Sciencemadness Discussion Board

Lab-scale continuous fractional distillation

DavidJR - 19-4-2018 at 17:45

Has anyone had any success using a continuous fractional distillation process at a lab scale?

I want to distill a hundred litres or so of a crude/dilute product. Aside from buying/building a bigger distillation setup, or doing countless small batch runs, a continuous process, like used in industry, seems like a good option.

I tried cobbling together a crude setup yesterday (using what limited glassware I have) but honestly it didn't work very well at all. Probably not surprising considering the setup:

Essentially I had a 500ml 2-neck round bottom flask atop my hotplate on its side, with a tubing adapter in the lower neck (to draw off the bottoms) and a 300mm air condenser in the other, extending at an incline. The end of the air condenser was attached to an angled vacuum adapter, and a 300mm vigreux column was attached to this, mounted vertically, followed by a still head w/thermocouple probe, another angled adapter, and a condenser mounted vertically. A dropping funnel was connected with a hose to supply the feedstock, and a siphon was used to withdraw the bottoms.

With this setup I found it very difficult to control the rate of addition of feedstock and takeoff of bottoms. Also, the vacuum adapter in between the two reflux 'columns' tended to flood. I didn't use a completely vertical column setup because I ran out of room, and also it's obviously not desirable to have the feedstock just drip down into the flask rather than slowly go down the walls of the column, and the only thing I had on hand to use was an air condenser. I suppose I could possibly have packed it with something.

Sigmatropic - 19-4-2018 at 22:19

Easiest would be to rig a rotavap with two collection flasks which can be removed from the vacuum system without removing vacuum from the main system. This would allow continuous take off. Next would be to connect a straw to the relief valve so that it can be used to suck up more crude product and inject it down the vapor duct without taking the vacuum off.
If that isn't fast enough consider something truly continuous such as agitated thin film or falling film distillation. It all depends on what you want to do (collect the distillate or the residue) and how often you'll be going to do it.

Sulaiman - 19-4-2018 at 23:58

I have 20 litres of kerosene waiting to be distilled,
so I considered continuous distillation and decided that it is too difficult for now,
based mainly on that thinking;

. your 'stripping' section is equivalent to very few theoretical 'plates' - I guess one
. your 'rectification' section is not much better - I guess two 'plates' equivalent

Are three or less 'plates' enough to separate your mixture ?

. ideally the feedstock should enter the junction of the stripping and rectification sections at the temperature that would exist at that point under steady-state full reflux with no feedstock or product flow.
Your setup is introducing cold feedstock - completely upsetting efficient separation.

. most difficult of all is maintaining a state of dynamic equilibrium between feedstock rate and takeoff rate,
if not automatically regulated then a continuous distillation rig will require more continuous attention than a batch system.

. IF I made a continuous distillation setup I would want it to be semi-permanent
to justify the time required to set up and fine-tune - and the cost.

Is your product higher or lower b.p. than the solvent ?

If lower then I suggest that you just get the largest pot that you can
and use a column of sufficient(i) theoretical plates in a batch system.

If higher then maybe a continuous distillation setup would be quicker,
but unless your product is quite valuable,
I doubt that the extra cost would be easy to justify.


(i) Decide on the allowable concentration of product in the pot at the END of the batch run - that will be discarded.
Decide what concentration/purity of product is required.
Use sufficient theoretical plates.

Can you divulge solvent and product b.p. plus initial and desired concentrations ?

Fulmen - 20-4-2018 at 00:09

Actually the feedstock should probably be added to the bottom. IIRC in industrial setups the feed is often in the middle of the column somewhere, but at a temperature that's in equilibrium with the column. Adding cold, dilute feed to the top would throw everything out of whack, reducing the concentration of the product.

But your biggest problem is probably how to regulate this properly. How did you regulate the takeoff, ideally it should be self-leveling (like a side-takeoff, don't know if a siphon would work with the boiling). Trying to balance feed and takeoff manually sounds like a very difficult task. You also want to measure the takeoff concentration somehow, the simplest would probably be the temperature in the boiler.
If you get this sorted it might be possible to get some useful results. I assume you don't have any automatic regulation, this will of course limit the results somewhat. What's your goal, maximum recovery or maximum purity?

DavidJR - 20-4-2018 at 09:04

For the purposes of discussion let's say, hypothetically of course, that the feedstock was around 15% v/v ethanol in water. Therefore the distillate would be the product, ideally the ethanol/water azeotrope, and the residue/bottoms would be waste.

