H2S water pump, low flow, high discharge.

[ywchua]

Hi all,

We currently have a PD pump on our Hydrotreater pumping high concentration H2S (above 2000 ppm)water at 6 m3/h. It pumps from a suction of 3.7 kg/cm2 to 70 kg/cm2. Unfortunately we are only using single packing on this service, which I believe is specified wrongly during design. Any recommendation/experiences in similar situation on seals to use?

Pump is not reliable at all because of its service, valves keep damaging causing severe water hammering (believe a pipe elbow failed once due to fatigue), but thats another matter.

 

[micalbrch]

If you can exclude cavitation (which can cause valve and packing failures), try a v-shaped packing or another pump type.

 

[JJPellin]

I would generally consider a reciprocating pump to be unacceptable in high H2S water service. We had one such pump when I started at our plant 22 years ago. We eliminated it. It was replaced with a multistage centrifugal pump with dual, pressurized mechanical seals (API Plan 53). The seals system is instrumented and monitored. This would be my minimum expectations for a service with this much risk.

The poor reliability of the existing pump may be a result of cavitation. Designers who work with centrifugal pumps sometimes neglect to account for acceleration head. This term depends on the length of the suction line and cannot be negated by making the line size larger.


Johnny Pellin

 

[ywchua]

micalbrch:

Maybe cavitation, will be looking deeper into with the manufacturer. Most probably at this point need to consider a different pump type.

JJPellin:
Can't do multistage, think flowrate is too low for that. Had a chat with SunDyne and I believe they can work with such low rates, or simply a minimum flow controller. I've not heard of acceleration head, will look deeper into it.

Had another idea, to use PTFE diaphragm pumps, according to brochures they can generate pressures up to 300 bar or more. Plus side, no problems with sealing anymore.

 

[micalbrch]

The PTFE diaphragm pump is indeed a good option but bloody expensive for 6 m³/h against 70 bar.

A multistage centrifugal seems to be the better choice if they are available for that flow.

 

[JJPellin]

If I did the math right, this comes out to about 26 gpm at about 1000 psi. You could get a centrifugal pump from Sundyne, RotoJet or Sulzer that would meet this requirement. The Sundyne would be a high speed integrally geared pump with very low efficiency. The RotoJet would be a pitot-tube pump. The Sulzer would be a long multi-stage barrel pump or possibly a tie-rod stacked diffuser pump. The pump I referred to above was about 90 gpm (20 m3/hr) at 1000 psi (70 kg/cm2) and it ended up with a 13 stage stacked diffuser Sulzer pump.

There is a good write up on evaluation of NPSH and pressure pulsations in plunger pumps in Cameron Hydraulic Data, Copyright 1994, Ingersoll-Dresser Pump Company.


Johnny Pellin

 

[ywchua]

Thats what I am contemplating. Our experience with high speed vertical pumps have been that its very restricted in flows. For good control of them a minimum flow controller is usually required, or just an orifice back to suction, that should control pretty well. When it deviates from its operating envelope, slight induced vibration have great impact on the high speed shaft. Technically the minimum flow controller will work for multistage pump options as well.

Currently looking at diaphragm pumps. Questioned one of the suppliers about the use of PTFE in this service and apparently they can withstand most corrosive products? It will need to be larger cylinders Im guessing to allow a slow speed. The valves as well needs to be ball so theres no wear parts. Current PD pump has poppet valves, wears the holder rapidly because of the service its in.

Will look into acceleration head and see whether the pump is cavitating and destroying the poppet valves.

 

[micalbrch]

You can easily find out whether cavitation is the reason for the poppet valve failures or not: If the failure is related mainly the suction valves, cavitation will most likely be the reason. If the discharge poppets are affected as well, it will not be cavitation.

PTFE diaphragms will for sure work but are expensive when they need to be exchanged. EPDM looks good for wet H2S and is for sure much less expensive.

Yes, slow speed is important for a piston diaphragm pump. No. of strokes should not exceed 150 rpm. That sounds a lot but is o.k. as you do not have solids.

About the valves you should trust what suppliers offer but you should inform them about your concerns with poppet valves. When you use ball valves you need metal balls due to the high pressure and they are pretty noisy. Poppet valves are also available with plastic insert and might be a good choice, too. But discuss that with the possible vendors.

 

[ywchua]

Managed to speak to the manufacturer couple of days ago. Modification proposal to have larger pump, larger pistons to reduce pump speed. Valves will be changed to a semi slurry type valve, no details of that yet. Im not too sure of the proposal for packing, they suggested double packing with collection of leakage back to suction, honestly not very confident with that because suction is 3.5 kg/cm2. Plus we've been introducing double seals for hazardous services on site, having such a high concentration H2S with packing just doesn't make sense.
And most important of all a piston life is only approximately 8 months on a good service according to the manufacturer, PD pumps not my field any comments on that?

On diaphragm pumps, good idea with sealing but at the end of the day its still a PD pump, think we have a sensitive piping configuration to fit a PD, may not have sufficient NPSH.

Centrifugal pumps seems like difficult to control in terms of flowrate. Because Operations vary flowrate between 4 -7 m3/h and pressure required is 70 kg/cm2 at least. A flow by pass seems good but its heck of a job to throttle 70 kg/cm2 back to suction of 3.5 kg/cm2.

This service is horrible, and I've been in contact with another refinery in Netherlands, they have the same service but they require larger capacity so multi stage works for them. They came across pitot tube pumps, any comments with that? Just wanted to share some ideas.

 

[micalbrch]

8 months lifetime on the pistons is normal for a 24/7 (or similar) operation. You can't expect much more. About the leakage collection back to the suction side: Your vendor propose it? They shall guarantee that it will work. The rest of their modification recommendations make sense to me.

I can't tell you whether a diaphragm will work with your NPSH as long as you don't tell us how the NPSH is.

 

[ywchua]

Yea thatas what I thought about the leakage collection back to suction, unless they're expecting high pressure leakage from the packings, if not they'll flood the entire stuffing box. And yes it was recommended by the manufacturer.

Im trying to work out NPSHa at the moment, but the process side confuses me a little, need to determine vapour pressure and stuff, then will try the acceleration loss calculation, havent tried that before. Process side of things should be done by process guys, but everyones hectic at the moment so Im kinda a one man show. Will let you know NPSH once I do.