independent vfd of 3 pumps feeding system at separate locations 11

[electricpete]

I am asked to provide input on the following project.

We have three 75HP 1800 rpm well water pumps. They pump into a common header, except that the tie-in points for the three pumps are widely separated (up to a mile apart I think).

The plant load varies widely. Up until now these machines have been fixed speed. Also the pumps are far from the operator’s stations so often the wrong number of pumps is running for the load (far to the left). We have had very low reliability. Presumably we are also wasting a lot in energy costs.

What has been proposed is to install VFD for each motor. Each motor will be controlled independently by sensing it’s discharge pressure and varying speed to match a target.

Also there is some kind of lead/lag feature where one pump would be designated to turn off during periods of extremely low load. There would be a time delay before the pump turned off and minimum off-time before it turns back on.

I am not very familiar with control schemes for this setup. I asked the question how do we control the load sharing and ensure the pumps are operating near BEP. I was told that all of the above parameters (pressure setpoints and time delays) will be adjustable. We will watch and see what it does and tweak as necessary.

The well water system is deemed relatively low priority to the plant and the project is expected low-budget and quick turnaround. We have quotes for the VFD system installed to feed existing motors at somewhere around $12,000 per pump. (Outdoor installation with cabinets).

My two questions:
1 – Is individual pressure control suitable for this type application?
2 – Is a detailed study of the system operating parameters required first, or is it ok to trust we can tweak it later.
3 – What do you think about the 3rd pump shutoff feature? I am worried that under certain conditions we are doomed to have constant cycling as soon as the time delays expire.
Thx.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[itsmoked]

That big distance between pumps would worry me.. I can see pressure waves running all over the place with confused results..

Whether a "detailed" study is needed or not, I would ask a disinterested party (paid even) to at least illuminate the possible/expected problems so that they could be addressed before modifications.

 

[ScottyUK]

ePete,

I suggest a cross-post into forum830 where the controls guys hang out.

My personal opinion, not having done an application quite like this:

1) Probably not the best option in an ideal world, but it may be the best you can get in the real world without serious expenditure. You can minimise the problems by using a decent industrial grade PT - e.g. a Rosemount, Foxboro, etc. Cheap sensors will doom you to a world of frequent calibrations and lousy control.

2) If you keep your loop tuning fairly placid I think you could probably tune this on the fly. Remember that a closed hydraulic system can develop large pressure swings from virtually no flow, so be wary of very low or zero flow conditions.

3) Shutting down the third pump should be fairly trivial if you have this lot under control of a DCS, and could be arranged even without one. We use a similar control system to bring in a second pump when the first one reaches 70% of capacity, and it doesn't revert to single pump operation until demand drops back to 20% or so. Figures are approximate - that job was installed quite a while ago.

That's three questions!


----------------------------------

One day my ship will come in.
But with my luck, I'll be at the airport!

 

[Barry1961]

Hitachi has a free savings calculator for using a VFD on pumps. It is in software downloads.

http://www.hitachi.us/Apps/hitachic...p/hitachi/forbus/powerequipmentsystems/&nId=2

Using VFD’s and sequencing multiple pumps is a very common application but the distance is a little far. Fincor makes packaged multiple pump controls for varying demand times but I don’t know if they allow multiple pressure inputs. The distance would probably kill it though since the package would have the VFD’s in one panel.

I would put a VFD/sensor at each pump that has Ethernet(ABB, Hitachi, AB) and run it back to a switch, (N-Tron) to a local PLC and do my own program.

You should be able to find modes of operation for multiple pumps by doing a google for “pony pump” vfd control.

Barry1961

 

[wschel]

Do you have a hydrostatic tank in the system? That & very slow acting controls should lessen the problem with pressure waves.
Will there be any communications between pump controls? You'll probably need to use a combination of pressure & total flow in order to match the number of pumps to the demand.

