Low Flow, High Pressure Water Injection (Refinery) 1

[RefineryRR]

For those operating low flow, high pressure water injection pumps in crude oil refineries, can you please share with us:

1. What type of pump you are using (i.e. type, manufacturer, model)
2. Differential Head, Suction Head and Flow Rate
3. Average MTBF achieved

I am operating a reduced crude desulphuriser (RCD) plant with an exceptionally outdated set of water injection pumps with poor reliability. The required differential head is 200 bar (from 0.0 bar suction), at 6.5 m3/h. I would like to do a survey of what is being used in similar services around the world, and what types of pumps are being used to achieve exceptional reliability. My aim is to upgrade my system, but I want to first see who is having success in such high DP, low flow services, and what pumps are used.

For those using centrifugal pumps (single/multistage), please describe your control system.

Thanks

 

[lilliput1]

Check hydraulic (oil) pumps in the internet. Confirm they have model suitable for water service. You should increase the available suction head and check available NPSH will meet that required. Your discharge piping should be hydraulic service type suitable for at least 3000 psig.

 

[DubMac]

Check out Danfoss' axial piston pump. Originally designed for hydraulic applications, the Danish converted this design into water and/or glycol pumping for RO systems and certain O&G services. If I recall, the pump is built per API 674/5 and can get up to your flow ranges. Not sure if they even make it anymore, but it sure worked well in RO applications.

 

[georgeverghese]

Also take a look at the Lewa pump range - metallic double diaphragm pumps.

If you are hesitant to use a recip type pump and dont mind wasting some power, try a larger capacity centrifugal pump assembly on recycle. Else, if power comes at a premium in your plant, install a surge vessel that operates at 200barg to feed the reactor continously at 6.5m3/hr on flow control, while the higher flow centrifugal pump assembly runs start stop (between level control low and level control high) to feed the surge vessel. Use HP nitrogen ( or some other HP gas at 200barg) to act as a gas cushion on this surge vessel.

 

[JJPellin]

You asked this same question a few months ago. I don't have much to add to the comments I made then. As I noted at the time, all of our high pressure wash water pumps are currently centrifugal. You did not mention what you use for your source of water. Most of our use stripped sour water which is considered process water. We would have a great difficultly justifying the use of a non-API pump in this stream. Rotojet recently came out with a fully API, integrally geared pitot tube pump that would meet your process conditions. That would be my first choice.

If you decide to convert from PD to centrifugal, you should account for the drop in efficiency when analyzing the economics. You driver size is likely to go up dramatically and the ongoing energy cost will be much higher.

Johnny Pellin

 

[RefineryRR]

JJPellin,

I am looking for more detail on what systems are being used at refineries worldwide; "centrifugal" pumps is not sufficiently detailed. As I said in my intro, I am looking for pump manufacturers & models. I am also looking for details regarding the achieved MTBF, lessons learnt, ongoing/new reliability issues & required control system(s).
(My suction vessel is elevated at approximately 20m, unpressurised)

If I were to install a multistage centrifugal pump, what would the control system look like? Obviously the system would need a recycle/spillback line, but since your process reservoir (e.g. reactor, HX, separator, etc.) that the pump is pumping to, will operate at a higher pressure, how do you prevent all the flow bypassing the process and choosing the lower-resistance recycle path? What control valves would you use? Where would you have them? What would each control valve be cascaded to?

Also, how would one determine the size of the suction vessel required?

 

[JJPellin]

As I noted in my reply in February, most of our pumps are Sundyne, single stage, integrally geared, high speed centrifugals. The largest are model LMV-346 running at 124 gpm and 4460 feet of head. These pumps have a simple spill back to the supply vessel with a control valve designed for high pressure drop. I know almost nothing about control valve design and cannot comment on the particular design used. A flow transmitter measures the total flow from the discharge of the pump and opens the spill back control valve to maintain the single set-point. In our case, we run the pump as close as possible to exactly the best efficiency point (124 gpm). As long as the process demand is lower than 124 gpm, the control valve will spill back the excess. This control has no need to cascade with any other controls as far as I know. I have attached a crude sketch to show the basic arrangement of the spill-back control valves.

These pumps have a mean time between failure of less than one year. The most common failures are gearbox seal leaks, product seal leaks and oil leaks. When we repair the pumps, we tend to find cavitation damage to impellers, inducers and diffusers. This is true despite the fact that we have an NPSH margin of more than 50 feet and run right at BEP.

I do not have any knowledge about how the suction vessel was sized. We have pumps in this service that come directly off of the stripped sour water header and have no suction vessel. For the ones with a suction vessel, I assume that they were sized to provide some minimum residence time or to allow for operation for a certain period of time in the event that the SSW header is taken out of service. The suction vessel for the LMV-346’s has a capacity of about 2000 gallons.

Our main lesson has been that high speed Sundyne pumps in this service tend to have poor reliability. Our best examples (LMV-331 running at 90 gpm and 2800 feet of head) probably achieve close to plant average reliability (4 year MTBF). Our worst examples would probably be the LMV-346’s that I referred to above.


Johnny Pellin

 

[RefineryRR]

With such high differential pressures across a single impeller, is the machine not invariably exposed to discharge-to-suction internal recirculation and resulting cavitation? One would be inclined to say that if the API-specified min wear ring clearances are maintained this shouldn't be a concern, but I am skeptical that achieving a differential head of over 2,000m with a single impeller will not lead to recirculation.

Have you been able to ascertain the root cause(s) of your equipment failures (i.e. gearbox seal leaks, mechanical seal leaks, etc.)?

 

[shvet2008]

to RefineryRR

I use this kind of pump as wash water pump in hydrocracking unit (~170 barg@10 m3/h) for ~10 year. We have bad experience with this pump. It is unreliable and we have fix and repair it rather frequently. As JJPellin said Sundyne would be the best choice.

 

[RefineryRR]

shvet2008

What failures do you experience? Have you identified the root causes of your failures? What is the average MTBF for these pumps?
Do you use any other types of wash water or water injection pumps?

 

[bhaskar5150]

For our wash water applications in HDT unit, we are using single acting triplex reciprocating pumps (Make: Pernoni, Model:Titano 3130); discharge pressure : 160 bar, but you have to ensure a full flooded suction. So far we are not facing any problems. Periodic preventive maintenance is sufficient.
Paralley, I would suggest sundyne pumps, provided if you have proper flow control at the pump discharge end. (Remember : Sundynes cannot handle process variations @ +/- 10%, otherwise stay prepared to pay money)!!! You know pump pumps money.