Structured or un-structured solver
i hope someone will be able to help me.
I would like to understand the difference between structured and un-structured meshing. What are the main differences of the 2 methods ? What are the pro & the cons of each method ? Are physics models the same ? Is there any sector which uses more a model than another (aerospace, turbomachinery, automotive...) In the future, which model will win ?
Thanks for your help
Re: Structured or un-structured solver
The fundamental difference lies in how the adjacency of neighboring cells and nodes are provided.
In a structured grid, the cell or node adjacency is implied from the ordering of the nodes. The grid definition is given in terms of a network of points with distinct i,j,k planes, like a 3-D block. There is no need to store cell connectivity definitions. Every node (i,j,k) in a structured grid has a fixed number of neighbors (6), and those neighbors are always identified as i+1, i-1, j+1, j-1, k+1, k-1. Generating grids for complicated geometries can be more difficult or time consuming using structured grids, but it is easier to construct higher order schemes versus the unstructured approach.
In an unstructured grid, cell and node adjacency is completely arbitrary. A given node j may have any arbitrary number of neighboring nodes, and these neighboring node numbers cannot be implied from the node order. Rather the neighbors must be explicitly provided by providing the cell connectivities. The key advantage here lies in the relative ease of using arbitrary meshes of tetrahedra, pyramids, prisms, and/or hexahedra, allowing for flexible, rapid grid generation compared with block structured grids. Key disadvantage being these methods are inherently more memory intensive owing to the need to store the cell definitions.
Which is "better" depends largely on who you talk to.
|All times are GMT -4. The time now is 22:10.|