system to help with various scenarios, again...I just left them out!!

"I'm not sure what you mean by this...? Are you talking about if you have more than 1 motor connected to the same VFD?"

Yes. You can run multiple motors -- lets say 3 for example. Running and starting all three together is best. You cannot start just one and then another and then another to meet increasing load.

Dan Bentler

I'm still confused...

I am talking about starting one, then another, then another...except each motor will have its own VFD, 4 VFDs total.

shoelesscraig said:

I'm still confused...

I am talking about starting one, then another, then another...except each motor will have its own VFD, 4 VFDs total.

Click to expand...

That is the preferred way of doing it. I was fairly sure that is what you wanted to do so in hindsight maybe I should have not mentioned it.

Dan Bentler

Hi.
What I usually use is 1 PID.
The first pump starts on the PID and tries to maintain pressure.
If it runs at max speed for say 10 secs (after startup stabilisation) (this means that 1 pump can't keep up)then I start the second pump on a very slow ramp up.
As the second pumps speed increases the first pump will back off via the PID.
When the 2 speeds match, I drop the second pump out of ramp up mode to follow the speed of the first pump (ie both are controlled by the same PID)
Do the same for any additional pumps.

The reverse is the same if the pressure is high for a time period and all pumps are running on minium speed.
Drop 1 pump out of the PID and slowly ramp it down. The PID pumps will speed up to take up the slack. When the pump speed is at zero then stop the pump.

You have to play around with the speed and time settings but it works good.

Regards Alan Case

I like that layout Alan.

Alan Case said:

Hi.
What I usually use is 1 PID.
The first pump starts on the PID and tries to maintain pressure.
If it runs at max speed for say 10 secs (after startup stabilisation) (this means that 1 pump can't keep up)then I start the second pump on a very slow ramp up.
As the second pumps speed increases the first pump will back off via the PID.
When the 2 speeds match, I drop the second pump out of ramp up mode to follow the speed of the first pump (ie both are controlled by the same PID)
Do the same for any additional pumps.

The reverse is the same if the pressure is high for a time period and all pumps are running on minium speed.
Drop 1 pump out of the PID and slowly ramp it down. The PID pumps will speed up to take up the slack. When the pump speed is at zero then stop the pump.

You have to play around with the speed and time settings but it works good.

Regards Alan Case

Click to expand...

I would do a slight variation to the above. Only have one pump controlled by the PID ( the first one) when the output of the PID reaches 80% start the second and ramp it slowly up. This will cause the PID control to back off. When the PID gets to about 60% stop ramping the second pump.( freeze it in place) The first pump will try to control
via the PID. Continue this until the second pump reaches max then do the same for the thrid and forth.

The reverse is true at about 20% from the PID start ramping down
and shutting off pumps.

Shoeless,

My intreprtation of what Dan meant is that if you are using ONE VFD, then all the motors need to start together. Otherwise, if an attempt is made to bring on an additional motor after the others have started, then that motor will probably cause an overcurrent trip of the ONE VFD.

Since you are planning to have a VFD on each pump, then you don't have to worry about the above.

I don't have any experience with controlling several VFDs like you propose, but I remember that it has been discussed in the past here at least 2 or 3 times, and there were ways to do it.

Like Dan says, you will need to set a minimum speed for each VFD, so that it doesn't try to run a pump below its effective speed curve.

I think the control problem can be summarized like this:

If two or more pumps are running, then only the last pump needs to control speed, as the other pumps need to run at full speed.

So your PLC program should start the first VFD-controlled pump, and let a PID loop control its speed until it is runnig at 100% for some time period (in other words it can no longer keep up), then that VFD speed is locked at 100 and the second VFD and pump is brought on, with Pump 2 speed controlled by a PID loop. Repeat for all 4 pumps.

Then as demand slacks off, reverse the process: If the PID-controlled pump (the last one started) is running at minimum speed (as determined by pump curve) for some time period (say 3 minutes) then shut off that pump and transfer PID control to the next online pump.

In the above method, you only need 1 PID loop, because it can be transferred to the pump that is doing the final tweaking of the pressure.

Check the link.
http://www.danfoss.com/BusinessAreas/DrivesSolutions/Frequency+Converters/VLTAQUA++Drive+FC+200.htm
It has "Standard cascade controller"
I have never use it.
Maybe you need something like that

The best advicde so far is from Dan - "CHECK THE PUMP CURVE".

This is a problem with lots of wrong answers and several right answers. The mechanics of your system will be determinant on how to control this.

A problem with keeping one pump at max speed when you bring on another is that you have a big bump in flow rate. The second pump is going to have a minimum flow that is higher than 0%. If its min flow is, say 25% of design, you would be going from 100% flow to 125% flow - a bump in pressure is sure to occur and cause instability.

To get bumpless transfer you need to to have both pumps run at 50% capacity after starting and then have them ramp up, or have one ramp up and begin ramping the second when the first hits 100%.

If you use one PID control for all pumps you need to use different tuning constants when one, two, or three pumps are on since situation each will have different dynamic responses.

Note that pump speed and flow rate aren't a 1:1 correleation, and that pressure rise from friction varies as the square of the speed. If the system is mostly static head or you have a large standpipe, this doesn't matter.

There may be some help at this site: http://pumpsystemsmatter.org/

Again, programming the control is the easy part - figuring out the mechanics of your system is the tough part, and mostly a mechanical engineering function.

Lancie1 said:

If two or more pumps are running, then only the last pump needs to control speed, as the other pumps need to run at full speed.

Click to expand...

The only part I do not like this statement is exactly when one pump is at full speed and the one entering the game is at less that 10% of speed and stays there for a while. Last one is going to overheat.

A few questions about your setup.

Are all of the pumps centrifugal?

Do they share the same discharge header?

How important is the control of flowrate compared to pressure? What I mean is, which is the PID trying to control, pressure or flowrate?

In your current setup, assuming that all of the pumps are centrifugal and all share a common discharge header, all three big pumps come online at the same discharge head and contribute roughly equal flowrate to the system. This, I imagine, gives a noticeable bump in pressure/flow. Adding VFDs to the big pumps may not change that. Assume pump 1 is running at 100% speed. Now bring pump 2 online at 50% speed. Because discharge pressure follows speed, pump 2 will contribute nothing to the system until it is ramped up to 100% speed, matching discharge pressure of pump 1. Now the PID must backoff the speed of both pumps until either pressure or flowrate reaches the desired setpoint. You will still get a bump. Perhaps slowing the already running pumps as you ramp the incoming pump. This may not be desirable as a drop in pressure will occur and that is what it seems you want to avoid.

Are you trying to achieve less of a bump as the individual pumps come online? Perhaps a very slow ramp up of speed to the incoming pump would help prevent the pressure/flow spike. I do imagine that it is a very narrow speed band where the pump coming online starts to contribute to the system until it matches the output of the other running pumps.

Shawn

When I get a job like this that is math based, I like to start in Excel. Build a Scroll Bar tool that goes from 0 - 100% to represent the Load. Then work out the math and logic to get the control performance that you desire.

Once it all works, then transfer the math into the PLC program. When you are finished, save the spreadsheet to remind yourself how the job was done (and for training). You can also use the spreadsheet model as a troubleshooting tool by seting up the inputs that you see in your online system and let it tell you what outputs that you should be seeing in the field.

This forces you to get good with Excel, but once you are, it pays huge dividens down the road.

Best Regards,

Bob A