Sundyne Model LMV-341

[JJPellin]

We have had three failures of a Sundyne pump model LMV-341 running in stripped sour water service. These three failure had a common pattern that we have never seen before. First, a little background may help. We have about 175 Sundyne pumps in our refinery. About 75 of those are gearbox driven pumps running in all sort of services. The particular pump that is giving us trouble has only been in service since September. It has had three complete wrecks during that time. All three failures involved the seal sleeve spinning on the high speed shaft, allowing water to flow under the sleeve and directly into the gearbox. Each time, we thought we found and corrected the problem, but obviously we were wrong.

Failure 1 – The pump had been running for about 3 months since the start-up of the unit. The pump tripped off on low oil pressure. When the operators got out to the pump, they found water shooting out of the gearbox vent. When we opened it up, we found that the inducer had unscrewed, the impeller had dropped and rubbed in the case. We checked all of the likely causes and ruled them out. The motor rotation was correct. The pump was running at the time of the failure. All of the parts were the correct parts. The flow rate from the pump had been stable and just slightly above the BEP flow for the pump. There was no indication of any problem with the source of water they use. We basically felt we eliminated all possible causes except for manufacturing or assembly errors. We filed a warranty claim against the manufacturer. They accepted the claim and paid for all of the repairs.

Failure 2 – The pump had been running for about 3 months again when the operators noted that water was shooting out of the gearbox vent. They shut the pump down. When we opened it up we found that the seal sleeve had spun on the shaft, burning up the o-rings on both ends and allowing water to pass under the sleeve and directly into the gearbox. This time we found a definite machining error. The threaded hole in the inducer was not threaded all the way to the bottom. The manufacturer’s specification states that the hole should be threaded at least 0.800” deep in a 1.0” deep hole. We found approximately ½” of thread. We screwed the parts back together. We found that the inducer stud was reaching the end of the threads at about the point where the face of the inducer contacted the impeller. In other words, it felt like we were tightening it up, but instead we were just bottoming out the stud in the incorrectly threaded hole. It made perfect sense. Even though we had torqued the inducer as per procedure, it was not providing the full clamping force against the impeller, seal face, sleeve, seal face, shaft step stack up. We filed another warranty claim against the manufacturer and they are still evaluating that one.

Failure 3 – The pump shut down on low oil pressure. When they drained the oil out of the gearbox, they found large quantities of water in the bottom. Once again, the sleeve had spun on the shaft allowing water to enter directly into the gearbox. Everything was assembled correctly. The threaded holes in the output shaft and the inducer were the correct depth. The inducer stud had at least 1/4” of extra threads before it would bottom out in the holes. We know from the previous repair that we had torqued the inducer to the correct value and all parts in this high speed shaft assembly were new.

There are two identical pumps in this service. The other pump has run when ever this pump has been out for repair. It has had no problems and has never failed. We obviously have some other problem that is causing these high speed assemblies to loosen up. In all of our other 175 Sundynes, running for the past 20 years, we have never seen an inducer come unscrewed or loosen up before. Does anyone have experience with this type of failure or have any theories that might explain whey this is happening. Thank you in advance for any information you may have to offer.


Johnny Pellin

 

[unclesyd]

I wasn't involved but we had a similar pump installed as a recirculating compressor that failed in the same manner. There were two pumps running 24/7/365 so it was decided to add an installed spare. When called on the spare would run for some period of time and start leaking H2. This happened several times until it was discovered that something was threaded wrong, something like right hand instead of left hand. This problem was corrected and a far as I know the pump is still in operation.

 

[rota1]

Hello Johnny Pellin
Some suggestions
1)check the depth of the recess in the impeller that meets the shaft shoulder,if the recess is not deep enough, torquing the locknut would not give adequate compression on the O rings/sleeve/rotary seal to achieve the required friction to prevent spinning of the sleeve on the shaft.A not deep enough recess, would inrease the impeller back clearance. I think Sundyne should provide you with the "as built specifications " ie; front and back impeller clearances, then you can check and compare.
Is the O ring compatible with the product pumped?

 

[JJPellin]

Thank you for the suggestions. I have verified that all threads are correct (left hand to tighten with rotation). I checked the stackup of all spinning parts. There is no axial interference between the impeller and shaft, inducer and impeller, sleeve and shaft or seal faces and shaft that would keep the clamping force from acting on the sleeve. The particular stackup of this pump does not rely on the o-rings to driven any of the rotating parts. We have wear that proves to us that the sleeve actually spun on the shaft. But, to answer your question, the o-rings are Viton and are holding up well with no indication of material compatibility problems in this stripped sour water stream.

