Making good CFD meshes and dealing with skewness and orthogonality is cumbersome for complex domains. Producing a castellated mesh is simple and guaranteed to produce good elements, at the cost of introducing an error in the geometry. Can anyone provide references on the quality of CFD simulations on castellated meshes compared to the respective body-fitting meshes?
It strongly depend on the geometry. Make some test with OpenFOAM. All the tutorial are availabe both with castellated and body fitting mesh.
Franco
This paper provides a good analysis of, and proposes a solution to, the issue of accurate boundary representation in Cartesian grids.
The paper uses NURBS and SPLINES to represent the true boundary of the geometry. The quality of the NURBS surfaces is then varied to quantify the error introduced by incorrect boundary representation. It's not quite the answer to the question you asked, but does include some good references, and a good solution to the problem you are exploring.
http://onlinelibrary.wiley.com/doi/10.1002/nme.4914/abstract
- Badges
- Science topic