3 replies to this topic

[FishingforAnswers]

Hi,

 

I am trying to simmulate separation of ethanol and water( xEth = 0,85 molar % in feed to xEth = 0,99 in bottoms ) at 2000 kPa in a distillation colum. I am using NRTL /SRK as fluid package in HYSYS 8.8

 

This mixture has a positive azeotrope. Therefore the alcohol rich stream should be the bottom stream of the distillation column. While the colum does converge, I am not able to concentrate up to xEth = 0,99 molar %.

 

Can anyone help?

 

[Bobby Strain]

The last time I checked ethanol vapor pressure is greater than that of water at lower pressure; hence it is more concentrated in the vapor phase than the liquid phase. And the azeotrope contains about 5% water by liquid volume at low pressure, less under vacuum. I never designed such a separation to operate at such a high pressure. The distillation in industrial practice is conducted at near atmospheric pressure where water is separated as the bottom product and ethanol overhead is near the azeotrope composition. Then the overhead is dried with molsieves.

 

Bobby

Edited by Bobby Strain, 25 August 2016 - 12:53 PM.

[MrShorty]

According to NIST's Refprop program, ethanol has a higher vapor pressure (lower boiling boint) at 2 MPa, and the azeotrope for this system is really close to 85 mol% ethanol. I think my first question for analyzing this simulation is to determine at exactly what composition my thermo model (you claim to be using an NRTL/SRK gamma/phi model) predicts or estimates the azeotrope at my chosen pressure.

[serra]

you may find many documents discussing this separation (I attach one which may be useful for a preliminary review),

I presume you do not wish to obtain (only) a numerical solution from a process simulator (which may be, or not, useful)  but to investigate about a possible "real" solution,

for that you need to define accurate models for VLE and enthalpy (starting from some reliable source, ddb has data for pressures up to a few bars but there are many other sources),

some thermo models (for example EOS with complex mixing rules) allow to extrapolate data from lower pressures but you should verify final accuracy,

I don't know if your SRK-NRTL model allows such extrapolation,

I have used the SRKX-NRTL-HV model available in Prode Properties (SRK-NRTL with Huron Vidal mixing rules),

possibly you have something equivalent in Hysys...

 

Attached Files

•  ethanol_distillation.pdf   551.78KB   206 downloads