11 replies to this topic

[Wendy]

Hello Everybody,

I have been asking couple of questions on PSV fire sizing and I found people here are very helpful. I have got one more question wrt PSV fire sizing. Hope someone can shed some lights for me.

PSV fire sizing is required on a vessel containing wellfluids from the wellheads. The PSV set pressure is above the cricondenbar, ie. 600 barg. If it was for a single component fluid, it would be 'supercritical'. Are the procedures/equations under section 5.15.2.2 of API 521 5th Edition (Jan 2007) applicable? Since this section is for vessels containing only gases, vapours or super-critical fluids? If so, would it be appropriate to use the surface area of the entire vessel as the surface area exposed to fire?

I understand two phase relief sizing is not normally required for fire sizing. However, if the PSV lifts the pressure drop across would cause flashing. Is two-phase sizing thus required rather than using section 5.15.2.2?

Many thanks
Wendy

[JoeWong]

Hello Everybody,

I have been asking couple of questions on PSV fire sizing and I found people here are very helpful. I have got one more question wrt PSV fire sizing. Hope someone can shed some lights for me.

Hope you enjoy reading and learning here...

PSV fire sizing is required on a vessel containing wellfluids from the wellheads. The PSV set pressure is above the cricondenbar, ie. 600 barg. If it was for a single component fluid, it would be 'supercritical'. Are the procedures/equations under section 5.15.2.2 of API 521 5th Edition (Jan 2007) applicable?

Since this section is for vessels containing only gases, vapours or super-critical fluids? If so, would it be appropriate to use the surface area of the entire vessel as the surface area exposed to fire?

Please refer to 5.15.2.2.2. It is for vapor and supercritical fluid.

I understand two phase relief sizing is not normally required for fire sizing. However, if the PSV lifts the pressure drop across would cause flashing. Is two-phase sizing thus required rather than using section 5.15.2.2?

First, PSV lift and re-seat more-or-less maintain the pressure at blowdown and maximum accumulated pressure. Thus, not necessary internal pressure will drop.

Regardless of internal pressure drop, there is a potential of 2 phase in case the liquid level is too close to PSV outlet. This critical level can be estimated using DIERS method.

[Wendy]

Thanks Joe,

I am not familiar with the DIERS method, I would be grateful if you could explain a bit more or pass on some references.

What I also do not understand is your last statement on the 2 phase - "Regardless of internal pressure drop, there is a potential of 2 phase in case the liquid level is too close to PSV outlet." The PSV set pressure is above the cricondenbar and the entire fluid is dense phase, which means no distinction can be made between two phases, and no phase transition is observed with temperature. I do not see how location of the PSV matters?

The vessel is spherical having a diameter of 1m.

[rxnarang]

Wendy,

Try this method. It works for all kinds of fluids, supercritical or not. It requires a simulator to calculate the enthalpy at various flashes, and the assumptions are the same as in the DIERS method ( HEM, no slip). See Appendix D of API 520 as reference for the omega method, which, in my opinion, is not as robust as the method proposed in the paper.

http://people.clarks...gn/reliefv2.pdf

See also Ron Darby's article on two phase sizing for PSV orifice.

Yes, you are right on the fluid transitioning to two phase as it moves through the orifice and the pressure drops. The above method takes this into account.

The above is for sizing the orifice.

For calculating the relief load, the standard API 521 formulae for heat transfered and height of flame (25 ft) may be used, but of course, use your engineering judgenment.

Regards
Rajiv

[JoeWong]

I am not familiar with the DIERS method, I would be grateful if you could explain a bit more or pass on some references.

As expected, you will come back with this question.

Let me give you an example. A liquid filled vessel exposing to external fire. initially it is at subcooled condition. It absorbed heat and birng the temperature upto the saturation point (IPB). During this heating process, there is expected thermal hydraulic relief due to liquid expansion.

