Maintain velocity while reducing cavitation on a gravity return line to a sump tank

[Rputvin]

I've got an interesting situation on a customer installation. The system uses a roof-installed cooling tower, being supplied with 1400-1500gpm of warm water (90-105°F). The return is a 12" gravity drain, which terminates slightly above the water level of the 3500gallon sump tank (approx 70-85°F).

The main issue is that the customer located that drain near a pump suction (one of five) and the velocity of the return water combined with the entrained air in the partially-flooded line is pushing air directly into the pump suction. As a result that one pump is cavitating as the case fills with air and it can't make pressure to push open its check valve. Everything else is operating as it should. These are end-suction, top-discharge, centrifugal Bell & Gossett e1510 5GB & 6BD pumps and happen to have an air trap in the case as the discharge volute comes from the side/bottom

http://bellgossett.com/pumps-circulators/end-suction-pumps/e-1510/

We attempted to counteract this scenario during design with baffling and a suggested setup for the return pipe. The customer/installation contractor didn't get the memo and here we are with the return in a bad spot that bypasses the internal baffles, and dumping the full force at the surface of the water in the tank and creating a big, agitated mess. We'll address this with some small rework changes to put that baffling in the right spot.

During the process of talking with the customer we had a contractor rep from the pump manufacturer do an independent evaluation. He concluded it was entrained air. The curious part of his report was his continued issue with air being in the gravity return line. His suggested fixes include moving the tower drain from the side of the basin to the center of the bottom (cannot physically happen without changing towers), or staying with the side outlet and installing a vortex breaker as well as a cap of sorts to prevent air from entering the outlet from siphoning. I don't believe air can be removed from the drain line and have it still function, though I may be wrong, and the tower basin must drain to empty during normal operation- so there's no way to ensure the return is flooded regardless. But it got us talking about maybe making changes to this setup to help ensure success on future systems, as this is not the first time issues have arisen from improper installation of the gravity return.

So on to my actual questions;

Does anyone have experience with similar gravity draining setups that might have a good approach? I'd love to hear thoughts on possible changes to equipment or practices, or information related to the topic. Are breathers or vents of some kind advisable? Would a diffuser at the end of the return line remove a lot of the agitation/air being introduced into the tank? Do you use a stillwell or build more containment into the tank to handle the messy return water?

Process diagram with some information removed is attached.

 

[bimr]

The best approach is to install a separate section on the tank where the drain flows. Have this separate section overflow into the tank where the pumps are pulling suction.

 

[georgeverghese]

Would agree in general with this independent evaluator's concern with entrained air in this 12inch gravity line from the tower bottom to the pump cold well.

Have a read of Perry Chem Engg Handbook 7th edition pages 6-28 and 6-29. To me it looks like you then have 2 options

Option a) If you must retain the drain line exit at the tower to be flush with liquid level(as you say to maintain minimal liquid level at the tower bottom), then you've got to enable self venting at the gravity drain line, for which the Froude number must usually be kept at less than 0.3. Which results in a max permissible gravity line velocity of 0.5m/sec for this 12inch line. At 1500gpm, this line is currently running at 1.4m/sec, which explains why this line isn't self venting at the moment. So add another 2 12inch drain lines with from the tower (each with it own exit nozzle at the tower)down to the cold well to get 0.47m/sec.

Option b) Increase the liquid level at the tower bottom as suggested by the Kalinske equation 6-137 in Perry. For this, I get a submergence h of 0.22m for the current 12inch exit nozzle. You can maintain this level by installing a suitably sized level control valve near the discharge of this line at the pump cold well and a level controller at the cooling tower bottom. Obviously, this 0.22m means submergence of the top of the exit pipe. Add a vortex breaker at the entrance to this gravity line at the cooling tower bottom.

 

[katmar]

I mostly agree with George's analysis. In any gravity driven drain line where there is a possibility of air entrainment the Froude Number must be considered. There is just one additional point that needs considering in George's Option a).

