Jump to content



Featured Articles

Check out the latest featured articles.

File Library

Check out the latest downloads available in the File Library.

New Article

Product Viscosity vs. Shear

Featured File

Vertical Tank Selection

New Blog Entry

Low Flow in Pipes- posted in Ankur's blog

2

Two Phase Psv Calculation Through Direct Integration As Per Api 520

two phase psv direct integration

17 replies to this topic
Share this topic:
| More

#1 emma14778

emma14778

    Junior Member

  • Members
  • 10 posts

Posted 11 January 2021 - 02:07 PM

i have some experience with single phase PSV sizing as per API 520 methodology but am new to two phase sizing. I have started looking into two phase PSV sizing using API's method of direct integration of the isentropic nozzle flow. 

 

My system is a glycol/water system at 202degC and has a PSV with a set pressure at 375PSIG and 10% overpressure. The system is highly subcooled at the inlet and discharges to atmosphere, saturation pressure of the system is 220PSIA.

 

  1. Following the direct integration method in API I complete a table at a 4% integration step size which determines the maximum mass flux. I was surprised to see the highest mass flux had a mass quality of 1 (no vapour). What does this mean for the PSV relieving condition? Does this mean the PSV is a liquid only PSV?
  2. Is there an accepted method for calculating discharge line losses in a two phase system? 
    1. I have done some two phase line loss calculations before using either the friedel, chisholm-baroczy, or lockhart martinelli correlations. I've only used these calculations for fairly small pressure drop systems and have not been concerned with the critical pressure of the system. I believe I usually make the assumption that the vapour fraction remains constant which is likely not a good assumption for a PSV relieving scenario with higher delta pressure. 

 

Please let me know if you need any additional information. 

 

Thanks in advanced!

 



#2 Bobby Strain

Bobby Strain

    Gold Member

  • Members
  • 2,941 posts

Posted 11 January 2021 - 04:05 PM

You can find software at my website for two-phase flow calculations. If you have HYSYS I can share software to easily size PSVs for two-phase flow. Prode Properties also has calculations, and it's free.

 

Bobby


Edited by Bobby Strain, 11 January 2021 - 04:57 PM.


#3 breizh

breizh

    Gold Member

  • Forum Moderator
  • 4,893 posts

Posted 11 January 2021 - 09:53 PM

Hi Emma ,

Consider the resource (LESER)  attached to support your work  .

You will find formula and examples.

Hope this is helping you.

Breizh 

Attached Files


Edited by breizh, 11 January 2021 - 09:56 PM.


#4 latexman

latexman

    Gold Member

  • Forum Moderator
  • 1,081 posts

Posted 12 January 2021 - 08:09 AM

1.  4% increment steps are larger than API recommends (1%). Did you test if that was adding any excessive error?  Finding a maximum flux indicates choked.  Being choked with no vapor phase does not feel right to me, but I have no details.

2.  My company uses HEM (mainly in U.S.) and the Omega method (mainly in Europe).  It depends how people were trained.

     1.You are correct, the volume fraction must be allowed to change with each pressure increment, if needed, to be of value in a PSV sizing.



#5 emma14778

emma14778

    Junior Member

  • Members
  • 10 posts

Posted 12 January 2021 - 12:17 PM

Hi Everyone - thanks so much for the responses!

 

Bobby - I have access to Aspen Plus V10 for physical properties. Can you provide a link to the tool you are referencing and I'll see if I get similar answers.

 

Breizh - Thanks for the reference. I will take a read through it this evening and get back to you with any follow up questions.

 

Latexman - I have updated the calculations to a 1% increment and posted the data below. The highest mass flux is still right before the bubble point with no vapour fraction. I am not sure if this isn't working well because I am so highly subcooled on the inlet. When switching to 1% increments I get a maximum mass flux of 11,224lb/sft2, using the 4% increment I got 5660lb/sft2. 

 

I also tried the Omega highly subcooled method and calculated a mass flux of 10,150lb/sft2 which is fairly consistent with the 1% increment. 

 

I am still unsure whether the direct integration method is telling me the valve is liquid only. I tried calculating the tail pipe pressure drop which resulted in an estimated PSV valve outlet pressure of 187PSIA which is two phase, but I'm honestly not sure if Aspen is the right method to try to estimate the tail pipe losses.  