Now at saturation point, liquid entering the PSV is at boiling liquid and 2 phases relief is expected. Infact, 2 phase relief is expected much early as partially subcooled liquid passing PSV, 2 phase relief will occur in the nozzle when the liquid is expriencing pressure decreases.

2 phase relief via PSV will continue, there will be significant content in the vessel relief via the PSV. The liquid level may drop (maybe ONLY...there are more complication behind which i don't intends to discuss and confuse).

I am not sure if you have heard of Kelvin-helmholtz instability (click for wikipedia quote), as the vapor is relieving through the nozzle with high velocity, there will be a driving force to "agitate" the liquid and liquid tends to leave the liquid surface entrain into vapor. Those 2 phase relief occur in the PSV eventhough there are some gap between the liquid surface and PSV inlet nozzle. Now the difficult part is how to determine the liquid level where you will have complete vapor relief where liquid droplet is too big and cannot reach the PSV inlet nozzle (on vessel).

There are (at least) Churn-turbulent and bubbly-viscous disengagement model to determine the critical level.

Write up to here... I hope you are nervous with your previous job. Don't worry, accorinding to Fisher report, there are more than 50% of PSV in the world are UNDERSIZED and still in service NOW !

I always take Phil advice, if you are have minimum experience in PSV sizing, just make sure some credible guy more experience than you overseeing your work.

One book your may consider is
Emergency Relief System Design Using DIERS Technology (Click HERE to read more)
The Design Institute for Emergency Relief Systems (DIERS) Project Manual

Phil has more experience in this area... Hoping he will provide more input to you.

What I also do not understand is your last statement on the 2 phase - "Regardless of internal pressure drop, there is a potential of 2 phase in case the liquid level is too close to PSV outlet." The PSV set pressure is above the cricondenbar and the entire fluid is dense phase, which means no distinction can be made between two phases, and no phase transition is observed with temperature. I do not see how location of the PSV matters?

My response is to tell you that 2 phase relief is possible in FIRE scenario in general.

You are right if your system is at supercritical region.

The vessel is spherical having a diameter of 1m.

More prompt to 2 phase relief as compare to horizontal vessel.



Rajiv has provided you with a very good article (http://people.clarkson.edu/~wilcox/Design/reliefv2.pdf) for supercritical fluid in fire scenario. This paper use step-wise method. Pretty good but tiem consuming. However it use some simple thermodynamic model. What i suggest to you is model it in the HYSYS to get the properties and size the PSV accordingly. You will see some surprises.

As for the PSV sizing, it use the the critical mass flux model base on enthalpy. Pretty good method and has been used by many people like Simpson, Ron Darby, myself, etc.

[Wendy]

Thanks to you both, Joe and Rajiv, for the great advises and references.

I have sized this PSV following API 521 section 5.15.2.2.2 which is for vessels containing only gases, vapours or super-critical fluids. It resulted in a very small orifice, less than 0.01 in². The calculated relieving temperature is more than the recommended maximum vessel wall temp. and hence I have assumed a low relieving temperature to be conservative. Even with the relieving temperature the same as the operating temperature, the required orifice is just over 0.01 in². Btw, for those of you who are experienced in this field, what would do you do if the calculated relieving temperature using API 521 is more than the vessel wall temp for supercritical fluids? Would you assume a conservative value. What I normally do with other fluids is get the temperature at the relieving pressure using a simulator, but does not quite work with supercritical fluids.

I will set up a calculation following the reference Rajiv provided. Interested to see and compare.

Many thanks
Wendy

[rxnarang]

Wendy,

That means a nominal 1D2 PSV; whcih is typical of expansion cases.

What concerns me is the relieving temperature you computed. If you compute temps is excess of 1100F, then the vessel wall will fail before the PSV relieves. The PSV is of no use .

You now have to use Passive fire protection like fire insulation to prevent vessel failure.

And also use vapor depressuring to reduce stress.

