PD pump discharge PSVs

[Scipio]

Here's a little repeating nuisance problem that I keep seeing, thought I'd throw it out in case someone here's come up with a practical solution;

On a lot of gas wellsite facilities involving compression, produced water and hydrocarbon condensates (C5+) are knocked out of the gas stream prior to compression. The two liquids are then either dumped to tank(s) on site and are trucked out, or one or both are injected back into the gas downstream of the compressor for separation downstream. This is usually accomplished with a blowcase or a pump, in the latter case usually a simple motor driven packed plunger pump if power is available at site, or a pneumatic powered pump if not.

Because of the potential pressures developed by the motor driven recip, (pneumatics are usually self limiting as long as the pneumatic air/gas supply has overpressure protection) PSV's are required on the pump discharge to protect against overpressure due to blocked flow. Where I've been having fun lately is discovering a number of these installed in the field with the a conventional PSV that exhausts back to pump suction. Picking some arbitrary numbers for the sake of 'for instance', let's say 300# ANSI suction piping, 600# ANSI discharge piping, operating at ambient temperatures with design pressures of 740 and 1480 psig, respectively. I arrive on site and discover a conventional PSV set at 1440 psi exhausting back into the suction. Since a conventional PSV operates on differential, the discharge pressure could in fact be as high as 2180 psig before the PSV opens.

I've employed a variety of means to correct this, depending on the situation. In one case, due to a delining field the 600# discharge piping only operated at about 400 psig, and the 300# suction piping at 100 psig. I was able to drop the PSV set point on suction piping that I was able to set a conventional PSV at a practical pressure and still cover all operating cases. In another case, I was able to re-route the exhaust of the PSV to an atmospheric tank, in a third case, to a flare knockout equipped for fluid retention.

I'm thinking my only alternative in situations like this, which will allow me to discharge a pump PSV based on actual discharge pressure back into pump suction, regardless of suction pressure, is installation of pilot-operated PSVs. Any thoughts on the matter from the pump gurus? Bear in mind this is natural gas field production, so suction & discharge pressures are frequently all over the place.

 

[CHD01]

Yes, its preferable to go to a tank on the suction side; since you can get into difficulties going to the suction line due to pressure build-up due to flow. There is no easy answer. I'm not that familiar with your specific application, but going with the pneumatic source for pressure is as you say easier to deal with. I can offer you one more item to worry about. That is, you also need to size the relief valve for instantaneous flow which is a function of pi (3.1416) for simplex, duplex, etc type pumps. So, look out, you may find more pumps than you would like off by a factor pi on the relief calculated rate of release.

After being cruel, I owe you possible solution. It won't eliminate the problem of the relief valve, but it might help you; have you considered a shear pin for the motor shaft to disconnect it from the pump or Hi-Amp cut-off? Or do you consider this to refined for the rough & tumble environment you've got?


The more you learn, the less you are certain of.

 

[CHD01]

I forgot to address your recommendation for using a pilot operated valve.

It could be a good idea; normally you use a pilot-operated valve when you have excessive inlet pressure drop to the relief valve; this typically applies when the relief valve must be installed a good distance from the vessel being protected.

But in your case, what you want to prevent is the constant superimposed backpressure from the suction line on the conventional valve which adds to the r/v set pressure.

So I suppose you could try installing the pilot valve to sense the discharge pressure in the discharge line; and use actuation of the pilot-operated valve to open the main valve which is still relieving to the suction line. That way, at least the valve opens at the right relieving pressure; now your problem is restricted to whether or not the P.D. pump can overcome the suction normal operating pressure and buildup the suction pressure above normal - which could get you back to square one. That is of course a function of line size, line length, and system volume for the suction side.

Looks like you will continue to have fun with these installation; of course the most fun is tossing this out to those who don't have a clue what the problem is.

I'll have to think about it, since its easy to jump to a conclusion and forget something simple.





The more you learn, the less you are certain of.

 

[zdas04]

Scipio,
I've addressed similar problems by pulling PD pumps from most applications - why does our business love the things so? I've replaced them with vertical blow cases (which dump fast enough to keep up with the demand) and centrifugal pumps. Centrifugals will stall before you have much of a pressure problem so they are more-or-less self-limiting in the service you've described.

If money is too tight (in a $6 US/MCF gas market, go figure) to pull the PD pumps then I would pipe the PSV back to a tank or steel pit. Any other option is fraught with unintended consequences. Make sure you account for the piping in your sizing calcs, but it is a much more reasonable approach than a pilot-operated valve exhausting into a closed system.

David Simpson, PE
MuleShoe Engineering

http://www.muleshoe-eng.com

 

[aggieengineerrmb]

You might also want to consider operating your PD pumps with a VFD equippped electric motor and using the output signal from a pressure transducer to adjust the speed, therefore capacity, of the PD pump to compensate for the varying conditions and maintain a constant discharge pressure regardless of conditions. Most VFD's today contain built control logic circuits so that this is a very inexpensive and simply approach to implement.