Coker Charge Pump - Design Mods

[planck121]

We are refinery situated in Northern Canada and are expericing issue (mostly errosion related/vibration) with our coker charge pumps that pump heavy bitumen from the coker column. The issue seems to crop up from the fact that the material gets into the key way causing key way/key erosion and subsequently leading to impeller fretting and vibration on the pump. To add to this the impeller is fitted with a coke crusher to crush any heavy particulates entering through the pump suction or during start up from warm stand by.

Many companies propose various design modifications to get rid of the key way and the coke crusher (such as blinding the key way to completing getting ride of the key by by using spline interference fitting between the impller hub and the shaft) are there any tested design mods for such service that anyone might have had good experience with to get ride of the crushers and the key way all together.

Thank You
Cheers

 

[Artisi]

Not my field of expertise - but what is special if anything about this pump, is it a centrifugal pump, materials or design etc.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[planck121]

Aritsi,
It is a typical between bearing centrifual pump, two stage, double suction, running in high temperature application, delivering bitumen. The speciality is not so much in the pump itself, but the process is quite critical since it is pumping from the main upgrading bitumen feed column, heated bitumen....We have spare pumps to maintain availibity. However because of the high rate of errosion due to aggrigates (heavy bitumen) the pump sees all sorts of issues, especially errosion in the keyway and between the impeller hubs and shaft. The reason being at high temps due to hoop stress on a rotating disk (ie the impeller) we tend to lose some interferance and go into clearance causing particle migration between the hub and the shaft.....If we build too tight of a clerance that will cause issues in disassembly of the impellers when the pump is pulled out for repair....hence we are trying to address possible design modifications to increase MTBF for this application....

Hope that clarifies some of the context.

Thanks

 

[Artisi]

Planck121,
Just trying to get a better idea of the application, is 2 stage necessary for the head generation due to high head or because of low speed operation (in an endeavour to reduce wear rates)therefore requiring 2 stage to achieve the operating head?

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[planck121]

Artisi,
Its mainly because of high head generation, plus also the first stage is double suction (NPSH issues) also as part of high temperature application but the second stage is single suction.

Cheers

 

[Pumpsonly]

For a double suction impeller, the pressure on both side the impeller hub would be balanced and should not have product flow through the key way to cause erosion as you suspect.
The second stage impeller mounting is also located by a shaft step and would not possible for product flow through the key way.

See attached link to a typical pump construction.

http://file.seekpart.com/keywordpdf/2011/5/21/20115211183911.pdf

I would to suspect the key material or its size may be not good enough for the torque.

 

[Compositepro]

Erosion of the keyway is due to clearance and movement in the keyway. Clearance cannot be allowed or you will have wear. For this type of application the shaft is usually tapered so installation of the impeller is easy but clearance can be adjusted to zero.

 

[GPRnD]

You should not loose interference between the impeller hub and shaft due to "hoop stress". The stresses in the impellers of this pump type are typically very low. If you are loosing interference due to high temperatures, then that indicates an incorrect material choice for the impeller and shaft.

I think it would help if you can post a picture of the pump cross section.

Compositepro, 2 stage BB2 pumps in this application do not have tapered shafts at the impeller location (at least I've never seen on or designer one that way).

 

[planck121]

Pumpsonly,

I concur with you that we cannot have issues on the first stage key way because of simliar pressure profiles on both sides of the impeller. The issue in our case is happening on the key way on the second stage impeller. Perhaps possibly as a result of high temperature and the coke crusher itself causing instability with the shaft during operation.

Bradshi,
I have posted a cross section drawing....


Many pump companies are proposing a so called hexagonal shrink fit between the impeller hub and the shaft. So the shaft has a hexagonal profile around the area where the impeller will fit. This will help get rid of the keyaway and actually help transit shaft torque without the requirement of a key. I am wondering if there are any other such modification (along the same lines) or even if this hexagonal profile fit been tested in similiar applications elswhere leading to any succeess...

 

[planck121]

Picture attached, sorry missed it on the previous reply

 

[Artisi]

Not knowing the full hydraulic spec it is difficult to make any real contribution other than to say that a the use of a double suction pump of this configuration running with fine tolerances wouldn't be my first choice.

