PSV Fire Case vessel full of liquid 6

[JASR76]

Hi all!!

I have a question about Fire Case in a vessel full of liquid (HX, etc..) engulfed by fire using API 521 recommendations.

a) At the beggining the liquid will have a thermal expansion and the PSV will relieve liquid.

b) After that, an increase in temp will relieve vapors generated with some liquid.

c) Finally (some hours?), level in equipment will reduce and only vapours will leave the PSV.

My question is:

1/ Which scenario do you normally use to size the valve?

2/ Do you calculate all cases and compare? or do you normally consider a typical case, for example, take the worst case in terms of duty/latent heat and consider ALL as vapour in worst case.

I dont know if there is a normal procedure as good engineering practice and I would like to know your always interesting opinions.

Regards

JASR76

 

[zdas04]

In a blocked-in liquid full vessel, pressure will change about 100 psi for every degree F the liquid increases. Therefore your potential to do serious harm during a fire is huge. The good news is that it takes a very small volume of liquid removal to fix this problem. So in terms of urgency of the problem, the PSV can be very small to handle the liquid-full case. If it is very small, then you'll be in the liquid-full case for a very long time--i.e., if you remove a tablespoon of water the pressure will drop below the set point and the PSV closes. Then a few degrees higher and the PSV lets out another couple of tablespoons. Sizing the PSV for this case is both safe and conservative. Eventually, you'll reach the point where there is not enough liquid to take away the heat of the fire and the steel in the vessel will flow. I don't think you need to worry about sizing the vessel for vapor.

On the other hand, if you size the PSV for the steam case then you'll empty it very quickly by dumping too much mass each time it pops and you'll be empty much sooner.

This scenario assumes a liquid-full, blocked-in vessel. If that is not your scenario then you have to look at how much mass can flow to the vessel during the event, and where the heat of the fire can go. It gets really complex.

David

 

[pipehead]

You can't have a full liquid leaving the PSV during a fire, it may be partially a liquid and vapors, but not 100% liquid. A PSV can pass a constant stream of fluids on a mass basis, so liquids have a higher density than a vapor so a PSV will pass more cubic feet per minute in the liquid phase than the gas phase. So relieveing a vapor phase requires a larger PSV.

Here is what happens during a fire, I'll pick a tank with propane and butane mixed. The pressure in the tank is 140 psig at 70 F and the PSV is set to relieve at 200 psig. The tank cannot be 100% full becuse you should have a vapor are, however, we'll assume 99%.

As the fire starts heating the liquid it will expand and BOIL. Bubbles will rise to the high spot as the vapor will increas the pressure and the liquid will stop boiling because the system is at pressure temperature equalibrium. This will continue until the temperature pressure equalibrium is finally at the set point of the PSV and the corresponding temperature which would be around 150 F for this case. This may only take a few minutes or even seconds depending on heat from the fire. Some liquids may come out because of entrainment. Even if we assumes the vessel was 100% full and liquids exited the PSV, it only takes 10 cc of liquid to lowe the pressure a 100 psi on a 100% full vessel.

Now the heat from the fire will generate vapors at a calculated (or assumed rate) and the PSV will relieve the vapors at the rate they are generated. Now, because we asumed a mixture, the temperature will rise slowly in the vessel because propane will boil off mostly at first leaving a mixture with a higher boiling point liquid (butane). Over the next few hours, the temperature would rise to about 200F.
Finally the vessel will only have vapors and they will expand and exit the PSV. Since there is only a few pounds of vapors to absorb the heat, this expansion of vapors will happen very quickly and the vessel will most likely fail because of high temperature loss of strenght. The good news is there is only a few pounds of material and its fuel enrergy will not had much to the existing fire. Also, the vessel failure most likey will just rip apart and not fly into missles.

 

[zdas04]

pipehead,
I don't think you have your thermodynamics right. Say you have a blocked in liquid-full vessel at 100 psig and 60F. Let's say it is water. If you raise the temperature of the water 10F then you will raise the pressure around 1,000 psi. The result is you are even farther from boiling point than when you started. If a PSV opens, then you will have 70F liquid water leaving the valve. Nothing about boiling is even close to happening.

If you have a vessel near saturation temperature and pressure like in your example, then the dP across the PSV will result in boiling, but not all fire cases are anywhere close to saturation conditions.

David

 

[pipehead]

why would anyone want to operate a vessel 100% full of water at 100 psig and 70F? You are desribing a liquid full line ony. Water cannot exist in a two phase system at those conditions, you need a gas blanket to get there.

