[Weiqiang Liu]

Hi all,

I am using ANSYS ICEM to draw a structured mesh and tried to import this mesh into fluent. I followed instruction in tutorial and transform structured mesh into unstructured mesh.

All the cells in the domain is quadrilateral. I did not change shape of any cells in the domain. However, the mesh is changed from structured to unstructured.

I am wondering what is the criterion to determine structured and unstructured mesh?

Best regards

Weiqiang

[sapujapu]

Your grid is topologically structured (it can be mapped to a cartesian coordinate system with a conformal function), but the data structure of the grid is unstructured because that's the only data type that Fluent supports.

The difference between structured and unstructured data structs in grids is that structured grids contain the information about a cell's neighboring cells already in the node numbers. So if a cell has the node numbers 1,2,3,4 and another cell has the node numbers 3,4,5,6 the solver knows that these cells are next to each other and share the face 3,4.

An unstructured grid does not have this information because the nodes are numbered more or less randomly. To find out which cells share a face, unstructured grids contain a so called connectivity matrix. During the solving process, the solver always has to look up this info from the matrix. That's also the reason why unstructured grids usually have longer solving times than structured grids.

[Weiqiang Liu]

Originally Posted by sapujapu

Your grid is topologically structured (it can be mapped to a cartesian coordinate system with a conformal function), but the data structure of the grid is unstructured because that's the only data type that Fluent supports.

The difference between structured and unstructured data structs in grids is that structured grids contain the information about a cell's neighboring cells already in the node numbers. So if a cell has the node numbers 1,2,3,4 and another cell has the node numbers 3,4,5,6 the solver knows that these cells are next to each other and share the face 3,4.

An unstructured grid does not have this information because the nodes are numbered more or less randomly. To find out which cells share a face, unstructured grids contain a so called connectivity matrix. During the solving process, the solver always has to look up this info from the matrix. That's also the reason why unstructured grids usually have longer solving times than structured grids.

Hi，sapujapu
thanks very much for this detailed explanation. Can I ask a question? If unstructured data type takes a longer solving time, why fluent does not use structured data type as default?

Best regards
weiqiang

[sapujapu]

It's an unstructured solver because those solvers can deal with unstructured and structured grids. Structured solvers can only deal with structured grids, so it's quite limited in its capability. As a commercial solver, it makes sense to have as many options as possible, even if it might increase the computation time.

[Weiqiang Liu]

Originally Posted by sapujapu

It's an unstructured solver because those solvers can deal with unstructured and structured grids. Structured solvers can only deal with structured grids, so it's quite limited in its capability. As a commercial solver, it makes sense to have as many options as possible, even if it might increase the computation time.

thanks very much for your detailed answer

Best regards

Weiqiang