 

 

Attached Files


Edited by emma14778, 12 January 2021 - 12:29 PM.


#6 Bobby Strain

Bobby Strain

    Gold Member

  • Members
  • 2,941 posts

Posted 12 January 2021 - 12:48 PM

Emma,

    My software only works with HYSYS. So, you should get Prode Properties. If you supply the fluid composition and property set you use, I'll check your calculation, and others are likely to do the same. Your uncertainty is compounded by using Excel formulas of unknown validity. Prode Properties is well validated, so results using it with Excel will assure confidence in the values.

 

Bobby



#7 breizh

breizh

    Gold Member

  • Forum Moderator
  • 4,893 posts

Posted 12 January 2021 - 11:22 PM

Hi Emma ,

Are you sure about the description of your system ?  2 phases ? 

With the data supplied I got 11232  ; 11178 and 11224  lb/s-ft2 depending of the methods described in the paper attached .

My understanding is that it is a subcooled liquid .

 

Good luck

Breizh  

Attached Files



#8 latexman

latexman

    Gold Member

  • Forum Moderator
  • 1,081 posts

Posted 13 January 2021 - 09:07 AM

427 psia  sizing pressure

220 psia  saturation pressure

187 psia  back pressure

 

Predominately subcooled liquid flow, but I see your concern, there might be flashing before the end of the PSV nozzle.  Have you researched "frozen equilibrium"?  Basically, it takes some time to reach equilibrium, and this liquid will be moving fast.  With predominately subcooled liquid flow as you have I think it is highly probable that liquid will be ejected by the PSV flow nozzle into the tailpipe where it will attain equilibrium and start flashing.

 

Are you modelling the inlet pipe, PSV, and outlet pipe in the Safety Environment in Aspen Plus?  And, you've indicated two phase flow is possible in the PSV and tailpipe?  It's been a few months since I've done this and I didn't open Aspen so I may not have the nomenclature 100% right.


Edited by latexman, 13 January 2021 - 09:10 AM.


#9 emma14778

emma14778

    Junior Member

  • Members
  • 10 posts

Posted 14 January 2021 - 06:04 PM

Hi All - thanks again for all of the detail responses and support. This is a great learning opportunity :)

 

Bobby - My system is 83wt% water, 17wt% monoethylene glycol at 202degC. The set pressure is 375PSIG with 10% overpressure. Typically I use the property package HYSGLYCO in Aspen Plus. If you're able to run the calculation on your end to compare that would be greatly appreciated. I have never heard of prode properties before but it looks like an excellent resource. I'll be sure to download.

 

Breizh - I am definitely a subcooled liquid on the inlet of the PSV but I flash somewhere along the PSV and/or the tail pipe. From the Aspen piping calculation it is predicting two phase in the PSV itself but I am not sure if the results from the direct integration calculations showing the highest mass flux has a vapour fraction of 0 contradicts this.

 

Latexman - Yes that is exactly my concern, I am not sure how to determine if I am really flashing in the relief device and if the results from the direct integration method are trying to tell me I am liquid. I haven't heard of frozen flow before - even though the pressure and temperature may be indicating two phase it may not have the time to reach equilibrium for the phase change? I'll have to do some research and get myself up to speed on that one. Thanks for the hint. I am not using Aspen Plus for the PSV sizing - I don't currently have access to their PSV tool. I am trying to calculate from fundamental principles as per API 520. My company has a PSV tool built in house that calculates single phase relief. I can calculate the inlet line losses using this tool and they are significantly less than 3% (approx 0.2%) with no phase issues as expected given the degree of subcooling. I am trying to calculate the PSV capacity through the API method and then use Aspen Plus pipeline to solve for the two phase line losses. After a little more research I see API 521 (7th edition) section 5.5.10 explains a method of two phase line losses - have you ever used this method?

 

Thanks again everyone!


Edited by emma14778, 14 January 2021 - 11:15 PM.


#10 latexman

latexman

    Gold Member

  • Forum Moderator
  • 1,081 posts

Posted 14 January 2021 - 10:09 PM

What is the controlling scenario, sizing flow rate, size of the PSV, it’s flow nozzle size/diameter, and it’s liquid and vapor flow coefficients?

