0.5 ppm max impurity specifications for a distillation Product; Validity of design procedures 3

[plantprowler]

If the specifications on a certain lower boiling impurity in the product is only 0.5 ppm is it feasible to reach such tight specs using distillation as a separation procedure?

Or is distillation typically not the right seperation-operation choice for achieving such tight ppm level specs in products. The two substances are quite far apart in normal Boiling Points. 150 C (impurity) vs 220 C (main product). Actual distillation is under high vacuum so the BPs may be closer together at 10 mm Hg. In any case, Antoine equations are available for both.

Do the standard Distillation Design procedures (e.g McCabe Thiele or more refined variations ) hold for designing and calculating the right number of stages?

 

[ash9144]

Yes distillation can be used to make that split. It may not be the best option but is an option. Standard design methods will work as long as there is no non ideal interactions.

 

[plantprowler]

@ash9144

Thanks! What unit operations may be better explored for ppl separations? Adsorption?

I know this may be a tad generic but I'm brainstorming for ideas.

 

[ash9144]

Other options would depend on the components and the process and ultimately economics. I am guessing it is your job to evaluate those.

 

[plantprowler]

Thanks! I am in the process of evaluating. Will report what I find.

 

[apetri]

as noted by ash9144 the selection of process and operating conditions will depend from the mixture you wish to separate,
also the ideal model for fugacity, (based on Antoine vapor pressure equations) may work (or not),
you can provide more detailed information and someone may help ....

 

[moltenmetal]

Nobody can tell you the answer to the question you've asked without a heck of a lot more information than you've given.

If the stuff which can't be in the product is in the feed in the % range and your target is ppm levels, you're likely out of luck with many processes just due to the risk of short-circuiting, or you'll need multiple steps in series just to handle the short-circuiting risk. Do a calc and see how many mL of feed are needed to pollute a L of product beyond the spec, then ask yourself the question again.

Fundamentally, the right way to remove a small amount of mass from something isn't to do something dramatic like evaporation to the entirety of the BULK material- if there's any way that you can go after the low mass stuff in a more strategic way, letting the bulk material just pass through, that will likely be more efficient. As an example, you may use distillation to do bulk drying or water separation, but to get rid of traces of water you tend to use adsorption/absorption or chemical reaction with a dessicant.

Whether or not some other more selective process (adsorption or other) is viable for your separation depends entirely on what compounds A and B actually are.

 

[Compositepro]

"Stripping" is a suitable process to remove trace contaminants which are more volatile than your product. It is a type of distillation process.

 

[georgeverghese]

From a distillation process simulation, 0.5ppm spec in one of the products is typically beyond the mass tolerance limits for simulation iteration convergence, but in Pro/II- Simsci for example, you can respecify these tolerance limits in the global simulation specs to enable tigher tolerances. But you may need to set the mass tolerance limits for iteration loop convergence to be say 10% of the spec purity for this product for obvious reasons. Another parameter that may need tightening is on heat balance. Have a careful read of these in the simulator guidance / reference manuals. It will then take much longer to converge, so a closer approximation of the column internal molar compositions and temperature profiles to the final converged solution will be also required to prevent the simulation from crashing out.

 

[StoneCold]

I think the first question is. Have they actually produced this material, at this specification in the lab? If the answer is yes then how?
That will give you some insight as to what to do. Then it is just a mater of scaleup. I think that is a better approach then just guessing that a simulation is giving you the right answer, but it is not my money.

Regards
StoneCold

 

[apetri]

yes, without detailed information (see my previous post) it's hard to provide answers...

George,

with modern procedures (full/partial Newton solvers) tolerance in convergence is not (normally) a problem, I attach an example of distillation column solved in Microsoft Excel with the free version of Prode Properties, as you see errors are really very low,

however for ppm values a solution couldn't be available, even with 200 or more trays...
also we must caution users (see my previous post) that improper selection of thermodynamic models can introduce very large errors,
for example if you select ideal / vapor pressure model for very non-ideal mixtures,
the same for enthalpy etc.

 

[delittle]

maybe Separation Process Principles (Henley, Seader...) and / or similar textbooks can help

 

[georgeverghese]

@petri,

This is indeed new to me, but am curious what Prode / MS Excel means by lumping both mass and energy balance into a single tolerance output. If my memory serves me right, mass and enthalpy errors are reported individually for each iteration loop in Pro / II. I'd also check this simulator manual for default global convergence tolerance limits to see if these match this output.

Am assuming the thermo model applied for this simulation does predict VLE behaviour for this mix reasonably well.

@plant prowler

In my limited experience, the tightest column product purity spec I've come across is a 10ppm(mole) spec for oxygen in a -195degC overheads nitrogen stream, which I've verified (by manual lab chemical analysis) is actually achieved in plant operation. All refinery and other oil and gas column product purity specs I've come across so far are no where near this. Pls check this 0.5pmm limit for "relevance" in column simulation - there may be some other parameter that may be the dominant - relevant purity specification for this product. As others have said, without knowing the details of your simulation, it is not possible to say much more.

Antoine equations would be fine for getting VLE data for low pressure pure component streams, but not for most multicomponent mixes.

 

[apetri]

@George,
the Excel page reports the errors as (Win/Wout-1.0) mass balance and (Hin/Hout-1.0) energy balance, it's in Excel VBA code, you can show separately the errors and/or adopt different formulations, there is a free version (for students) "

http://www.prode.com/en/distillationcolumn.htm"

if you wish to test and/or compare, as you can imagine there are limits (comparing the free with commercial versions), no global convergence (required for difficult specifications), no reactions, no batch etc.
but the 64 bit version is very fast and I use it regularly with Excel and Matlab.


I agree about purity spec's for distillation columns, my experience is limited to about 10 ppm for a ethanol separation...