PSV Discharge Header - Sizing Assumptions

[bhzimmer]

I'm sizing a PSV discharge header for a multi-stage natural gas compressor. Each individual PSV has been sized to relieve the entire capacity of the compressor individually.

My question is this, should I assume that only one PSV will relieve at one time, and size the discharge header accordingly? Or should I size the discharge header for a worst case scenario (3 PSV's discharging simultaneously)?

Also, are there any standards on this issue for compressor packages specifically? If not, what standards/specifications should I be using as a guide?

Thanks in advance for any advice.

 

[BigInch]

Assuming that you only have one compressor here, It is not obvious why you should size the line for a flow of any more than 1 PSV ie. = the flow of one compressor. From where can you get more flow?

It is quite common to have 100% standby for PSVs, but your mention of a third PSV is a little confusing. Do you have a 150% standby requirement? If not, why the 3rd PSV? I'd suggest you have another look at your capacity design requirements and make sure you've got the right understanding of the standby requirements (just to be sure).

If you are planning on having some future higher flowrate, then size the line for the future higher flowrate. If not, I can't see why you would not just size for the flow from one compressor.

Going the Big Inch!

http://virtualpipeline.spaces.msn.com

 

[bhzimmer]

Sorry, I must have been unclear.

Each stage of compression within the compressor has a PSV designed to protect the vessel in that stage with the lowest MAWP. So the 1st stage PSV is designed to protect the 1st stage discharge pulsation bottle and so on.

My concern is whether I should consider the possibility of 2 or even 3 stages requiring relief. I think this may be excessive and could be considered "double jeopardy". However, I would like to know what is the "accepted practice" for this situation.

Also BigInch, I'm not 100% sure what you mean by standby. I think that if a PSV has 100% standby that it would have a maximum capacity equal to (or greater than) the max. capacity of the entire system. Is that right?

My concern about the sizing of the PSV Header is due to back-pressure caused by several PSV's discharging into the same header simultaneously. If that back-pressure was high enough, then it could reduce the capacity of a PSV, and damage some of the equipment.

 

[BigInch]

Ya that was not clear.

I would assume you must blow down not only stage 1, but the whole compressor and any gas bottles around it, if not blow down everything in the whole station. I'm viewing this as a blowdown required for some reason not necessarily initiated by the compressor itself. Ie. general fire condition, etc.

I can't imagine a client objecting to a blowdown line sized for the entire simultaneous sumation of volumes either.



Going the Big Inch!

http://virtualpipeline.spaces.msn.com

 

[BigInch]

I was going to clarify my standby notation. I think its a typical definition. The clarification is in order, if not a little difficult now, since I made a bit of a mistake when I first mentioned it. I meant the 150% above to = 200%

100% Standby would be one valve operating, one down for maintenance at the moment when a relief event occurs, both valves individually sized for full flow capacity. Installation hence contains 2 full capacity valves. Generally this would give a reliability of approximately 95% for typical equipment installations. Note: Reliability is highly variable and and stated for illustration purposes only. It must be determined for any given type of equipment, considering its service and maintenance environment.

200% Standby would be would be one valve operating, 2 not functioning, all sized for full capactiy = 3 valves. This would typically give a reliability of 98%, same note as above.

300% Standby follows the same logic, but might typically give a reliability of 99.5% = 4 valves ... same note.

Going the Big Inch!

http://virtualpipeline.spaces.msn.com

 

[Montemayor]

Zimmer:

Your original post was clear in stating that you are dealing with only one natural gas compressor, and not 3. However, what you fail to state is the TYPE of compressor - i.e., is it a recip or a centrifugal? The type of machine is critical for the PSV capacity analysis of each stage requirement since a recip has 2 inherent check valves built into each stage while the centrifugal has none. I assume you are sizing the PSVs on the basic "blocked discharge" scenario on the last stage.

Please define the type of compressor and if you have any side streams or recycle streams entering any of the stages. All this information has to be defined or understood before any header sizing requirement is discussed.