Please read this paper by Scandpower and you can apply the findings to your case.

http://www.scandpowe...xposed_to_Fire/

Regards
Rajiv

[JoeWong]

Wendy,

In my opinion, vessel wall may be good for high temperature BUT there are other instrument, nozzle, flanges, etc attached to the vessel which may NOT good for high temperature...

Apart from Rajiv recommendation on PFP and Depress...there are other protective measures you may consider. Read More in "Protective Measures against FIRE other than Pressure Relief Device (PRD)"

[Wendy]

Thanks to both of you again.

I tend to find relieving temperature is above the maximum recommended wall temperature in most cases if we calculate it from API 5.15.2.2.2 equation 11 (T1=P1/PnxTn). What I normally do with hydrocarbon mixture is to use simulators like VMGSim to obtain the relieving temperature at relieving pressure by setting vapour fraction = 1. This normally results in a much lower relieving temperature. However,I can not use a simulator to do it for dense phase mixture.

I just tend to find the API 521 method using ideal gas equation is far different from simulated results. Do you use the API 521 method to calculate the relieving temperature and justify whether a fire protection on the vessel is required? or there is other ways?

Many thanks
Wendy

[rxnarang]

For a vessel containing liquid I use bubble point as relieving temperature. For a vapor expansion case I tend to rely on the API 521 method as described by you. In case the relieving temps are above failure then other methods are necessary.
Regards
Rajiv

[JoeWong]

I tend to find relieving temperature is above the maximum recommended wall temperature in most cases if we calculate it from API 5.15.2.2.2 equation 11 (T1=P1/PnxTn). What I normally do with hydrocarbon mixture is to use simulators like VMGSim to obtain the relieving temperature at relieving pressure by setting vapour fraction = 1. This normally results in a much lower relieving temperature. However,I can not use a simulator to do it for dense phase mixture.

Please allow me to make a quote from excellent explanation by Mr. Montemayor in other post.

"A Polytropic Process is one in which changes in gas characteristics during compression are considered. This is considering that an Adiabatic Process is one during which there is no heat added to or removed from the system (Q = 0), and Isentropic Process is one wherein the entropy remains constant (Delta S = 0), and an Isothermal Process is one in which there is no change in temperature (Delta T = 0). A Polytropic Compression Process is typical of a dynamic-type compressor – such as a centrifugal compressor. A reciprocating compressor typically follows an Adiabatic Compression Process very closely.

A Polytropic Process is defined in Thermodynamics as an internally reversible process, which conforms to the relation PVⁿ = Constant.

If you plot the compression curves for Isothermal, Adiabatic, and Polytropic compressions on a Pressure versus Volume graph, you will find that the Polytropic curve falls between the other two, but closer to the Adiabatic curve."


In my opinion, all based on PVⁿ. It depends on which path you take...

For API, n=1
For Simulator like HYSYS (compressor model), the default value is n=0.75. I am not sure what figure has been used.

I just tend to find the API 521 method using ideal gas equation is far different from simulated results. Do you use the API 521 method to calculate the relieving temperature and justify whether a fire protection on the vessel is required? or there is other ways?

For first part "Do you use the API 521 method to calculate the relieving temperature ...", it is difficult issue...i reserve to comment now.

Second part "...justify whether a fire protection on the vessel is required? or there is other ways?", API is one of those standard, recommended practices, etc demand for additional fire protection from overpressure & safety perspective, there are other code (e.g. NFPA, GB, etc) and local code which are subject to you project requirement and local requirement.

[cmp74]

Hello Joe,

You mentioned above that you have used the article (above link) for supercritical fluid in fire ase.

Im trying to build an excel spreadsheet for similar case involving a light parraffin.But im stuck at step:8, which is to calculate maximum flux, G .

The author lowers the pressure arbitrarily (which is backpressure) and find sp volume/enthalpy isentropically.Im using P-H diagram for this, but it is very tedious and time consuming.

is there any shorter method for this??Can HYSYS or Aspen be used for it??

An answer is greatly appreciated.

regards,
C