Volume I - Section V - CFD Methodology 
Page V - 5 
5. 1.6.1 Grid generation 
Grid generation is the process of specifying the position of all of the control volume cells that 
will define both the simulation’s physical configuration and the space surrounding it. Grid 
generation is one of the more challenging and time-consuming aspects of CFD because it 
involves creating a description of the entire configuration that the computer can understand. The 
model thus defined must include the relationship with the space surrounding the chosen model as 
well as the surfaces and processes contained within it. In both cases the most important factor is 
to maintain a suitable number of control volume cells in areas where there will be large or rapid 
changes occurring. These changes may be changes in geometry, such as a sharp comer of an 
object, or they may be sharp changes occurring in the flow field around the object, such as the 
edge of jet issuing from the diffuser. This is called maintaining a suitable grid resolution. 
Without a suitable grid resolution, valuable information can be lost in the numerical simulation 
process and the resulting solution can be misleading. Determining what exactly constitutes 
enough grid resolution is one of the most important jobs a CFD scientist or engineer performs. 
While too few control volume cells can lead to useless simulations, too many control volume 
cells can lead to computer requirements that cannot be fulfilled. A perfect example of this 
situation is trying to run the latest version of Microsoft Word on a 286 chip. 
5. 1.6. 2 Numerical simulation 
As with every other aspect of CFD, the numerical simulation process can also be broken into two 
steps, as follows: 
1) Modeling the Physics 
If the user does nothing else, then the boundary surfaces of the solution domain are "zero flow" 
(i.e., symmetry surfaces). These have zero mass flow, zero surface friction, and zero heat 
transfer. The interior of the domain contains only fluid as defined by properties such as density, 
viscosity, and so on. Anything else, such as inflow or outflow, walls, internal objects, or heat 
gains or losses must be specified explicitly by the user. These are known as boundary conditions. 
The locations of boundary conditions are defined in terms of six spatial coordinates (xS, xE, yS, 
yE, zS, zE), in meters, referenced from the origin located on one comer of the solution domain. 
In the case of a two-dimensional planar (flat) boundary condition (the shelves) the orientation is 
specified and the six coordinates degenerate to five. Additionally, some planar boundary 
conditions should only affect the fluid (e.g., an external boundary wall has only one surface 
present in the solution domain). 
For accurate geometrical representations, the grid lines (surfaces of the control volume cells) can 
be forced to align with a boundary condition. If this is not done then the boundary condition will 
