CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Tecplot (http://www.cfd-online.com/Forums/tecplot/)
-   -   Post-Processing -> Interpolation of cell-centered values to vertex (http://www.cfd-online.com/Forums/tecplot/77550-post-processing-interpolation-cell-centered-values-vertex.html)

PCFD June 26, 2010 14:02

Post-Processing -> Interpolation of cell-centered values to vertex
 
Hi,

I am using Tecplot as a post-processor to visualize CFD-solutions which are stored within an _unstructured CGNS_ (cell-centered). Loading the CGNS-file into Tecplot, I noticed that contour-plots of (all) quantities like mach-number etc. are not smooth. (The simulation converged very nicely. Therefore, I believe that it is only a visualization-problem).

1. Could this be a Tecplot-interpolation-problem? Are there fixes available??
2. I thought that the problem might be the data-storage within the CGNS. Maybe the contours become smooth if I transfered the data from the nodes (=cell centeres) to the vertexes???

Has anybody experience with this issue??? Are there free tools available to interpolate the solution from the nodes to the vertexes???

Thanks in advance.

DoHander June 26, 2010 16:09

There is a simple methodology to interpolate the data from the cells adjacent to one node: use the inverse of the distances as weights. You can search for the articles of Neal T Frink:

http://aaac.larc.nasa.gov/tsab/cfdla...ssertation.pdf

Page 17 formula 3.10

Hope this will help.

Do

praveen June 26, 2010 23:54

Quote:

Originally Posted by DoHander (Post 264660)
There is a simple methodology to interpolate the data from the cells adjacent to one node: use the inverse of the distances as weights. You can search for the articles of Neal T Frink:

http://aaac.larc.nasa.gov/tsab/cfdla...ssertation.pdf

Page 17 formula 3.10

Hope this will help.

Do

You will lose accuracy by doing that. You must do atleast linearly-consistent (exact for degree-one polynomials) interpolation (Frink proposes a way to do that). It should be possible within tecplot but I have never done it. You might getter advice by posting on a tecplot forum, see

http://www.tecplottalk.com/


All times are GMT -4. The time now is 22:58.