Keeping the Fr No below 0.3 will allow the drain to be self venting - provided that the discharge is flooded. If the discharge is not flooded the air will still be entrained with the falling water and making the drain pipe bigger (or having more of them) may actually make the problem worse because the water is going to fall at its terminal velocity anyway and there will be less frictional resistance to the air flowing with it.

The drain line (either 1x 20" or 3x 12") from the cooling tower basin must continue vertically downward to about 6ft below the surface in the pump sump, and then gooseneck back up to the inlet to the sump. It is in this flooded gooseneck section that the Fr No works its magic and allows the air to disengage from the water.

Another consideration is that even if the water draining from the tower basin has no air in it at all, if it enters the sump tank above the water surface it will entrain air as it enters the body of liquid. If you can implement a self venting drain it would be much better to have the inlet to the sump below the surface where no air will be entrained.

Katmar Software - AioFlo Pipe Hydraulics

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[LittleInch]

Well basically this sounds like a near vertical return pipe so your pipe partly full is probably doing a lot more than 1.5m/sec. there are some threads on here that talk about vertical drops and the most you can get down a pipe and still remain steady state is about 7/24 full or something like that.

So the issue is how to slow down the flow at the base of pipe.

you could do it a few ways.

Empty the pipe into a smaller holding tank a metre or so above the main tank and then have a large pipe drain it into the bigger one.

Add a large pipe, say 24" in a U bend with air vents on both the inlet side and the top of the second U bend to prevent syphoning / surging

Add an orifice plate or similar to create a full pipe of x meters above the plate / valve at your rated flow and hence you only send out flow at a much mower velocity because you have a full pipe.

Dead end the inlet pipe ( maybe increase diameter to say 24" ) and drill lots of small holes in the end to diffuse the water. You will probably be able to submerge this diffuser as the water will emerge sideways and hence not have the velocity to entrain air all the way down to your suction pipe

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[Rputvin]

Appreciate the feedback so far.

bimr - We typically don't know the final configuration of the field pipe, but the attempt was made on this install to dictate location and setup. Containing the return may have unintended consequences in terms of getting the tank fully mixed. The nice part about a violent return pipe is that our tank stays relatively clean and it allows our separator downstream on the process side to pull off debris coming out of the tower.

George - the tower must drain completely empty. It's outdoors with water, operating 24/7/365. If we cycle the tower off for temperature control or maintenance, etc, any water in the tower would freeze. Not that we haven't had towers freeze, but they tend to not work as well in that condition. Adding more drain lines isn't an option for this particular situation.

we could control the level in the tower if needed, but as stated the tower is designed to not retain water. The less water in the basin = the less splashing we get through the air intake louvers, the less mineral deposits in the harder-to-access tower basin, the less volume change in the sump tank, the more suction head we have on the pump suctions from the height of the water in the tank, the less headaches we get from the customer.

katmar - the discharge is not completely flooded, the tower starts from an empty condition.

I'm leaning toward coming up with a good vent design we can implement that would help facilitate draining, then terminate below the water level in the sump tank with an arrestor of some kind.

LittleInch - in this case, yes, we are near-vertical. We implement the same basic system for most of our cooling tower installations (at least a dozen annually) and most operate without much concern. Some include horizontal sections that cut down on drain velocity and we end up backed up in the (relatively small) tower basin and overflowing it - when this happens it's again typically the customer or installation contractor not following recommendations for the installation, or we were not given the detail of how far the tower and sump will be from each other and the customer can't build in enough drop across 200ft of horizontal run.

Taking velocity off at the end of the drain is also where I'd prefer to focus. We're trying to get as much velocity as possible at the outlet of the tower drain to help ensure it fully drains. Most installations are not as ideal with the tower located roughly directly above the sump. The downside of this particular installation site is that the mezzanine the pump station/sump tank is located on has all of 6" to spare in any direction, so no dice on an intermediary tank, though I like where you're going.

