Cavitation damage to casing on suction side - Why?

[wht66]

I am investigating three treated water pumps (in the same installation) that have experienced significant cavitation damage to the pump casing on the suction side. The pumps typically operate with about 8m NPSH, have a cast iron casing and bronze impeller. There is no noticable cavitation damage to the impeller.

The pumps are conventional dual-suction split casing centrifugal water pumps from a reputable manufacturer. The inlet conditions are not the best, with three 90-degree bends (in the same direction) in quick succession leading to the pumps. Individually, the pumps are operating at about 500L/s at 30m head.

The pumps are operating at a specific speed (in US units) of approximately 3000, and a suction-specific speed of around 8000.

My initial thoughts are that this is due to a combination of the relatively high suction-specific speed combined with the poor inlet conditions, but I am puzzled by the fact that there is damage to the casing but not the impeller.

Any thoughts?

 

[Scipio]

First off, you say the pumps operate at approximately 8 meters NPSH, is that 8 meters available or the pump requires 8 meters? Also, what's your pump speed? A quick calculation backwards on your suction specific speed, assuming an 1800 rpm pump, says your flowrate requires about 10 meters.

You're right about the piping configuration, not the best layout. That kind of configuration could be asking for prerotation, but I thought that usually leads to radial load problems, not casing cavitation.

Where's your pump operating with respect to the best efficiency point? Might be a matter of low flow recirculation in the casing causing the cavitation.

 

[wht66]

The pump is operating at 1450rpm. 8m difference between manufacturer's NPSHR and NPSHA, ie should be plenty of head available. The pump is operating significantly to the right of BEP, apparently due to differences between the "design" condition and the actual operating condition (ie less head actually experienced than designed).

 

[BobPE]

wht66, you answered you question in your second post. Put the pump back to no more that 15% to the right of BEP and you will solve the problem. You are probably cavitating at the dual entrances due to a high velocity condition without suficient NPSH due to losses in that strange suction arrangement. After the fluid passes the dual suction entrance to the pump, velocity reduces and cavitation ends before the impeller.


Look to modify the impeller, or get the correct pump since explaining the mistake will only get harder every time the owner repairs the pump.

BobPE

 

[wht66]

Thanks BobPE. The 8m head (NPSHA-NPSHR) is just before the pumps, ie after the bends, so I'm still a bit puzzled. I was interested to see your comments on why there would be cavitation on the casing and not the impeller.

I haven't actually measured the head, but at a tapping point about 1m from the inlet to the pump I can get a reasonable fountain (at least 4-5m) if I open the valve on the tapping with the pump running.

The BEP is about 410L/s @ 41m. Typical operation 520L/s @ 31m, so about 25% to right of BEP. I agree with your comment on "bringing it back", but probably we are looking to pump replacement.

One other issue (I was trying to minimise complication on the original post!) is that there is also an abrupt contraction on the inlet taper where the contractor apparently found he had a DN600 to 450 taper discharging to a DN400 suction and just decided to bolt them together (don't ask me how, exactly!)

Agree with choosing the right pump! That's why I'm here! Some better design and workmanship wouldn't have gone astray either!

 

[Scipio]

Are the suction piping bends in the plane of the pump shaft or perpindicular to it? One thing that can happen with multiple turns like that is the water literally gets thrown to the outside of the bends. If the piping is in the same plane as the pump shaft in a double-suction pump, once the flow is split inside the pump suction you get a very unequal split of flows and static pressures to each impeller - instead of your 520L/s being split into 260L/s through each side of the impeller one could be doing 300L/s, the other 220L/s.

I have to admit though, I'd expect cavitation to show up in the impeller, though bronze is more cavitation resistant than cast iron. Maybe they're both experiencing the cavitation, but it's mild enough the bronze is able to handle it without any damage? Where exactly in the casing is the damage? Is it just inside the suction nozzle, or between the impeller and casing near the impeller eye?

 

[BobPE]

Well now WHT66, that gives us a lot more information to work with....I have to make an assumption on your fittings and assume that the DN450 is right to the DN400 pump suction flange ( I restated this because I am thinking the same way you are, just how did the contractor do that!!!) With this arrangement, your pump shouldnt be the problem, and I would venture a further assumption that operating at 25% above BEP isn't the problem either. The problem is the suction dynamics. Your flow velocity is in the range of 10 fps meeting the transition between the pump flange and that reducer has to be looked at as an orifice loss. Headloss probably isn't the problem (although I doubt you summed up the loss for an orifice type sudden decrease in diameter), but rather the flow development length becomes the problem. The vena contracta from this disturbance is probably extending right through the pump suction casing and causing cavitation on what ever surfaces are there within the boundries of the fluid in a rather uniform manner. The vena contracta is diminished prior to the impeller, so no damage shows there. No matter what, the suction velocities seem rather high. Lower the suction velocities, fix the pump suction piping minor losses, and throttle the pump to its BEP and it should work.

