Since the seminal work of Osborne Reynolds regarding the observation of turbulent flow in a pipe (1883) and Ludwig Prandtl with the introduction of the boundary layer concept (1904) and later his presentation of the mixing length idea for the description ofEckert, M. turbulentRotta, J.C. boundary layer flow (1925) [1, 2], a huge number of researchers and scientists has tried to understand the physics of turbulence and were engaged to find theoretical approaches and mathematical descriptions of the turbulence phenomenon. But it is a matter of fact, that the physics of turbulence are still not completely understood. The consequence is that based on empirical experience various models were generated. One class of such models consists of the Reynolds-averaged statistical approach using the Boussinesq hypothesisBoussinesq eddy viscosity approximation, given for exampleWilcox, D.C. by the Turbulence model algebraic mixing length ideaalgebraic mixing length idea, the Turbulence model one equationone equation models as well as the Turbulence model two equationtwo equation models. This group of models has delivered solutions accurate enough for many practical industrial applications. Therefore we focus in this chapter our interest on this group. Of course, there are other approaches for the computational simulation of turbulence like the Turbulence simulation Large Eddy Simulation (LES)“Large Eddy Simulation” (LES), the Turbulence simulation Detached Eddy Simulation (DES)“Detached Eddy Simulation” (DES) or the Turbulence simulation Direct Numerical Simulation (DNS)“Direct Numerical Simulation” (DNS) [3]. All these methods try to resolve the turbulent eddies partly (LES,DES) or completely (DNS) by fine or very fine numerical grids, while the remaining eddies (LES,DES) are still modeled.