PSV sizing for vapor-liquid mixtures 5

[daniellouw]

Can anyone please tell me how to size a PSV (Pressure Safety/Relief Valve) for vapor-liquid mixtures. Some say that you calculate the area for the vapor and the liquid seperately and add the two values to get the total area (A certain PSV distributor did some calculations for me using this method). However, according to API RP 520, p55 you should use other methods (Appendix D). I calculated the are using both methods and I got answers which differ a lot. Thank you very much.

 

[TrevorP]

Two phase flow through a nozzle, as in relief devices, has been reasonably well studied in recent years, although the methods are not perfect and disagree with one another somewhat, we are certainly much further forward in this area than 20 years ago. What is clear is that calculating an area for the gases, and an area for the liquid and adding the two together, as used to be done before the 1980's, is no longer considered to be sound practice, as it nearly always results in a relief size that is too small.

The DIERS institute (Design Institute for Emergency Relief Systems) has been involved in a lot of work to address this issue, and to develop new methods for sizing 2 phase relief valves. These include the HEM model (Homogeneous Equilibrium Model) and the HNE (Homogeneous Nonequilibrium model).

Some useful published articles on this issue include:
"Select the Best Two-phase model for two-phase relief sizing", Darby, Meiller and Stockton, Chemical Engineering Progress, May 2001. (

http://www.aicheme.org/cep/

)

"Easily size relief devices and piping for two-phase flow", J.C. Leung, Chemical Engineering Progress, December 1996.

and "Properly size pressure-relief valves for two-phase flow", Darby, Self and Edwards, Chemical Engineering, June 2002 (

http://www.che.com

).

If you have used one of the methods described here for your calcs, your answer is almost certainly better than the suppliers answer, if they are using the addition method as you describe.

 

[MortenA]

The newest rev. of API 520 has a set of formulas for two phase relief.

Best Reagrds

Morten

 

[TD2K]

Concur with MortenA's recommendation. The old method of adding up the areas required for the vapor and liquid was only for sizing saturated water PSVs but was used for other fluids simply as there were no other methods. However, API 520 now includes methodology for sizing relief valves handling 2 phases

 

[ddkm]

Can someone share the equations to be used for the DIERS 2-phase flow method? I'm still searching the literature and have not found (as yet) a useful source of understanding for this topic.

I'll share as soon as I have something. Thanks.

 

[Assumptions]

The September 2005 issue of Che has a good article on the topic, Size Safety-Relief Valves for Any Condition.

http://www.che.com/

 

[EGT01]

DDKM,

I can assure you that having the equations will not provide you with an understanding of the subject.

However, as previously recommended, a good source for the equations can be found here...

http://apistandardsonline.api.org/s...lsPage&SiteID=api&productID=33222700&Env=BASE

Besides the references cited above, other resources that will be helpful are offered in this thread...
thread124-127879

 

[phorth]

I have just joined this forum, so my reply is unfortunately two years too late for the original query. However, I would advise that the best method to use for sizing two-phase relief systems is the HEM (Homogeneous equilibrium Method)by Profs Hewitt, Richardson and Saville of Imperial College (UK).
This method is based on thermodynamic principles and has been extensively validated by a programme of experimental research at Imperial. It is easier to use than DIERS and much more accurate than the old API method. It has been established as the standard for the offshore oil industry in the UK since at least 2001.
Refer to "Guidelines for the Safe and Optimum Design of Hydrocarbon Relief & Blowdown Systems" published by the Institute of Petroleum, August 2001, for a description of the method, calculation procedure, and supporting research results. There's also a lot of other useful design information in this book.
Ref also: Hewitt, Richardson, Saville, Weil: Two-Phase pressure relief sizing: an assessment study, Imperial College report MPS/73, June 1995

 

[Bourbon103]

Ironically, I found this thread today, so perhaps you could help me with a similar situation:

I'm sizing a relief valve on a feed water heater. Water enters the tube side @ 700 psig and ~259 deg. F. Steam enters the shell side @ 75 psig (320 deg. F). Sizing for a tube rupture and subsiquent relief to an atmosphereic holding tank, the water will flash exiting the relief valve, choked flow, etc. Can this be treated as above? I assume this falls into the DIERS type of discussion.

 

[EGT01]

Bourbon103,

For the water conditions you have stated and based on the relieving pressure on the low pressure side of your exchanger, I think you could assume that flow through a failed tube should remain liquid as it enters the relief valve. In that case, assuming you are designing for a conventional type relief valve, you should be able to use the procedure as outlined in Section D.2.2 of API RP-520, 7th edition, and as described for a sub-cooled (or saturated) liquid that enters the relief valve and flashes.

 

[Bourbon103]

EGTO1,

Yes, that is the conclusion I had come to as well resently (just seeing your post today). Assuming the inlet will be all liquid, I calculated the required orfice area and inlet piping requirements. Unfortunetly, I do not have the newest API 520. We have the 6th ed. and the appendicies are different and do not include the sections on 2 phase flow. I found a text version online, but lists no equations.