BobPE

 

[wht66]

Thanks for your comments, BobPE and Scipio. I am sorry if I seem to be drip-feeding information - that's not my intention.

BobPE, there is a short length (about 500mm) of DN400 leading to the suction flange of the pump after the contraction. The contractor tried to reduce the contraction problem by tapering the cement mortar lining in the DN400, obviously with limited scope for any success.

The contraction loss has been factored into recent hydraulic calculations, but I cannot say whether this was done when it was originally designed.

Scipio, the suction piping bends are in different planes, but rotate the flow the same way. There is general agreement that there is significant pre-rotation of the flow, and this would be compounded through the conservation of rotational momentum in the reduction from DN600 to DN400.

The cavitation damage is predominantly outside and adjacent to the static wear ring.

It does seem to me from these discussions, however, that there is a relatively high-velocity and highly rotational "jet" into the suction side of the pumps - hardly good conditions but difficult to quantify the likely effects.

My problem is trying to quantify the effects so that the same mistake is not made with the new pumps, which are required to meet the same duty. I am considering some slower pumps (960RPM) with 600mm suction which would reduce the inlet velocity and pre-rotation, but cost a lot more money. If I can quantify the effects, then I can have greater confidence in the selection of the replacement pumps.

 

[BobPE]

if you have the ability to design the pump suction over again, use no greater than a 2.5 feet per second velocity to size the suction pipe and use an eccentric reducer to meet the pump suction flange. Keep the bends our of the suction if you can but at this design velocity, you can tolerate some bends as long as they are not close to the pump. This is a typical suction design that serves its purpose very well in many applications.

You are not drip feeding us LOL....We love this stuff as you can tell and we like asking questions in order to feel you out as well as help you....

Take care

BobPE

 

[PUMPDESIGNER]

Bronze is more ductile than iron. Shock waves emanating from imploding cavitation bubbles therefore have less effect on bronze than iron. My own experiences indicate that this difference is considerable, more than one might intuitively expect. Have seen many cavitation situations where the bronze impeller was far less damaged than the iron.

It is usually safe to assume cavitation by the damage you describe. However I am curious if you have proved this to yourself. One very quick test is: If you can hear the cavitation at full flow, reduce flow and see if the cavitation noise decreases or stops. Many centrifugal pumps become very quiet at shut off when they were obviously cavitating under max flow.

PUMPDESIGNER

 

[wht66]

This is an issue I have not been able to explain. The noise from the pump continues during discharge valve closure, and if anything seems to become worse. This does not make any sense to me at all, especially given that the pumps are starting at 25% right of BEP. From leaning on the counterweights on the discharge check valve, it would also appear that flow is relatively unsteady.

Does this shed any light?

 

[BobPE]

the flow will tend to be unsteady with conditions like yours...Starting and/or operating at +25% BEP is not good either...I would venture a guess that the suction you have is not even good for 50% of the flow that is occuring, so if you have a butter fly valve, I would think it would be almost closed. Then you will put the pump into discharge cavitation...so you are right, it may not go away no matter what you do with that system...

BobPE

 

[PUMPDESIGNER]

BobPE is correct about valve closure putting the pump into discharge re-circulation cavitation, however that 3,000 Ns number indicates that the pump would not vibrate or make noise at shut off. A 3,000 Ns pump will usually become quiet and vibration free at shut off, even if it is recirculating.

I am puzzled by this. Look forward to a resolution, perhaps can learn something. What about chemical damage BobPE, think that could be it. (Just kidding).

PUMPDESIGNER

 

[wht66]

It's been two months, I know, but I'm back with more details and questions!

Some additional particulars:

1/ The cavitation damage is all on the suction side and predominantly on the inside and outside of the wear ring support spigot, although there is also significant damage to the flow splitter.

2/ Site tests have shown that under normal operating conditions NPSHA-NPSHR is about 2.0m, so although it is low it is probably not a direct cause of cavitation.

I have heard suggestions that the cause might be recirculation to the right of the BEP. Recirculation on the left of the BEP appears to be well understood - eddies occurring in the impeller because of the mis-match between vane angle and speed/flow at that particular point where the radial flow can't "keep up" with the blades. What I have read suggests that there are no problems to the right of BEP, but it has been suggested to me that the reverse conditions occur when the flow is well to right of BEP, with the blades unable to "keep up" with the radial flow with the result that recirculation also occurs but on the other side of the blade.

The resulting eddies on the inlet side of the pump, exacerbated by low NPSH, might be the cause of the problem.

Does this hypothesis have any basis?

 

[Artisi]

In one of your posts you asked about chemical damage, this could well a contributing factor to the problem.
You could have minor "cavitation" problem, not necesssarily a NPSHA/NPSHR consideration but flow separation or other problem already well addressed - with this being the case, it is possible that you have the insidious problem of cavitation - corosion- erosion especialy if there are any fines in the pumped liquor.