The traps and diffuser options have been discussed in the office. We haven't gotten much traction on it as too much flow/velocity is rarely the issue and when we do have an issue it's more often a lack of flow/velocity. Do you have any details on the double U-bend setup?

I do agree that some method to deal with the excess velocity at the end of the drain pipe is a good course of action.

Thank you

 

[katmar]

This is not a new or unique design problem. There are tried and tested methods that you can use and you do not need to re-invent the wheel.

Putting a valve or orifice anywhere in the pipe cannot reduce the velocity to less than 4.3 ft/s for 1500 gpm in a 12" pipe and this is enough to entrain air. The only way to prevent air entrainment is to use a pipe sized for self venting flow (Fr < 0.3) in conjunction with a flooded section.

Using an intermediate tank is an unnecessary complication. A simple self venting gooseneck solves this problem in a gazillion installations every day.

If you cannot install more or larger outlets from the tower basin, use as short a section of horizontal 12" as possible, and tee it into a vertical 20" section. The 20" pipe should extend about 2 ft above the tee, and then have about 6-10 ft of 4" vent to atmosphere above that. The 20" section must extend down to 6 ft below the surface of the lower (destination) tank to provide the necessary flooded section where the disentrainment can take place.

If you have problems with flooding tower basins, it is better to go down vertically with the drain pipe as soon as possible after the outlet nozzle. After this vertical section you can go across horizontally - with some small slope if necessary. To get 1500 gpm through 200 ft of horizontal 12" pipe requires a bit less than 2 ft of head. If this head cannot be provided by the ullage in the basin, the basin will overflow, but if the first part of the drain line is vertical it is easy to develop 2 ft of head in this vertical section. It is important to get the air out by sizing for self venting flow in the vertical section because any air being carried along with the water in the horizontal section will dramatically increase the required head. For 1500 gpm you would use 20" pipe for the vertical section and then you could safely use 12" for the 200 ft horizontal section.

If the pipe is uniformly sloped over the whole 200 ft it will entrain air and not run full of water. This means you do not get any pressure recovery from the drop in height over the length of the pipe and all the head must be provided in the basin. This will likely lead to cyclical surging as the basin level builds up sufficiently to flush the air out of the pipe.

Katmar Software - AioFlo Pipe Hydraulics

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[Rputvin]

katmar - Thank you for the info. I'll probably visit this early next week and digest your post then, I need to get moving on another project for the time being.

Have a great weekend

 

[georgeverghese]

From what you've said so far and the constraints you have to work within with this existing drain line, I dont see why you cannot maintain a level at the tower bottom - at 1500gpm, would imagine there simply isnt sufficient residence time to freeze the water if you maintain say 250mm submergence above the 12inch drain nozzle. Maybe if there is roaring blizzard outside at -20degC. (Is this facility somewhere in Canada?). Agree submerging the entrance at the pump cold well by a few feet as @katmar suggests will help.

@katmar, From what is in Perry re the Kalinske study on page 6-28;

"For heads greater than the critical, the pipes will run full with no entrainment ". The Kalinske study accounts for cases where Nfr > 0.3 in the drain line in the expression for critical head. In this case, with 1500gpm with a single 12inch line, Nfr=0.84.

The critical head in this case is most likely 225mm. Since there is then no (or minimal) entrainment in the drain line , I dont see the need for the line to be self venting ( ie to operate at Nfr < 0.3).

And conversely, if the line is set to operate at Nfr<0.3, given the line is then self venting, there is no need to submerge the inlet nozzle at the cooling tower.

Else, if you mean the discharge is to be fully flooded - at the pump cold well end - agree with you.

 

[katmar]

@georgeverghese - Yes, when I referred to the necessity of a flooded section I meant that it should be at the discharge end of the drain line.

