How to establish proper computation in first time steps?

acukrovacukrov Member

Hi,

I am solving planar interface evaporation using "Multi-Fluid VOF" model implemented in ANSYS Fluent 2020 R1 Academic.

The interpolation of volume fraction field near the interface is conducted using "Geo-Reconstruct" model.

To obtain a correct solution, I have to enter "solve/set/expert" in TUI and answer "yes" on "Keep temporary solver memory from being freed?" This is preceded by one or more iterations where is this option set to "no".

The question is: "Is there anything additional that one should do in order to avoid the initial discrepancy shown in the bottom left corner in the attached figure?"

Many thanks in advance!

Best regards,

Alen

Comments

  • DrAmineDrAmine GermanyForum Coordinator
    edited March 5

    Why do you need that solve expert thing? 

  • acukrovacukrov Member
    edited March 5

    Dear Mr. Amine,

    I am convinced that this is necessary for more accurate consideration of volume fraction gradient used in calculation of interfacial area.

    Best regards,

    Alen

  • DrAmineDrAmine GermanyForum Coordinator
    edited March 5

    Actually it is just to get the gradients before they get removed. They do not influence the gradient calculations at all. Only if you require in UDF to access some gradient in general purpose macro then yes it should be activated. Alternatively is to allocate gradients without having to do that expert thing.

  • acukrovacukrov Member
    edited March 6

    Dear Mr. Amine,

    thank you for response.

    You are right. I tried again by setting up the case from the beginning, completely in ANSYS Fluent 2020 R1 Academic, and there was no difference between two approaches.

    Something else is causing the initial discrepancy.

    Best regards,

    Alen

  • acukrovacukrov Member
    edited March 6

    Dear Mr. Amine,

    I solved the issue by performing initial 100 time steps with two order of magnitude lower size of the time step than the one used in simulation afterwards.

    The results is shown in the attached figure.

    Best regards,

    Alen

    Errata

    In the legend should write: "Numerical solution 100 cells".

  • DrAmineDrAmine GermanyForum Coordinator
    edited March 6
    Do reducing the size at start is helpful. Thanks for the update.
  • acukrovacukrov Member
    edited March 9

    Dear Mr. Amine,

    yes.

    The results for different mesh resolutions are displayed in the attached figure.

    Best regards,

    Alen

Sign In or Register to comment.