Is mosaic meshing really good?
 

As you may know, ANSYS has introduced a new type of meshing called mosaic meshing. They claim that this type of meshing is 45% faster, requires 30% less RAM, increases the accuracy and needs much less cells(about 40% less) compared to the other meshing types.

I just wanted to know that is it actually good or its just for advertising? What are the strengths and weaknesses of mosaic meshing?

https://www.digitalengineering247.co...saicResize.jpg

https://encrypted-tbn0.gstatic.com/i...onBUw&usqp=CAU

thin boundary layers -> better boundary layer capture.
cubic cells throughout the domain -> fast solution due to simple cell shape, accurate solution due to good cell shape and smooth progression of size of nearby cells.
polyhedral cells connecting the boundary layers and cubic cells -> it just connects stuff together.

One caveat is there. Hexcore meshes generally need to be refined very much to capture a particular region of high gradient (i.e a shockwave, refraction wave etc). In such cases, the number of cells can increase significantly high.


NASA has solved this issue, in two ways in the past:
i. use a cartesian grid, where the grid is rotated, stretched and aligned with the shockwave.
ii. use a spring-analogy adaptation method in tetrahedral grids to make skewed cells that are less in number, but give good performance and accuracy ( https://fun3d.larc.nasa.gov/example-33.html https://fun3d.larc.nasa.gov/example-51.html )