I don't have access to a rotovap, so that's not an option at the moment. And I'd kinda like to try a truly continuous process just because I think it would be fun. So I'm not terribly concerned if it would be more practical to just get the biggest still I can and do batches.

The columns are definitely not very good in terms of theoretical plates. I will probably experiment with packed columns soon, so should be able to do a lot better that way.

Preheating the feedstock would be a good idea. I had thought I might be able to get away with a slow rate of addition, at least as a first test. I will play about with adding the feedstock at the bottom.

My background is not chemistry but electronics/software engineering, so automating it is not a huge barrier and actually part of the attraction to trying this out. I'm thinking of using peristaltic pumps to control addition/takeoff rates.

Fulmen - 20-4-2018 at 10:04

Using pumps for both feed and takeoff is tricky as it requires monitoring the boiler level. A side drain keeping a constant level will be much easier to control, basically you should be able to regulate the feed from the boiler temperature. The feed could in theory be preheated either by the waste or the column/condenser, but I suspect the heat losses will be to great in a small-scale setup.

Magpie - 20-4-2018 at 10:08

Here's some suggestions:

*use the 500mL 2-neck rbf as pot
*use a column of 1" ID, 300mm long (better longer)
* pack the column with a Cu or ss scrub pad, or broken glass.
*Feed at the still head with a constant feed, pre-heated to 78°C. Ideally, your feed would enter the column where the concentration of ethanol is the same as the feed. A compromise might be to use two shorter columns and feed between them.
*Try to get a bp of 78.1°C at the stillhead
*Use any old condenser, cold water cooled, angled per the stillhead adapter
*withdraw water continuously from the pot (at same rate as the feed rate)

If you can feed at 1L/hr you will be done in 100 hrs.



[Edited on 20-4-2018 by Magpie]

[Edited on 20-4-2018 by Magpie]

DavidJR - 20-4-2018 at 14:44

Stainless steel pot scrubbers seems like a good idea.

If I bought 3x 300mm columns, I could stack them and supply the feedstock via an adaptor between the bottom and middle columns. Then at the top of the column an arrangement of adaptors and a coil reflux condenser that vaguely resembles a dean stark trap. That way I can control the amount of reflux by changing the rate that product is removed from the trap.

Obviously this is going to end up pretty tall so there's no way I'll be able to put it on my usual workbench. And I don't think my clamp stands will cut it either...

Any thoughts before I go off and spend all my pennies on new glass?

Sulaiman - 20-4-2018 at 15:36

Quote: Originally posted by DavidJR  
Stainless steel pot scrubbers seems like a good idea.

If I bought 3x 300mm columns, I could stack them and supply the feedstock via an adaptor between the bottom and middle columns. Then at the top of the column an arrangement of adaptors and a coil reflux condenser that vaguely resembles a dean stark trap. That way I can control the amount of reflux by changing the rate that product is removed from the trap.

Obviously this is going to end up pretty tall so there's no way I'll be able to put it on my usual workbench. And I don't think my clamp stands will cut it either...

Any thoughts before I go off and spend all my pennies on new glass?


1) From the SM library download
Vogel: Practical organic chemistry (third edittion)
and read the chapter "Improved apparatus for fractional distillation" starting on page 94.

2) Only one (tall and sturdy) vertical rod for boss heads/clamps, and far less space is required, if you arrange your product condenser(s) vertically.

3) The glassware will be a surprisingly small part of the cost.

4) The temperature difference between successive upper theoretical plates is a fraction of a degree !
So very stable operation is required, this requires;
- Controlled external coluimn heating per Vogel, or good thermal insulation
- Stable pot heating, e.g. use a variac or pwm, not a simple on/off bi-metalic thermostatic regulator

5) One of the hard and/or expensive parts is the variable take-off head,
which you have already considered.
I am attempting to do this by controlling the reflux condenser water flow rate and temperature,
allowing a little vapour to pass through the reflux condenser to the product condenser.
Maintaining such a fine balance has so far eluded me :(

P.S. I doubt that this will be a quick project so you will have plenty of time to look for 'nice' glassware such as mirrored vacuum insulated hempel column(s), variable take-off head, heating/cooling/pumping equipment etc.

