Fire case for safety valve on 3 phase separator 1

[Guidoo]

A amine flash drum is a 3-phase separator that separates a stream into water, oil and off-gas. The "oil" could contain C4s and C5s. When calculating the relief flow for the fire case, I find a very large flow when using heat of vaporisation of n-butane. Since vessel is mainly filled with water, it is my assessment that relief valve could be much smaller. Current calculation assumes that all heat input from the fire is used to vaporize n-butane, while in reality much of this heat will be used to further heat-up the water instead (no equilibrium).
In other words, the current calculation is too conservative. Problem is how to convince authorities etc. that a smaller relief valve would be acceptable, and what should then be the basis for the calculation. Any suggestions?

 

[TD2K]

First, conservative is good for safety valves.

That said, if you feel the HC vaporization is not the sizing case, you are going to need to be able to show that by the time all the light ends have evaporated (due to the heat input from the fire), the vessel will be under set-point and the only way to further increase pressure is to vaporize water.

Don't get yourself in a situation where you are not adequately protected.

 

[Rosalynn]

I recently tackled a similar problem. Mine was a light hydrocarbon condensate mixed with heavy bitumen (big boiling point gap between the two), lots of water, and the vapour phase. I ran into similar concerns. After discussion with two more senior colleagues, this is what we did:
- We calculated the wetted surface area of the vessel for the hc phase and for the water phase separately, and calculated the two corresponding fire case heat inputs. This required assuming that the water phase and the hc phase did not mix during heating. Don't know if it would apply for your case, but we figured it was ok in our case (relative viscosities, relative densities, etc.). We felt that it was reasonable to assume that the vessel area that the water sees would heat the water phase, while the vessel area that the hydrocarbon sees would heat the hydrocarbon--but only if they didn't mix appreciably.
- We looked at the boiling points of our "dominant components". We used hexane to approximate the hydrocarbon phase (quite conservative in our case, due to intermixing with the heavy bitumen) and water as our dominant components.
- Water and hexane were far apart in boiling points at relieving pressure that the majority of the hexane would flash off before much of the water did. We calculated this relieving rate, then we calculated the water vapour relief rate.

In our case, based on our relative vessel areas, the hexane relieving rate governed. The calculated relieving area for the PSV was less than if the whole liquid volume had been occupied by hexane (again, the exposed areas come into play). This was the PSV area we went with. Sort of makes sense, eh?

Hope this helps. In any event, it'll provide food for thought. (We would have loved to simulate this using dynamic Hysim modelling, but didn't have access at the time.)