CFD Steps and Mesh Generation
(1) First of all, you must create or choose a computational domain, and then a mesh generation (also called a grid); after that, the computational domain is divided into many small elements called cells (mesh). For 2D domains, the cells are areas, while for 3D domains the cells are volumes. You can assume that each cell behaves like a tiny control volume in which conservation equations are solved. The quality of a CFD solution is highly dependent on the quality of the grid. Therefore, it must be necessary that the mesh should have high quality before proceeding to the next step.
(2) The type of fluid (water, air, etc.) is specified according to your requirement, with fluid properties like temperature, density, viscosity, etc..
Source Fluid mechanics by cengel 
(3) After that, the numerical parameter and solution models or algorithms selected based on whatever you analyze in your problem. In ANSYS, There are many models available for laminar and turbulent fluid flow as shown. From that the most popular & widely used models are Laminar, Kepsilon, & Komega.
various model available in Ansys 
(7) Once the solution has converged, flow field parameters such as velocity and pressure are plotted and analyzed graphically. Most commercial CFD software has builtin post processors, designed for quick graphical analysis of the flow field. ANSYS is also standalone post processor software packages available for this purpose. Since the graphics output is often displayed in vivid colors, that’s way CFD is also known as the nickname “colorful fluid dynamics”. For example, the below image shows the velocity analysis of the nozzle in which the magnitude of velocity at different points at any location is indicated by different colors.
Mesh type :
People also search “what is meshing ?”
There are two main fundamental types of mesh

 Structured mesh
 Unstructured mesh
Structured mesh
Fig – a 
Fig – b 
Unstructured mesh
Fig – a 
Fig – b 
Mesh Generation :
Fig – c 

 To construct this grid, nine nodes are specified on the top and bottom edges; these nodes correspond to eight intervals along these edges. Similarly, five nodes are specified on the left and right edges, corresponding to four intervals along these edges.
 The intervals i = 1 to 8 and j= 1 to 4 are marked in Fig. The mesh is generated by joining nth i node corresponding to nth j node. In a 2D structured mesh, each cell is easily identified by an index pair (i, j). For example, the shaded cell in Fig. is at (i =4 , j= 3).
Fig – d 

 For complex geometries, an unstructured mesh is usually much easier for the user to the generation of mesh. In ANSYS, you can create both types of mesh.

 Do a comparison between structured mesh and unstructured mesh ( show above images figc and figd), you can notice that the righthand side edge indicates in red curves is not identically generated.

 At boundary layers, where flow variables change rapidly near to the wall and highly resolved and finer mesh are required near to the wall, structured mesh enables much finer resolution than do unstructured mesh for the same number of cells.
 In particular, you must always avoid highly skewed cells, as this can lead to convergence difficulties and inaccuracies in the numerical solution.
Fig – e 
Fig – f 
Source Fluid mechanics by cengel 

 Where θmax is the maximum angle between any two edges of the cell and θmin is the minimum angle between any two edges of the cell. θequal is the angle between any two edges in the ideal cell. For quadrilateral cells, θequal = 90° and for triangular cells θequal =60° as shown in figf.

 Generally, the structured mesh has low or zero skewness and the unstructured mesh has high skewness compare to structured mesh.
 Generally, high skewness occurs mostly near to a sharp corner in bodies. To avoid this high skewness just chop off or apply little fillet at a sharp corner as shown in fig below. This does not affect very much in a result of the analysis.
Reduce skewness 
Hybrid mesh

 Sudden changes in cell size can may lead to convergence difficulties in the CFD. For complicated geometries in which boundary analysis is important, in this condition the goal is to create a mesh in such a way that no cells in the mesh are highly skewed. To avoid a sudden change in mesh and highly skewed cells, we can use hybrid mesh with a combination of structured mesh and unstructured mesh. fig show example of hybrid mesh.
 In hybrid mesh, we create finer (structured) mesh near to boundary or most focusing area of analysis and course (unstructured) mesh in a computational domain far from boundary. This technique has two advantages (1) sudden change in meshing is avoid so the probability of formation of a highly skewed cell is reduced and (2) Computational time is reduced as we create finer mesh only at near to boundary which is our focusing analysis area.
Helpful