Mainly the CFD procedure can be divided into 3 stages:

Pre-processing

Processing/Solver

Post-processing

Among these 3 processes, the pre-processing and post-processing stages are more customizable and are more controlled by the user. The solver/processing stage is where the numerical solution is calculated using computing power. The user does not have much control in manipulating the solver if he/she is using commercial tools. The procedures are already written in the form of computer programs and are not open source (Hence sometimes it is called black-box).

So, the user should mainly concentrate his efforts on pre-processing and post-processing in case of common thermo-fluid problems.

The CFD simulation is done in the following steps:

**1. Pre-processing**

**I. **A fluid domain is chosen and mesh is generated. This is one of the most important and most time-consuming processes in CFD analysis. The mesh generation is also called grid generation.

**II. **Boundary conditions are applied on the respective faces of the domain.

**III. **The fluid properties are chosen.

**IV. **The solution algorithm is chosen. Usually, the default settings give a 'good enough' solution to regular fluid flow problems.

**V. **The initial conditions are provided.

**2. Solver/Processing**

The discretized form of governing equations that is conservation of mass, momentum, and energy are solved iteratively by the computer until the residuals approach zero (or predefined values). This process takes a long time depending on the computing power available and the complexity of the problem (number of mesh elements)

**3. Post-processing**

**I.** Once the solution is converged and monitors (such as drag, pressure drop, etc) have become steady. The critical sections are taken and different visualizations/contour plots are plotted for important fluid parameters like pressure distribution, velocity distribution, temperature distribution, etc.

**II.** Animation of streamlines is created to identify the flow separation zones. Flow separation can also be identified on velocity contours.

**III. **Pressure drops between sections, lift and drag values, and average velocity on the boundaries are calculated. Post-processing varies from problem to problem. Which parameters to be analyzed is completely dependent on users' interests.

**IV. **Finally a report is created with suggestions for design improvements. This process is repeated over a period of time until an optimized design is found.