Based in the API 521, Guide for Pressure Relieving and Depressuring Systems
Fourth Edition, March 1997 following points:

Section 3.20.2 Location of a pressure relieving device in a normally liquid System:

The paragraph at API 521, Section 3.20.2, states, "Where valves or other devices are sized to relieve vapors caused by vapor entry generation of vapor in a normally all liquid system (see 3.10, 3.12, 3.13, and 3.19), care should be taken to locate the device so that it actually relieves vapor and is not required to relieve the volumetric equivalent of the vapor as liquid".

Section 3.15 External Fire at point 3.15.3.2 for Liquids:

The paragraph at section 3.15.3.2 states that: " There is an interim time period between the liquid expansion and the boiling vapor relief in which mixtures of both phases need to be relieved simultaneously, either as flashing, bubble, slug, froth or mist flow until sufficient vapor space is available inside the vessel for phase separation. This mixed phase conditions is usually neglected during sizing and selecting of the pressure relief device.

Other paragraph at this section states: "For some vessels, particularly overfilled steam drums or polymerization reactors, the interim relieving conditions may be the factor that would determine the size of the relieving device"

I have the following Questions:

In a Liquid filled System (i.e. LPG filled vessel), where the pressure relief device is located in either the top head (for vertical vessel) or on the overhead line from the vessel exposed to fire:

Will be the liquid expansion relief rate enough to allow sufficient vapor space inside the vessel for phase separation during fire case?

Did you know how could be the sufficient vapor space for phase separation estimated?

Since in almost all cases, vapor will be generated as soon as enough space is available for vapor to form, to relieve the pressure on the vessel, if this space is not enough for phase separation, should be the pressure relief device during a fire, be able to pass a volume of liquid equivalent to the displacement caused by vapor generated by the fire?. It means that pressure relief device shall be designed to pass the volumetric equivalent of the vapor as liquid?

The volumetric equivalent of vapor as liquids meaning is equivalent to the volumetric fire gas rate? If yes, shall then, the relief device, designed in terms of liquid mass rate, to the equivalent volumetric fire gas rate times liquid density at relieving conditions, besides it will occur during a short time ( seconds) period?


Thanks, and best regards,

Ana Lorena Marrero ochoa

Ana:

This is such an important pressure relief subject that you have brought up that I find myself adding comment before Phil Leckner has a chance to add his valuable and expert opinion to this thread. I anticipate that Phil will be joining this thread – mostly because I believe he will detect the hidden, important safety features it contains.

First of all, I suspect you are alluding to a standard LPG (Liquefied Petroleum Gas) storage tank which usually takes the form of a “bullet” vessel – a cylindrical, horizontal vessel that usually has 2:1 ellipsoidal heads on each end. These types of vessels normally have two pressure relief valves mounted on the top vapor space of the vessel. The PSVs are separated by a 3-way valve and allows for one PSV to always be in operation while the other is on “stand-by service. One, 90 degree-turn of the valve handle puts one PSV immediately into service, and isolates the other. These tanks are filled to a maximum liquid LPG content of 85-90% of the total water internal volume. The LPG is stored as saturated liquid product and the vapor space above the liquid is in equilibrium with the parent liquid below it. I am stating all of this basic data – most of which should be apparent to everyone - to make sure we all agree on the application and what is involved.

Section 5 of the API 521 Recommended Practices deals, as you have noted, with an external pool fire. The API literature has always written its recommendations in a very general manner, in my opinion, and it behooves all engineers to carefully take note of the intent and the purpose behind the API’s recommendations when one is trying to apply them. Although API states that the system is “liquid filled”, you and I know that the system is NOT 100% liquid-filled. There has been ample and sufficient vapor space incorporated into the storage tank design and operating instructions due, primarily, to the fact that the contained liquid is saturated. This is very important basic data to take into consideration in your design. The API is forced to deal with the subject in a general manner; we are free and able to deal with it in a specific manner. The basic intent of the API is to advise you that the liquid contents of a vessel under fire is what is indirectly protecting you (and the surroundings) from an impending vessel structural failure – assuming that your relief valves work properly. The liquid content in the vessel (while it lasts) is a virtual heat sink that converts latent heat input into relief vapor and keeps the vessel cooler than what it would be forced to undergo were it filled with vapor (or empty).

Consequently, you have no reason to be concerned with liquid thermal expansion – as you would be were the vessel to be 100% liquid-full. The percent of volume taken up by the liquid expansion is miniscule compared with the amount of internal vapor generated during a fire. The subject is purely academic and not practical for the purpose of protecting the tank and the surroundings during a fire. The normal 15% vapor volume designed into such a tank is the inherent expansion chamber built-in for the sake of handling a saturated liquid and it also serves to cushion any further liquid expansion.

The important feature you bring up is the possibility of entraining liquid into the inlet of the relieving PSV. Again, the minimum 15% vapor volume designed into the tank as well as the location of the PSVs also lends sufficient vapor disengagement space to deter any 2-phase mixture reaching the PSVs. These factors should normally protect you and the tank from undergoing any undue or unexpected stress during a fire. Sure, the tank is under extreme potential stress during a fire; but while the liquid reservoir lasts, the vessel can continue to function as a pressure vessel while relieving excess pressure and receiving the cooling effect of evaporating liquid. However, once the liquid inventory is exhausted and the fire continues, the vessel starts to undergo the possibility of total structural failure due to an inability to relieve pressure build up faster than the heat accumulates in its structural shell and supports.

For that purpose, API recommends depressurization techniques and other protections procedures such as water sprays and insulation. These methods are not 100% effective however, and good, conservative, engineering judgment should always be employed in studying the proper way to combat the fire case scenario.

