Jump to content

Dr. Flow

Members
  • Content Count

    14
  • Joined

  • Last visited

  1. The VWT online help contains a tutorial, showing how to set up an external aero analysis for a car, file -> Help -> Tutorials
  2. I surface meshed the model in HyperMesh, exported it in Nastran(Fluent) format and brought it into VWT. Worked smooth for me, VWT ran the case and generated a report. Attached is the nas-model I used. wing.zip
  3. instead of "-np 48 -ppn 12" just use "48 12".
  4. the path posted above at 12:08 (today) seems to be correct. xvfb needs to be installed on the cluster, since the VWT automation calls AcuConslole in the background to setup the case.
  5. Looks like while bending one end of the body you introduced some intersections. Use "Check 2D mesh" in the CFD tetramesh panel or the "Intersection" utility to find them.
  6. for transient runs you can average the drag over several time steps to get the overall drag of the body. How many time steps you need to include depends on the time increment and the fluctuation frequency of your flow. For steady you can use the last time step or average of the last 5 time steps.
  7. When generating a surface mesh in HyperMesh for importing into VWT, it is recommended to use a tria mesh (e.g. generated with surface deviation) and export it as Nastran(Fluent). This export format will also add some dummy property cards into the exported *.nas mesh and ensures a proper import into VWT.
  8. You need a certain run time to "wash out" the initial phenomena. The idea behind this formula is, we need 30 sweeps of air over the car to reach a stationary "clean" flow (NOT steady, stationary means a flow which has some kind of periodicity in it). The formula for the time step is based on the expected shedding frequency determined by object height and flow speed. You want to have 30 result points in one cycle for good accuracy.
  9. in VWT 12.1 the drag is averaged over the last 50 iterations. This is good for transient runs, but might not be the best solution for a steady state run. As far as I know in VWT 13.0 there will be user defined number of iterations used for averaging. If you go to the run directory, there you will find a file called Drag_PLots.Drag_Coeff.dat (on Windows), this contains the history of the drag.
  10. In the run directory you should find a *.nas file, which contains the wind tunnel and the body. Check if your wing is included. Do you see similar behavior with a simple box which is intersecting the ground ?
  11. The problem come from the fact, that you didn't generate a BL on the bottom face of the box. So the interface between the two volumes is not connected anymore to the bottom face of the second box. Therefore you get a "non-enclosed" warning. Check for free edges after you have meshed the first volume.
  12. What is the error message HyperMesh brings up after failing ?
  13. You have a pretty rough transition between boundary layer and surrounding tetra mesh. Increase the number of boundary layers The volume mesh in the refinement zones looks coarse, try to refine it. the frontal area computed in VWT is only an approximation. For the sphere you can compute is analytically and use that value 50m/s inflow speed is a lot. Maybe the result you are comparing to is for a different Reynolds number. I ran a sphere with 0.1m diameter and inflow speed of 1m/s resulting in Re-number of 7000. The results I got matched very well the experimental data. Attached is the case Sphere_tutorial.zip
  14. Hi, the tutorial model didn't get included into the installer, I am sorry for that. You can find the Nastran model attached. Regards, Dr. Flow roadster.zip
×
×
  • Create New...