Cooling Tower Make-up Pumps

[macmet]

I'm having an issue at my plant with our cooling tower's make-up water pumps.

What we're seeing is our cooling tower calls for make-up water, solenoids valves open downstream of the pump, five seconds later our make-up pump turns on, and then instead of pushing water up to the tower, the pump appears to just spin, eventhough a pressure gauge near the outlet reads 160psi. The expansion tank is always full as designed and the fill control loop seems to work well on it.

Some times the system will work right away, other times it takes an hour or two of manually overriding the control system to get the fluid up to the tower.

Other system details,
- cooling tower is 250 ft above the pump outlets
- expansion tank's surface is about 6 ft above the pumps
- fluid is water at common city supply temperatures
- the system worked last summer without issue.

I've read in our troubleshooting manual that the pumps may need to be primed. An online search seemed to point in that direction too, but my operator is adament that they do not need to be primed and that it is not possible for air to be causing the problem. I'm not sure I agree, but I'm not a pump expert and would like to hear what other people think may be causing the problem.

As always, appreciate any feedback or thoughts.

 

[GPRnD]

Ok, a few things don't make sense to me so I'll make what might be some (maybe) stupid observations:

The way you describe it the pumps have to overcome a 250 ft static lift at the outlet, equivalent to about 7.5 bar or about 109 psig. Presumably after the static lift it discharges into an open system.

Thus if you are measuring 160 psig, and the system was once working, I would suspect that you have a blockage or closed valve somewhere in the system.

Did you ever take pressure readings when the system was operating correctly ? If so what did they show ?

It sounds like you have a decently flooded suction (no suction lift), so really the only way for air to get into the pumps would be for the suction inlet to vortex from the free surface of the tank. However since you appear to be generating pressure at the pump outlet, I don't believe that is happening.

 

[macmet]

Thanks Brad,

You're correct, our system is open.

And we have also looked for any valves that are sticking, but the only valves that we can find appear to actual properly when the logic tells them. I base this on the sound they make, but perhaps they need to be completely taken apart.

Unfortunately, we do not have any pressure readings from when the system was functioning.

Thank you for your feedback.

 

[BigInch]

I think Brad is on the right track. What's the pump's BEP head. As it is you may be generating 30 to 50 psi more than you need for normal flow, depending on head losses in your discharge piping to move your design flowrate in that pipe, so I'm not totally sure. And a higher discharge pressure than needed for that might indicate two things, your discharge is blocked, forcing the pump back on its curve and causing a higher head to be generated at lesser flow, or perhaps the pump just might have been selected with too much head, again not sure. It would be convenient to know if there is any flow in the discharge pipe, but apparently we can guess that there is no flow, primarily because you'd see it shooting as high as 369 feet into the air. 50 psi would be a typical shut-off head increase you might find for a pump designed for a near 110 psig discharge pressure, so that would also tend to confirm that flow is completely blocked and the pump is at shutoff head. Is the pump overheating when it is doing nothing but spinning?

What controls are you having to work exactly? So far it does kinda point to a malfunctioning solinoid on a valve, or perhaps a roughly sheared valve stem that catches sometimes and operates the valve, but sometimes not.

Let your acquaintances be many, but your advisors one in a thousand’ ... Book of Ecclesiasticus

 

[DubMac]

macmet,

You might also want to check ALL of the distribution valves up on the hot water deck of the tower. Many of these are really crappy valves and although the stem may make it look like it is open, one of the primary valves may actually be stuck closed.

Also make sure you trace the discharge line completely. Many of the half*ss tower rebuilders put valves in the weirdest places.

With the 160psi pressure, your's does not sound like a pump problem.

 

[Pumpsonly]

Some times the system will work right away, other times it takes an hour or two of manually overriding the control system to get the fluid up to the tower

.

Perhaps you could further elaborate on your above statement what you did with the control system to get it working.

 

[ione]

Could you post the performance curve of your pump. Your pump could run deadhead against an obstruction. Since you’ve noticed sometimes the pump run properly while other not I’d check whether the solenoid valve you’ve mentioned in your OP is working properly.

 

[KiwiMace]

I agree with the comments above - you are pumping against a closed valve or less likely some misc obstruction. I would add:

On butterfly valves, commonly the taper pins that retain the disc to the shaft can shear. The disc stays still while the actuator/shaft rotates, and can cause the intermittent opening. On the outside everything will look normal.

The valve will need to be taken out to check the internals, and also check the seats. Rubber seats may be pinched and deformed, causing the pins/shaft to shear in the first place.

