Relief Valve Fire Sizing - Pressure Vessel 4

[butelja]

In evaluating the pressure relieving requirements for a pressure vessel in the fire sizing case (UG-133(b)), the pressure is allowed to 21% above the vessel MAWP. There is no mention of vessel temperature, although Appendix M-14(a) suggests it should be considered.

If a high boiling point liquid is in the vessel, the case may arrise where the design temperature is exceeded well before the MAWP is reached. In this case, which of the following is correct?

1.) The vessel must stay below the design temperature and below 121% of the MAWP when relieving.

2.) The vessel must stay below 121% of the MAWP, and temperature is not considered.

3.) The vessel MAWP is derated for temperature, and the pressure must stay below 121% of the derated MAWP at the elevated temperature?

 

[sanjay]

how emergency venting is decided on fixed roof tank

 

[MJCronin]

This "vessel temperature" topic was dicussed and evaluated in an article published in "Chemical Engineering"last year.

As I recall, the author of the article (a PHd as I recall)suggested that the cooling effects of the liquid are considerable and should be considered in the vessel strength/RV settpoint calculations.

Contact Chemical Engineering magazine and request a reprint..... perhaps...???

Good Luck !!!................................MJC

 

[kstaylor]

Be careful! With vessels containing high boiling point liquids in a fire relief case the yield strength of the vessel shell material may be exceeded before any appreciable boiling begins. As we all remember from thermodynamics, the boiling point will be elevated with increasing pressure. With no boiling occuring, all the heat from flame impingement just goes into the vessel and raises the temperature of the steel as well as that of the contents. Vessel rupture is a very real possibility in these cases.

If you determine that the yield strength of the vessel material will be exceeded before the contents begins to boil and generate vapor to be relieved you must protect the vessel from becoming overheated. This is usually done with either fireproof insulation & jacketing (i.e. stainless steel jacketing, not aluminum like is normally used) or with some sort of deluge system.

See API-520 & API-521 (available from

http://www.api.org)

for further guidance. Good Luck!

 

[varunpant]

Dear Butelja,

Generally, the API generates the formulas / correlations by simulating the worst scenario, and there is inherently a factor of safety considered, but, as you said, there can be situations that the yield strength of the vessel shell material may be exceeded in case no precaution is taken!
The point is, the onus is on the design engineer that such a situation does not arise. YOU must ensure that the vessel does not fail. Please consider any of the following while designing for a vessel exposed to fire (I am pulling it straight out of API, so no credit to me ;-)


1 Provide excellent drainage, and do not allow accumulation of any flammable material for a substantial period of time.
2 Provide External insulation to limit the heat transfer
3 Provide earth-covered storage.
4 Provide jacketing.
5 Limit the Fire areas with diversion walls
6 Last but not the least, in case you anticipate any major hazard, make suitable arrangements for depressurization of the vessel.
Hope this helps,
Best regards

 

[TD2K]

Sanjay, for determining emergency venting of fixed roof tanks, look at API 2000.

 

How to size a depressurizing valve?

 

[TD2K]

YS-1, Fisher will do this for you. Also, Pro/II and Hysys has depressuring modules built into their simulation programs.

I've also done it on a spreadsheet. Start off with the system volume to be depressured at the initial pressure and temperature (and mass). Calculate the flow through the valve and piping based on the valve Cv or Cg. After a few minutes, calculate the mass that has passed through the valve and recalculate the inventory in the system. Use the new mass in the system to adjust the pressure. Continue in steps until you reach the final pressure you want. You then can try different sized valves till you depressure the system in the required time.

Depending on the initial conditions, you may need to include some terms to adjust for the change in gas compressibility and/or temperature in the spreadsheet as the system depressures.

 

[rusman]

YS-1.

A restriction orifice (RO) is normally be installed to limit the flow, instead of the depressurisation valve. In this case, the RO will be installed together with the blowdown valve (which will be actuated automatically upon detection of fire, or during ESD system).

As mention by TD2K, both process simulation of PRO/II and HYSYS have its own depressurisation module to determine the maximum relieving rate and hence you can calculate the required orifice size taking into account the flow is critical. Please refer to API RP 520 Part 1 in determining the orifice size formulae for the critical flow regime.

A normal spreadsheet can be develop taking into account the isothermal behaviour of the gas system (but bear in mind that the depressurisation process within the vessel is near isentropic, say 60 to 70 % isentropic, while the process across the orifice plate is adiabatic). Grote has give some guideline on how to calculate the blowdown rate). Pls let me know if you need a copy of that paper and the spreadsheet.