You are correct about a LIQUID FULL system, and I stated that it would take just a few cc's to relive 10,000 psi from such a system.
A thermal relief in your scenerio is 3/4 c 3/4 valve only. If you didn't design for the vapor release from a boiling liquid you,d have a failed vessel in 3 seconds after the fire started.

 

[JASR76]

Pipehead, zdas04,

Thank you very much for your answers.

The question is as simple as a heat exchanger. Imagine a heat exchanger full 100% of Hydrocarbon that is heated (with the help of some steam )from a pressure P1, temp T1 (liquid phase) to a Pressure P2 (P1-dP) and T2 (liquid phase too, but a higher temp). This would be a case of a typical heat train. During all time, HX is totally full of liquid .

With this statement, the FIRST question is:

1/ Do you normally use a procedure to size a safety valve for this kind of cases?

In my opinion, initial liquid thermal expansion would lead to a small PSV compares to fire case, so we can forget about it in this exercise.

BUT inside this "non-typical 100% full liquid" fire I have several concerns:

a/ Use a flash to simulate fire evolution across time is a good idea to get 'latent heat values' and 'vaporization massrate' and decide which is the biggest vapour rate in this contingency

b/ Once you have this vapour rate:

b.1/ Vessel (or Hx in this case) is almost or totally full of liquid

b.2/ When PSV begins to open some liquid will go through the PSV nozzle

The SECOND question is:

2/ How do you take into account this liquid? It seems like if level decreases inside vessel (Hx) liquid is not going to relieve as easily.. DO you study a proportion of liquid vs gas across the valve according to Two-phase equations in API 521?

Thank you very much for your answers, I think all opinions helps to make me clear with this question..


JASR76

 

[zdas04]

JASR,
I try to avoid this altogether. First I look for any credible scenario other than fire case and thrermal case (e.g., can an upstream pump exceed MAWP, or can the reservoir pressure get to this vessel), the size of the PSV for that case will always exceed the releiving requirements of the fire case or the thermal case.

If there is no other credible scenario then I'll look at how close the vessel is to the boiling point. For example if your liquid has a coeeficient of thermal expansion that works out to 50 psi/degree F, it is 100F below its boiling point at normal operating pressure, and is within 200 psi at normal operating pressure, then a 4F temp rise will lift the PSV and you should size for 100% liquid (i.e., very small PSV). On the other hand if you are very close to the boiling point and a few degrees of added temp will boil the liquid before it raises the pressue enough to lift the PSV then you need to size it for the vapor case (much bigger PSV).

Whether the liquid flashes in your releif header is a pipe-sizing exercise and can be important in the dP across the valve, but is not important to the full-flow PSV calc.

David

David Simpson, PE
MuleShoe Engineering

http://www.muleshoe-eng.com

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

The harder I work, the luckier I seem

 

[JASR76]

Thank you very much zdas04

but my question is about considering if this particular fire case in a full liquid vessel(considered as critical case) should be consider as:

a) Gas relieving

b) Biphasic relieving

You know that API 521 size with different equations so, it would be possible that considering biphasic would lead to higher orifice than vapour. Thats my point.

Thanks

JASR76

 

[Latexman]

Can you prove to yourself that the liquid venting and two- phase venting will be over at 10% accumulation or below, before the PSV atarts vapor-only venting for the "steady-state" fire scenario in the 10 to 21 % accumulation portion of the allowance?

Good luck,
Latexman

 

[dcasto]

You must document what scenerios you investigate. MAX FLOW, THERMAL EXPNSION (from a heat source or sunshine), FIRE, and in an heat exchanger a single tube failure (or more if plausable).

zdas04 is correct on blocked flow for most exchangers being the limiting case, but fire is plausable for low flow exchangers and a ruptured tube becomes controlling in chillers and some steam heaters.

 

[JASR76]

LAtexman, casto,

Thank you for your answers.

First of all, the critical case in this safety valve is FIRE, all rest scenarios are very reduced in comparison to this FIRE case.

And, by the way, this exchanger in fire is only an excuse to treat an important case for me. FIRE in vessels FULL of liquid.

How do you analyse them? vapour phase discharge? biphasic?

Perhaps it's a good oportunity to give our opinions over this issue (API suggests something about it but its not very clear)


THANK YOU

 

[dcasto]

I see a fire case as only vapor, if liquids get carried out, it reduces the valve size, so 100% vapor is the most conservative.

