2 replies to this topic

[Tschannen14]

I am trying to simulate a packed distillation column using the RadFrac block.  I want to model the separation using NRTL, a Rate-based distillation model, and specified structured packing.  Many of the compounds I am trying to model are not in any Aspen databases and have been User Defined.  To run the Rate-based distillation model, Aspen requires vapor and liquid viscosity.  I have managed to estimate the viscosity for all my compounds and fitted them to the DIPPR model (DIPPR Equations 101 and 102).  However, the model requires an estimation for the mixture viscosity to simulate the separation.  I do not have model parameters for any of the mixture models (Chapman-Enskog-Brokaw-Wilke, Chung-Lee-Starling, TRAPP, Lucas, etc.).  I would prefer to approximate the mixture viscosity by using a mole fraction weighted average (Sum[y_i*visc_i]).  A model such as this does not seem to exist in Aspen for vapor mixture viscosity.

 

Does anyone have any suggestions to get around this issue?  I know it's possible to write my own mixing rules, but I'm not sure I will be able to install a Fortran compiler on my computer as it has restricted access.  If this is possible to accomplish using Excel, that might be an option.  Any insight is greatly appreciated.  Thanks!

[MrShorty]

I am not familiar with exactly how Aspen implements different viscosity estimation techniques. I know that many of these estimation methods are based on a mole fraction average with various correction terms. See, for example, this description of Wilke's method: http://profmaster.bl...s-mixtures.html  In these kind of estimation methods, the correct choice of "mixing rule" parameter will result in an "ideal" mole fraction average like you want. Using the above example of Wilke's method, you should be able to show that, if phi(i,j)=1 for all i,j pairs, Wilke's method simplifies to a simple mole fraction average.

 

I would suggest further research into the exact methods Aspen employs to see which ones will reduce to a mole fraction average with the right choice of mixing rule parameter(s).

Edited by MrShorty, 19 March 2015 - 10:04 AM.

[rrozz]

Hi there,

 

The Wilke's method is for estimation of viscosity mixture of gases.

There is also better method of Reichenberg as seen in Properties of Gases and Liquids.

 

There's unfortunatelly no good method for viscosity of liquids.There's an error of about 5% average.

 

Some are pretty good as UNIFAC-VISCO or ASOG-VISCO and they base on structural contributions.

but none of these is good for water solutions.

There are also methods like McAlinster (?) that demands 3 parameters or more for accurate correlation, and not many are included in different sources.