Octree with adaptive technique seems very simple and easy for complex geometry. Why this method is only popular in the field of compressible flow. Although some applications exist for incompressible flow, these cases consider low Reynolds number. Often turbulent model is not taken into account. Is there any problem for incompressible flow? Seldom commercial package employ this technique.

Viscous flows, especially at high Re, require high-aspect ratio cells in the near-wall region. This is not easy to generate with Cartesian grids. Moreover since the cells are not aligned with the boundary, you end up with arbitrary shaped cut-cells and there is loss of accuracy and stability. Even for compressible flows, it has been used for inviscid and laminar flows only.

thanks, can you detail the problem of high-aspect ration. Another question is that according to my knowledge,poor indeed, fluent and phenics can use cubic to fit a arbitary boundary, but not including an adaptive technique. I really seldom heard that the bad stability of the package. If I choose a proper tubulent model,like les, things will become better?

Please see the following thesis

http://hpcc.engin.umich.edu/CFD/publ..._coirier.ps.gz