 

[Latexman]

dcasto,

Since the liquid phase carries very little enthalpy out of the system compared to the vapor phase, a two-phase relief device is usually larger than a vapor-only relief device for the fire case.

At least that is my experience dealing with aqueous mixtures with surfactants in them.

Good luck,
Latexman

 

[25362]

Visit thread378-141036 with a list of threads on the subject.

 

[dcasto]

Latexman, there is not much energy from the fire removed by the lquid phase, but the amount of expansion and pressure build up by the liquid phase is nothing. Once a fire has started near the vessel, the liquid will expand from thermal and the PSV will relieve liquids at the design mass rate which will empty the the vessel parially as a vapor space takes it place, from that time on, the PSV will only vent vapors.
I don't have the insite on the basis of the standards, but I assume the 121% overpressurre allowed in the fire case is to compensate for the instantaneos opening of the PSV during a sudden fire and there maybe a mixed flow. I've never seen anyone claim there is constant liquid and vapor venting for a fire case in hydrocarbon service.

 

[djack77494]

My approach to this problem is as follows:

Using your example of a liquid filled S&T heat exchanger, it is totally credible to imagine the exchanger (shell side) to be blocked in (e.g. for maintenance) and for a fire to occur. I do not assume that the exchanger or vessel will always be immediately drained.

I like to locate the PSV at/near a high point in the system. Then it is reasonable to expect only vapors arriving at the PSV inlet.

The heat input first warms and then vaporizes the trapped liquid contents. So the volume to be relieved is due to vaporization of liquid (except for a very small amount of fluid expansion that initially occurs). Sizing is then based on the calculated heat input divided by the latent heat at relieving conditions. For mixtures, over time the temperature may increase and the latent heat decrease. Hopefully, the effect will not be very noticable, else things may deteriorate badly.

 

[Latexman]

dcasto,

My experience is not in hydrocarbon service, so this may be why we are at odds here.

In a conservative fire scenario design, the PSV removes more heat than the fire puts in, right? Any answer, except yes or [Δ]H_PSV = [Δ]H_fire, does not meet Code, so I assume we are in agreement so far.

Now, let’s exaggerate what you said to try to show my point. Since “there is not much energy from the fire removed by the lquid phase”, let’s say that no energy is removed relative to that removed by vapor-only venting. Therefore in the steady-state, an infinite, or very large, flow of liquid out the PSV is needed to remove more heat than the fire puts in, right? This requires an infinitely large, or very big, PSV. The same would be true of a two-phase mixture of almost all water and nearly no steam.

Is this realistic? No, but consider a real experiment documented in The Design Institute for Emergency Relief Systems (DIERS) Project Manual. A tank was 95% filled with water at 150^o C under it’s own vapor pressure of 58.5 psig. A 2” full port ball valve is opened quickly. About 28% of the tank contents vented. The experiment was repeated, except 1000 ppm of a liquid household detergent was added. About 96% of the tank contents vented.

If a PSV had to remove the same amount of heat/unit time in the steady-state (like in a fire scenario) in both experiments above, which would require the largest PSV, the one that 28% went through or the one 96% went through?

Basically, most of the heat is carried away by the vapor in a two-phase mixture and the liquid portion goes along for the ride, therefore a two-phase PSV is usually larger than a vapor-only venting PSV.


Good luck,
Latexman

 

[dcasto]

Hydrocarbons have a 150 BTU/lb heat of vaporization and are always at bubble point in process, so the first 150 BTU's make 1 lb of vapor, so at least 10 cf of vapor. Water, I'd go along with what you state. To understand where I operate, take a boiler, making steam and the drum is 95% full and the fuel gas ruptures and the heat input tripples, will the PSV ever see liquids or just steam.

 

[Latexman]

No idea, but if it had 0.1 % of soap in it you'd have a roof full of shaving cream!

Good luck,
Latexman

 

[djack77494]

dcasto,
Through optimal location of the PSV, it should see only vapor unless your vessel is really TOTALLY full of liquid at the start. Most vessels and tanks would have a vapor space incorporated into the vessel. The PSV inlet should originate in that vapor space. I fail to see any concern about 2 phase flow. Unless your vapor velocity were very high indeed, no appreciable liquid would be entrained with the vapor. So design as suggested and don't worry about a liquid phase in the relief valve (IF THE ABOVE ASSUMPTIONS CAN BE REALIZED). There of course will be liquid in the lower portion of the vessel, but not at the relief valve.
Doug