Sundyne LMV311 process leak out from port 1 1

[Rreach]

We have a Sundyne LMV 311 with single seal plan 21, operating in hydrocarbon service. We recently overhauled the pump.
Both port 1 and port 2 drains are extended out of the noise hood and routed to a nearby drain. Port 1 drain length from the pump flange is about 3m while port 2 is about 6m. Both lines have multiple bends and are opened to atmosphere at the end of the pipe.

This time when we ran the pump, we observed process leaking out at 12 dpm from port 1 and nothing from port 2 after 12 hours. When we disconnect the port 1 and port 2 flanges on the pump, we do not see any physical leak but did observed vapor coming out from port 1 but again nothing from port 2. This indicate that whatever that is leaking out is the process vapor and it is condensing along the drain line of port 1 and coming out as liquid leak at the end. We tested this liquid and it is indeed process, not the lube oil.

Now my question is; according to the Sundyne configuration/drawing (refer attached) any process leak will leak out to the cavity at port 2 first and must pass through the throttle bush before it can leak out from port 1.
Why am I not seeing anything out from Port 2? Anyone has seen this before?

 

[JJPellin]

As shown in the drawing you attached, there is normally a deflector above the throttle bushing. This is a spinning disk that may be causing some amount of a pumping action. This deflector may be encouraging air to be drawn into port two and expelled from port one. Since you are condensing a very small amount of vaporized product, this tiny flow might be enough to send all of the leakage from the port one connection.

Why do you have both ports open? This defeats one benefit of the throttle bushing. In the event of a catastrophic seal failure, the throttle bushing could reduce the size of the release. With both ports open, the total amount of the leak might be greater. If you are more concerned about the risk of pushing pump product up into the gearbox, then leaving both ports open is probably a good idea.


Johnny Pellin

 

[JJPellin]

One other possibility occurred to me. If the pump process is cold, there could be some thermo-siphon effect. The heat from the gearbox heats the vapor between the seals. Hot vapor tend to rise and pass out of port one, while cooler ambient air tends to be drawn in through port two. The combination of the pumping action of the spinning deflector and some amount of thermo-siphon could explain this. Keep in mind, this throttle bushing is a close clearance carbon bushing. It is far from a vapor tight seal.

Johnny Pellin

 

[JJPellin]

Any comments, Rreach?

Johnny Pellin

 

[Rreach]

Hi Johnny,

Sorry for taking so long to reply. We were checking a few things over the past few week to determine whether this vapor leakage is excessive or not.
We measured it at 2000 ppm, and so far it has been stable, not worsening. The plan 21 flush and temperatures are normal and sufficient.

This is actually a hot pump running at 200 degC. Your first point is valid. My thought will be that vapor tend to rise and somehow the deflector is drawing the vapor up to the upper cavity.
The reason both ports are opened is indeed to separate the two leaks and reduce the risk of pushing the product leak into the gearbox.

Can I ask another question on your thought on this vapor leak whether is it excessive? I do not see this vapor leak on the other standby pump that we have. The operating point and the temperatures profile are the same for both the pumps.

 

[JJPellin]

I would consider a continuous leak of 2000 ppm to be excessive. At that temperature, a single seal should be able to seal to less than 500 ppm. Because of the design of these pumps, the leak could be coming from the mechanical seal faces, or it could be leaking past a number of different gaskets. It is unusual for this model pump to be used at temperatures that high. Some of the o-rings can be replaced with high temperature gaskets. But, some of them cannot. What o-ring material are you using. For that application, I would recommend Kalrez or something equivalent.

Johnny Pellin