#11 emma14778

emma14778

    Junior Member

  • Members
  • 10 posts

Posted 14 January 2021 - 11:14 PM

Controlling Scenario: blocked discharge

Sizing flowrate: 32,510 lb/hr

PSV Size: 1 1/2" x 3" 325mm2 Orifice

Vapour Kd: 0.627

Liquid Kd: 0.491

*Discharge coefficients are ASME rated from NB-18, they are slightly lower than usual for this design valve.



#12 PaoloPemi

PaoloPemi

    Gold Member

  • Members
  • 500 posts

Posted 15 January 2021 - 04:39 AM

emma14778,

in case of doubts, in this forum you can find examples of VBA code for solving isentropic nozzle with HEM method, see https://www.cheresou...ng-temperature/it should be easy to adapt these procedures to your preferred software.

Take care that you need to solve also phase equilibria, I am not familiar with Twu-Sim-Tassone EOS which, according your msg #1, predicts a saturation pressure of about 220 psia at 220 C for the mixture 87% water + 13% Ethylene Glycol,

if I select the PRX-NRTL package (with Huron Vidal mixing rules) in my copy of Prode Properties I get higher values (about 270 psia) , I have no access to experimental VLE data to verify which is the correct value.

Other points to consider are the methodology itself (depending from application you can prefer homogeneous equilibrium as HEM method or non-homogeneous equilibrium, non-homogeneous non-equilibrium etc.) plus the contributes of inlet / outlet lines as noted by Latexman.

I am not familiar with the models included in Aspen, my copy of Prode Properties includes the mentioned HEM, HNE, HNE-DS, NHNE plus a custom model which I created for specific applications), note that different models can predict different values...

As you see sizing a PSV for two phase flow can require some specific skill.



#13 latexman

latexman

    Gold Member

  • Forum Moderator
  • 1,081 posts

Posted 15 January 2021 - 09:10 AM

Excellent advice, PaoloPemi.  Step 1 is always the selection of the best thermo/VLE model for the problem.

 

Nicole, I have Aspen Plus V10 and I was researching your problem using the Methods Assistant.  It seems to me that HYSGLYCO was tuned for Triethylene glycol/water pair at temperatures lower than your range in your pressure range and at pressures lower than your range in your temperature range:

 

"The Glycol property package should be applicable over the range of temperatures, pressures, and component concentration encountered in a typical TEG-water dehydration system: between 15°C to 50°C and between 10 atm to 100 atm for the gas dehydrator, and between 202°C to 206°C and 1.2 atmospheres for the glycol regenerator."

 

The Cubic-Plus-Association EOS may be a bit better and a more general purpose method if you have it in your version:

 

"The CPA method represents the Cubic-Plus-Association EOS modeldeveloped by Kontogeorgis and co-workers (Kontogeorgis, Voutsas, Yakoumis, Tassios, IECR 1996). The model combines the SRK cubic EOS with an association term similar to that of SAFT, as present in the PC-SAFT model. Mixing rules apply to the cubic, whereas combining rules are used for the association term. The model’s applicability covers the VLE and VLLE of mixtures containing hydrocarbons and polar/associating chemicals such as water, alcohols, glycols, esters, and organic acids."

 

The CPA method seems to have been developed to handle hydrogen bonding, which is exactly your problem's main non-ideality.

 

Using CPA in your problem definition, I got a saturation pressure of 226.56 psia at 202 C.  Only 6 psi greater than your saturation pressure; 3% difference.  Not much to worry about here, IMO.  Paolo, can you double check that 270 psia saturation pressure?

 

Anyway, using my CPA properties, I used my company's HEM PSV software, similar to Breizh's article above, and got the following:

 

 EMMA

  Estimate of flow rate with 2 data states.
  Areas and coefficients for nozzles.
 
  Reservoir pressure             =  427.20 psia     
  Surrounding pressure           =  187.00 psia     
  Nozzle area                    =   0.503 in2      
  Nozzle coefficient             =  0.4910
 
  ==================== PROPERTIES ====================
    DATA STATE                      A           B
    ==========                   =======     =======
    Pressure, psia                226.60      187.00   
    Wt. fraction gas             0.00000     0.01993   
    Gas density, lb/ft3          0.51020     0.42360   
    Liquid density, lb/ft3        53.650      54.260   
  ====================================================
 
  REAL NOZZLE RESULTS:
    Mass velocity   =  122288.0 lb/hr/in2
    Mass flow rate  =     61511 lb/hr    
    Stored energy   =   1287.18 Btu/ft3  
    NOZZLE THROAT:
      Pressure      =    226.57 psia     
      Thrust / area =     136.0 psi      
      Velocity      =      91.3 ft/s     
      Wt. frac. gas =   0.00001
 
And the next API nozzle size down (0.307 in2 = 198 mm2):
 
EMMA
  Estimate of flow rate with 2 data states. 
  Areas and coefficients for nozzles.
 
