Entrapped air in horizontal pipe 1

[orlandobill]

Does anyone have experience with this? Here's my situation:

Vertical pump sends 15500 gpm of water into an empty 30" diameter 120' long horizontal pipe that turns down 10' vertically, turns and runs another 60' horizontally and then turns up another 20' vertically. The 60' long "trap" holds water every night when the pump is turned off, but the 120' long horizontal pipe drains completely.

Is the pump flow (~7 ft/sec) enought to push all of the air out of the horizontal pipe? Is there any information out there that might help? Should I see huge slugs of air coming out of the discharge? I have flow issues and I'm thinking that a huge air bubble in the horizontal line is restricting the flow. Thanks for any information.

 

[zdas04]

If you look at monometer theory, you'll see that uphills and downhills in a line completely filled with liquid are immaterial and the only thing that matters to hydrostatic head is the elevation difference between the inlet and the outlet.

In real life, a pump discharge in hilly country is rarely related to final outlet pressure. The deviation from theory is air bubbles prevent accessing the rebound effects of the downhill sections and your pump discharge pressure is usually dominated by the height of the highest hill. This phenonenum is the most common version of "gas lock" or "gas interference".

In a line with minimal elevation differences and your huge flow rate, the liquid being pumped will drag this tiny amount of air (600 SCF is an infetisimaly small number) out of the line within the first few seconds of pumping. Gas interference just can't be causing you flow problems.

On the other hand, my handy-dandy reference book shows the onset of erosion to occur around 5 ft/sec, Your Reynolds number works out to around 1.5 million. I wouldn't be suprised if your flow problems were related to excessive friction.

David

 

[rconner]

Not claiming to have experience in this, but (if I understand your layout correctly) it would seem at least one air release valve say at the top of the horizontal 120' section of pipe exiting the pumps but maybe right where/before the flow dips down in the 90? and 10' riser, might allow the air to exit without need to scavenge it out with flow?

 

[timbones]

The whole situation sounds a little strange, but that may be because the information we have may not be complete. Where does the water drain to? Is this a wet-pit pump? Is there a check valve on the discharge side of the pump to prevent water backflow? Is there a discharge isolation valve?

I you are indeed filling an empty pipe every time you start the pump, then you will have to purge the air out of high point vent every time. There should definately be a high point vent before the pipe turns down the 10' run.

Tim

 

[Artisi]

What is the rated head capability of the pump?

 

[orlandobill]

rconner- I agree that an air release valve at the point where the horizontal run transitions into the down vertical run would be a good idea, but I need to have extreme confidence that there is air being trapped and reducing the flow. The cost of installing one now would be very significant and management would want to be confident that the air release valve would help.

timbones- The pump is situated in a wet-sump without a check valve. When the pump is shut off for the night, the water in the horizontal pipe drains back to the sump leaving the 120' horizontal pipe empty. The entire 10' down vertical, 60' horizontal and half of the 20' up vertical remain full of water. Does that help?

Artisi- Shut off head is about 65 ft. According to the pump curve, the pump is capable of pushing 15500 gpm @ 21 ft of head.

Has anyone ever used software to model strange pipe flows? How about physical modeling?

 

[Artisi]

what is the difference in elevation from the sump water level to the discharge point, is it 10ft?
What are the flow issues, you need to expand a little so we better understand the likely problem.

 

[quark]

I get about 11.5 ft pressure drop including 10' elevation in the discharge (only discharge side). Velocity and Reynolds number match with that of David's. You may have excess flow, if your pump is designed for 21 feet head.

 

[quark]

I agree with rconner. A vent with a discharge leg higher than the 20' discharge pipe section should vent the air.

 

[Artisi]

I also came up with similar total head - hence the question of elevation before addressing what could be excessive flow and possible cavitation, inlet vortices or other inlet problems.

 

[katmar]

I believe that the trapped air is the problem. It looks to me like the cause is a combination of two conditions. The first is that your design velocity is in the transition zone between the self venting and siphoning conditions. The second is the transient start-up conditions which we often overlook.

For vertical downflow of liquids the Froude number must be less than 0.31 to ensure that self-venting is possible. On the other hand, a Froude number of greater than 1.0 will ensure that bubbles are swept down and out of the system (i.e. siphon flow). A flow of 15,500 USgpm in a 30" pipe gives a Froude number of 0.78 so the bubbles do not easily rise upwards in the 10' downward-flow section, but neither are they easily swept downwards and out. The trapped air will greatly reduce the density in this section so you do not get the expected pressure recovery. Plus the water has to accelerate through this section because the bubbles take up a considerable fraction of the cross sectional area of the pipe and this will raise the pressure drop further.

