Pump pressure at low flow

[BigBill53]

Hello,
I have a pump system that is giving me some headaches and I can't figure out why. This pump is controlled by a VFD, because the load varies quite a bit. The VFD has been set up to vary the pump speed in order to maintain a constant pressure. The pressure sensor is downstream from a non-return valve.

The whole thing works quite well when valves are open. The problem is when all valves are closed. The system was designed so that when the VFD slows down to around 40% of its maximum speed for 15 seconds it goes into sleep mode. The idea here is that if all valves are closed the VFD keeps slowing the pump down to maintain the pressure setpoint until it reaches a point where it can turn off. The non-return valve ensures that pressure is kept in the pipe so that the VFD does not cycle from being in sleep mode to run mode.

The problem here is that the VFD is not going into sleep mode when the valves are closed. There are no leaks and there is no water flow, but the pump just keeps pumping in order to maintain the pressure. Because there is no water flow the impeller is overheating.

Originally we thought it was the programming in the VFD but that has been tested and proven to work.

Any help would be greatly appreciated. I have installed a number of similar systems to this and never had these problems.

 

[NGLENGR]

Centrifugal pumps are fun like that.

Is your pressure sensor downstream of the check valve? Does the check valve leak? Might be that the pump is running at 50-60% speed to keep the pressure up and never goes into the idle mode.

Check the pump curve and calculate what the "deadhead" pressure is at 40% speed.

 

[MortenA]

Just my understanding - you have a check valve upstream the pump - on the suction line?

Question: How do you know the check valve dosnt leak?

Best regards

Morten

 

[BigInch]

How much does the pump slow down before it overheats? Or, in other words, what percentage of full speed is the pump running when overheating begins.

Answer that above and in the meantime you might want to think about this,

I don't think you are using the correct control philosophy to do what you want to do, or your equipment isn't operating within the parameters you need to do it. I assume that your pressure sensor is the signal you use to adjust the pump RPM and that this is a typical centrifugal pump.

Why should the VFD slow down the pump?

If you have a pressure sensor between the pump and the check valve, it will never read a low enough pressure to slow and turn off the pump when the pump is running, unless all valves are wide open and you have very, very low resistance in the discharge piping when at maximum flow.

If you have a pressure sensor after the check valve, basically the same thing happens as in the above.

The only thing that will ever happen to the pressure sensor no matter where it is installed (on the discharge side of the pump) is that it will read pretty high values.

As a downstream valve closes, that just makes your pressure go higher and higher as the pump moves to the left on its chart.

In order to make a pressure sensor that slows or stops the pump work when pressure is increasing, you must have a pump discharge pressure when flowing at around 50%, or less of BEP flow, that is higher than the discharge pressure you need at BEP flow. That is possible with the right pump curve, but most pump curves get flat in that range, so there will be little pressure difference to work with. But you could conceiveably do this and slow the pump down as pressures approached a pressure setting in that high range when flows were reducing and the operating point moved to the left of the curve. But the vfd would not slow the pump down more than the speed that it needed to run the pump holding directly onto that pressure set point. If the downstream valves opened and flow increased, the discharge pressure would try to drop and the vfd would pick that up and speed up to hold at the pressure set point again. This appears to be what you want to do, except

As the pump reaches its slow down pressure, maybe it does slow down, or maybe it doesn't (answer the first question) but in any case it does keep running. Why? Because the pressure set point is still too high for any speed the pump is running in those flow ranges to ever start slowing down, or it does not slow down enough, runs at a still too high rpm and overheats.

As I mentioned above, if the pump curve is flat in the low flow range, it would be very easy to set the slowing pressure too high. You might even have it set totally above the maximum pressue that the pump can produce, or the curve is very flat and the sensor can't pick up enough change in discharge pressure to send a big enough slow down signal.

I think what is probably happening is that a vfd controlled pump discharge pressure drops very fast when a vfd is slowing a pump down. If the vfd is slowing the pump down, it slows it so much that the maximum pressure it can produce at the slow speed drops below where you have the pressure set point and it just stays at that rpm, but that rpm is still too high to prevent overheating before sleep mode kicks in.

If you can set the timer (if there is one) that begins the count-down period before sleep mode begins, or (if no count-down timer) just to begin sleep mode at a higher rpm, you might solve your overheating problem.

All of that is pretty dependent on what your system curve is doing (if valves are open) as the pump slows down too. If your system curve drops too fast as flow reduces, your pump will again just keep on trying to run at the highest rpm it can as it trys to keep up the pressure. You might not ever reach your sleep mode pressure.

 

[BigInch]

Forgot to ask the other questions.

What is,

1.) the shutoff pressure of the pump, ie. pressure at zero flow,
2.) discharge pressure and flow at normal operating point and
3.) the discharge pressure and flow at the 40% speed opeating point
4.) pressure set point.

 

[tr1ntx]

"I have installed a number of similar systems to this and never had these problems."

How similar?

"The whole thing works quite well when valves are open. The problem is when all valves are closed."
". . . the VFD is not going into sleep mode when the valves are closed."

What is the pressure reading when the valves are closed?

With no flow, what would make the non-return valve close? Is it spring actuated or mounted vertically or something? The flow is deadheaded, as the pump slows, the pressure drops and probably goes below the setpoint. From that point, you'll never get to where you're trying to go, you're chasing your tail.