Thanks

rusman

 

[ys1]

My message is to Mr. Rusman.
Can you please give the guideline of Grote & spreadsheet you have mentioned. Basically in the column for which the depressuring valve is to be provided, there is no liquid. Column contains molecular sieves through which ethylene gas will be flowing.

Regards

 

[rusman]

What's your operating pressure and the initial depressurisation pressure of your column?

Thanks

 

[ys1]

My message is again to Mr. Rusman.
The operating pressure is 15 kg/cm2g. Operating temp is 25 deg C. Vapor is ethylene with carbon dioxide which will be adsorbed by molecular sieves. Vapor flow rate is 40000 kg/hr. What does initial depressurisation pressure mean?

Regards

 

[gunnarhole]

Dear TD2K,

I also use the spreadsheet approach you describe above. One interesting thing I have noticed is that if you use the geometric mean of the starting and final pressures in the valve equations (P1 x P2)^0.5 you get an "effective mean" flowrate. This eliminates the need to iterate, which is good enough for many cases.

Regards,

Gunnar

 

[ys1]

My message is to TD2K.

Could I have a sample calculation for a column containing only vapor. Somehow I'm still confused & unable to proceed.

Regards

 

[TD2K]

YS1, just came across your request. Drop me an email at testdog2000@yahoo.com for a copy of the spreadsheet.

 

[Samiran]

TD2K,

I've not seen your spreadsheet but I assume you account for the drop in pressure which will result in a gradual reduction in depressurisation flow. What I have done in my spreadsheet is to do a valve rating calculation in every iteration which ensures the installed Cv value of the valve is never exceeded. It reduces the flow thru the valve to match the valve Cv. What I am unable to do is to predict the drop in temp with the depressurisation ? Any hints?

Samiran

 

[TD2K]

I haven't refined the spreadsheet to that extent Samiran. One could simulate the gas expanding (adiabatically?) inside the vessel and develop a correlation for the temperature change with pressure and then put that into the spreadsheet.

On the other hand, if you don't include the temperature effect (and it isn't enough to create problems with hydrates, freezing, metallurgy, etc), the density you will get with a constant temperature is lower than it will be if you included the temperature decrease of the gas, thus the flow rate through the valve will be lower than actual and you'll have a conservative sized depressuring valve.

 

[Gilzow]

While all of the topics above are indeed interesting, I have a new but related question. I work for a firm that sizes relief devices, and I was wondering if there was some published literature that gives the mathematics on how to predict the rate of boiling with respect to time? I realize that you will have liquid heating ((Q/m*Cp)+T1=T2, until the temperature reaches the bubble point at relief pressure. What happens when it reaches the bubble point? I realize to get the maximum vaporization rate, I would divide the heat duty by the latent heat of vaporization, but how do you determine how long it takes to get to this maximum boiling rate, i.e., does it go through boiling regimes. I feel that it will start boiling slowly and eventually reach a maximum boiling rate, but does this only occur due to a changing heat flux, or does it occur due to incomplete mixing. If anyone has any literature or has some comments, I would love to hear about them.
Thanks,
G

 

[grt12]

Gilzow,
To do what you are asking we use a simulation in HYSYS which applies the heat input to a series of flash vessels and over time the temeperature rises and boiling is predicted. By using this approach we get a better estimate of the relief temperature and a good idea of how the mixture in the vessel will boil with time.

 

[Gilzow]

Thanks grt12, but I am trying to write my own program to predict how the relief load will increase over time as boiling increases from pool boiling to nucleate boiling. I have a physical properties package that I can access in order to run a series of flashes. This will allow me to predict the bubble point for a range of pressures.

My problem is that API 520 only allows for constant heat flux equations and is only concerned with the maximum (vapor) relief rate (Q/latent heat of vap). (I am not trying to model 2-phase flow with respect to time; yet.)

The way I believe heating of a liquid and how the boiling rate increases is as follows: (of a single component)

1. Heat goes into the fluid to heat up the liquid to the bubble point and the temperature rise is not uniform throughout the vessel. The liquid that stays in contact with the sides of the vessel the longest retains more energy and reaches the bubble point first.

2. Once some of the molecules near the sides of the vessel reach the bubble point, some to the heat input goes into boiling the liquid and some of the heat input goes into heating the other molecules that are not yet at the saturation temperature.

3. Eventually all of the liquid reaches the bubble point and all of the heat input goes into vaporizing the liquid, i.e. the maximum relief rate.

My question is, does anyone know of how I can model this mathematically. If there is any literature out there I would love to hear about it.
Thanks,
G