pump cavitation 3

[vodeni]

I am trying to figure out the reasons behind apparant cavitation and failed impeller on double suction split case centrifugal pumps. NPSH available is way higher than requested. Pumps are fed from a common 48" suction manifold. The only reason for the cavitation I can come up with appears to be a relativly short suction piping from 48"x48"x16" "T", through 16" isolation valve and 16"x12" reduction directly connected to a 12" suction flange. Could that be a reson for such a violent impeller failure. Any ideas?

 

[BigInch]

I wouldn't say that cavitation failure is normally violent. Usually its discovered by an inspection precipitated by an audiable noise and an otherwise unexplained performance drop and typically amounts to no more than repair or replacing the damaged impeller and correcting the condition that caused the low NPSH. If it is true that NPSH is higher than necessary, you should first try to account for other factors that could have caused the failure. It is possible that a calculated NPSHA not taking into consideration short lengths of bending and turning flows just before entering the pump suction does not give a true and representative NPSHA value and indeed could be the trouble, but first could you describe the exact nature of the failure in detail?

http://virtualpipeline.spaces.msn.com

 

[bimr]

The problem is due to poor piping layout. The tee is causing the fluid to swirl from the close location of the tee as the fluid enters the pump. The tee will cause the flow profile to distort towards the side of the pipe causing unequal loading across the pump. These effects become magnified with increasing pipe sizes.

Suction piping for double suction pumps should not be installed so that there is an elbow close to the suction nozzle except when this elbow is in a plane at right angles to the pump shaft. There is always uneven flow in an elbow and this unequal flow causes more water to enter one side of the impeller than the other side. The unequal flow may cause impeller damage due to cavitation or flow separation.

Suction piping should have 3-4 diameters of straight pipe between the pump and the elbow prior to the pump. The reducer may be included within this length. Long radius elbows should be used on the suction side.

The solution is to somehow straighten the flow entering the pump with a flow divider (straightener) of some sort inside the pipe. Not sure that this is possible.

Another option is to use a 45 Deg Y instead of the Tee.

Another option is to increase the diameter of the 48" header to reduce the velocity and flow distortions in the header

A full port isolation valve would be preferable to something like a butterfly valve as well.

Refer to the Hydraulic Institute Standards for arrangement of piping.

 

[vodeni]

I already checked everything I could, and that was the only explanation I could come up with, that made sense and had some sort of theoretical support. I guess, I just could not get a real sense (gut feeling) that such a relativly low velocity (barely 2 ft/s at maximum flow) could cause turbulence at the existing allignment (Tee branch flow and than straigth through a ball valve and reducer into the pump suction (double). Since the space is rather limited I thougth of installing some sort of vane that would "streamline" the flow but how to do that without dynamically modelling flow (If this is the reason for cavitation)

Thank you all

 

[BigInch]

Don't get in such a hurry. First, tell us what was the nature of the "catastrophic" failure. Did you examine the impellers for evidence of cavitation?

If the impellers show evidence of cavitation, you can bet the ranch you don't have enough straight length of pipe before the pump. Recommended straight runs are 5 D minimum to 10 D, especially after a 90 or T. Actually velocity head (v^2/2g) helps NPSHA, although its seldom considered, so your low velocity doesn't help that situation any either.

http://virtualpipeline.spaces.msn.com

 

[bduane]

You never mentioned what the fluid is. Is it possible there are entrained gases that are coming out of solution at the impeller?

In Florida ground waters there is a lot of dissolved H2S that tends to come out of solution when you start pumping.

Are you injecting any chemicals like chlorine or ammonia in front of these pumps for disinfection/residual control?

In either of these two cases, the problem could be corrosion, not cavitation.

Bduane

 

[bimr]

A few more questions:

Where in the piping is the 2 ft/sec flow measured?

Are all of the pumps experiencing the damage?

Is the last pump also experiencing the damage? The last pump should have an elbow and the flow through the pump should be different than the ones with tees. Although you still have a double suction pump with uneven flow in an elbow.

 

[Artisi]

Can you post any pictures of the "damage" as pictures can save a lot of to-and-fro in trying to explain what / where /why is going on.

 

[BigInch]

Artisi, I agree. While cavitation from poor flow preparation is a good guess, I'm still not 100% convinced w/o further evidence or confirmation of cavitation induced impeller pitting or its characteristic noise.

http://virtualpipeline.spaces.msn.com

 

[Artisi]

Likewise- it sounds like poor entry conditions resulting in mis-match of flow onto the impeller - but without seeing the impeller/s it's anyone guess at the moment

 

[vodeni]

Whow! Thank you all so much. Everything you say makes a lot of sense. First of all, it is effluent water in a wastewater treatment plant. Secondly the moment I was asked to help on this project, I requested to see the failed impelers (two out of three were replaced last year), the answer was no impellers, no pictures. So no help here either. I requested some flow/pressure monitoring to determine exactly what the pumps are pumping at what pressure, but client has been kind of slow.

I think I made a mistake saying that it was 48" suction when in fact it is a 36" common suction. There are also three "low pressure effluentpumps" drawing water off the same suction but they have not been a target of this project. There are four pumps in question, three of which run coninuously in parallel. Only one of four pumps has somewhat better suction piping arrangement (several feet of straight line prior to suction flange). Three pumps provide anywhere from 4,000 to 6,000 gpm at measured system pressure around 300 feet TDH. When I took instantenuous measurements of system's flow and pressure it appears that pumps are pumping 30% less flow (for the recorded system pressure) than they should, based on the existing pumps' design curve.

 

[BigInch]

Well if you can't confirm its cavitation and you have good NPSH, go with the prerotation idea and put 5 to 10 diameters of straight pipe in there. 5 might not be enough with a T.

http://virtualpipeline.spaces.msn.com

 

[Artisi]

One other possibility, as these are effluent pumps in a waste water treatment plant is erosion from entrained solids together with poor entry conditions.

But it's all crystal ball gazing at the moment without any hard evidence.

 

[stanier]

I suggest you get a copy of the ANSI/HI 9.8 standard and consider the recommendations therein. These are generally in line with what has been expressed above.

The standard describes the need for physical modelling if the design does not comply with criteria such as straight lengths on pump suctions. Also you may care to check the Pumping Station Handbook by Sanks. In particular the section on "Blunders to Avoid'. One of which describes your scenario.

Geoffrey D Stone FIMechE C.Eng;FIEAust CP Eng

http://www.waterhammer.bigblog.com.au

 

[ltho98]

Do you have any pictures to show what the mechanism of failure was?

If you have plenty of NPSH then it might be entirely possible that air is being introduced into the system prior to pumping. Do you have any mechanism of air relief?