The volumetric equivalent of vapor as liquids does not normally enter into the design of the appropriate PSVs protecting the vessel because of the factors mentioned previously. However, if you are really and truly considering a vessel 100% filled with LPG, then it is a far more serious situation and one that should be considered as a thermal relief application where conventional vapor PSVs do not apply.

You are wise in noting that you should take into account the possibility of liquids entering the PSVs and also concerning yourself with sufficient disengagement space to ensure that pure vapor enters the PSV. However, unless I am mistaken in my assumptions of what you are facing, the application seems more benign than what you describe as a possibility.

I look forward to your comments and especially to Phil’s.

Art:

Thanks for you’re soon and complete answer, yes if LPG is in equilibrium (saturation) an enough vapor space is available, then liquid relief will not occur.

But, I am talking about a fire case in vessels or columns located in a LPG DEA treating Unit:

Where LPG from gas plant is treated in a liquid liquid PP/DEA extraction column at 40°C and approximately 26 barg, treated liquid PP is the overhead extraction column product, and rich DEA is the bottom extraction column product. Downstream this extraction column is located a DEA/PP mixer, a caustic sealing vessel and a caustic washer tower, where liquid PP is the top product.

I think that if vessels top or columns overhead products are liquid (LPG), then it should be handle like a 100% LPG liquid filled vessel, where vapor space is not available. Based in that, and the API recommendations,

Should I size the relief valve in a fire case scenario to pass the volumetric equivalent of the generated vapor as liquid?

The volumetric equivalent of vapor as liquids meaning is equivalent to the volumetric fire gas rate? If yes, shall then, the relief device, be designed in terms of liquid mass rate, to the equivalent volumetric fire gas rate times liquid density at relieving conditions, besides LPG relief at bubble point will occur during a short time ( seconds) period, while a enough vapor space is reached ?

Thanks, Ana

If I understand you correclty, you are questioning whether you should size the relief valve for liquid at the rate equivalent to the amount of vapor you would produce in a fire.

By all means check the relief valve size for this scenario. But also check it for a straight fire relief case as well i.e. do both calculations. I believe you fill find the relief valve size based on the vapor generation will be greater than the equivalent liquid rate. Then specify the valve for the largest size.

My concern is even greater and that enters the two-phase relief arena. DIERS had shown that in totally liquid full vessels, the bubbles generated can actually induce a two-phase relief until you get that vapor space expanded. The bubbles expand the liquid and are entrained as the liquid relieves, thereby creating a two phase relief. This would then fall under the non-flashing two phase relief scenario.

Many years ago while working on a GE project, I remember that GE had a criteria that if their vessels can normally operate at some specified liquid level, then all fire relief cases were to be considered two-phase. I think this magic number is greater than 80% full but I can't be sure.

Hello Phil, thanks for your response, yes I am questioning whether I should size relief valves for liquid at the rate equivalent to the amount of vapor volumetric rate.

Like I explain before, my case could be considered like a 100% full LPG vessel. My interpretation for the API recommendations for a fully liquid vessel exposed to a fire is that the relief valve should have the capacity to relief a liquid volumetric rate equivalent to the vapor volumetric generated rate. Additionally, liquid is at bubble point, then relief valve should have the capacity to relief the vapor volumetric rate as liquid at bubble point, taking consideration that liquid will flash across the relief valve. When I compare the relief valve required capacity to relief an equivalent vapor volumetric rate as liquid at bubble point, using HEM method, I find that relief valve capacity area shall be 5 to 7 times bigger than the required relief valve area when is sized only for vapor fire relief load. I used HEM method because I have information from papers and companies design criteria recommending using HEM method instead DIERS method to size relief valves for two phase flow relief.

I just need from anyone with LPG relief situations experience, which confirm that my considerations are ok and that relief valves must be sized considering all mentioned above or not, if not I need to know the reasons for that. My concern is because like I mentioned above, relief valve capacity area based in a liquid volumetric rate equivalent to the gas volumetric rate, and considering two phase flow will result in a relief valve 5 to 7 times bigger than the relief valve area sized only for vapor using the relief load generated by fire.

I really appreciate your help in this matter.

I don't have direct LPG experience but as I wrote before, DIERS has shown that for an external excessive heat scanrio (i.e. fire), you can expect a two-phase relief if the vessel is 100% liquid full. Assuming flashing two-phase relief is a worse case over just non-flasing two-phase flow (entrained vapor) and using the HEM should be adequate. Point of information, HEM is now the method of choice by API (not that I think API is necessarily correct in everthing they publish).

The fact that the relief valve is considerably larger than vapor relief alone does not surprise me nor does it scare me nor would I let it influence me in maintaining a safe design. To me, you are using your best engineeing judgement and good engineering practice to be safe. Unless someone can show you AND document differently for LPG specifically, I say your analysis is very resonable and go with what you feel is correct and safe.

Your approach is absolutely correct. Iterating Art's oft repeated warning; there are two aspects
a) Calculating the relief flow rate. And your approach is correct. API 521 section 3.15.3.2 clearly specifies that if the PRD is located below the liquid level, then the PRD needs to relieve liquid equivalent of vapor generation. So, even if there is vapor space avalaible in the vessel, if the PRD nozzle can see liquid, say through liquid expansion, then the PRD needs to be designed for liquid relief.
Sizing the orifice area. The final size is normally left to the vendor, but remember for two phase PRD, no vendor will certify the valves. The HEM method you have used is correct. As a warning, please note that the method specified in API 520 sometimes leads to higher orifice area than through HEM method. This is true when the relieving conditions are near critical values, and property predictions are tougher. I would say use both methods and then choose the higher area calculated.

I have sized such valves in LPG service and your conclusion is correct. The size of the relief valve is significantly higher than for vapour only service.

Regards
Rajiv