Subject to meeting the hydraulic spec, I would be looking at a hard metal slurry type pump with screw-on impeller running with wider clearances which would be a lot more robust and less likely to suffer the problems you are having.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[GPRnD]

planck121, thanks for post the X section.

Looking at it, it appears to be a fairly conventional 2 stage BB2 design.

Sorry to ask more questions but here goes:

1. What are the shaft, sleeve and impeller materials ?
2. The sleeves and impeller hub look quite thin radially, is the 2nd stage keyway open or exposed to the fluid at any point and if so where ?


At this stage, my concern is that the design relies on a stackup of sleeves and impeller hubs all the way through from the 1st stage to the 2nd stage. This is not good paractice in high temperature pumps due to the potential for loosening due to differential thermal expansion. Much better is to have the second stage impeller located on its own shaft shoulder.

I've seen polygon shafts used for severe erosion situations at room temperature, but I'd be concerned with using them in conjunction with high temperature. It seems to me an easier way would be to improve the sealing at the 2nd stage impeller, through some geometry, material and tolerance changes.

 

[planck121]

@bradshi
Impeller : A487+CASMW+TMT+ Stellite
Shaft: A275- 410
Sleeve: A275- 420

The second stage key way was intially exposed underneath the sleeve area. This has been addressed by bliding the keyway to flush at the step so there is no particles flowing through this gap. The gap is no more after the bliding process.

 

[JJPellin]

We have a total of 6 coker charge pumps similar in many ways to yours. The primary difference is that ours have single suction first stage impellers. Ours were originally constructed in a similar manner to yours with the impellers and sleeves stacked up with nuts on both ends. We made a number of modifications to our pumps. But in the area where you express interest, we changed to individually located impellers. Each impeller has a split ring to locate it. The split ring is trapped under a step machined into the impeller such that it cannot come out. The impellers were changed to an interference fit.

I am not familiar with the materials you listed. Our impellers are CA6NM and our shafts are 410 SS. So, there is very little difference in thermal expansion rates. If your impeller material expands much more than your shafts, then this seems like the main problem to me. Thermal expansion results in clearance. Clearance results in movement. Movement results in fretting and wear which results in more movement.

We have converted a number of pumps to polygon fit impellers. However, none are at such high temperatures. We have converted two large multi-stage charge pumps and two liquid ring vacuum pumps. All of these conversions have worked very well. I like a polygon fit for ease of assembly/disassembly and elimination of the keyways. But, they are expensive and add to delivery time for the relevant parts.

We are preparing for another round of modifications to our pumps. We will be eliminating the coke crusher wear rings during this round. We have strainers in the bottom of the fractionator that limit particle size to less than 0.25 inch. The pump should be easily able to pass these particles, so no crusher is needed.

I would try and optimize the materials to minimize differential thermal expansion between the shaft and impeller. I would convert to a polygon fit. I would choose a triangular polygon with a steep angle to limit the unlocking torque. I would eliminate the coke crusher wear rings if you feel that your pump can pass the likely particle size from your fractionation column.


Johnny Pellin

 

[planck121]

@ JJPelin,
Appreciate your post. I am in agreement with you, we are considering removing the coke crushers and infact are making modifications to the frac strainers to limit the particle sizes (however, when the pump is on standby operation or if stopped) we will have to ensure that solidfied chuncks do not enter the pump somehow (ones that might have solidfied in the line,etc).

Would it be possible to post a generic xsection drawing of the split ring mods you have been mentioning. Maybe this is something we will have to look into rather than a stacked arragement. You mention that the polygon fits are not made on pumps at high temperature. Is it because of limitations with the polygonal fits due to temperature or simply because your organization chose not to uprade the fit on the charge pumps.

Thanks Again

 

[GPRnD]

JJPellin, I think your advice is spot on.

I believe the OP omitted some of the material numbers for the impeller, but I am guessing what he meant to say was ASTM A487 CA6NM with overlays for critical wear areas.

This makes the pump pretty much a standard C6 construction, similar to yours. (Although A487 CA6NM is normally reserved for pressure boundary parts). Thus I am not sure much further optimization of thermal expansion coefficients is possible.

My suggestion would be to go with individually located impellers. With the stackup design the OP has, there is the possibility of something getting loose as the expansion of each part will be slightly different. The 2nd stage impeller will tend to thrust to the left, opening up a gap if such a possibility exists.