If your pump has a 65' shut-off head, and you are operating at 21' head, you are probably in quite a steep section of the pump curve. At start up, while you definitely still have the air in the system and you are attempting to flush it out, you will not have the pressure recovery in the 10' section and your static head will be the full 20'. Add to this the friction and acceleration losses and you probably have a pump discharge pressure significantly over 21' and a flow of much less than the desired 15,500 USgpm. This start-up flow is very likely not enough to flush the air out and you never get to your 15,500 USgpm. What discharge pressure do you actually measure at the pump?

Under ideal conditions (no trapped air) your friction and exit losses are likely to be about 2 or 3 ft of head. Adding this to your net static head of 10' indicates that the pump should be OK.

How to fix it? An air vent near the end of the 120' horizontal section is probably the way to go. I would also like to have a foot valve or non-return valve in the sump to avoid the venting being required at every re-start, but there may be process reasons why you have not done that.

Katmar Software
Engineering & Risk Analysis Software

http://katmarsoftware.com

 

[BigInch]

With a typical electic motor, I would not think its start-up torque related, however it may be such that air is never totally evacuated and a resident air pocket is restricting flow. Vent with a foot or discharge check should eliminate these problems.

Quick and dirty model: Is there any way you can measure the volume of water entering the system at pump start vs volume of water exiting to see if there is some resident volume of air remaining trapped in the system. Remember to reduce the 120' of 30" air volume to compensate for the increased pressure on that air volume when starting.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[orlandobill]

The sump water level is 19' below the discharge level.

Elevations:
sump water level = 81'
120' horizontal @ pump discharge = 93'
60' horizontal run = 83'
discharge = 100'

katmar- Our pumps do not currently have gauges for pump discharge, but we are working on getting them installed next week. I'll update after the installation.

 

[rzrbk]

If there is Significant air at the top of the pipe, (enough to reduce the cross-section) it may be detectable by sounding the pipe with a hammer, or looking for a condensation line on a humid day.

 

[katmar]

The 12' height difference between the sump water level and the level of the 120' horizontal section is also part of the head that the pump has to provide. The design head of 21' leaves no margin of error with the theoretical head of 19' static plus 2' friction. If there are any valves, or more than the three elbows you have mentioned, you will be undersized on the pump. On this basis you cannot even consider installing a foot valve at the inlet.

Katmar Software
Engineering & Risk Analysis Software

http://katmarsoftware.com

 

[orlandobill]

katmar- Thanks for all your input, but I'm confused. The BEP of my pump curve is around 31' and 13500 gpm. Is this what is refered to as "designed for" as others are indicating? My understanding is that the pump will operate at any point along the curve. In our case (with only ~21' of head) the pump should discharge 15500 gpm @ >80% eff. We will be fine as long as we are pushing more than 13000 gpm, but we can't even get that. My thought was since we are not getting the expected 15500 gpm that something is definately off. Please confirm whether you still think the pump is undersized.

 

[BigInch]

What Harvy is saying is,

Static heads add or subtract to get total static head

= 19 ft - 10 ft + 20 ft = 29 feet total static lift

Now you must subtract flow losses of 15500 gpm in a 30" diameter pipe that is a total of,

120' long + 10' vertically, + 60' +20 = 210 feet long

If that's more than 31' - 29 ft, you've run out of head. In any case this is a marginal installation... at best!


BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[lilliput1]

Check your inlet condition. If you see whirlpool, your submergence is not enough and you will be drawing in air causing cavitation. Also check the pump required NPSH. I estimate total 32.1 ft wg loss but I included (2) gate valves & a check valve. I did not have a strainer though. This also includes 19' differential head, borda entrance and exit out the 30" pipe. The friction loss/100' of pipe is estimated to be 0.452 ft wg/100' of pipe. This is bases on an open system & additional 15% allowance for aging. The valves came out to 161 ft equiv length. Friction & head loss to pump suction is 14.3 ft. so available NPSH is 33.96-.783-14.3 = 18.87 say 17' to be safe. Compare w/ NPSH required

 

[orlandobill]

We've gotten a little off track. I must have confused everyone. My total static head is only 19' (100'-81'). Add another 2' for dynamic losses and I'm up to 21' or so. I'm wondering if trapped air in the long horizontal pipe might be causing my reduction in flow (from 15500 according to the curve to <12000). I'm investigating other possibilities (incorrect pump rebuild, obstructions, inproper sump conditions, etc.), but I wanted to hear if other felt as though entrapped air might be contributing to the issues.

 

[BigInch]

If you don't mind some little fountains (its water heh?) Drill tiny little holes along the top of the 120'. Kidding... Or maybe not.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com