Sundyne LMV-322 Barrier Fluid Leak (MTBF ~20 hours)

[RefineryRR]

Good day All

We have a Sundyne LMV-322 pump with a Plan 53B seal system, operating in HF Alkylation service (product is a combination of propane, butane & HF acid). This set of pumps are our refinery's number 1 & 2 bad actors. We experience a high-frequency of failures on these pumps will failures occurring within 24 hours of start-up. We experience barrier fluid leakage (both to atmosphere, and into the process).

Some notes:

The pump's rated operating flow is 30m3/h, and the pump runs steady at 28 m3/h
Design suction & discharge pressures are 10 bar & 25 bar respectively
There is approximately 2-3m of positive NPSH margin
The pump has an external propane flush at 11 bar
I have noticed a very erratic discharge pressure on the discharge PI (fluctuations from 20 - 28 bar)

Questions:
1. What could be contributing to these extremely frequent failures?
2. What can contribute to such an erratic discharge pressure while the observed discharge flow is steady?
3. What aspects of the Sundyne's bowl & diffuser geometry could contribute to possible turbulence, internal recirculation/churning which could be causing these regular seal failures?

 

[JJPellin]

This model can operate at very high rpm. What is the impeller speed of your pump? Is the Plan 11 truly Plan 11? Sundyne would often use a plan 31 with an internal cyclone seperator. The Plan 53B would be adjusted to about 30 psi above seal chamber pressure. Are you adjusting the barrier pressure based on measured seal chamber pressure or the value provided by Sundyne? These are the possible causes that occur to me based on your description:

The Plan 31 is passing much more flow than expected because of leakage around the tip of the orifice fitting.

The motor is running backwards. The LMV-322 is the only Sundyne model that would have the motor rotation in the clockwise direction.

The barrier fluid pressure was set based on the expected seal chamber pressure rather than the actual seal chamber pressure.

The barrier fluid used is a poor choice for the seal design and operating conditions.

The seal uses an o-ring material that is not compatible with the process.

Is this a bellows or pusher design?
What is the barrier fluid?
What is the shaft speed?
I don’t recall if LMV-322’s have integral diffusers or removable ones. Which does yours have?


Johnny Pellin

 

[RefineryRR]

Impeller Speed: 8,400 RPM
There is no plan 11 or 31 on this pump; only 53B
The barrier fluid pressure is 15 - 20 bar (measured operating stuffing box pressure is approximately 11 - 12 bar)
The motor direction is correct
The barrier fluid type has not changed. This barrier fluid is used successfully in similar services elsewhere in the plant
The o-ring material is perfluoroelastomer, and is compatible with the process

The seal is of the pusher type
Barrier fluid used is BP Transcal N
Input Shaft Speed: 2,900 RPM, Output Shaft Speed: 8,400 RPM
LMV-322's use integral diffusers. This pump is standard and therefore also has an integral diffuser.

 

[JJPellin]

You stated that the only flush plan is 53B. But, you said you have an outside propane flush. That sounds like a plan 32. That rpm is low for this model. I assume that the seal is in a double configuration with both seals running on opposite sides of the same rotating seal face. So, any error in the position or run-out on that face would cause both seals to fail. Verify the running length of both seals. Verify face run-out. Lap all of the parts in the seal stack-up. Verify that the backside of the impeller is flat. I would expect one of these causes:

The thrust bearings in the gearbox are shimmed incorrectly resulting in incorrect seal running length.
One of the sleeves is the incorrect length also resulting in incorrect running length.
One of the seal sleeves has the ends not parallel within specification of 0.0003 inch.
Sundyne is providing you with the wrong part for the rotating seal face.
The backside on the impeller is damaged and not flat causing distortion of the seal parts.
The seal is installed incorrectly. Some arrangements require a thin spacer under one of the seals. The Sundyne drawings are often unclear on this point or completely wrong. We had a pump in the shop recently that required this spacer, but the drawings provided by Sundyne did not show the spacer. The spacer was listed on the BOM.

Johnny Pellin

 

[RefineryRR]

My apologies we do have an external flush (plan 32) on this pump in addition to the plan 53B.

We will do those specification checks as you recommended.
With regards to the gearbox issues you raised, I doubt these will be the cause as the gearbox used in this service is used in multiple other locations with success. Also, the seal components and design are common (with only material changes to cope with the low pH environment). So, again, I doubt these will be contributors.

You will see in my original post that I mentioned that the discharge pressure gauge needle was very erratic, while the measured flow was stable. What could be the cause of this?
I am strongly leaning to the cause of failure being linked to some aspect of the process which the Sundyne is disagreeing with. An unstable discharge needle with a relatively stable flow suggests to me that the fluid density (and potentially vapour pressure as well) is varying. What are your thoughts?

