AnsweredAssumed Answered

Homogeneity in continuously stirred tank

Question asked by Xenofon Kalogeropoulos on May 7, 2014
Latest reply on May 26, 2014 by Jared Conway

Like • Show 0 Likes0
Comment • 7

Hello everyone,

I have got a model of a mixing vessel with a stirrer inside. After making sure the model is properly set-up for transient analysis I want to do some simulations on the mixing time (blend time) it would take for the fluid to become homogenous after injecting a specific amount of substance inside it (say for example powder or another fluid).

Can solidworks offer such a feature? I would expect it would either be under Tracer Study or Particle Study?

I would welcome your thoughts and suggestions. I have attached a snapshot of the tank setup.

No one else has this question

Outcomes

7 Replies

Jared Conway May 8, 2014 3:54 AM
Mark Correct
Correct Answer
Both tracer and particle studies are post processes. You could try them but it really depends on how you are going to measure homogeneity. More common would be to create a new flow stream in the system and measure volume fraction In multiple locations. But then you need an outlet.
Like • Show 0 Likes0
Actions
- Xenofon Kalogeropoulos @ Jared Conway on May 8, 2014 5:59 AM
  Mark Correct
  Correct Answer
  Thanks Jared, once again you are being very helpfull.
  
  What about introducing a fluid subdomain (e.g. a small sphere) and specifing the same fluid but at a different temperature?
  Running a transient study with this setup, I could use the temperature field and track the response to that temperature disturbance as a function of time at various points inside the vessel?
  
  Your idea with inlet/outlet sounds very good but for a Continuously Stirred Tank Reactor type setup. Mine is more like a Batch reactor.
  Like • Show 0 Likes0
  Actions
  - Jared Conway @ Xenofon Kalogeropoulos on May 9, 2014 1:10 AM
    Mark Correct
    Correct Answer
    definitely worth a shot
    Like • Show 0 Likes0
    Actions
    - Xenofon Kalogeropoulos @ Jared Conway on May 13, 2014 12:04 PM
      Mark Correct
      Correct Answer
      Jared, thanks for the reply. It took some time to set-up and run some scenarios but the idea seems to work wonderfully. In fact I am defining 2 fluids in the domain (both are ideantical fluid though, just for "tracking purposes") and run the transient study for 45 secs (physical time) and track the mass fractions at a number of points inside the tank.
      
      There is 1 thing I am a bit surprised though. I am putting my secondary fluid to the top of the tank in the shape of a cylinder which has a volume equal to 20% of the tank's total volume. So in theory, after infinitely long time of mixing the mass fraction should be 0.2 everywhere inside the tank (assuming perfect immisibility, but again the 2 fluids are ideantical so this should hold).
      
      The problem is that when I run my 45 sec of physical time simulation, the mass fraction evens out at about 0.22. I know 45 secs might not be enough time, but there is good convincing evidence because getting data from about 20 different points inside the tank the statistical set has the following characteristics:
      
      mean: 0.22118
      St Dev: 0.00089
      Var: 0.000001
      
      Before running into this kind of problem I had in mind assume a converged solution when the concentration is within plus/minus 5% of the theoretical 0.2, but now this kind of screws it up for me.
      
      I know the problem sounds a bit tedius, but if you have come across this before or have any suggestions it would be great.
      
      Thanks again!
      Like • Show 0 Likes0
      Actions
      - Jared Conway @ Xenofon Kalogeropoulos on May 13, 2014 9:18 PM
        Mark Correct
        Correct Answer
        there was someone else that was having trouble with some diffusion problem
        my recommendation would be to setup some test cases to make sure your method works right under other conditions
        for example no rotating region would eb the first part
        Like • Show 0 Likes0
        Actions
        Xenofon Kalogeropoulos @ Jared Conway on May 15, 2014 10:30 AM
        Mark Correct
        Correct Answer
        Hi Jared, thanks for the suggestion.
        
        I have crated a simple geometry (hollow cylindrical tank) with no rotating region, and a fluid initial condition at the top of the domain (20% of tank's total volume), as I was doing before (same fluids but separate for tracking difussion). Instead of rotating region I apply some initial velocity to the secondary fluid. The solution looks great and the mixture becomes homogenous after some seconds with even mass fraction 0.2 inside the domain.
        
        I repeat now the above but with introducting a rotating region in the tank (very simple 2 blade object) and obviously removing the initial velocity from the initial conditions. Sadly after simulating, the mixture become completely homogenous but with mass fraction 0.24 this time.
        
        Is there something going wrong when trying to couple rotating regions with difussion tracking?
        Like • Show 0 Likes0
        Actions
        Jared Conway @ Xenofon Kalogeropoulos on May 26, 2014 6:48 PM
        Mark Correct
        Correct Answer
        It's possible. There are multiple flow regions when you use rotation. I would make sure that you are following all of the recommendations for rotating flow, make sure your mesh is well refined, make sure steady state works first and make sure your convergence criteria are tight enough and that you have the right goals defined.
        Like • Show 0 Likes0
        Actions

Homogeneity in continuously stirred tank

Outcomes

Related Content

Recommended Content