No label defined (Q4542973)

The paper describes two classes of incompressible turbulent closure models which are coupled with the RANS (Reynolds-averaged Navier-Stokes) equations through a turbulent eddy viscosity. In the first section, within the framework of linear EVMs (eddy viscosity models), a hierarchy of closure scheme is given, ranging from the zero-equation to two-equation models. Within the framework of nonlinear EVMs, the characterizing feature is the polynomial tensor representation of the second moments of Reynolds stress. The second section presents the Reynolds averaging procedure and the resulting form of the mean momentum and continuity equations. The third section describes linear EVMs, where the RANS formulation is closed by using a Boussinesq-type approximation between the turbulent Reynolds stress and the main strain rate, given by an isotropic relation. For zero-equation models, the eddy viscosity is defined by an algebraic relationship rather than by a differential one. In this category, the authors present the Prandtl's mixing-length model, and Cebeci-Smith and Baldwin-Lomax models. The Baldwin-Lomax model bases the outer layer length scale on the vorticity in the layer rather than on the displacement thickness as in Cebeci-Smith model. For the ``half-equation'' models, advection effects are deemed essential, whereas turbulent transport and diffusion effects are assumed to have much less importance. The Johnson-King model has the characteristic that the eddy viscosity changes in a prescribed functional manner from the inner layer to the outer layer form. For one-equation models, the authors introduce a transport equation with an algebraic prescription for turbulent length scale. Spalart-Allmaras model constitutes a local model based on dimensional analysis, while the model by Baldwin-Barth has its origins in \(k-\varepsilon\) formulation. The two-equation models allow to derive transport equations for both the turbulent velocity and length scales of flow. There are presented \(k-\varepsilon\) and the \(k-\omega\) formulations. At the end of this section, the authors give some formulations with the capability of being used in wall-bounded flows: the Spalart-Allmaras model, and the Durbin model based on an elliptic relaxation approach. The fourth section displays the nonlinear EVMs which are able to capture some effects of stress anisotropies that occurs at the differential second-moment level of closure. The first example is Speziale model which assumes that the anisotropy tensor depends on the Oldroyd convective derivative. The model of Shih and the model of Craft are the next. The remainder of this section presents algebraic stress models in which a polynomial expansion of anisotropy tensor is considered, the expansion coefficients being derived in a mathematically consistent fashion from the full differential Reynolds stress equation. The authors also exhibit an implicit algebraic stress model based on two assumptions proposed by Rodi such that the turbulence reaches an equilibrium state and anisotropy of the turbulent transport, and viscous diffusion is proportional to the anisotropy of Reynolds stresses. Finally, the authors give an explicit solution, and examine the equilibrium assumption, as well as turbulent transport and viscous diffusion assumptions. For the entire collection see [Zbl 0984.00032].