Mixing Time Studies Not Converging To Same Values

ryandryand Member Posts: 3

To whom it may concern,

Hello. My name is Ryan and I am a beginner user of ANSYS Fluent. I am running transient CFD simulations in ANSYS Fluent (a.k.a. mixing time studies), using previous Fluent files of steady state CFD simulations that have converged. These studies use a mesh created from a geometry I built in SpaceClaim. The geometry represents that of a bioreactor.

I am conducting these mixing studies essentially at 5 different agitation speeds (speed at which the impeller in the bioreactor revolves), since the previous steady state simulations were run at 5 different RPM's. In the mixing study simulations, a tracer is added into the tank and mixed throughout. The results display graphs showing the mass fraction of the flow (Vertex Average of Tracer) at various points, over time.

For the first three agitation speeds (66 RPM, 84 RPM, and 101 RPM), the Vertex Average of Tracer converges to 0.000985. However, at 107 RPM and 114 RPM, it converges to 0.0009909. 

Do you know what may be causing the discrepancy in what number to which the results converge?

If there is any more information you need, please do let me know. Thank you in advance for your help. 

Answers

  • KremellaKremella Admin Posts: 2,326

    Hi,

    Are you monitoring some kind of species concentration here? I'm trying to understand why this value should be the same at different RPMs? Because each RPM will likely change your flow field in your mixing tank. Also, are these multiphase VoF simulations?

    Karthik

  • ryandryand Member Posts: 3

    Hello Kremella,


    I am monitoring the mass fraction of a tracer that has been injected in the top of the bioreactor. So I would expect the volume fraction to be the same since I am injecting the same volume of tracer in each mixing study. And no conditions change, other than RPM, so the tracer would only get mixed faster. The volume fraction should not change.


    And these are single phase simulations. Let me know if there is anything else I can clarify.


    Sincerely,


    Ryan Doyle

  • KremellaKremella Admin Posts: 2,326

    Hello,

    The greater the RPM, the more agitated the flow becomes, and this enhances mixing. Therefore, I'd expect the flow to change with RPM. Because the flow is different, the distribution of the mass fraction would also be different.

    I'd suggest that you compare the flow parameters (velocity and turbulence) at this cut plane and check this.

    Karthik

  • ryandryand Member Posts: 3

    Dear Karthik,


    Okay. I computed the volume fraction of tracer before starting the simulation. I thought that, since each simulation adds the same amount of tracer, the volume fraction of tracer at the beginning should be the same, and not be impacted by RPM. But I can look at the XZ plane of the simulation to see if there is unusual flow at the injection point of the tracer.


    Thank you for your time.


    Sincerely,


    Ryan Doyle

  • RobRob UKForum Coordinator Posts: 8,371

    Can you do a mass and volume integral of the tracer, that should show the same result (but not the same number!) but may be more reliable than the vertex average. The issue may be convergence in the solution over time, the faster impeller speeds could mean you need a slightly lower time step. I assume you used the same base mesh in all cases? How did you run the flow and tracer calculations?

Sign In or Register to comment.