 

[JJPellin]

Measure the suction pressure, if possible. A plugged suction strainer, at each pump or in the bottom of the vessel, could explain all of this. Has any change been made to the inducer? Sundyne tends to obsolete old inducers and replace them with new designs. About half of the time, the new replacement does not perform the same as the original design. Does this pump use a diffuser throat extension? This is only common on pumps with very large diameter impellers. Inspect the diffuser throat for cavitation damage. For very small throats, this can be difficult. Sundyne insists that the discharge control valve should be within 5 feet of the pump to avoid pressure pulsations. In the real world, this is usually impractical. We are replacing these pumps in our sulfuric acid Alky with canned motor pumps to get rid of the chronic Sundyne problems.

Johnny Pellin

 

[georgeverghese]

How is it possible to have barrier fluid loss to atm for plan 53B - this has a pressurised seal tank with seal tank press > process fluid pressure, and there usually is a low pressure trip transmitter included at the bladder top N2 space pressure; and you have a positive propane flush on the secondary seal face ? Otherwise, this could only be a plan 52 arrangement, with the seal pot floating on LP flare header pressure ??

Does seem like Sundyne havent got their act together on many of these pumps-In my experience, we deliberately avoid this manufacturer with their high speed, high head pumps - Exxonmobil had a bunch of these many years ago, but these had the Operations team up in arms at many locations - not any more.

 

[JJPellin]

georgeverghese,

The barrier fluid pressure is higher than the process pressure. This should prevent process from leaking into the barrier fluid. But, it does not prevent barrier fluid leaking to atmosphere. As described by the OP, the barrier fluid pressure has a higher differential relative to atmosphere than it does to process. Leakage across the outer seal would be leakage of barrier fluid to atmosphere, which is actually more likely given the higher differential pressure.

Johnny Pellin

 

[georgeverghese]

JP, Agreed, barrier fluid cannot leak out from the seal tank top, and neither can process fluid reverse flow in to the seal tank. If there is a positive liquid propane flush flow on the external / outer seal face, how could barrier fluid leak out this way to atm ?

 

[JJPellin]

The propane flush is on the process side of the inner seal (API Plan 32). Any leakage across the outer seal faces would be barrier fluid leaking to atmoshere.

Johnny Pellin

 

[RefineryRR]

We've measured the suction pressure and it is as per design - 10 bar.
There are no strainers at the suctions of these points. Any significant obstruction would create an NPSH problem, which we do not see evidence of on the impeller, inducer or diffuser throat.
The inducer is unchanged.
There is no diffuser throat extension.

Any thoughts on the impact of density change on the erratic discharge pressure? My thoughts are that the mixture of flush and product could be creating a product of inconsistent density, leading to the Sundyne struggling to maintain a constant discharge pressure. We do not see this phenomenon on other non-Sundyne centrifugal pumps with similar pumped products & the same flush, but with the high specific-speed Sundyne, could it be more susceptible to a phenomenon such as this?

I do not suspect cavitation as I do not see any evidence of cavitation damage on the impeller, inducer or diffuser throat. We also do not see a similar erratic behaviour on the flow.

 

[georgeverghese]

Dump the Sundynes and get some other make - not worth the agony.

 

[JJPellin]

The absence of visible cavitation damage on the impeller, inducer or diffuser throat, does not suggest that you do not have any cavitation. With cavitation in Propane and Butane, I would not expect any visible damage. I would still verify the suction pressure, if possible. I would still verify that there is no strainer in the bottom of the vessel that these pumps draw from.

The difference in the mixture of this product includes specific gravity from about 0.5 to 1.8 (propane to pure acid). This would definitely result in wide swings in discharge pressure. I was assuming, based on experience with our Alky units, that the amount of acid in the stream was very small (trace). If that is not the case in your unit, then this could be the complete explanation for the pressure swings. Drastic changes in discharge pressure could result in thrust movement in the high speed shaft assembly. As the discharge pressure changes, the seal chamber pressure will change. If the Plan 53B is adjusted to a fixed pressure, then the differential pressure across the inner seal will also be changing. This could be beating your pumps up quite badly.


Johnny Pellin

 

[RefineryRR]

JJPellin, I am encouraged by your previous post as that is my theory of what is going on. My suspicion is that we have significant quantities of products with density extremes, causing the thrust on the high-speed shaft to vary.

If this is the failure mode: what damage would you expect to see on the seal components (particularly for a double seal, face-to-face)?

 

[georgeverghese]

A review of the capacity control scheme on these pumps would tell you what control mods would be necessary to enable a constant discharge pressure.