Liquid full vessel - relieving temp? 5

[GWright303]

I would appreciate comments on the following:

I am examining an existing emergency vent, (on a funda filter), to ensure its capacity is suitable for a fire case. The existing emergency vent line is 3", vents to a dump tank, and the Bursting Disc is set at 4 barg. This work is a HAZOP review action.

I have surveyed the existing line (length, bends etc)such that I can compare its capacity vs the calculated relief rate.

I have identified that the worst possible case for a fire to occur, would be during cleaning, when the filter is completely full of acetone.

Using API methods, and the equipment GA drawing, I have calculated the heat absorption as a result of the fire.

As the vessel (filter) is liquid full, the initial effect will be thermal expansion. Correspondingly, (in conjunction with the heat absorbtion from before), I have used the hydraulic thermal expansion equation (from API 521) to calculate the initial liquid relief rate, (as API 521 suggests on pg18), due to the liquid swell.

What I have calculated so far is the initial liquid relieving rate only. However, if heat input continues, and the acetone boils, then vapour generation would take over as the determining factor for relief sizing.

The bursting disc will pop at 4 barg. I need to determine what the temperature is at this point, so I can judge whether relief will be liquid only, or two-phase.

My question is - how?

I dont know if I have somehow missed a section, or whatever, (perhaps I am barking up the wrong tree?), but there seems to be zero guidance on this in the API documents. There is guidance re relieving temperatures in the case of vessels which are full of gas ("unwetted") but not for this case.

I would gratefully listen to any advice on this, as well as how I have approached the problem so far.

Thanks for reading.

 

[SeanB]

Well rightly or wrongly the way I was taught to do this is as follows. Input your composition into HYSYS or another simulator. Set your pressure at the PSV set pressure and your vapor fraction to zero and let the simulator calculate the temperature. This is your bubble point temperature - which will be your relieving temperature when you begin to generate vapor.

 

[skearse]

I would check the bubble temp for acetone @ 4 bar, and use that as a basis to start with. But, there is a paper floating around that suggests that even for liquid filled vessels, the final release for a fire case is vapor generation and that the two-phase release period can oftentimes be neglected.

Simpson, L.L. "Fire Exposure of Liquid-Filled Vessels." Paper D-29-190-1, presented at the 29th DIERS Users Group Meeting, Las Vegas, NV, April 29-May 2, 2002.

 

[786392]

Dear, Are we not inviting a 'BELVE'to occur?

Best Regards
Qalander(Chem)

 

[SeanB]

Dear,
You are going to have to explain your statement a little further. How do you think a BLEVE will occur? The PSV will relieve and protect the vessel, but only for a certain amount of time. IF the fire is not extinguished then the vessel will eventually rupture and then a BLEVE can occur. However, I am not sure how your statement is related to the discussion on relieving temperature.

One more follow up on my post above - the more conservative approach is to find the bubble point temperature at the set pressure + 21% allowable overpressure.

 

[GWright303]

Thank you for replying everyone.

Re: possibility of equipment failure - the provision for fire fighting is pretty good, so it is reasonable to expect any fire to be extinguished quickly.

The possibility of solvent (for a pool fire) being able to form is low anyway, due to drain gullys in the area.

(the above has also been considered in my calculations)

If necessary I will use HYSIS to gauge the bubble point temperature, and use these relieving conditions for flash calc (for produced vapour fraction) at relief conditions, as suggested - thanks.

However, points from API docs:

API 521 3.15.3.2 says if the relief system will see liquid (as in this case)then the determining sizing should be equal to the liquid displaced by vapour generation. It also says that "mixed phase conditions are usually neglected during sizing".

API 3.19.3.2 explains how to calculate the vapour load (simple: Q / latent heat). (I think any potential load from flashing should be added to this also?).

So from that interpretation, it would seem the way ahead is to ignore two-phase flow and and proceed to calculate vapour load, and the system would have to cope with the equivalent volume of liquid (displacement).

Why does API suggest mixed phase can be ignored, given vapour could form by flashing liquid, or heat input? Is that statement from before 2-phase methods became established good practice, or is it just considered over-onerous?

Thanks.

 

[786392]

Thanks Everyone including SeanB & OP Since It was a somewhat wild thought.
Sorry for unable to continue discussions as I was commuting from workplace to the residence and no means of communication.

Best Regards
Qalander(Chem)

 

[JoeWong88]

This subject to fluid characteristic i.e. foamy, reactive and construction of the equipment i.e. narrow-flow passage like jacket vessel.

You may be interested in "How Fluid Characteristic affect 2 phase Relief via PSV on Liquid filled Vessel Exposing to External Fire"

JoeWong
Chemical & Process Technology

 

[GWright303]

Many thanks for that interesting link JoeWong.

Accordingly, I have dug out a source regarding determination of flow regime etc and will read up before proceeding.

 

[skearse]

"Why does API suggest mixed phase can be ignored, given vapour could form by flashing liquid, or heat input? Is that statement from before 2-phase methods became established good practice, or is it just considered over-onerous?"

The reason that API gives in their API training course is the reference I cited above. Sorry, I should have mentioned that in my previous post.

 

[MortenA]

Going back to your original question - temperature.

Your asumption is that the vessel is full of liquid and we are only dealing with thermal expansion.

This could either be done the simple way: Thermal expansion of liquid in an enclosed vessel will lead to very rapid pressure increase for an enclosed vessel (could propably happen without a fire if the liquid that was loaded is colder than the room - you may wanna check this option for the OP part of HAZOP - and the temperature will thus not increase much

Og more complicated: Use hysys to calculate the density at initial conditions (remains constant since your volume and trapped mass does not change. Use hysys to solve for fixed pressure of 4 barg and find the temperature at wich the density is equal to your original density.

Be aware that liquid density is not the best estimated properti for many EOS's - so maybe a refernce book value would be better. But since you dont NEED the density as such we can hope that the error remains more or less constant at a fairly narrow P/T range.

Best regards

Morten

 

[chem101]

Hi Engineers:
in API book it is mentioned that for Thermal Relief Valve use 3/4D1 Psv. Can some explain me that do we have to use this size all the time. Even if the Temperature can go up to 500F or 1000F.
Thanks Allot

 

[MortenA]

Just for thermal relief i think its because its always small volumes - but its not true if its a flashing liquid.

Best regards

Morten

 

[SeanB]

Generally I don't bother to calculate the relief load for thermal contingincies because as said above, the load is very small and the 3/4 D 1 valve is more than capable of providing the required relief. However, on long (and large) lines that can be blocked in I usually do the calculation just to be sure.