Which turbulence model is the correct one for your CFD simulation
Computational fluid dynamics is generally based on 2 sets of equations. The first one is called the continuity equation which basically describes the conservation of mass flow rate in a fluid volume. The second set is called the momentum equations. There is 1 momentum equation for each spatial dimension (x,y,z) and they are based on the second law of Newton.
Furthermore, the combined set of equations are called the Navier-Stokes equations. In order to simulate the smallest scales of eddies, it would be necessary to generate high-resolution meshes and apply very small time steps. Turbulence modeling consists of adding transport equations that model turbulence rather than numerically resolving them.
Moreover, there are different types of turbulence models which you can use to solve a relatively high Reynolds number CFD case. The list below indicates the weakness and strength of each available model and its engineering applications. You may have used some of them in different software packages whether commercial, educational or open-source.
|Name of Turbulence Model||Number of transport equations||Characteristics||Covergence|
|Spalart-Allmaras||1||1- Used for aerodynamic problems such as airfoils|
3-Bad at predicting separation of flow
|K-Epsilon||2||1- Used for external flows|
2- Used for an initial solution in preparation for further analysis
3- Not convenient for swirling flow, flow separation, or high-pressure gradients
5- Good for low-velocity aerodynamics
6- It requires a wall treatment function
7- Not convenient for Jet Flows
|K-Omega||2||1- It is good at modeling near wall scenarios such as internal flow problems (pipe flows)|
2- It might over-predict shear stress for high-pressure gradients
3- It doesn't require a wall treatment function
4- It is very sensitive to boundary conditions
5-It can be used to model turbines, pumps and general turbo-machinery applications.
|K-Omega SST||2||1- Good at predicting flow separation and flow reattachment |
2- Good choice of model for high velocity and complex aerodynamic simulations
3- Good choice for boundary layer problems (If you want an accurate model that predicts accurately pressure loss due to friction at the wall etc..)
4- In free stream flow, the model works like K-Epsilon avoiding the problems posed by the normal K-Omega model.
Other models like the large eddy simulation are more advanced and require more processing power. The LES model is a combination of turbulence modeling and direct numerical resolution for the larger eddies in the flow. We may talk about LES in detail in a different article.