8 replies to this topic

[_luis]

Dear Experts,

I am currently sizing a relief valve for a liquid-liquid extraction column for fire case. Some key information are listed below:

 

• The LLE column is liquid full. It is a tall tower, with around 30m in T/T length
• The extract (mostly organics) comes out at the tower overhead. Heat of vaporization is around 300 kJ/kg
• The raffinate (mostly aqueous) comes out at the tower bottoms. Heat of vaporization is around 1500 kJ/kg

In the event of a fire, the bottoms portion will mostly be the one experiencing the heat flux from the external fire. However, since the column is liquid full, I imagine the tower to have an immediate heat transfer via convection/conduction from the heavy raffinate to the light extracts, with the overhead extracts being the first to vaporize at the top of the column (after preliminary expansion of course) before any vaporization of the bottoms occur. Thus, I plan to calculate the relief load based on the heat of vaporization of the extract (dHvap = 300kJ/kg).

I'd like to ask if this is a safe method to take, or is it overly conservative? Is there something else that could happen aside from what I imagine.

PS: I have read API 521 but it doesn't have any specific discussion on this type of problem.

Warm regards,
Luis

[fallah]

Luis,

 

Please specify if the column would be isolated at inlets/outlets in the case of a fire occurence...

[_luis]

Fallah,

The two inlets (feed and solvent), and one outlet (raffinate), do not close during fire case (high pressure detection).
The overhead (extract) valve closes at the detection of low pressure at the top of the column (so I expect it opens up at the detection of high pressure in the column).

[fallah]

Luis,

 

You should arrange a simultaneous energy and material balance around the column to be solved for unknown outging line as relieving path through the PSV...                  

[_luis]

Thanks Fallah.

I'll try to calculate based on your suggestion.
Just a background though, this system is made by a licensor so we were not given the simulation files as the contractor - only the relief properties for the specific fire case. Unfortunately, the relief properties that they gave is based only on the bottoms heat of vaporization (1500kJ/kg). They also provided the data based on the top column set pressure (instead of adjusting for the pressure gain across the height of the column). This made me doubt the values given.

[Jiten_process]

_Luis,

 

It would be better if you had attached P&ID snap. Understanding the isolation philosophy is very important and process specific for safety studies. 

 

Anyway, just imagine, the moment fire occurs, your extraction process will no longer be in stady state condition. That means virtually you wont have extract and raffinate in seperate phase. Understand the fire flames are considered only 8m long vertically, so it will continuously boil off whatever remains at bottom be it raffinite or extract or mix feed. 

 

make sure that the heat of vaporization you have considered is at set pressure and find the equilibrium at set pressure condition to find exact property at relieving condition. It is good approach to calculate relieving rate based on top liquid product and bottom liquid product seperately and take a conservative case to firm the sizing. Also, check and study the relieving process incase of fire, as you said, your inlets and outlets are normally not closed in case of fire and hence this would affect the composition/temp of actual stream goint to PRV. See if you can take credit of overhead outlet. But this is strongly as per licensor guidlines. 

 

you said,

" Unfortunately, the relief properties that they gave is based only on the bottoms heat of vaporization (1500kJ/kg). They also provided the data based on the top column set pressure (instead of adjusting for the pressure gain across the height of the column)"

 

 

What do you mean pressure gain across the height of the column? do you mean pressure drop across the column ?? when fire occurs, its vapor (not 100% vapor as liquid entrainment will be there) which is travelling from bottom to top and release through PRV. pressure drop across the column in such cases is not so appreciable that it causes pressure difference from top to bottom. Normally in my experience we assume constant pressure from top to bottom in case of fire. pressure drop and bottom liquid head is important only in case of steady state process. 

[Bobby Strain]

Luis,

      Don't be mislead about the questions of isolation. The whole purpose of the fire relief case is for a blocked in vessel! So don't take credit for any other outflow. And, your basis for latent heat is the correct one. Looks like you did a good job with your analysis.

 

Bobby

[_luis]

Thanks for the responses guys:

It seems there are two perspectives whether to consider an open path or a blocked vessel.

Reviewing API 5.15.2.1:
 

 

It is typically assumed that the vessel is isolated during a fire in order to simplify the analysis, although a more

detailed analysis can be warranted in certain cases. Crediting for alternative relief paths that remain open during

an overpressure event is generally an acceptable practice. However, it should be recognized that operators

and/or emergency responders attempt to isolate certain lines and vessels during a fire condition in order to limit

the fire spread and to safely shutdown the unit. There can also be actuated valves that fail in the closed condition

when exposed to a fire. It can be difficult to establish with a degree of certainty whether a particular line will

indeed remain open under all fire conditions. Further, unless the line is open to atmosphere, consideration should

be given to the potential that the fire-relief flow in the alternative relief path will overpressure other equipment.

Hence, it can be necessary to add the fire-relief load elsewhere. Ultimately, the user shall decide whether a

scenario is credible or not.

 

I guess it is up to me to determine the more conservative path.


Jiten_process,

Yes, pressure drop/pressure gain (depends on how you view it). For this case, the column is pretty tall and is full of liquid, so the effect of the liquid static is evident. But I see your point as to how it should apply for a steady-state or non-steady state case. I can put up a rough sketch next time when I get my hands on the P&ID.

[Jiten_process]

hi _luis

 

 

 

Reviewing API 5.15.2.1:
 

Quote

 

It is typically assumed that the vessel is isolated during a fire in order to simplify the analysis, although a more

detailed analysis can be warranted in certain cases. Crediting for alternative relief paths that remain open during

an overpressure event is generally an acceptable practice. However, it should be recognized that operators

and/or emergency responders attempt to isolate certain lines and vessels during a fire condition in order to limit

the fire spread and to safely shutdown the unit. There can also be actuated valves that fail in the closed condition

when exposed to a fire. It can be difficult to establish with a degree of certainty whether a particular line will

indeed remain open under all fire conditions. Further, unless the line is open to atmosphere, consideration should

be given to the potential that the fire-relief flow in the alternative relief path will overpressure other equipment.

Hence, it can be necessary to add the fire-relief load elsewhere. Ultimately, the user shall decide whether a

scenario is credible or not.

 

I guess it is up to me to determine the more conservative path.

 

yes it is up to you as you the best person to know about the process and make sure to take concurrence of client/consultant/licensor. Specially relief rates of columns are always tricky one as it involves change in equilibrium conditions. 

 

moreover, as I understand from your reply, your column is full of liquid across its length. I have come across so far strippers, absorbers and distillation columns where liquid is accumulated only on bottom part of column except trays or packing hold up. And we follow the practice as i have mentioned in my previous post for those columns. But yes, evaluate the scenario carefully for fire case by running the simulation, that will give u perhaps more picture. 

 

Good luck...