Design optimization tips for CFD

[Anonymous]

Computational fluid dynamics (CFD) is a branch of fluid mechanics that uses numerical analysis and data structures to analyze and solve problems that involve prediction of fluid flow, heat and mass transfer, chemical reaction and related phenomena.
The geometry and physical bounds of the problem can be defined using computer aided design (CAD). From there, data can be suitably processed (cleaned-up) and the fluid volume (or fluid domain is extracted).
The meshing is done such that, number of nodes and elements are optimum and any complexities are avoided by making the cad geometry simpler. This is done in order to reduce the solution computation time required by solver thereby reducing the number of iterations.
The meshing could be quad/hex or tri/Tet.
Hybrid meshing can also be performed and it could be uniform or non uniform.

The main aim of design optimization is to make the design simpler such that only essential surfaces taking part into the fluid transfer remains.
Some of the design optimization methods are:

1. Removing unwanted features such as flanged portions, fillets, bolts, aesthetic features such as grips, bumped patterns, etc.

2. Looking for any symmetry or periodicity helps to drastically reduce the number of elements and nodes making the mesh more simple. Axisymmetric models such as nozzles, bottle, pipes, etc. can be meshed using axisymmetric concept.

3. If we want to predict external flow, generally use 2D surfaces for analysis.
If not then use Boolean operations on 3D enclosure shapes.

Convergence achieved depends on:
1. Mesh resolution and independence
2.accuracy of physical model
3. Assumptions made