Aspects of CFD computations with commercial packages

Abstract

The purpose of this chapter is to give some insight into the steps that are needed toobtain a CFD solution of the flow field inside or around an object with the use ofa commercial CFD software package. Note that it is not the intention to comparedifferent commercial CFD software packages. The applications that are shown canbe computed with most of the available software packages.A CFD solution involves the following basic steps:- Creation of the geometry (or import of the geometry from a CAD package)- Grid generation- Choice of the models- Application of the boundary conditions- Flow field computation- PostprocessingThe first step is the creation of the geometry. Usually this is done with a separateCAD package. However, since the grid generator has some specific demands on theimported geometry, the imported geometry often has to be 'cleaned up'. Most CFDpackages provide a CAD tool together with their grid generator. The geometry createdwith this embedded CAD tool is directly suitable for the grid generator. However,design engineers are using specific CAD packages for their needs and thereforethe most common way to obtain the geometry in the grid generation package is theimport from a CAD package. The 'cleaning up' phase is treated in Sect. 12.2.The next phase is the grid generation process. A choice has to be made as towhich kind of grid will be used: structured, block structured, unstructured, hybrid.For viscous calculations, a boundary layer mesh also has to be constructed. For turbulentflow calculations, the distance to the wall of the first cell in the boundary layermesh depends on the near-wall treatment of the turbulence model. In cases wherethe grid is not optimal for an accurate solution of the flow field, grid adaptation canbe used in order to adapt the grid to the computed flow field features, such as shocks,slip lines, etc. . .These aspects will be discussed in Sect. 12.3.The choice of the models depends on the kind of flow to be computed, andwill have an impact on the grid generation process. The flow can be two- or threedimensional,steady or unsteady, incompressible or compressible, laminar, turbulentor both and heat transfer can be important. These are the models used in the examplesin this lesson. Other models that are often used, but which could not be dealtwith in this introductory CFD course are mass transfer, chemical reactions, combustion,multiphase flows, discrete particle flows, flow in moving geometries, etc.,and combinations of the above. Some modelling aspects on turbulent flows will bediscussed in Sect. 12.4.The next step is the application of the boundary conditions. Since the flow fieldis only computed in the region of interest, adequate boundary conditions have tobe provided at the boundaries of the computed region. Frequently used boundaryconditions are inlet, outlet and wall boundary conditions. Different implementationsof these boundary conditions are considered in Sect. 12.5. More complex boundaryconditions can be defined through user-written routines (Sect. 12.6).The computation of the flow field with the solver becomes of less and less concernto standard users of a commercial CFD software package. So, the user can focuson the fluid dynamics without caring too much of the numerics behind it. However,the more experienced user who intends to write user routines that can be coupledwith the software package needs to have a basic understanding of the underlyingalgorithms of the discretization and solution techniques, which is the subject of theother chapters in this text. Some solver aspects are discussed in Sect. 12.7.Once the flow field is computed, it can be analyzed in the postprocessing phase.Many postprocessing means are available today. It is not the intention here to gointo much detail on postprocessing features, but a short overview of possibilities isgiven in Sect. 12.8. If the user is not satisfied with the solution, a grid adaptationstep can be performed as mentioned before.More complex flow calculations e.g. with moving meshes and fluid-structure interactionscan also be performed these days and will have an influence on the differentsteps outlined above, but are beyond the scope of this chapter. © Springer-Verlag Berlin Heidelberg 2009.