All of the above is just for operation at atmospheric pressure,
if you think that you may do reduced pressure distillations then you will also need a cow or a pig or something like a Perkin triangle.

AND
as you will spend a significant ammount of time and effort in constructing your still,
you will want a space for its permanent installation.

P.S. I think that ss scrubbers in the column is a good idea - high surface area, low holdup, high throughput,
if just for drinking ethanol I have read that copper wool helps to reduce the quantity of 'nasty' smelling/tasting compounds in your product.
I have not progressed beyond glass spheres for column packing yet,
copper or ss wool has the great advantage of not requiring a glass spiral to hold up the packing.

P.P.S. lots of good information here https://homedistiller.org/

[Edited on 21-4-2018 by Sulaiman]

Magpie - 20-4-2018 at 16:55

Don't be in a hurry to buy equipment. Do some testing first with what you have.

The most important thing to establish is: what is your requirement for product concentration? Does it have to be the azeotrope or will a lesser concentration suffice?

A shorter column may be adequate. Try what you have first. I specified a 1"ID column to help prevent flooding. A slower feed rate (boilup) will also help reduce or eliminate flooding. Don't pack the column too tightly.

Insulate the column.

Plunkett - 20-4-2018 at 16:59

Quote: Originally posted by DavidJR  
If I bought 3x 300mm columns, I could stack them and supply the feedstock via an adaptor between the bottom and middle columns.

What about using a piece of iron pipe as the column instead of glass? It would take some creativity to mate the pipe to a glass flask but it seems like it would be much easier to and cheaper to attach addition ports, thermocouples, etc to a steel pipe.

[Edited on 21-4-2018 by Plunkett]

Sulaiman - 20-4-2018 at 17:42

I think that an iron pipe column is a poor idea as the iron will conduct heat between theoretical plates.

Magpie - 20-4-2018 at 19:58

A Glass column will be very useful for detecting flooding. Stay with glass at least until you get the system working.

What product concentration do you require?

Fulmen - 21-4-2018 at 01:31

A standard dean-stark trap would actually be an excellent choice as it is a bit more versatile. As for the proper feed location this should cover the basic theory: http://www.separationprocesses.com/Distillation/DT_Chp04i.ht...

DavidJR - 24-4-2018 at 12:11

Quote: Originally posted by Sulaiman  

1) From the SM library download
Vogel: Practical organic chemistry (third edittion)
and read the chapter "Improved apparatus for fractional distillation" starting on page 94.

Very interesting read, thank you. I have a PDF copy of the fifth edition which it seems does not contain this information in the same depth.

Quote: Originally posted by Magpie  

The most important thing to establish is: what is your requirement for product concentration? Does it have to be the azeotrope or will a lesser concentration suffice?

To be honest I don't have any hard and fast requirements. I'll have to go off and do the calculations to figure out what a good compromise would be.

Quote: Originally posted by Plunkett  

What about using a piece of iron pipe as the column instead of glass? It would take some creativity to mate the pipe to a glass flask but it seems like it would be much easier to and cheaper to attach addition ports, thermocouples, etc to a steel pipe.

Hmm I probably wouldn't go with an iron/steel pipe, but have been considering using copper pipe due to it being readily obtainable at reasonable cost, easy to work with (solder etc), and with many types of fittings available. I would still need to figure out how to interface it to glass though...

Quote: Originally posted by Sulaiman  
I think that an iron pipe column is a poor idea as the iron will conduct heat between theoretical plates.

If the outside of a (thin-walled) copper pipe was well insulated then the conduction in the copper may still be acceptable. There'll surely still be the desired temperature gradient though perhaps might take longer to reach equilibrium.

Quote: Originally posted by Fulmen  
A standard dean-stark trap would actually be an excellent choice as it is a bit more versatile. As for the proper feed location this should cover the basic theory: http://www.separationprocesses.com/Distillation/DT_Chp04i.ht...

Yes, but currently I don't own a proper Dean-Stark trap, because I have been using a makeshift one made from various adapters. Didn't see the point in buying one when I could already do the same thing more flexibly with what I had.

Also, that website is a fantastic resource on this. The index page seems to be lost to the digital graveyard, so had to poke about in the source to find the working main page: http://www.separationprocesses.com/Mainmenu.htm

The distillation info is here: http://www.separationprocesses.com/Distillation/MainSet1.htm

[Edited on 24-4-2018 by DavidJR]