Pressure Relief at heat exchanger shell and tube for fire case 1

[wenchop]

Hello, I have to design a pressure relief valve (PSV) for fire situation for both the shell and the tube side of a heat exchanger. The reason is that we have many trains, and they can be isolated/blocked during routine maintenance, and we are assuming that they will not be drained at all times.

Both tube and shell are liquid full (water and oil). My concern is that I am getting huge valve orifices if I assume that vapour generation will displace an equal volume of liquid. Also, I don't know what to assume about heat flux from fire into the tubes (as they are behind a shell).

How do I go about calculating the relief load for fire for this shell and also for the tubes? I am doing basic engineering and do not need to dive into so much detail at this point.

 

[Latexman]

If the PSVs are on top of the shell and tubes (higher elevation), it is unlikely the vapor will displace much liquid.

Let's discuss what happens from start of fire to end of scenario. The fire starts. Liquid heating happens first. Liquid expands and activates the PSV. Flow will be very low due to thermal expansion of liquid. Next, the liquid may/will start boiling and generating vapor. The high density liquid will seek lower elevations (down in shell and tubes) and the low density vapor will seek higher elevations (up at PSV). There may be some two phase flow for a short while until enough vapor space is created to disengage liquid droplets, but this is usually small. Most people ignore this liquid to vapor transition period. If there are special circumstances that will sustain this two phase flow period, it must be accounted for though. Once liquid droplet disengagement has become efficient, there should be vapor only venting until the heat exchanger runs dry. After that, vapor only venting continues as the vapor expands from the heat of the fire.

For the tube side, the most conservative approach is to use the heat exchanger’s external surface area (shell and heads) as the exposed surface area. If this approach results in an unacceptable Pressure Relief Device size, a more exact approach is to account separately for the direct heat transfer through the heads and the indirect heat transfer through the shell/shell fluid to the tubes. The limiting factor through the shell/shell fluid to the tubes may be either the heat transferred from the fire to the shell side or the heat transferred through the shell fluid to the tube heat transfer area. Evaluate the heat transferred through the shell fluid to the tube with the shell at it's sizing temperature and the tubes at their sizing temperature.

For the shell, just use the heat input due to the area of the shell only (no heads).

This is a quick introduction to what can be a quite complex problem. Seek training! Read up on it! Talk to the experienced Engineers! Let an experienced Engineer peer review your work!

Good luck,
Latexman

 

[poli60]

Hi Latexman.
In my refinery experience, the fire load to the tube side PSV is calculate by only considering the exposed surfaces of the exchanger heads.
And I can add that up to 15 years ago almost no one considered the fire case for the tube side. It can be easily seen in a lot of plants designed in the past.
Regards.

 

[Latexman]

Hi poli60,

Yes, I too remember those days when the area of the heads only determined the tube side relief. For me in the chemical industry I recall it being 20-33 years ago. I started in 1979. But, in the Pressure Relief Device training, certification, and experience I have had the past 15 years, that is no longer acceptable to the three chemical companies I have worked for.

Good luck,
Latexman

 

[wenchop]

Thank you Latexman for your excellent post, it is very helpful. Another point to consider is whether the bubble point of the tube side liquid is less than the bubble point of the shell side liquid. If it is, then there can be another kind of heat flux from shell to tube side.

Again, to all readers, note that in this case the assumption in this case is that the heat exchangers can be blocked while there is a fire incident (therefore a PR valve is probably required). This is usually not the case. In general, administrative procedures should be in place to have drained and depressurized the exchangers before blocking them (mandatory if they do not have PR valve). Caution signs are recommended at the block valves of all exchangers that can be isolated.

Pressure relief design is a safety issue, and just as any important field of engineering, its design should be reviewed by experienced/qualified engineers. I know that my post may seem to be that of a beginner, but I would never practice safety design without guidance and detailed review from more experienced members in my organization.

 

[wenchop]

Depending on the location of the tube side PR valve, vapour generated could displace liquid, causing a liquid load to reach the PR valve inlet. So the PR valves should be located at a high point. The manufacturer should be involved in the design, and PR valves designed for vapour relief should be placed in the correct locations. The tube relief path is more elaborate than the shell relief path.



I have reviewed some information on how to calculate latent heat of vaporization for multi component systems. There is not a clear consensus about the method to be used. Some use a conservative 50 BTU/lbm but this may result in an oversized PR valve that may chatter (hard to avoid in a fire relief valve with multi component system as the latent heat varies so much, so more than one valve may be a better solution).



I found that most people think that lighter hydrocarbons have lower latent heat of vaporization than heavier hydrocarbons, but the opposite is true (at least at low pressures where I checked). Lighter hydrocarbons have lower "boiling" points, so they will flash first, but they have higher latent heat of vaporization.



Also many people think that latent heat of vaporization is not a function of pressure, but it is.



Also, as the flashing progresses, the wetted area is less because of the liquid that has changed phase (if the liquid inlet is blocked).