  Reservoir pressure             =  427.20 psia     
  Surrounding pressure           =  187.00 psia     
  Nozzle area                    =   0.307 in2      
  Nozzle coefficient             =  0.4910
 
  ==================== PROPERTIES ====================
    DATA STATE                      A           B
    ==========                   =======     =======
    Pressure, psia                226.60      187.00   
    Wt. fraction gas             0.00000     0.01993   
    Gas density, lb/ft3          0.51020     0.42360   
    Liquid density, lb/ft3        53.650      54.260   
  ====================================================
 
  REAL NOZZLE RESULTS:
    Mass velocity   =  122242.0 lb/hr/in2
    Mass flow rate  =     37528 lb/hr    
    Stored energy   =   1287.18 Btu/ft3  
    NOZZLE THROAT:
      Pressure      =    226.57 psia     
      Thrust / area =     135.9 psi      
      Velocity      =      91.3 ft/s     
      Wt. frac. gas =   0.00001
 
The next size down API nozzle size (0.196 in2 = 126 mm2) was too small.
 
Notice I used the liquid coefficient, because in both cases my software predicts the pressure at the nozzle exit is essentially the saturation pressure with almost no vapor (Wt. frac. gas =   0.00001).
 
Now, I realise this mixes the methods I used with your methods and two-phase backpressure calculations, but it was the best I could do quickly.  The real answer will be a convergence of the PSV and tailpipe calcs providing the backpressure on the PSV outlet using the same methods and property model.  I hope this helps you. 

Edited by latexman, 15 January 2021 - 02:41 PM.


#14 Bobby Strain

Bobby Strain

    Gold Member

  • Members
  • 2,941 posts

Posted 15 January 2021 - 12:50 PM

I get essentially the same value as Latexman using HYSYS PRSV dataset.

 

Bobby



#15 PaoloPemi

PaoloPemi

    Gold Member

  • Members
  • 500 posts

Posted 15 January 2021 - 02:46 PM

Latexman, as written in my post, I was calculating saturation pressure at 220 C (not 202 C),

herebelow some saturation pressures calculated for (water 87% Ethylene Glycol 13% molar fractions) with some models  available in Prode Properties

CPA-PRX with VDW mixing rules -> 205 psia

CPA-NRTL-PRX with Huron-Vidal mixing rules ->  207 psia

PRX-NRTL with Huron-Vidal mixing rules ->  202 psia

PC-SAFT ->  208 psia

 

since these results are strongly affected by BIPs in Prode database (calculated at low / moderate pressures) I am not  sure about the real accuracy and I have no experimental VLE data points at these conditions, however I guess the values shouldn't be far from real data



#16 latexman

latexman

    Gold Member

  • Forum Moderator
  • 1,081 posts

Posted 15 January 2021 - 02:58 PM

Sorry, I missed that 220/202 issue.  Thanks for pointing it out.

 

Another issue.  You said you used mole fractions and Nicole said "83wt% water, 17wt% monoethylene glycol ".   :wacko:



#17 PaoloPemi

PaoloPemi

    Gold Member

  • Members
  • 500 posts

Posted 15 January 2021 - 03:08 PM

thanks, now the results are about the same, water 87% Ethylene Glycol 13% mass fractions

saturation presures at 202 C

CPA-PRX with VDW mixing rules -> 220.5 psia

CPA-NRTL-PRX with Huron-Vidal mixing rules ->  223 psia

and so on... I guess both softwares calculate about the same values



#18 breizh

breizh

    Gold Member

  • Forum Moderator
  • 4,893 posts

Posted 18 January 2021 - 02:16 AM

Hi ,

To add to my previous post a doc published by ASPEN where you can find the calculation methods and examples .

Good luck

Breizh 

Attached Files






Similar Topics