One thing that might make this work (if I have the problem pictured correctly) is a pressure storage vessel downstream from the non-return valve. Maybe an air-over-diaphragm type or maybe just a standpipe if you can. Increase your pressure setpoint a little bit and then when it trips and the pump slows, the backflow from the pressure storage will close the check valve and maintain pressure in the system. You'll need 2 pressure setpoints, the slightly higher one to initiate slowdown and the originally desired pressure to restart.

 

[BigInch]

It seems apparent that some high pressure setting slows the pump.

"Increase your pressure setpoint a little bit".

If he makes the set pressure higher than it is now, I don't understand how will that slow the pump down at all.

 

[NGLENGR]

It may be that his VFD is looking for <40% speed as the "idle" speed. The program is also looking for a certain pressure setting. The VFD/Pump combo is probably running at some higher speed to maintain deadhead pressure.

We had a new VFD driven compressor that was programmed to "idle" at a certain speed and then it would trip. Another loop in the program forced it to maintain a certain minimum speed because of a pressure requirement, so it would never go below that value after start. Essentially, it would never go through a normal stop sequence!

That's why he needs to find the pump curve and ballpark his idle speed based on what the deadhead pressure would be. Don't arbitrarily assume 40% on one pump will be the minimum speed when it can easily run 41% all day just to deadhead.

There's also the factor of what the suction is...does that vary?

 

[BigInch]

In the attached typical looking pump system, set point at redline, you can see that rpm could be reduced for any operating point above that line, but to hold at least that pressure, it couldn't go below about 85% of full speed.

You can see that a 40% speed makes a head so far down there that the pump will run at 40% speed all the night and day long trying to reach redline, but it never, ever will reach it.

You'd need sleep mode to kick in at anything below 85% speed in this diagram.

 

[tr1ntx]

BigInch,

Not by itself. I think he needs a trip point and a restart setpoint, probably along with a little pressure storage downstream of the check. My thinking is that either the check is not closing, and thus the VFD is oscillating up and down across the setpoint, or if the check is closing, the slight backflow drops the downstream pressure and the VFD picks back up, again oscillating across the setpoint.

 

[BigBill53]

Thanks for your responses guys. Being an electrical engineer I wouldn't say I am a pump expert - that's why I needed some expert help from mechanical engineers.

I wasn't aware of this "deadhead" pressure until you mentioned it. From what I can find this is the pressure at zero flow. The system has a pretty big head under certain situations, and looking at the pump curves it needs to be running at 80% of full speed to maintain this pressure. In other situations the head is a lot smaller and so to keep the pressure the pump only needs to run at 30% speed. I will have to change the system so that the valves can only close when there is this smaller head.

I made an incorrect assumption that because there was a check valve (sensor is downstream from it) the pressure in the line would be maintained as the pump slowed down if all the outlet valves were closed. Looking back there probably isn't enough pressure differential in this case to close the check valve, and so the VFD keeps running at speed to maintain pressure.

 

[ccfowler]

BigBill53,

Deadhead operation of a pump more than momentarily is an excellent way to create excessive repair costs. Internal pump damage starts very soon. Control inherently gets messy as the pump becomes an expensive mechanical water heater. As the water within the pump warms quite rapidly, its density decreases, and the discharge pressure drops due to the reduced fluid density. The key is that a centrifugal pump produces "head" with "pressure" being the result of both the head and the fluid density.

It would be no surprise to find that after a seemingly brief period of deadhead operation your pump may be unable to produce enough "pressure" at full speed to lift the discharge check valve due to the water in the pump having been cooked to too low a density. (I've seen pumps inadvertently operated deadheaded long enough to turn into mechanical steam generators rather than just mechanical water heaters.) Since your pump is seeing deadhead operation by intent, its internals are being subjected to progressive damage in repeated incidents (increments). This internal damage results in increasing internal recirculation and progressively deteriorating performance. Your control scheme presumes "constant" pump performance characteristics that it actually turns into "variable (deteriorating)" performance characteristics.

It would be worth considering the installation of a "minimum flow" recirculation line from the pump discharge to some point far enough upstream in your system to provide adequate cooling of this flow. This recirculation flow would commonly be controlled in a simple ON and OFF manner based on assuring a suitable minimum flow through the pump. The cost of such a system is usually readily justified by the avoided repair costs and improved process system performance and availablility.

Valuable advice from a professor many years ago: First, design for graceful failure. Everything we build will eventually fail, so we must strive to avoid injuries or secondary damage when that failure occurs. Only then can practicality and economics be properly considered.

 

[BigBill53]

Hey guys,
Thanks for your help. I know this thread is a little old now but I just thought I should tell you the whole result of this project (and maybe stoke up the fire a little bit...)

The system works perfectly now. The VFD maintains a pressure of 700kPa. When all outlet valves are closed the VFD goes into sleep mode. When the valves open the VFD comes out of sleep mode and starts pumping again.

It turns out somebody had tapped a 1" line (off a 12" pipe) to fill a small water tank. This wasn't on any drawings and that section of the pipe is buried (which was nice and convenient). The VFD was not going into sleep mode because this tank was filling.

Now here is the part that I can't explain: I have set the sleep mode for the VFD to 25Hz (50% speed). The pump curves show that to maintain a pressure of 700kPa at no flow I would need to run the pump at 85% speed.

The whole thing has worked fine for 3 weeks now.

So, am I reading the curves wrong? Or what's on the curves at zero flow not what happens in the real world?

 

[BigInch]

What happens in the real world when Q=0 can depend quite a bit on if the Q is the numerator, or the denominator. In some equations it appears in both places.