Hi everyone! I'm trying to implement an extractive distillation process in Aspen Plus. The DSTWU model works completely fine. When I use the results of the DSTWU results to transform the model to a rate-based RadFrac model, the model does not work as well as expected (in numbers = the purity of the desired product after the first column decreases from 0.99 to 10^-4).
As a recap: In the first column, the solvent is added to the azeotropic mixture. The solvent and the desired product go to the bottom of the column. In the second column, the solvent is separated from the desired product (top of the column).
The problem is in the first column where the solvent is added to the azeotropic mixture. I tried several things: adjusting the convergence method, adjusting some parameters by executing some sensitivity analyses (for example: on the reflux ratio, the position of the feed stage), etcetera. However, nothing seems to work.
My question is: does anyone has any idea how to improve the performance of the first column? The convergence method is currently "strongly non-ideal liquid" since the convergence method "azeotropic" gives the following error: "RATESEP CALCULATIONS FAILED TO CONVERGE VARIABLES STAY ON BOUND"
Thank you in advance!
Does the shortcut design allow you to have two feed streams (feed and solvent)?
Dear Manish Pokhrel
Thank you for your response! In the DSTWU (the shortcut design), only one feed stream is allowed. Therefore, I added a mixer before the DSTWU column to mix the azeotropic mixture and the solvent. In the RadFrac (the rigorous design), I have implemented the number of actual stages, the feed stage, the actual reflux ratio, and the distillate to feed ratio obtained from the DSTWU design. Also, I have kept one input stream, but it did not make any difference when I divided the azeotropic mixture (placed on the obtained feed stage from DSTWU design) and the solvent (feed stage varied over the length of the column) in two different input flows into the RadFrac column (most probably because the obtained parameters are based on the DSTWU design with only one feed stream).
Thank you in advance for further advice!
With kind regards,
Fabiënne Lucas
"(most probably because the obtained parameters are based on the DSTWU design with only one feed stream)"- absolutely true. Therefore, we are not supposed to believe the results obtained from shortcut design which is meant to design distillation column with only one feed stream.
If the solvent is a heavy boiling component (usually is the case), solvent goes to the top part of the column and azeotropic feed towards the middle to lower part of the column. First, it would be wise to converge the simulation using an equilibrium model rather than a rate-based model (personal experience).
Dear Manish Pokhrel
Thank you for the advice! I tried to implement an equilibrium model using RadFrac, although rate-based models are more desirable to use since they give a more accurate representation of the process without the use of efficiencies in comparison with an equilibrium model.
However, if I implement an equilibrium model, the yield of the desired product is still very low in comparison with the DSTWU model. Even the last stage of the last distillation column (separation of the solvent and the desired product) dried up. Most probably because also in this situation, the obtained parameters are based on the DSTWU design with only one feed stream.
Is there any way to find appropriate parameters for the RadFrac model with a short-cut method in Aspen Plus for extractive distillation (or any other process with two feed streams)? In case not, at least I know why it does not work, so thank you very much for that insight!
Thank you in advance!
With kind regards,
Fabiënne Lucas
We don't usually use shortcut model for components involving azeotropic mixture. Some of the questions before we proceed.
1. what is the feed composition?
2. Type of solvent?
3. Thermodynamic model?
- Badges
- Science topic
- Similar topics
- Engineering
- Materials Engineering
- Powders