If the inlet is to be maintained in a flooded state, with the submergence greater than the critical head calculated from the Kalinske study, then the pipe certainly can run with Fr > 0.3 and without any air entrainment. I generally try to avoid this type of arrangement on a low-tech installation like a cooling tower because it requires a control valve and level sensor to maintain this critical depth. For this sort of application I prefer the self regulating, no power required option of a self venting pipe. In the end it comes down to the relative costs of the control system vs the bigger pipe, and to some extent the maintenance and operations policy implemented at the plant.

Just to make it absolutely clear that we are on the same page:

1. If the inlet to the drain line is kept flooded by controlling the level in the tower basin then the drain line will be flooded all the way to the control valve, and the line can safely discharge into the sump above the water level (although this may cause some entrainment with the waterfall effect). This arrangement will result in a smaller diameter drain line being required.

2. If the drain line is sized for self venting flow and the discharge end of the pipe is kept flooded by discharging into the bottom of the sump or with a gooseneck then the tower basin outlet will not be totally submerged and the line will contain air and water down to the sump tank surface level. This will require a larger diameter drain line than Option 1, but will not require any control equipment.



Katmar Software - AioFlo Pipe Hydraulics

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[georgeverghese]

@katmar, Thanks, that clears the air. Agreed a self venting line with the pump cold well end submerged with this gooseneck would be preferable for a greenfield application.

A simpler way to implement option 2, given that flow is fixed at 1500gpm, the line is fully flooded, and level in the cold well is also constant, would be to use a butterfly valve locked into a throttled position to keep level in the tower at >225mm. And a sight glass or plain level guage at the tower bottom end. That may possibly address @rputvin's concern with level fluctuations with the previously suggested LIC - LCV control loop.

 

[nondimensional]

Is the return line vertical from the roof mounted towers back to the basin? If so, not sure if Froude number analysis does anything. This is mostly applicable for sloped pipe.

That one pumps seems to be air binding - lower density air at the impeller eye is not pushed out. I recommend you consider a APCO air release valve (model 50, or model 200) on the pump suction. If pressure is low, then ask for a 'soft seat'. To able to capture air, add this on the largest available diameter pipe (slowest water velocity) in the pump suction pipe.

 

[katmar]

@nondimensional, The Froude number is important, no matter what the orientation of the line is. Different Fr No targets apply for different orientations, but if the line can contain air and water then Fr No should be considered.

Katmar Software - AioFlo Pipe Hydraulics

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[LittleInch]

This is one of my archived threads about vertical pipe flow and itself has a number of other references.

https://www.eng-tips.com/viewthread.cfm?qid=409750

Basically you are probably very close to the problem zone of slug flow and your pipe if vertical is operating somewhere close to 4-5m/sec at close to 7/24 full.

I think if the Froude no goes>1 then you hit a mixed, glugging flow.

I'm pretty sure Katmars gooseneck and my double U bend are one and the same thing. Basically a section of pipe which is always full and allows velocity to be reduced. Making it say twice the diameter will help a lot also.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[Artisi]

I would look closely at the free-fall of return water from the exit of the return line driving air into the tank, can the return line exit well below tank level to eleviate any air-entrainment.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[LittleInch]

I think though that some of the air is coming down the pipe with the fluid.

If the pipe was completely full before it reached the pond then sure, submerging it would help, but if the air is entrained in the pipe before it gets to the pond then it could drive the air lower.

The original description makes it sound like there is high velocity liquid / air mixture hitting the surface to create "an agitated mess".

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[Artisi]

LI, having re-read the OP, looks like your comment is correct, probably highly aerated at the source - the free fall certainly isn't helping.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[Rputvin]

Sorry I haven't gotten back to this yet. It's been a bit of a crazy week.

I do appreciate the discussion. The current setup of just a open drain pipe is the simplest, cheapest solution. It works well enough, but I'd rather put more thought into it and try and avoid some of the issues we've encountered when the install specifics or site geography puts us into a less-than-optimal situation.

I'll be taking this info and having a discussion with our team and figuring out what approach we want to take, and start incorporating it into our systems.

Thanks again!
-Ross