Crack the manual drains/vents or check gauges along your pipework. Seeing where the deadheaded pump pressure drops off will indicate your problem valve. Check the pressure ratings of your valves as 160psi exceeds the lower classes.

 

[macmet]

Everyone, please see attached pump curve. I apologize for the poor quality of the scan, but it's a scan of a faxed copy.

I'm not a pump expert and would be interested to hear anyone's thoughts on how appropriate a pump with such a curve is for this application.

I'd also like to ask if pipe systems such as this typically have check valves in them? Our system does not.

Thanks

 

[BigInch]

Its pretty much what I guessed you had in my post above. The BEP is at 150 psi (with water) and its shutoff is 33 % higher at 200 psig more or less. It is probably what you should be using, but really can't tell unless we know your flowrate and piping lengths and diameters. I'd guess its pretty well on the right head numbers though.

We will design everything from now on using only S.I. units ... except for the pipe diameter. Unk. British engineer

 

[GPRnD]

Honestly, the suitability depends on what your system resistance curve looks like. You have 250 ft of static head + some unknown quantity of frictional head loss from the piping after the pump discharge.

Say for example your frictional head loss was very low, the pump would runout to 480 USGPM which is 130% of BEP. This is higher than preferred but acceptable on a smallish pump like this assuming NPSHa is adequate.

Since in reality you will have some frictional losses, it is likely the pump is operating closer to BEP. In which case I would say it looks fairly well matched.

 

[DubMac]

If you are reading 160psi near discharge flange then the pump is running at shutoff. In other words at 0 gpm flow, your shutoff head is right at 375 feet which is 162 psi.

Certainly sounds like the pump is doing what is told to do and isn't the problem.

Sounds like you have traced the discharge line and there are no closed or obstructed valves??

Looking at the performance curve for this pump, I note that it is extremely flat; meaning that there is very little change in pressure as flow varies from shutoff out to design point (absolutely flat from 0 to 250 ft head).

You only have a 10-15 psi variance from shutoff to operating point (all of that is above 250gpm) and you are using pressure to try to control the pump's flow with an extremely narrow range of pressures to work with.

Imagine if the steering wheel on your car only had 1/8 turn to control the entire car; the smallest movement would cause you to swerve all over the road. OK, that is a cheesy analogy but is somewhat similar to what you seem to have.

A pump with a steeper curve would have been much better for this service, but I realize you are probably stuck with the one you have.

Your system curve is also somewhat flat as it is mostly static head (250ft). WIth just a little bit of line plugging from scale or obstructions, the friction portion of your curve will rise dramatically and will move the operating point of your pump back to the left (pump will ALWAYS run at intersection of the performance curve and the system curve).

Once the operating point moves left of the 250 gpm point, your pump will essentially "wander" back to shutoff and flow goes to zero.

This is pumps 101 stuff to most of the guys on this post, I'm putting it in because you said you were somewhat new to pumps. Maybe some of the other guys (or gals) could expound upon how your system curve is going to dictate where the pump operates. I would like to hear some opinions on how this could be solved given the above assumptions are correct.

 

[rmw]

Here is what I see in the curve and in the information you presented.

First the curve gives no NPSHr curve or value. Curious. It is a pump basic.

I notice that the suction velocity is horrendously high or at least it is in the pump world(s) I have lived in.

You state that the liquid level is 6 ft above the surface.

You state that the operation is intermittent. Does the water in the expansion tank (this is the suction source??? right???) remain at city water temperatures or does it rise (sunlight heated) while not pumping?

What is the duty cycle?

I suspect that your pump is cavitating. Can you hear any unusual noises associated with the pump operation, especially when it is not appearing to pump?

Is the expansion tank surface open to atmosphere? If so, can there have been any trash that has entered into the piping between the tank and the pump? (dead birds, pipe scale, etc?).

If closed, is it pressurized at all - I can't see where you mention that.

Even though you state that the expansion tank flll control works well, it is not instantaneous. Does the level pull down when the pump initially starts?

Where is the check valve that prevents emptying of the line between the pump and the CT? Is the discharge guage you are reading upsteam or downstream of the check valve?

What the heck size CT is it that wants 300 GPM instantaneously as makeup? You don't have to answer that question if it is personal - just curious.

I suspect that at a zero flow condition when your pump starts, before the flow losses begin to kick in, this pump runs out on its curve and cavitates the pump. If you can obtain a NPSHr curve for this pump, I believe that you will see it begin to rise steeply in the region that you are trying to operate.

