11 replies to this topic

[TunTun]

Hi everyone. This is my first time using Unisim software. I am simulating the swing pressure distillation process to separate ThF and water. The system will consist of 2 columns: low pressure column and high pressure column. When I simulate the low pressure column, I activated the reflux ratio and distillate rate values, then the system converged. But when I simulate the high pressure column, the system only converged when I activated the reflux ratio and reboiler duty. And then the distillate rate value was 71.33 while the value I wanted was 77.35. Please show me how to fix the error or tell me where I went wrong.

[Pilesar]

I think you will not be able to fix reflux ratio, reboiler duty, and distillate rate for a static feed condition. One variable will need to float. If you want to fix distillate rate, then adjust your reflux ratio or reboiler duty until you get the rate you want. Alternatively, you might adjust the feed rate, feed temperature, or feed composition to match the answers you want. The thermodynamic method might be tweakable, but this will be more frustrating for you.

[TunTun]

Thank you very much. I simulated this process using Unisim software based on an article using Hysys software. I don't know if you know the results of these two software are the same or maybe a little different. I attached the image of the blue text I wrote, can you review it to help me see if the results are acceptable?

Attached Files

•  IMG_0895.jpeg   75.3KB   2 downloads

[Pilesar]

Unisim was formed to satisfy the courts. There were some software companies that merged and to keep from unfair competition, the Hysys and Unisim products started from the existing source code for Hysys. Many of the best developers of Hysys left that company and went to further develop Unisim while Hysys development slowed during the merger time. This was years ago and I don't know the progress of the software since the merger time. The calculated results of both products should be similar given the same inputs. I have not used Hysys for simulation except for predicting hydrate formation and I have never used Unisim. Both are good packages. My background is Pro/II and AspenPlus. Steady state simulation principles apply no matter what your software package.

  It looks like you have closed the loop on your model by sending the overheads back to feed the first column. This magnifies any differences since they are compounded in the recycle stream. 

  The steady state model converges. Congratulations! Notice how the raw feed to the first column is the same for your model and the standard model you are copying. The product stream of the first column removes 6 kmol/hr less than the standard. There is only one other place for that mass to leave the simulation... the bottoms product stream of the second column. So if you do not match the mass flow of the first product stream, it is impossible to match the mass flow of the second product stream.

  I suspect that you are trying to match the steam flow of the first column. That is incorrect! You should try to match the product temperature and composition of the bottoms stream instead. The temperature and pressure of the streams leaving the column represent the stream composition. It is the composition you want to match. The reboiler duties should match also in a perfect world but it is much less important than the composition. The composition is important because the product stream has specifications required for it to be sold! That is where the value is to the company owners. Try to match the THF product stream composition better.

  No pressure is given for the bottom product streams of either column. The pressure affects the temperature at the correct composition of the streams. Did you give a pressure drop inside the columns for your model? If not, add pressure drop and you may find that your bottom temperatures are increasing closer to the standard model. The product temperatures after the cooling exchangers are not very important compared to getting the columns producing the correct product compositions.

  There is no need to make major changes... your model is close. A tweak here and there will make it better. Simulation takes time. In my experience, good models are not made without some rework. And more rework.

[TunTun]

I tried adjusting as you said and everything seems to be fine. I activated the product's component fraction and distillate rate but I got a reflux ratio in the second column of 0.9992 which is very different from the 0.6 in the article. Does it matter? Can you help me figure out where my problem is?

Attached Files

•  IMG_0913.jpeg   82.57KB   1 downloads
•  1J7J87LG7_37VFQG.jpeg   24.52KB   1 downloads

[breizh]

Just a comment,

Pressure at the bottom should be higher than pressure at the top of the column.

Are you using the same thermo package?

my 2 cents

Breizh

[Pilesar]

The liquid at the top of the column falls down the column by gravity. The vapor at the bottom must push its way through the falling liquid to get to the top of the column. The resistance for the vapor results in the pressure at the top of the column being lower than at the bottom. The column pressure is controlled at the top of the column so that pressure usually is measured. The bottom pressure is not directly controlled. It is often useful to measure the pressure at other locations than at the top. Operators may monitor the pressure drop to recognize or prevent problems. The pressure profile and temperature profile of the column helps the engineer confirm how the composition varies inside the column. Each stream will have its own boiling point temperature at a specific pressure which can be used to deduce its composition.

[breizh]

Hi,

Should be NRTL based on document underneath.

Edit: You will find optimized parameters for the columns (number of stages, feed location, Reflux and pressure) if needed.

Microsoft Word - 211.docx

 

Breizh

[TunTun]

I used Wilson model based on this. Should I change the model or continue using it?

Attached Files

•  1-s2.0-S0957582023001015-main.pdf   18.42MB   23 downloads

[breizh]

Hi

Let see what Pilesar and others have to say!

Making a trial changing the thermo package won't harm.

another reference where Wilson and Peng Robinson were used:

https://www.research...nd_optimization

 

 

My 2 cents

Breizh 

[Pilesar]

Wilson is often used as a default thermo where there is much liquid interaction. The engineers at the last company I worked for used Wilson as the go-to thermo method for most chemical simulations. I had never used Wilson previously since I was 'raised' on using NRTL and preferred NRTL for chemical distillations. Some of the choice depends somewhat on the software you are using. There is only so much testing the software company can do -- that is why there may be several undiscovered bugs in the software. The companies have standard sets of problems they run during their testing. These test sets may have more test problems with some thermo packages than others. This results in the experts at one software company recommending similar-but-different thermo packages than experts at another company.

  The above is just background info. The goal is to match the real world. This may require trying more than one thermo package to see which is a closer fit. The 'standard' in your case is a simulation someone else did. This will be very difficult to match exactly unless you use the exact parameters on the same software. Keep in mind that the other 'standard' simulation is just an approximation of the real world results. Similar-but-different simulations might be just as close to the real world results while not matching the other simulation.

  I do not know the best thermo for this system if starting from scratch. I would likely choose NRTL as my first attempt. But you are trying to match the research paper which is a different goal. Reading your attachment gives this quote: 

"The separation of THF and water, which form a minimum-boiling

azeotrope, was used as a case study. The crude feed is 2000 kmol/h

with 6/94 mol% THF/water at 30 ↑C. The purity demands for THF and

water are 99.99 mol% and 99.999 mol%, respectively. The Wilson

equation was used as the thermodynamic model in Aspen Plus V12, and

the built-in binary interaction parameters are given in Table S1."

 With this input, I suggest you continue using Wilson if you want to match the research paper.

See the purity demands in the quoted section. Those MUST be achieved in your model. Set your specifications to achieve these! The product flow rates will be almost exact to the research paper.

Once you get convergence at these specs, then see where the temperatures and duties do not match the research paper. Adjust your model to try to get closer. Try not to lose patience... simulation takes more time than you want to give it!

[breizh]

Hi,

Other publication to comfort your choice of using Wilson .

Another one using NRTL

Comparison of Controllability Features of Extractive and Pressure Swing Distillations on the Example of Tetrahydrofuran Dewatering | ACS Omega

Breizh

Attached Files

•  thf.pdf   3.68MB   5 downloads