5 replies to this topic

[racerX]

I have a problem I am working on, brief description is as follows

Basically the facility has a vent scrubber that is sized to knockout all droplets above 900 microns in diameter, however API Std 521 says that it should knockout droplets uptill 300 to 600 microns. Now the gas from the vent scrubber is vented to the atmosphere ( not flared)

what I have to show is that this insufficient vent scrubber size does not pose a hazard, ie liquid pool accumulating on the ground

the procedure i have folloed is

1) calculated the liquid coming into the vent scrubber during relief scenario ( due to low temperature condensation due to venting from PSV)

2) used a normal distribution for finding the liquid droplet distribution ( with mean = 900, variance = 200, both random as no specific guideline to chose the mean or variance)

3)now all droplets above 900 will be knocked out and below will vent through atmosphere ( mostly higher alkanes in composition, dodecane, undecane etc)

4) now how do I find out how these droplets will flow in the air , i.e. thier distribution on the ground below ( spread) at avg. wind speed conditions



Any help is appreciated

[Zauberberg]

Not sure if this can be calculated manually but there is a software package called FRED that is used for dispersion modeling.

As far as separator/scrubber efficiency is concerned, a recommendation is to make a field test and do the following - if the conditions/design allow you to do so:

1. Calculated the amount of liquids entering the scrubber
2. Use some kind of enclosed pot or bucket to capture the liquid phase flowing out from the bottom outlet of the vessel. The difference in these two quantities is the overhead liquid carryover.

We have done similar field measurements in the plant in Africa, where the target was to calculate the amount of liquid carryover from the K.O. drum upstream of Mol Sieve beds. We ended up with some 2% carryover. And make sure to evaluate all possible hazards in a proper way before you actually perform the field test.

[racerX]

Zauberberg, the problem is that its the full relief scenario, so a field test is not possible, however for calculations I am taking the conservative route and assuming maximum liquid possible,

The main issue is how to determine the distribution of the liquid condensed into different droplet sizes as the ones below 900 micrometer will be venting out, which will help to quantify what amount of liquid is going out, and for that droplet distribution is essential

[kkala]

Problem is understood to be the distribution of droplet "concentration" on the ground, even though rate of liquid going out of the scrubber is not known, nor its droplet size distribution. A full rational estimate based on assumed realistic data may be very difficult, but one could try to estimate boundaries. For instance:
1. Max liquid carry over to atmosphere is possible to estimate for the case of full relief, at least as understood from the post (if not, consult scrubber manufacturer).
2. The two boundaries (extreme cases) might be: (α) all liquid carry over has uniform particle size of 900 μm (β)all carry over has uniform particle size of 100 μm.
3. Gas dispersion models give satisfactory results for particles lower than 100 μm, so apply one of them to estimate ground level concentration for case (β). This according to a Greek student book "Risk analysis" by M. Assael (2008), reporting Lees (2003) as reference.
4. Ground level concentration for 900 μm particles might be based on Lees (2003) according to mentioned book. It seems to use Stokes low for precipitation velocity. Ground level concentrations resulting from gas dispersion models (results according to Atmospheric stability, A to F per Pasquill) are multiplied by a correction factor (below 1.0 ??) depending on source height and distance of point considered. Applicable upper size limit of droplets is not reported.
5. Original source is expected to give more clarifications, which is: Lees F.P., 2003, "Loss Prevention in the Process Industries", Elsevier Science Ltd, Oxford.
6. If above has been managed, distribution assuming uniform particle size of (say) 300 μm could be also tried, to see results for an intermediate size too.
Hope it gives some help.
It is supposed that in the long term discharge will be directed to flare, since legislation does not (or will not) accept hydrocarbon emissions to atmosphere.

Edited by kkala, 08 July 2010 - 04:56 AM.

[racerX]

Thanks kkala for the inputs
I was wondering that if I were to suppose a uniform distribution of all the liquid droplets to be 900 micrimeters or anything below it , then the knockout drum would be rendered totally useless, ( designed for droplets above 900)

Now considering the different droplet sizes I find out how far and least they will fall out on the ground, i.e. it will give a range on the ground where the falling droplets will collect, Is there any way to quantify or give an idea what kind of potential that liquid has to start a pool fire, considering that we know the liquid composition

[kkala]

I was wondering that if I were to suppose a uniform distribution of all the liquid droplets to be 900 micrimeters or anything below it , then the knockout drum would be rendered totally useless, ( designed for droplets above 900)
Now considering the different droplet sizes I find out how far and least they will fall out on the ground, i.e. it will give a range on the ground where the falling droplets will collect, Is there any way to quantify or give an idea what kind of potential that liquid has to start a pool fire, considering that we know the liquid composition

Assuming no settling in knock out drum may be too conservative, former idea (for upper boundary) concerned carry over downstream the drum. This could be supposed of uniform droplets of 900 μm in one case.
Droplets will have a size distribution and drum will roughly retain those of above 900 μm size. This distribution is not known, but there must be a way to estimate quantity kept in the drum and quantity entrained, even by direct measurements and extrapolation.
Concerning the probable events, pool fire needs a small "lake" of fuel to be developed, so it could occur only in knock out drum rupture. As droplets are distributed in air from the stack, probable events seem to be either flash fire or vapor explosion. Both are developed in regions of space where gaseous fuel concentrations are between low and high flammability limits. But this supposes vaporization of fuel droplets, at least partially, making the model rather complex. Or the droplets can fall on the ground and be vaporized there. There may be models predicting degree of vaporization, but there are already a lot of uncertainties. It seems difficult to have a meaningful result.