16 replies to this topic

[farid.k]

hye team,

 

as we know outlet of relief valve pressure drop must be calculated carefully. below is my situation:

 

service: ethylene gas

relieving flowrate: 14000 kg/h

selected orifice size: 50H80

set pressure: 36.5 barg

outlet relief: to flare header, operating pressure 0.2 barg

governing case: fire case

location: top of pressure vessel

 

1-is it true to calculate the pressure drop of the outlet of relief valve by assuming outlet of relief valve during valve pop-up is same as per set pressure? meaning properties of the ethylene gas is at pop-up condition (P=36.5 barg at T= 141 oC) or P= header pressure which is 0.2 barg and T= 141 oC. this 2 condition will give 2 different properties and 2 different pressure drop value.

 

2-calculation of the pressure drop up to what point? is it until KO drum?? one of my sifu said is until 2 time pipe size change.

 

 

 

 

 

[latexman]

I assume a conventional PSV and outlet piping sized for appropriate backpressure. 

 

1 - Work the pressure drop backwards from flare header to relief outlet flange along the tailpipe.  Remember there will be choked flow at end of PSV's H flow nozzle.  Critical pressure ratio applies.  Then, there will be a pressure discontinuity across the shock wave to the backpressure from tailpipe.  It may take a few trials to get convergence.

 

2 - until the pressure drop is not significant to the problem.  This could mean all the way to flare or KOP or header.

Edited by latexman, 11 January 2017 - 11:12 PM.

[farid.k]

dear latexmen, thanks for the response.

 

yes, it is conventional.

 

pressure drop backwards mean from main header assuming main header pressure is 0.2 barg, up to flange of relief valve outlet? thus need to use properties of ethylene gas at 0.2 barg for pressure drop calculation?

 

after getting the differential pressure of the piping (which is build-up back pressure), that figure+superimposed must not exceed 10% of the set pressure isn't?

 

the other things is, what is allowable velocity for the discharge relief valve piping? sorry for my simple question.

[latexman]

Use 0.2 barg at main header flange on first iteration, until and if you see it needs to increase due to increased flow in main header thru KOP to flare.  You will have to iteratively converge pressures, temperatures, and properties.  I hope you have compressible flow software.

 

Yes, BP <= 10% of set pressure.

 

This is probably governed by company rules.  My company allows me to design up to Mach 1, if needed.  Some companies restrict Engineers to Mach 0.6, or 0.7, or 0.8, it depends.  You'll need to find that out.

[Bobby Strain]

You said that the sizing was for a fire case. So you can size based on 21% overpressure with an allowable backpressure of 21% of the set pressure. Otherwise Latexman gave you good advice.

 

Bobby

[farid.k]

Thanks bobby and latexman for reply..i just calculate pressure drop using vmg sim. just that need to have clear picture on the basis of calculation

[latexman]

Thanks Bobby, I always forget that little nuance because our in-house software does inlet and outlet checks at 10% OP, even the fire case where sizings are done at 21% OP.

[Bobby Strain]

So much for in-house software. But I have seen much worse.

 

Bobby

[latexman]

Exactly, I've seen too many old PSV sizings that didn't even check inlet and outlet dP's.  What a pain!

[farid.k]

So much for in-house software. But I have seen much worse.

Bobby

Can i know example software??

[Bobby Strain]

I was not addressing commercial software. But the in-house spreadsheet kind. If you use one of the commercial process simulators, they do a pretty good job with most tasks. But you should remember that thermodynamic calculations are their main focus. Along with property packages and equations of state.

 

While with my first employer, I lead an investigation into a ruptured relief valve lateral. And, it was to relieve ethylene resulting from a blocked-in regeneration heater outlet. There were a combination of errors, but one of the errors was that nobody calculated the PSV backpressure. It still happens.

 

Bobby

[farid.k]

I was not addressing commercial software. But the in-house spreadsheet kind. If you use one of the commercial process simulators, they do a pretty good job with most tasks. But you should remember that thermodynamic calculations are their main focus. Along with property packages and equations of state.

While with my first employer, I lead an investigation into a ruptured relief valve lateral. And, it was to relieve ethylene resulting from a blocked-in regeneration heater outlet. There were a combination of errors, but one of the errors was that nobody calculated the PSV backpressure. It still happens.

Bobby

Thanks sir for the sharing!

[RRahman]

Dear Farid,

 

All the above explanations have answered your question properly.

 

From my limited experience, the best way to calculate PSV backpressure would be to consider all piping up to the flare stack/tip. From my understanding, it seems that you are referring to an existing plant thus all the relevant Process and Piping documents should be available e.g. P&IDs, isometrics, safeguarding memorandum, PSV as-built datasheets etc.

 

Next, it is advisable to use commercially available flare hydraulics softwares e.g. Aspen Flare System Analyzer (Flarenet), Farris iPRSM etc. But I would recommend to proceed using Aspen Flarenet. You can refer to the following steps as reference, it may be may not be applicable to your case. Of course, the detail tutorials should come with the software.

 

(i) Build your model in Flarenet from the PSV, tailpipes, subheaders, main headers, flare KOD, and flare stack/tip. The pressure drop for your flare riser & tip can normally be found from Vendor datasheet. For piping, input all the straight lengths and fittings as per as-built isometrics.

 

(ii) For the PSV, input the relevant PSV and relief scenario data at the relieving conditions. For the fluid compositions, you can define the compositions in Flarenet as you normally did in VMGSim. 

 

(iii) You can select the calculation method in Flarenet which suit your particular application the most. I would recommend using Peng Robinson VLE method and Lee Kesler for the enthalpy method. For pressure drop calculation, you can use Beggs & Brill.

 

(iv) Once all the inputs are correct, you can run the flarenet model and get your results for particular PSV backpressure, Mach no. etc.

 

From my limited 4 years experience, I have come across cases in which the original Process engineer does not calculate the correct backpressure for the PSV. A common mistake was to calculate the total backpressure only for the governing scenarios. Always remember, there might be different governing scenarios for the PSV orifice area, for the inlet line losses, and backpressures. 

 

You can refer to API Std 520 & 521 for more details.

 

Hope this sharing can help you.

 

p/s: I seem to recognize your face but I cannot remember exactly.

[shahidulislam48]

Hi,

 

Is it Built-up backpressure <= 10% of Set pressure or total backpressure (built-up + constant) <=10% of set pressure for selecting conventional PSV?

Section 5.3.3.1.3, API standard 520 Part-1 implies about the first one.

 

Many thanks.

[mirandomka]

The first is API requirement (built up<10% Set P).
Reason is conventional PSV is typically spring adjusted for constant superimposed back pressure.

Depends on project and preference, sometimes you don't want PSV set P to be adjusted. Then, you have to ensure your total back pressure is <10%, so you won't need to adjust the spring pressure.

[shahidulislam48]

The first is API requirement (built up<10% Set P).
Reason is conventional PSV is typically spring adjusted for constant superimposed back pressure.

Depends on project and preference, sometimes you don't want PSV set P to be adjusted. Then, you have to ensure your total back pressure is <10%, so you won't need to adjust the spring pressure.

Hi,

 

Then which one is recommended practice?

Or, I could use any alternatives and take necessary steps.

 

Thanks.

[mirandomka]

If the second one is met, then don't worry about the first one. If the second one can't be met, then check if the first can be met or not. If not, use bellow or pilot PSV