[CFDPosterBoy]

Here's a fun problem. I've built a mesh of a simple rectangular duct in Star-CCM+ with flow along the length (0.6477 m long, 0.1102868 m wide, 0.005588 m thick). With this being a thin channel (that will only get thinner in later models), I've elected for a Low y+ wall treatment, to allow resolution of the viscous sublayer. The use of a Low y+ model has, understandably, forced the need for very thin cells near the wall. This by itself would drive up the mesh density to absurd levels. Therefore, I've used volumetric controls to scale up the cell sizes in the length and width directions. The resulting mesh is a reasonable 4 million cells, and has been validated against analytic models.

Now the hard part. I need to validate against an experiment I recently completed. However, try as hard as I might, the walls of the channel in the experiment are not perfectly flat. I mapped them and want to capture these variations. In previous work in Abaqus, I've been able to edit the Abaqus mesh directly to move nodes slightly to match a given geometry. I want to do something similar to my Star mesh. As it stands now, if I know the x, y, z coordinates of a given vertex, I can calculate the new position of the vertex. I don't need to know the positions of its neighbors. Doing this for all vertices in the mesh would yield a meshed geometry that matches the experiment. However, in Star, there appears to be no way to access and change the coordinates of the mesh vertices using a custom user function. I've been told by CD-Adapco that this is impossible. My dissertation depends on this being doable.

Now, I've already tried mesh morphing, but the thin wall cells result in negative volumes after morphing. I've also tried meshing the experiment geometry directly (rather than adapting an ideal geometry). However, this didn't play well with the volumetric controls, and resulted in a mesh of over 75 million cells.

If anyone has experience with either 1) directly moving vertices within Star-CCM+, or 2) exporting, modifying, and reimporting a mesh in Star, I would greatly appreciate any help. Thanks!

[Bust]

I have unfortunately neither experience with directly moving vertices within Star-CCM+, nor exporting, modifying, and reimporting a mesh.
But I wonder if there is any specific reason why you do not use prism cells/layers to mesh the boundary layer? This should reduce the cell count remarkaby.