Fluids

Fluids

Topics related to Fluent, CFX, Turbogrid and more.

Target Field Data for Turbulence Model Optimization

    • Andreas Türke
      Subscriber

      Hello everyone,

      I would like to perform a turbulence model optimization in Fluent 2024R1. Therefore I have high-fidelty results from a LES, which I would like to use as target field data. According to the manual (47.2.2.5. Defining Observables for Turbulence Model Optimization), I can write and read target field data with the following text commands:

      • /adjoint/utilities/interpolate/write-data
      • /adjoint/utilities/interpolate/read-data

      Unfortunately, there seem to some problems when writing and/or reading the data. To show this behavior, I am presenting the workflow with the data of a channel flow with main flow in z-direction.

      Writing Target Field Data:

      When I export the target field data, like the velocity in x-direction, with the "write-data" command, I get the following output:

      //adjoint/utilities/interpolate> write-data
      Export field of> z-velocity
      Interpolation file name [] vel-z
      Writing data to vel-z.ip ...
      x-coord
      y-coord
      z-coord
      pressure
      Done.

       

      At this point, the behavior already irritates me as the output says, that the "pressure" is exported, although i chose "z-velocity".

      Reading Target Field Data:

      In the next step, I would like to read the exported data back into same case into the user-defined memory (UDM). In this example, the target is UDM 0. Like before, according to the output, the interpolation data "pressure" is imported, although this should be "z-velocity".

      //adjoint/utilities/interpolate> read-data
      User define memory id [] 0
      Interpolation file name [] vel-z.ip
      Reading vel-z.ip...
      Reading IP data ...
          x-coord
          y-coord
          z-coord
          pressure
        Done.
        Initializing values...
      Done.

      Data Comparison

      When I compare the results of a contour plot of the z-velocity and the contant of UDM 0, I notice that both datasets, have a similar "shape", but the values are diffently scaled. The original data ranges from 7.2 to 20.4 m/s, whereas the UDM 0 values range from -0.003 to 13.2 m/s. So the exported data is neither the pressure field, as I might expected, nor the actual z-velocity field.

      Countour plot of z-velocityContour plot of UDM 0
      When I create an adjoint observable (target-volume-integral) with z-velocity and UDM0, the value does not evaluate to 0, although both fields should be equal. When I repeat the workflow but export/import the Static Pressure (field name: pressure) instead, everything seems wo work properly. I also tried this on another computer with Fluent 2024R2 and the same error occurs.

      Could you please tell me if I am doing something wrong with this workflow or if that is bug in Fluent? And is there a workaround I could use instead?

       

      Kind regards,

      Andreas

       

       

    • Murari Iyengar
      Ansys Employee

      Hi, you have to disable the node values in the contour plot. Else, there will be interpolation for display which can give you a different range.

Viewing 1 reply thread
  • You must be logged in to reply to this topic.