5 replies to this topic

[Wendy]

Hello Everyone,

I am very confused now about PSV fire sizing calculations, hope people here in the forum could please shed some lights on it.

We have a PSV before a compressor scrubber that needs to be sized for the case when the scrubber is on fire. First of all, I calculated the relieving temperature per the API 521 method using the ideal gas equation but the result was unreasonably high - higher than the API recommended carbon steel max wall temperature (1100 degF). I know there have been people debating this method of calculating the relieving temperature, people found it unreliable most of the time. I have then taken the two phase goes into the scrubber and using a process simulator (VMGSim) to obtain the T at relieving P and vapour pressure =1 at constant volume.

When vapour starts to generate from the heat absorbed by the vessel exposed to fire, the latent heat of vaporisation varies and I was going to use the suggested differential vaporization method by Mr. Leckner in another post (get the heat required to vaporise just 1% of the mixture and divide by the amount of material vaporised to get the latent heat, and use the properties of the vapour to determine the relief valve size. and continue this procedure again and again till the entire mixture is vaporised). What I get stuck is again the relieving temperature. I put the scrubber liquid stream under normal operating conditions through a heater and set the vapour fraction of the outlet stream to 0.01 and calculated the latent heat of vaporisation. The temperature simulated was 183 degC, which means 1% of the mixture will be vaporised at this temperature. Once the vapour starts to mix with the scrubber vapour the relieving conditions will change. What I normally do with PSV fire sizing is to calculate the relief area required for heat absorption from the wetted and unwetted surface separately, and add them up for the total area required. I have got two different temperatures from the methods described above, should I use two different relieving temps? I used to use just one relieving temperature calculated from the Ideal gas equation and if its more than the max carbon steel wall temp then use 100 degC to be conservative.

I just want to quickly run through what I set up for the separate calculations.
Unwetted surface:
Following API 521 4th Edition, calculate the relief load directly from 3.14.2.1.2 equation (8), and calculate the effective discharge area of the valve from equation (5)
Wetted surface:
Calculate the extra vapour generated from the latent heat of vaporization.
Check if the flow is critical by assuming a constant back pressure
Calculate the relief area required from API 520 6th Edition, 4.3.2 Sizing for critical/non-critical flow.

The sum of these two areas = relief area required.

I am now thinking maybe I should sum the relief loads, and calculate the required relief area based on API520. How do I then determine the relieving temperature?

Sorry for the long post. I couldn't get my head around it now. would really appreciate if someone could throw some suggestions here.

Thank you all in advance

Cheers
Wendy

[pleckner]

If the vessel has any liquid in it at all you don't need to consider the non-wetted surface. API recommends you use UP TO 25 feet in vertical length for the wetted-surface area (datum is taken at the place where a fire can be sustanied. Saying this another way, where the fuel can pool.) If your liquid level is less than 25 feet then so-be-it. You don't need to consider any additional height. The rationale for this is that the liquid will absorb all the heat from the fire. If you don't believe this then try the unbelieveable experiment of filling a paper cup with some water and putting a flame to the bottom. The cup will not burn until all the liquid is boiled off!

One of the big problems in fire scenarios is calculating a "real" relief temperature at the relieving pressure. We often come up with extremely high temperatures! What to do? That's the fallacy of many fire scenarios, the vessel will be destroyed way before it actually relieves or the piping around the vessel will give way (flange gaskets melt and leak); ditto for instrument connections on the vessel. But we are required to put relief valves on the vessel if fire is credible so just take the relief temperature you calculate and go with it.

For multi-component mixtures, the latent heat will indeed change and so will the relief temperature and the vapor properties. Doing the calculation as you propose is actually the correct way because the PSV area should go through some maximum point as you continue the boiling process until you start reaching the significant heavies, if there are any.

[Wendy]

Thanks a lot for the explaination, Phil.

I am hoping if you could clarify a bit more the wetted surface calculations. I understand what you said the liquid will absorb all the heat from the fire until its 100% vaporised. and during the vaporisation process, PSV will experience a maximum relief area. I normally take the LSHH as the liquid level for the worst case.

When a vessel is on fire, before the PSV lifts the system undergoes a constant volume process and once the PSV lifts, the system becomes a constant pressure process. What I do not understand is when the PSV lifts, the vapour relieved would still be a mixture of the vapour generated from the liquid plus the vapour that was in the vessel. If we consider the wetted surface on its own and the vapour generated from the heat absorbed only, we haven't taken into account the existing vapour in the process?

The other thing is when the liquid contained is mainly water, the latent heat of vaporisation is much more than HCs, the vapour generated from the heat absorbed would be much less as a result. If this is the case, the worst scenario would be when the vessel is dry or the liquid is at LSLL. Please correct me if I am wrong - for fire sizing calculations on multicomponent, if there is liquid in the vessel, we do not consider the non-wetted surface, just do a differential vaporisation of the mixture. Get the heat required to vapourise 1% through to 100% of the mixture and get the latent heat of vaporisation. Use the properties of the vapour generated and the vaporisation temperature to calculate the relief area. However, this maximum area is to be compared with the calculation for the non-wetted surface (API 521).

Thanks in advance and I am looking forward to some hints.

[pleckner]

When a vessel is on fire, before the PSV lifts the system undergoes a constant volume process and once the PSV lifts, the system becomes a constant pressure process. What I do not understand is when the PSV lifts, the vapour relieved would still be a mixture of the vapour generated from the liquid plus the vapour that was in the vessel. If we consider the wetted surface on its own and the vapour generated from the heat absorbed only, we haven't taken into account the existing vapour in the process?

As the liquid boils it "mixes" with the existing vapor until the pressure in the vessel reaches set pressure of the PSV. It is not common practice to consider the vapor space prior to boiling for good reason as it really doesn't enter the picture. The non-wetted surface isn't getting overly hot because the liquid is absorbing all the heat. Once the PSV opens, it is the rate of boiling that governs the rate of relief.

Your last paragraph is stated correctly except you don't have to worry about the vessel once the liquid has boiled out. If the fire hasn't been put out by then the system will rapidly fail. The whole idea of this calculation is to "control" the rate of release of the energy stored in the liquid as it is boiled. If all the liquid is boiled off, there is no vapor and the vessel will collapse, shrivel up and just die. So ignore the non-wetted surface calculation in this particular scenario.

[Wendy]

Thanks for the reply again.

One more question if you bear with me. When do we perform calculation on the non-wetted surface? API 521 states 'unwetted wall vessels are those in which the internal walls are exposed to a gas, vapour, or super-critical fluid, or are internally insulated regardless of the contained fluids'

When you mentioned the non-wetted surface calculation can be ignored in my case, I do not think I fully understand the statement. We should still consider and calculate the non-wetted surface if the scrubber/knockout run dry is a credible scenario? The PSV has to be sized for its worst scenario and that would either be controlling the vapour rate from the heat absorbed by the liquid as you indicated or relieving the vapour in the system (if no liquid)?

I am still a little confused. Sorry to keep bugging you

Thanks

[pleckner]

First, no bother at all. Questions and answers are what we are here for.

If you have a credible scenario where a vessel is in full vapor service then that is when you would apply the API calculation for non-wetted surface (fire). It becomes just another sizing exercise to compare with other credible scenarios when determining what is controlling. The discussion up to now has been about a vessel that is (partially) liquid full.