 

[BigInch]

Why not size for simultaneous blowdown? Do you think it will make a cost difference in this compressor? Can you assume that if one stage blows, the others will or cannot? IMO sizing a relief line for less than what is contained in one compressor unit is NOT a safe practice...IMO, regardless of what any code says.

Going the Big Inch!

http://virtualpipeline.spaces.msn.com

 

[frogcurry]

If sizing for relief valves lifting on blocked outlet, then the gas relieving from compressor A can't be also simultaneously coming out compressor B, if the order is A>B>C in series. Since after all blocked outlet on A is blocked inlet on B. Its not feasible to have A and B both pushing open their relief valves fully. However I could see an argument that the relief valve on A could be chattering with some flow while B is relieving continually. Still the maximum gas rate to relieve from both both RV's together can't be sustained as any higher than the throughput through A in this case - unless someone can come up with a good reason why it could be larger.

However other relief cases could be simulatenous, i.e. fire case if its considered to apply.

 

[BigInch]

Agreed, however, "However... ", is Precisely my point.

Going the Big Inch!

http://virtualpipeline.spaces.msn.com

 

[bhzimmer]

Monte, it's a recip, though I didn't explicitly mention it I suppose it was implied by the fact that it was a 3-stage compressor. Regardless, I apologize for the misunderstanding. There is a bypass line, running from 1st stage suction line to final discharge line, though I don't see how it will affect the PSV header sizing.

Big, thanks for the explanation. As for the issue of sizing for simultaneous blowdown, it is certainly the safer way to go. My concern is that it's overly conservative, and that some cost can be reduced.

The 'blocked discharge' scenario is the design case that I think should be used. However, instead of just applying it on the final discharge stage, I'm assuming it's possible for a blockage to occur in any stage.

Is it possible for all 3 PSV's (or even 2) to blowdown simultaneously? In what scenario would such an event occur?

 

[BigInch]

If you cannot guarantees that it is IMPOSSIBLE for more than one to lift at a time in all conceivable reasonable scenarios, including a general unit fire, then by all means size the flow for all 3.

Curious. Exactly what does "overly conservative" translate to in $ terms? What are the flow rates and the possible line diameters that we are talking about here? Does this mean the difference between 2000 feet of 20", 24" or 30" pipe?



Going the Big Inch!

http://virtualpipeline.spaces.msn.com

 

[bhzimmer]

Big, I understand your argument that the actual $ savings in cost is probably low enough to justify the oversizing and eliminate the risk, however remote that risk may be.

Frog, I personally agree with your logic. The fire case does not apply, however, can anyone think of another possible scenario where I would need to size for more than one PSV discharging.

 

[BigInch]

?

Going the Big Inch!

http://virtualpipeline.spaces.msn.com

 

[JJPellin]

I am not an expert on relief scenarios or PSV sizing, but I do know a lot about recip compressors. Unless you are introducing additional streams between the stages (which is very unusual) the total gas flow available for relief cannot be greater than the inlet flow to the first stage. If the first stage discharge were to relieve at the full flow of that stage, then no gas could be delivered to the second stage and thus it could not also relieve the full flow at the same time. This is also true of the third stage. Spill-back streams shouldn't matter. All they can do is divert third or second stage flow back to an earlier stage. The total amount of gas available to relieve is still no greater than the inlet flow to stage 1.

 

[Montemayor]

JJ:

Great analysis. You beat me to the punch. I tried setting up the response with the direct question(s) establishing that this is a 3-stage recip machine with a straight flow through the 3 stages: i.e., no side streams.

Therefore, as you explained, the relief quantity in any stage is the capacity of the 1st stage. You need a PSV in every interstage (as well as in the 1st stage suction and the final discharge, but there is no need for more relief capacity than the capacity of the 1st stage - in Standard volume/time. This fixes the lb mols of gas/time and you can then convert to actual volume/time at the discharge of each stage. There is no back flow through the machine because of the inherent "check" valves in each stage.

 

[Yuman]

I agree with the analysis of Montemayor.
each stage has his pressure and his flowrate, therefor you nedd 4 RV (including the discharge compressor).

 

[bhzimmer]

Thanks for the insight everyone.