 

[Bourbon103]

I'm not sure how to calculate the outlet piping requirements for two-phase flow. Normally, the pressure drop/100 ft. is calculated, which is a function of the flow rate. I did a basic % flash steam calc (ASHRAE 2004, 10.2) and calc'd 13.5% flash. Is it reasonable to divide the liquid inlet flowrate by this % and calculate 2 different pressure drops for the 2-phase flow as an estimate? I don't need to be exact on this, only a reasonable estimate.

 

[EGT01]

Bourbon103,

I think you may have misunderstood what I meant by saying "remain liquid as it enters the relief valve". At 259F, the saturation pressure for the liquid is about 34.8 psia. For a conventional type relief valve, 75 psig set pressure, and non-fire case relief, your maximum allowed backpressure at your relief valve is only 7.5 psig or 22.2 psia. Consequently, the flow may enter as liquid but you will have flashing as the liquid passes through the relief valve. The flashing will reduce the pressure differential available for calculating the mass flux and ultimately increase the required relief area compared to an all liquid relief calculation.

You really should get a copy of API 520 7th edition or follow the procedures as outlined in one of the other references. As additional resource, I know Farris Engineering's Sizemaster program includes methods from the 7th edition

http://www.cwfc.com/farris/sizemaster/size.html

but again, you are going to need the 7th edition to understand what's needed.

I'm not familiar with the flash calculation that you referenced but your %flash seems to be high. Based on an adiabatic flash starting with water as a liquid at 259F and reducing the pressure to atmospheric (14.7 psia), I only get 4.8% (by weight) flashed to vapor which based on a flow of 100 lb/hr liquid at 259F, you would generate 4.8 lb/hr vapor, 95.2 lb/hr liquid at 14.7 psia, 212F.

Also, I'm not sure what you mean by...
"Is it reasonable to divide the liquid inlet flowrate by this % and calculate 2 different pressure drops for the 2-phase flow as an estimate?"

Why would you divide the liquid inlet flow by the %flash? That would give you a flow exiting the relief valve greater than what is entering. Your comment must be a typographical error.

As to how to determine the 2-phase pressure drop, consider using the Lockhart-Martinelli method. You can find it referenced in Perry's handbook.

With all due respect, I would suggest you seek guidance from one of your more experienced co-workers to make sure you are proceeding as needed for the task.

 

[phorth]

I basically agree with the last post (EGT01), and would add that the relief valve tailpipe should be sized to give no more than 10% back pressure at the flow capacity of the valve plus 10%, which will be significantly more than the calculated relief flowrate. The flash may be 4.8% wt but will be much higher by volume, giving rise to high velocity requiring a large tailpipe to limit the back pressure.

To calculate the back pressure, use Lockhart-Martinelli or something similar, and take full account of the bends and other fittings in the tailpipe. Or find someone who has done it before and get them to show you.

Exchangers are best protected against tube rupture not with relief valves (which can be too slow to respond) but with bursting discs or buckling pin valves.

The research carried out for the validation of the HEM method (referred to in my post above) has shown that the flow of liquid through a relief orifice is not significantly affected by flashing downstream, so a flashing liquid can be treated similarly to a non-flashing liquid. This result may be surprising, but the measured discharge coefficients for flashing liquids were about 0.58 - 0.60 compared to 0.62 typical for a non-flashing liquid.

 

[Bourbon103]

Thank you for your suggestions, they have aided my thinking process.

Basically we have simplified this case to just dealing with the outlet piping. In assuming an all liquid inlet to the PSV, I was able to calculate an orfice area and with a known liquid required relief capacity I sized an appropriate valve. My problem is the outlet piping.

I have reviewed the Lockhart-Martinelli method in Perry's and found a template and calculation guide:

http://mas.adeptscience.co.uk/mas/worksheets/TwoPhasePressureDropApplicationServer.xmcd

but this assumes that the % flash is constant, correct?

The relief valve is set to 100 psig, and the 100% liquid water will be ~ 250F at the valve outlet. As the water enters the outlet pipe, which vents to atmosphere ~90ft away, including 2 std. 90s and 1 thru, it will flash to different steam qualities along the pipe, correct? I attempted to use the Lockhart method but I can't assume that it remains at the 14% flash (going from 110psig to 0 osig) the entire length of the tube.

 

[phorth]

Bourbon103,

Yes that is correct, the flash will increase along the tailpipe as the pressure decreases. This means that you have to tackle the problem by considering the tailpipe in several sections, assuming a constant flash for each section.
Your task is to find the size of tailpipe which will result in a back pressure of less than 10 psig at the relief valve outlet, for a flow equal to the valve capacity plus 10%. Start by picking the size equal to the outlet flange of the valve. Calculate the pressure backwards from the outlet using sections say 20 ft long.The first section will use the flash percent for atmospheric pressure (end of the pipe). If the pressure gets above 10 psig then start again with the next larger pipe size.

If you use a balanced bellows relief valve, then a back pressure up to 30 psig (30% of set pressure) is acceptable.

 

[Ashereng]

daniellouw, thanks for starting the thread.

To all the posters, thanks for the information.

I recently also sized a PSV for 2 phase flow using the individual gas area + liq area method. I am now re-doing the sizing using the API 520 Ed 7 method, available on the Farris SizeMaster program (using the Omega scenario).

Thanks.