 

[LionelHutz]

I can see all kinds of potential problems using 3 pressure transducers and 3 different PID loops to control pressure in a single header. It may work when there's lots of flow but the PID loops may interact badly with each other. Here's an example of what I can see happening. Say one transucer/VFD is reading the pressure as being higher than what the other two are reading. It's PID loop will start dropping the pump speed to compensate. At the same time, the pressure begins to drop so the two other PID loops will speed up their pumps to compensate. You'll then end up with 2 pumps running full speed and the other one running at a much slower speed.

 

[electricpete]

There is unfortunately no surge tank at all in this system. There is always some demand in the form of partially open throttle valves.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

There is no communication planned between pump stations or from pump stations to plant DCS.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[ScottyUK]

You may be able to modify the control using a flow element to reduce the target pressure as flow increases to reduce any tendency to cycle. Think of it as something akin to three generators operating in parallel under droop control. Most of the more sophisticated drives have some PLC-type functionality which allows you to perform manipulation of the inputs and drive a PID block within the drive. You could even imply the flow as a function of pump speed rather than using a physical flow sensor. I'm pretty sure Control Techniques' Unidrive has enough processing horsepower to do this - you may need one of the add-on modules to get the full logic capability. ABB's products probably have similar capability too, likewise with Siemens etc.

----------------------------------

One day my ship will come in.
But with my luck, I'll be at the airport!

 

[itsmoked]

I also see a sort of current hogging analogy.

If you were to put a restriction on each pump's connection to the main line then control each pump with a sensor on its side of the restriction then they would be partially isolated from the vagaries of the main line and their individual VFDs would be less influenced by the others.

But this is all stuff a specialist would consider...

 

[wschel]

Could you provide more information about the header configuration? This has a significant Are the pumps connected to a single pipe (size?) at different points (How far apart?) or does each have a line meeting at some common point? What is the flow & pressure ranges? With a little more information we might be able to come up with a workable solution.

 

[schell44]

I want to preface my remarks by saying that I am no expert in pumping system but I have designed a few complex control schemes. This is strictly my own opinion of the situation.
Assuming that all 3 pumps have similar pump curves and problems of stability can be overcome, the closest pump will be operating at a lower pressure, higher flow and the furthest pump will be operating at a higher pressure, lower flow. Controlled by a fixed pressure setpoint, each pump curve will cross the BEP at a different point. So trying to operate all three pumps close to the BEP will not work with simple pressure control loop.

The following will require some testing at different flow conditions with pressure measurements at all three pump locations.

I think that the most efficient method of operating the pumps is to run the closest pump as much as possible (lowest system losses, therefore cheapest operating costs). When this pump reaches maximum flow, the measured pressure becomes the start setpoint for the 2nd closest pump. The speed setpoint would be some value above the start pressure, determined by trial & error. The stop setpoint would be close to the highest pressure with the 2nd pump running are minimum speed that still produces flow.

Etc. for the 3rd pump.

Depending on the system operating condition, you may not need a VFD on the closest pump. For low flow conditions, you can experiment with other operating configurations. I realize that this will result the the closest pump operating most of the time, but assuming the well can continuously produce sufficient water and the pump will not be overloaded, I think this is your cheapest solution.
(I hope this makes sense, it's after my bedtime).

 

[LionelHutz]

Generally, you'd use one common speed reference for all VFD's feeding a common header. But, you have no plans of sharing a common speed reference between pumps and it sounds like this application would make that very difficult. And, I don't think you have much if any hope of getting 3 pumps to run at the same speed using 3 different pressure transducers and 3 different PID loops.

So, this means I'm with schell. Many VFDs can be set to just stop once they reach a certain minimum speed. So, if you tell each successive VFD down the line to operate at a slightly lower pressure you should get a successive starting operation. I'll try to give the basic example here for 2 pumps.

