[NES81]

Hi all,

I have been trying to figure out why ParaView produces "erroneous" gradient values at the "boundary" of cell level (size) changes for a simple 2D case meshed with snappyHexMesh. Trying to produce a Schlieren type of image by taking the gradient of density (rho).

Basically I have the OpenFoam 2D supersonic wedge tutorial case running with rhoPimplefoam, at ~mach 3. It solves fine, and all is good, but when I use the gradient filter in paraView (version 5.10.0) it calculates very high gradients randomly at locations where there mesh elements change size due to the settings in snappyHexMesh dict.

I may have found a work around solution by using the grad(rho) postprocess command in OpenFoam, which is displayed on the left of the attached images. The ParaView gradient filter is shown on the right. Same exact data set. (see images).

command for openfoam to calc gradient: postProcess -latestTime -func "grad(rho)"

Any ideas on how to get better results from ParaView, or methods of improving the "cell level" transition zone within snappyHexMesh Dict, it would be appreciated, otherwise use the openfoam grad(rho) function.

The other method I've played with is to use the filter "Resample to Image", which appears to generate a structured "mesh" and resamples the field. Then take the gradient of this resampled field, and it does improve the visual results, but comes with potential issues with choosing sample size (grid).

Of course when you have more complex geometry, and use snappyHexMesh, the number of "erroneous" points can get quite large and make it difficult to visualize the real field.