12 replies to this topic

[curious_cat]

What is the general strategy for pilot plant testing of distillation? i.e. Of course, the pilot plant column can be considerably smaller in diameter than the production column but is there any way to scale down height as well?

 

e.g.  Let's consider a packed distillation column. The HETP of lab / pilot units seems approximately the same as the larger industrial column so (say) your target column needs 80 theoretical stages, the same height is needed for the pilot column as the final production column?

 

Am I missing the obvious? How does one validate distillation at pilot plant?

 

PS. Regarding the need to validate at pilot-plant: The components are not well validated petrochemicals with good VLE data. This is a fine chemical distillation with multiple impurities which are not well characterized for VLE. 

 

Ergo, there is a simulation for number of stages needed but we have limited confidence in it. 

[breizh]

Curious ,

Why don't you ask a lab to perform the pilot  tests ? Generally they have all the equipments including the analytical part .

my 2 cents .

Breizh

[Pilesar]

Validate your model with the pilot plant. Once the model is validated, the model can be used with confidence. With fewer stages, more reflux is needed to make the separation. Beware that there is a minimum number of stages below which the separation specifications cannot be reached. If the pilot plant distillation succeeds, then there should be an evaluation of risk when major parameters are modified for scale up.

[PingPong]

Breizh advise is worth considering.

 

Otherwise:

Let's consider a packed distillation column. The HETP of lab / pilot units seems approximately the same as the larger industrial column so (say) your target column needs 80 theoretical stages, the same height is needed for the pilot column as the final production column?

If you would use the same packing and the same reflux ratio (to get the same separation) as in the industrial column, then you would need the same packing height.

 

You can however split the column in multiple columns in series to limit the height as well as avoid high L/D ratios that would give support problems.

 

You could also decide to use a different packing type for the pilot column, for example use Sulzer DX or EX laboratory packing instead of BX and take into account the theoretical HETP ratio of BX versus DX or EX according to Sulzer, and apply a safety factor, just to make sure.

 

 

Note that HETP values that vendors report for their packings are based on a standard test fluid, usually cyclohexane/n-heptane mix, and possibly at total reflux. Those HETP's may not be realistic for your fluid mix, especially not if your fluid is more viscous than the test fluid.

[curious_cat]

Curious ,

Why don't you ask a lab to perform the pilot  tests ? Generally they have all the equipments including the analytical part .

my 2 cents .

Breizh

 

@Breizh 

 

Great idea. But I am curious to know what these tests will be. 

 

Let me explain better: Say one studies a reaction etc. one can do a pilot test on a smaller volume scale & then scale up to plant scale. i.e. The cost is lower because of the  smaller volume scale

 

In distillation I am trying to figure what this scaling dimension. For sure the diameter i.e. I can test a pilot with a 6 inch dia column instead of a massive dia production column. 

 

But what I'm trying to figure is whether there's some way to scale down along the height of column dimension as well. Or does the pilot essentially have to try out with almost the same height dimension (since the HETP does not seem to be much improved for a pilot column). 

[curious_cat]

@Breizh

 

Or do you mean let a lab generate very reliable VLE data. And then just depend on simulation runs that use this reliable VLE data. Thereby skipping the pilot scale distillation column entirely? 

[curious_cat]

@PingPong

 

Thanks for those suggestions! I totally missed out on noticing that the DX / EX lab packings seem to have a much lower HETP than even  the best at-scale structured packings like BX and CY.

 

From the Sulzer plots it looks like one can pack in 5x more theoretical stages at the lower loadings and 2x more at the highest loadings. 

 

If that is  correct then it means one can pilot a distillation column in approx. one fifth to one half the height of the final distillation column? That would offer a 1:5 scale up ratio. Not amazing but still better than nothing eh? 

 

Of course, there's the additional advantage of diameter scaling. 

[curious_cat]

 

You can however split the column in multiple columns in series to limit the height as well as avoid high L/D ratios that would give support problems.

 

Do you just pipe the vapor from the top of the (n-1) column to the bottom of column-n and pump liquid backwards from the bottom of the nth column to the top of the (n-1)th column?

[Bobby Strain]

You will probably find that you can have a competent laboratory determine VLE data which you can then use in a simulation. And much lest costly.

 

Bobby

[PingPong]

Do you just pipe the vapor from the top of the (n-1) column to the bottom of column-n and pump liquid backwards from the bottom of the nth column to the top of the (n-1)th column?

Yes, and the 1^st column has a reboiler and the n^th column has the condensor.

 

In the industry this is also sometimes done. For example: a propane/propylene splitter (C3 splitter) needs many stages and is often built as two columns in series.

[curious_cat]

You will probably find that you can have a competent laboratory determine VLE data which you can then use in a simulation. And much lest costly.

 

Bobby

 

Thanks Bobby! That makes sense. So you just eliminate the pilot-plant validation for distillation that way? 

 

i.e. Just generate good quality lab data for VLE & then design a production column by simulation skipping the pilot entirely? 

[Bobby Strain]

Yes. You need to find a reputable laboratory though. I recall that Brigham Young U did this once upon a atime. And you can ask for recommendations from your favorite simulator provider. And be sure that you cover the range you are interested in. You might find some help through AICHE, too.

 

But I would first estimate vle based on general correlations, some of which are in your favorite simulator. If you insist on a pilot plant, you don't need to provide a replica of commercial equipment. Just operate the pilot column at conditions that cover the concentrations of interest.

 

And look at companies that will do this for you; those who already have a pilot plant.

 

Bobby

[PingPong]

From the Sulzer plots it looks like one can pack in 5x more theoretical stages at the lower loadings and 2x more at the highest loadings. 

 

If that is  correct then it means one can pilot a distillation column in approx. one fifth to one half the height of the final distillation column?

The factors you estimate may be too optimistic.

 

Here are some Koch-Glitsch plots that look more accurate than the ones in Sulzer brochures:

http://www.koch-glit...re_Gauze_SP.pdf

Note however that the ones for BX and CY are for ortho/para-xylenes while the ones for DX and EX are for chlorobenzene/ethylbenzene (CB/EB). Moreover they are all based on total reflux.

It is not clear to me what fluid system the Sulzer graphs are based on.

 

The following may be an interesting read for you, if you decide to proceed with a pilot plant:

http://kolmetz.com/p...dardisation.pdf

The author Žarko Olujić is also the co-author of a recent book on distillation equipment:

https://books.google...page&q&f=false

Page 162 from that book is missing in google books, so I attach it here as it shows Sulzer HETP data for BX:

 page 162.jpg   137.11KB   1 downloads

Edited by PingPong, 29 June 2016 - 04:28 PM.