We are continuing to investigate. We are now considering the possibilty of differential thermal expansion between the shaft and sleeve, torsional resonance, cavitation induced vibration, and a few other options. I need to give Sundyne another chance to assist since we know have more information that we did not have a the last failure.


Johnny Pellin

 

[bingopin]

JJ,
I read enough of your posts to know you are pretty thorough. Having said that, I offer these random questions/thoughts:

Wonder if more than one problem exists.
Are the seal leakoff ports open? It's #2 and #7 I think. It's difficult to see how water leaking into the gearbox under the seal sleeve could be shooting out the gearbox vent with these ports open. Of course, if the sleeve is loosened, this affects the gearbox seal function.
Doesn't the back of the impeller rub the difusser cover until you install the pump in the casing? Anything going on there to keep you from getting things clamped while on the bench?
These pumps equiped with aux oil pumps/coolers? Another potential source of water.
What did the inducer locking tab washer look like on these failures?

My experiences with Sundynes are usually related to failures caused by off design operation. The operating window of these pumps is so small that it's easy to get into trouble.

 

[JJPellin]

I am more and more convinced that there is another problem that is resulting in this failure. Our particular pump has a single seal that uses a full length sleeve. This sleeve runs up against the back of the product seal rotating face and up against the bottom of the gearbox rotating seal face. Any product that leaks under this sleeve completely bypasses the leak-off port (Port 1 in this installation) and comes out directly into the gearbox. But for this to happen, both of those seal faces have to spin with the sleeve. Our Port 1 connection has a plain-end nipple so that diligent operators are not tempted to put a bull plug in there. We tested the water that was found in the gearbox and based on hardness it cannot be cooling water from the oil cooler. It is definitely pumped product.

Because of the extremely high pressure design of this pump, there is no removable diffuser or differ cover. So, for this pump, when the impeller is mounted, there is no contact between the backside of the impeller and the seal housing.

There is a strange issue with the locking tab washer. The first failure occurred as assembled from the factory. When we took it apart, the tabs on the lock washer were still locked, but the inducer was loose. We felt it must have been an assembly error. When we rebuild these pumps, we don't lock the washers. Since there are only two grooves in the inducer and eight tabs on the washer, it is very difficult to get the exact torque on the inducer and still get one of the tabs to line up. I have always felt that the torque is more important. And, as I noted, I have never seen this assembly come loose on any other pump.

The operating point of the pump is another issue. We were initially running this pump at 127 gpm with a BEP of 124 gpm. Running 2.5% above BEP would not normally be a problem. But, within three months we were seeing a degradation in pump performance. We opened the pump up and found cavitation damage inside the diffuser throat which typically indicates you are running at too high a flow rate. We reduced the flow rate to 124 gpm (exactly BEP) and this problem ceased. We have two independent flow meters that show we are holding 124 gpm (+/- 0.5 gpm). Both flow meters have been calibrated and check out perfect. I witnessed the test run of these pumps at the factory and am certain that the performance curve, efficiency curve and NPSH curves are correct and accurate.

The “other” problem that I now believe must be related to these failures is cavitation damage in the eye of the impeller. At each of the three failures, we found cavitation damage between vanes at the impeller eye. We found no cavitation damage on the inducer at all. The NPSH required (3% head loss) at BEP is 12 feet. We have just over 100 feet of NPSH available. With an NPSH margin of 85 feet while running at BEP, we obviously should not be seeing cavitation damage. Right now I have inspectors x-raying the suction piping to check for hidden obstructions. Something could have been left in the system at original construction that has restricted the flow to this pump (lunch box, hard hat, tape measurer, partially decomposed human skull, etc.) I would not normally expect cavitation to cause the inducer to come loose and the sleeve to spin. I have seen a lot of cavitation problems with Sundyne pumps without any looseness in this stack-up. But I can picture how it would be possible to rattle the inducer loose with cavitation. I also plan to add a suction pressure gauge connection so I can monitor the suction pressure at startup. Hopefully, this will find something to explain the cavitation problem.

Thank you for your questions and comments.


Johnny Pellin