And, if no check valve, and the system is depending on the solenoid valves to hold liquid in the line, then in the seconds before the pump starts, the line can be draining back into the flow tank, and emptying the line reducing the head pressure on the pump guaranteeing that it runs out on its curve.

That is my story and I'm sticking to it until someone pushes me off my soapbox.

rmw

 

[GPRnD]

I suppose anything is possible but...

If the pump is cavitating why is it putting out 160 psig ? That value is basically the shutoff head of the pump. Even if the pressure gauge is after a check valve it should read 109 psig if the pump were failing from cavitation, not 160.


Also to design a 360 USGPM pump that required a flooded suction (33 ft NPSHr) gives you a Nss of 4200. 4200 is impossibly awful performance IMO.

All of which means I tend to disagree with it being cavitation unless the OP has omitted some detail.

 

[Pumpsonly]

I agreed with bradshsi, with 160 PSI discharge pressure, the pump is obviously at shut valve condition as per curve. The pump is working perfectly OK.

macmet, you have not feed back on the question about having to manually over ride the control system to get the water up to the CT.
I suspect that the solenoid valve did not open.


Without a check valve in the discharge piping means the water in the vertical pipe will drain back to the suction tank every when the pump stop. Unless the solenoid valve you mentioned is closed before the pump stop.
It it very strange to use a solenoid valve when the pump is solely supplying water to the CT
Is your solenoid valve located close to the pump or at the out let of the discharge pipe?

It would be better to install a check valve near the pump (with by-pass valve for draining) and configure the control to start the pump motor directly from the CT level switch.

 

[DubMac]

From my dusty library, the B&G Series 80 pump here requires about 30' NPSH at operating point. With 34' from atmosphere and 6' from static above suction, cavitation doesn't seem a likely culprit. Your problem seems to be on the discharge side of the system.

If discharge piping is either scaled up or obstructed by partially closed valve, your friction losses may have grown too large for the pump to overcome.

Your pump was designed to support about 350' total head; 250' static head plus 100' friction head. The maximum head this pump could ever support when it was new = 380' total. Too bad you already have max diameter impeller already installed.

If your friction head has surpassed 130' for whatever reason, this pump will not be able to get water up to the CT basin.

Pump wear also eats into the pump's ability to support the 350' required.

The elaborate control system seems unecessary; maybe just go with float valve level switch in the cold water basin to control pump's on/off??

You could put a common HVAC triple duty valve on discharge flange to take care of shutoff/check/balance duties??

 

[macmet]

Brad, what does NSS mean? I'm not familiar with using that to size pumps.

Pumpsonly... I guess I tried to explain the manual override of the system to get the water up to the towers. But I will try again.

Our towers have a water level sensor in the basin that has a high, regular, and low sensor. When the sensor gives the input to the control system that the water level is low, the system sends a signal to the solenoids to open, and five seconds later the pumps get a signal to turn on.

When this happens, you can hear the solenoids "open". What I mean by this is, is just that you can hear them click. The pumps when they turn on, begin to produce 160psi.

Sometimes this sequence succeeds in getting water up to the towers. Other times the pumps just produce 160psi but no water gets to the tower.

When we notice this, we put the low level sensor into override and toggle the input between low/not low. When we do this we can hear the solenoids opening/closing (clicking). Usually at some point the flow will all of a sudden start. Sometimes this takes 5 minutes other times a few hours. We have not been able to determine the pattern.

To answer some of the other questions...

The solenoid valves I'd estimate are 50 ft downstream of the pumps and about 15 feet higher. They are in the horizontal leg near the point where the pipeline turns up to to the roof.

Our tank is open to the atmosphere, but the top is mostly covered so debris is not much of an issue. The water in the expansion tank is in the basement of the building and at what I'd consider "normal" room temperatures.

Also, our system is only a couple years old. I believe this is only the second summer with this chillers and last year feeding the water towers was not an issue.

 

[BigInch]

Hey flatliners. This pump does not have a flat curve. The shutoff head is 33% higher than BEP head. At least that's not flat in my book ... or in Colombus' book either.

We will design everything from now on using only S.I. units ... except for the pipe diameter. Unk. British engineer

 

[ione]

This has already been asked before, but not answered. When the system is running properly and water delivered to its destination what's the pressure reading?

It seems strange to me that there could be an obstruction on the discharge line that sometimes prevents water from flowing and other not, but the solenoid valve.

 

[macmet]

I'm told that it reads 160psi when water moves as well.

I have not seen this myself though. I will look on Monday to confirm.