I'd say go with polygon shafts for ease of disassembly if that outweighs the cost. I've seen them used when there was absolutely no way to fully protect the keyway from solids, typically on multistage BB5 applications. To my mind with a BB2 pump you can protect the keyway, so it is not necessary, but using a polygon won't hurt (aside from the price).

Similarly if the OP can get the coke particles down to 0.25", they can remove the need for a coke crusher. The pump will comfortably pass any particle 0.125" smaller than the smallest hydraulic passage, which in these pumps is measured in inches.

 

[planck121]

@Bradshi and JJPelin,
Quite an interesting and informative discussion and advice. One anomaly in our pump that I cannot seem to figure out is why the first stage (double suction impeller) is also seeing keyway errosion and particle migration across the key way. As pointed out prior as well. Given the double suction nature of the first stage there should not be much of a pressure gradient on both sides of the first stage, let alone bitumen migrating from one end to the other with higher velocities.

JJPelin,
You pointed out that the first stage in your coker charge pumps is a single suction impeller. I'd assume a double suction impeller would be more hydraulically stable (also in our case there are NPSH concerns) so the double suction....But I am not able to relate fully on how a single suction impeller would help with some of the issues in these pumps. What may I ask has the MTBF on your coker charge pumps been after these design mods thus far.

Thanks

 

[JJPellin]

I have attached a cross-section of one of our pumps. We did not use a single suction first stage impeller for any reliability reason. It is only because the height of our fractionator makes a double suction impeller unnecessary. We have not used polygon fits in high temperature service only because it has not come up as an option yet. The last round of modifications to our coker charge pumps were completed before polygons had come into common use in pumps. Also, we were not seeing any problems with keyways, keys, etc.

We converted to a split ring mounting arrangement to help us address our primary failure mode. We tended to wear the bushing between the two stages. When the wear would reach a certain point, the vibration would increases drastically. We believe that on loss of the stiffness of the center bushing, the rotor would go into resonance. We enlarged the shaft as much as possible within the limits of the existing impellers. This left little room for retaining nuts. So, we converted to split rings and interference fits.

The reliability of our pumps is now limited by bearing life. When we stiffened the shafts, we also reduced the bearing span. With the bearings closer to the hot pumpage, we have difficulty getting adequate cooling. And, seal reliability if difficult to achieve in this service.

If you are experiencing flow with particles and erosion along the key in the first stage impeller, then something very strange is going on. I would verify that the piping arrangement is proper for a double suction impeller. There should be no elbow turning parallel to the shaft, no valves with the stem parallel to the shaft and only concentric reducers for the first 10 pipe diameters before the pump suction flange. If the piping is not correct, you could have an imposed pressure differential between the two impeller eyes.

A split ring arrangement may not work as well for your application. In our pumps, the single suction impellers tend to have a significant and stable thrust in one direction. This helps to keep them in place in the event that the interference were to be lost. A double suction impeller would present other challenges for the trapped split ring configuration.

In their current arrangement, our charge pumps probably have a mean-time-between-failure of about 12 to 24 months which is well below average for our plant.

Johnny Pellin

 

[GPRnD]

Double suction 1st stages are selected purely for NPSH reasons. Typically we see specs for these pumps requiring suction specific speeds of 11,000 or less together with very low NPSHa. In such situations, the best solution is a double suction 1st stage.

I could bore you all with why 11,000 Nss is a pointless and obsolete limit, but it would detract from the general discussion.

As an aside, coke crushers are bad for pump NPSHr. My rule of thumb is they increase NPSHr by 15%, although I have improved on that with some designs.


You should not get any pressure differential across the two sides of a double suction impeller, which leads me to suspect pipework problems as JJPellin suggests, or leakage under the sleeve separating the 1st and second stages.

My final comment is JJPellin's experience with center bushing wear leading to high vibration, exposes a nastly little secret about these pumps. According to API 610, 2 stage BB2 pumps are supposed to be rotordynamically classically stiff (not reliant on bushing or wear ring stiffness).

In reality it becomes very difficult to achieve this on big pumps such as coker charge pumps. I've seen some companies not even try to meet it and hence their shafts are significantly undersize from what is appropriate for this difficult service. Asking to see the L3/d4 values at the tender stage helps prevent some of this, especially if you have a target L3/d4 value based on operating experience.