To begin with, you have a low flow condition so pump 1 is running at some reduced speed. It is creating X pressure in the header. Pump 2 is set to maintain Y pressure, where Y<X. At the present time, the pressure in the header is greater than the setpoint for pump 2. So, the pump 2 control loop is telling pump 2 to run at minimum speed. This means pump 2 will be stopped since the pump 2 VFD is programmed to stop the pump when a certain minimum speed is reached.

Now, the flow requirements increase. Pump 1 accelerates to full speed and reaches the point where it can no longer maintain X pressure in the header. So, the header pressure begins to drop, eventually going below Y pressure. The control loop for pump 2 will now begin to tell pump 2 to accelerate. The commanded speed for pump 2 will go above the minimum speed to stop and pump 2 will now start. Pump 2 will bring the pressure up to Y and it will be the variable speed pump controlling the pressure.

I hope this makes sense. I see a couple of issues. The first is that you may have difficulty if the flow requirements change rapidly. It would become difficult to maintain a constant pressure during pump start/stops. It would also cause the pumps to start and stop a lot, which may not be a big deal depending on the motor and pump.

The second is the difficulty when stopping a pump. You'll likely have to take the pump down below the speed where it's basically producing no flow before you shut it off. If the first pump is at full speed and the second is producing flow when it shuts off then the pressure at the second pump will drop and it will want to immediately re-start. Because of this, you may need some external controls that monitor the pump speed and not allow it to operate without any flow for more than a certain length of time before forcing it off. Or, possibly, just have some external controls with timing that holds minimum flow speed for a while before dropping it off so that the first pump has a chance to actually slow down. Then delay the second pump re-start so the first can accelerate again to make up the pressure loss. <- I hope that one makes sense

 

[DickDV]

Usually, with an arrangement like this, the three wells are not equal, ie, one is preferred for the main flow and the others are "trimmers".

This naturally affects the control scheme. If the original poster would advise on this aspect of his system, we could put together a more appropriate solution.

 

[electricpete]

There is no reason for us to prefer running one pump over the others (unless it makes the control easier in which case we'd be glad to do it).

The demand varies within a range that sometimes requires only one pump and sometimes requires to (in their current fixed speed configuration).

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

Let me correct a typo:

"The demand varies within a range that sometimes requires only one pump and sometimes requires two (in their current fixed speed configuration)."

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[thewellguy]

I would strongly suggest that you involve your locak well contractor in this project. I have had to fix many well pump systems because the person who programs the pump does not understand its operation. Depending on for you have a submersible motor or a vertical hollow shaft motor there are different requirments. Depending of brand and size there are different requirements. For example a Hitachi sub must be up to minimun speed in 3 seconds and the minimum speed is dependant on if its a 2 or 4 pole motor. The 8" and 100" 100hp and 125hp have a minimum speed of 42 hertz. This is for the lift on the thrust bearing so that it is properly lubericated.

There are also many different program requirement for vertical hollow shaft pumps depending on if it is a oil or water lube pump.

I know that Berkley pumps has come out with an ABB drive that already has a software package in it that takes care of most of these concern. Goulds pump also has a vfd package however, I perfer the Abb due to the more options available.

I think the easy way to do this would be three drives at three different pressure points.

Good Luck

 

[DickDV]

I agree with thewellguy if there is no reason to favor one well over another.

I would set up each pump motor with its own VFD and pressure setpoint loop (PID). Setting the lead well for the desired system pressure, the first follower pump for about 3 psi lower setpoint, and the second follower pump for about 3psi lower again.

You will go from minimum flow to maximum in about 6-7 psi.

I would not set the setpoint loops real tight. If the drive accel/decel ramps are inside the PID loop, I would set these out to around 15 seconds to avoid surging and one pump "playing" with the other two.

 

[electricpete]

These are vertical hollow shaft motors. I think the pump is "turbine type"

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[thewellguy]

Because they are tubrines then you don't have to be quit so careful. However when you set you minimun speed/sleep level make sure that the pump is actually pumping water (there is water coming out the discharge)and not just running. This is so the line shaft bearings have water for cooling.