Seal failure on hot water application 1

[CheemaJ]

Hi,

We're running a duty/standby hot water (180[sup]o[/sup]F) centrifugal pump arrangement on our plant hot water ring main. A mechanical seal failure occurred over the weekend on one of the pumps.

The seal faces are silicone carbide. The pumps have been running for less than 6 months continuously. I am a little suprised they have failed after such a short period.

I've attached a photograph of the seal that we replaced.

Does anyone have any thoughts on what the root cause of the failure could be?

Thanks

 

[JJPellin]

I am happy to help. But, I need more information. First, are both of these seal faces Silicon Carbide? Hard on hard is generally not recommended for hot water. What is the seal flush plan? This service would normally be Plan 23 in our plant. What is the configuration of the pump? Pending those answers, I can speculate.

The faces look like they could be carbon (on the left) against Silicon Carbide (on the right). That is a better combination that SiC /SiC. The flush plan may be dead-headed or Plan 11. Either way, there does not seem to be enough cooling. In, what appears to be very hard, hot water, the heat generated by the seal is causing minerals to plate out on the face. There is flashing between the faces. There may be metal particles in the water, as well.

Because the seal flush plan is inadequate, the seal runs hot. Flashing between the seal faces opens the faces, allowing hard particles (metal or minerals) to become embedded in the carbon face which grinds into the hard face until if fails.

I would convert to carbon versus silicon (if it is hard on hard). I would convert to a Plan 23 flush with a magnetic separator. Depending on the type of pump, I might recommend other changes.

Johnny Pellin

 

[CheemaJ]

Johnny, thanks for the prompt response and offer to help on this.

According to the pump supplier the mechanical seal faces supplied are Carbon graphite/stainless steel. The pump used is quite a basic industrial centrifugal and having had a look at the API seal flush plans (in this table

https://www.flowserve.com/sites/default/files/2016-07/FTA157_Piping_Plan_Poster.pdf)

I can't identify how it is flushed (if at all). Based on your insight it does sound as though there is flashing between the faces and possibly scale being deposited in the process leading to increased wear.

So I guess as long as we're operating at this elevated temperature with such a basic pump the seals are going to continue to fail, regardless of material used.

It might be possible for us to lower the temperature that we are using in the system to mitigate this flashing. Say we reduced the temperature to 160 degrees would we see a significant increase in seal longevity do you think?

Also if you have any recommendations on reading material for specifying mechanical seals I'd be grateful. Specifying seals is something I too often rely on pump suppliers to carry out.

 

[JJPellin]

Stainless steel is not a very good face material. That is the face that is getting the most damage. Normally, you would expect the softer material (carbon graphite) to wear more than the hard face. But, in reality, the particles embed in to the carbon which becomes the grinder and it cuts into the hard material. That is what the pictures look like to me. If the hard face was Silicon Carbide, it would last much longer than stainless steel.

Dropping the temperature of the entire process is usually not practical. I would concentrate on lowering the temperature at the seal. So, these are my recommendations:

Ask the seal vendor to propose a new design with carbon graphite versus silicon carbide. Tungsten Carbide would also work, but not as well.
Add a seal flush to drop the seal temperature.
The best option for seal flush would be API Plan 23 with a water cooler (if you have cooling water)
Next best option would be API Plan 23 with an air cooler.
Next best option would be API Plan 21 with a water cooler.
Next best option would be API plan 21 with an air cooler.
Next best option would be API Plan 11 which is better than nothing.

We have a seal selection guideline that we developed ourselves. I don't think I am at liberty to share that. Our document is based on API 682. That is probably where you should start.

Johnny Pellin

 

[JJPellin]

When I asked what type of pump, I was referring to the configuration: Vertical turbine, vertical in-line, horizontal overhung, horizontal between bearings, etc. If it only has one seal chamber, I am assuming it is probably horizontal overhung. Open face impeller or enclosed? Wear rings on both sides of the impeller? Balance holes through the impeller? Does the seal chamber run at suction pressure or at discharge pressure? A cross-sectional drawing would be nice. A manufacturer and model would allow me to look it up myself.

Johnny Pellin

 

[CheemaJ]

Hi Johnny,

The pump is horizontal overhung. The impeller is semi-open. It doesn't appear to have wear rings nor balance holes. I am assuming the seal chamber is running at discharge pressure - given the rapid leak that developed when the seal failed. I've attached a very basic exploded view of the pump type from the supplier.

I will request a replacement seal in carbon graphite vs. silicone carbide. You mention tungsten carbide would not work as well - what's the reason for this?

I'm not sure it's possible to retrospectively install a seal flush arrangement on this pump type, even type 11, though I will ask the supplier. The pump they've specified does not meet the duty requirements so I'm a little annoyed.

After looking at data from the inboard temperature probe, the pump is often running at 185 degrees. Our process allows to run a little cooler, so I intend to change the maximum operating temperature to 167 degrees. Hopefully this combined with the alternative seal material will prolong the life of the mechanical seal.

I will also have a look API 682. Thanks again for your help.

 

[JJPellin]

Much smaller and simpler pump than I was picturing. Tungsten Carbide will heat check more readily than Silicon Carbide. But, for this pump, you may have few option. From best to worse would be Silicon Carbide, Tungsten Carbide, ceramic, stainless steel, bronze. Go as far up that chain as you can.

Johnny Pellin

 

[Airsporter1st]

Just as an addition to what Johnny has said; check that the carbon face is antimony bound rather than resin bound. The former is more resistant to blistering and a better conductor of heat.

However, 82 degC is not so bad for hot water, but you have to remeber that the seals themselves generate face and viscous heat, which raises the temperature still further. Water is a very poor lubricant for mech seal faces. Some seal manufactures employ face modifying features such as 'lube grooves', hydropads, wavy faces etc. which can help to move the point at which the water flashes nearer to the face I.D. and/or increase the film thickness and thus lubricity. You might want to talk to your seal supplier and see what they can offer.

All that said though, the faces pictured suggest that the main problem is scaling. Nevertheless, scaling is also dependant to some extent on temperature, so if you lower the temperature there should be some improvement.

Also as Johnny says, API 682 contains a wealth of information and I'd certainly recommend a good read of that as a starting point for further edification.