5 replies to this topic

[jmcate]

I'm calculating PSVs for a Shell and tube heat exchanger. Fire case is the sizing case for PSVs on both sides. Shell side saturation temperature is higher than the one on the tube side for the relieving conditions. Then, I estimate the amount of heat that could be absorbed by the tubes, and it's bigger than the heat received by the shell from the fire. So, it seems I should transfer the entire heat from shell to tubes.

So to calculate relieving rates I'm doing:

Tubes: (Qhead+Qshell)/Heat of vaporization(tubes fluid).

Shell: Qshell/Heat of vaporization(shell fluid) I'm not taking credit for the heat absorbed by the tubes.

 

Do you think this approach is acceptable? Is it too conservative?

Thanks in advance for the help.

[fallah]

jmcate,

 

With limited info provided the approch seems to be acceptable and not so conservative, provided that both sides have been isolated when the fire case gettting started...

[aroon]

Jmcate,

 

What is the basis of heat absorption calculation from shell fluid to tube fluid? If you are using normal heat exchanger formula (UA*LMTD) then it is not correct approach. Because in the event of fire, we have to consider stagnant inventory inside any equipment due isolation. As both the inventories are stagnant, U is no more valid and you will have only heat transfer due to conduction, which shouldn't be significant.

 

Hope this is useful for you.

 

-Aroon

[jmcate]

Hi, Fallah, thanks.

Aroon, to do the estimation, I tried with the Heat Exchanger software with a very small flow to have an idea of heat transfer coefficients. With these numbers, the impression is that in case of fire the tubes could absorb enough heat so overpressure on the shell side is not enough to open the PSV. But then, the inventory on the liquid side (tubes) would decrease rapidly, and the tubes will absorb each time less and less, requiring the PSV on the shell side). That's why I went for this approach, to take into account worst scenarios on both sides.

I can estimate with more precision heat transfer coefficients for standing still liquid on the tubes and condensation on the outside of the tubes, but don't you think I should be conservative and consider everything can be transfered?

Still reasonable? 

[aroon]

Relief valve sizing should always be conservative but shouldn't be unrealistic. If you go with your approach then it is too ... too conservative and you will have your PSV highly oversized. In actual operation, it will give other problems like chattering, which eventually can lead to failure of relief valve/seating surface. I would suggest you to go with realistic approach instead of thinking much haphazard. 

[jmcate]

Thanks, aroon. I will follow your advice.