CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Cloud interpolation (

Pascale Fonteijn February 7, 2003 17:57

Cloud interpolation
Hi all,

I have a large geometry that I have split in two separate sections and simulate them indepedently. The velocites at a certain plane in the first part are written to a data file. I use the data for a cloud interpolation on the inlet of the second part. However, the velocities and the mass flow are not identical.

Does anyone have experience on this topic? I presume that the data that is written out, are conservative values. Is that true? Are there some parameters that I missed?

Thank you for reading, Pascale

Sharks February 7, 2003 19:07

Re: Cloud interpolation
Hi Pascale

The cloud of points interpolation was really only developed to provide users with the ability to specify more complex initial conditions. The interpolation itself uses distance weighted averaging to calculate a value at a specific node, so there will be a general smoothing of the original data.

I hear there are some enhancements in 5.6 which allow for more accurate interpolation, but until then I think the only workaround is to employ user fortran and code up your own interpolation routine.


Bob February 9, 2003 05:43

Re: Cloud interpolation
Hi Pascale, Yes I agree. We had the same problem where a client provided inlet data from 44 different locations along an inlet plane. Rather than create 44 individual inlets we tried the cloude interpolation. We found the same problem as you, that the values we smoothed out due to the distance weighting. We tried creating a more densley populated cloude, which did improve the interpolation, but still fell short of the desired boundary conditions. In the end I wrote a PCL script to read in data to generate my 44 inlet boundary conditions. Good luck Bob

Holidays February 10, 2003 03:12

Re: Cloud interpolation

The algorithm is a DWA one which will always smooth the profiles and miss extreme changes in gradients. But there's a different scheme to be used via a env. variable, that will limit the DWA scheme to neighbouring nodes which is sometimes less of a smoother. Ask your support team. I also know that the general schame will be greatly enhanced in CFX-5.6.

PS DWA = Distance Weight Averaging (search on the web for the details of the scheme)

All times are GMT -4. The time now is 15:03.