next up previous
Next: Discretization of the Up: Finite volume computation of Previous: Introduction

Equations governing incompressible turbulent flows in general coordinates

In order to formulate the equations governing turbulent flow in general coordinates, tensor notation will be used; for an introduction see [Aris, 1962].

The physical domain with curved boundaries is transformed to a rectangle G with the mapping

 

Here, are Cartesian coordinates and are boundary conforming curvilinear coordinates. The mapping is assumed to be regular, i.e. the Jacobian of the transformation does not vanish. Covariant base vectors , contravariant base vectors , and the covariant and contravariant metric tensors and are defined as

The square root of the determinant of the covariant metric tensor, denoted by , equals the Jacobian of the transformation. The following formulae for covariant derivatives of tensors of rank zero, one and two, respectively, are used in this paper:

    

where

is the Christoffel symbol of the second kind. The summation convention holds for Greek indices.

Turbulent flow is governed by the continuity equation and the Reynolds-averaged Navier-Stokes equations. The Reynolds stresses are related to the mean rates of strain through the isotropic eddy-viscosity , which is calculated by the standard high-Re k- model [Launder and Spalding, 1974]. The tensor formulation of these equations is given by

 

 

 

 

 

where is the contravariant mean velocity component, p is the pressure, is the kinematic viscosity, is an external force per unit volume, k is the turbulent kinetic energy, is the turbulent energy dissipation rate and is the production of turbulent energy, given by

 

Finally, , , , and are dimensionless constants which, respectively, are taken to be 0.09, 1.44, 1.92, 1.0 and 1.3. The values of these constants are recommended in [Launder and Spalding, 1974].

Specification of the boundary conditions is straightforward, except in near-wall regions where wall functions are adopted to avoid integration through the viscous sublayer and to obtain log-layer solutions [Launder and Spalding, 1974]. These wall functions are given by:

where

and

Here, is the wall shear stress, Y is the distance perpendicular to the wall, is the Von Kármán constant ( 0.4) and E is a roughness parameter, approximately equal to for a smooth wall. The quantity denotes the tangential velocity along the wall and is given by

Finally, the flux of turbulent energy through the wall is set to zero and the value of at the first grid point away from the wall is determined from

 



next up previous
Next: Discretization of the Up: Finite volume computation of Previous: Introduction



ISNaS ontwikkeling
Fri May 26 14:01:30 METDST 1995