LPG storage tanks vent system 1

[halizo]

I have an interesting issue:

We have three (3) storage atmospheric tanks that contain: propane @ -50°F, iso-butane @ -6°F and normal-butane@ 26°F each of one. The tanks are insulated and have a vent header, each tank has a vent valve connected to the header, the valves are control type and they open 100% in overpressure case. The header goes to a burning pit. When I do the hydraulic simulation, if I use isothermal model I obtain liquid, if I use heat transfer with the ambient (ambient temp 66°F) I do not obtain liquids. API 521 says that is more conservative an isothermal model and in cryogenics cases is better adiabatic model. The header is not insulated. Which of these models is more realistic for this case?

Thanks in advance

 

[Montemayor]

halizo:

Your query is confusing. Your say that your tanks have a common VENT header and that this header goes to a burning pit. Vent headers vent GASES, not liquids. Liquids are drained, not vented. Gases are burned using flares; liquids are burnt by way of burning pits. Your case seems mixed up.

Additionally, you state you are doing a hydraulic simulation and you insinuate that you are flowing LIQUID through the vent header. Again, this is very confusing since, as stated above, vent headers are designed to vent gases - not liquids.

Perhaps the above is a result of English not being your native language - but certainly engineering logic is international in nature and common to all engineers, regardless of language. Can you explain in detail exactly what phase you are "venting"?

 

[halizo]

i appologize for my english, is not my native language, i am trying to do my best.
Yes the vapors coming from the tank(from the roof) are going to a vent header and then to the burning pit (it is an old LPG plant. In the burning pit it burn liquid stream and vapor stream. The vent is vapor phase but like i said i wil obtain liquid or not depending the model that i use.

hoping clarify

regards

 

[Montemayor]

Halizo:

Something is wrong with your simulation model. That's why I don't believe in using simulation models for a venting exercise.

You are venting saturated vapors from each one of the tanks in question - when they relieve. I repeat: the vent coming from the tanks is a VAPOR. And that is regardless of what your simulation programs says. This has to be the case in order for there to be any relieving of pressure in the tanks. You don't relieve tank pressure by "venting" (actually draining) liquid. The phase coming out of the tanks has nothing whatsoever to do with what the simulation program predicts or "thinks". And regardless of whether you do an isothermal or adiabatic flow simulation, the flow rate will remain in the vapor state. Depending on the distance the vent vapor has to travel and the potential heat the vapors can pick up, the velocity and the pressure drop will increase as the vapors progress down the vent header and towards the pit where they will be incinerated. You should be able to run either case and I would take into consideration the amount of heat pick up through the header if this is substantial. I presume you are interested in estimating a correct value for the total pressure drop, and if that is the case then you should investigate the adiabatic run.

I hope this helps to clear up the information involved. Other forum members will probably be contributing helpful comments.

 

[halizo]

montemayor: you are rigth the vapors from the tanks are saturated, but when you mix propane vapors at -50°F with n-butane vapors (saturated) at 26°F or/and iso-butane (saturated) at -6°F it will form liquid unless you have heating from ambient (ambient 66°F). Remember the header is not insulated. As additional information the relief pressure is 0.469 psig and the pressure in the header is between 0.3 psig and 0 psig.

I am trying to predict if there are liquids or not in order to design a kod drum for a smokeless flare.

thank again

 

[Montemayor]

halizo:

You haven't stated the basic premise that the scenario you are studying is that ALL storage tanks are relieving simultaneously. I had assumed only one tank was relieved at a time. Now that we are getting more information and data: Yes, you have to make a mixing balance of the different enthalpies to obtain a resultant mixture temperature. This temperature could be so low that it might cause condensation of some of the mixture components. For this scope of work you do need something such as a simulator to get you down to the net resultant temperature and phases in a quick manner.

API 521 does not, I believe, address the operation you describe. However, even if it did, the adiabatic mixing of all the predicted cold vapors still has to be done and its ultimate product phase and temperature calculated. The amount of liquid phase is a critical data to estimate - as well as the duration of the event. Depending on its quantity, I would assume an adiabatic model while simulating the mixture. If the relieving is done for a relatively long time, isothermal conditions might prevail after the initial reliefs due to ice formation and metal cool-down. If you indeed are forming a liquid phase, then you have inherited another problem: what are you going to do with the liquid condensate that accumulates?

Presumably (you haven't stated this) you are calculating around a local pool fire case as the triggering event to relief. If the tanks are all enveloped in flames, are you taking the heat transfer through the header that comes from the surrounding fire? Where you decide to separate the liquids is going to be critical in your scenario.

 

[halizo]

montemayor:
you have a very good points

1) the relief duration, there are a long reliefs, so, as you say the fluid circulating cool-down the line and make the process isothermal. But how can i assume adiabatic model if the header is not insulated?

2)Considerate that the fire heats the header, in this case the liquids will not form

We will installate a KOD to separate the liquids.

thank and regards