Historically, there are two approaches to model flow near the wall:
wall function approach
low-Reynolds number model
The wall function requires the cell center of the first grid point to lay in the logarithmic
layer, whereas the low-Re formulation requires an integration to the wall using a resolution of
. Both approaches produce large errors if used outside of their range of
validity. The wall function method deteriorates under mesh refinement, whereas the low-Re
formulation results in inaccurate wall values for the wall shear stress (and heat transfer) for
meshes slightly coarser than
.
In order to provide less sensitive formulations to the CFD user, wall models have been
developed that are insensitive to . This means that the computed wall values (shear stress and heat transfer) are
largely independent of the
value provided by the mesh. Any
-insensitive wall treatment reverts back to its underlying low-Re formulation if
the mesh is sufficiently fine, and reverts to a wall function formulation for coarse meshes. One
of the main obstacles in the formulation of such wall treatments for the
-equation is that no suitable low-Re formulation has been available. While many
such models have been developed and published, each formulation suffered from one or more of the
following problems:
The formulation is complicated, involving numerous highly nonlinear damping terms.
The formulation is not numerically robust for complex applications.
The formulation produces multiple solutions for the same application (that is, non-unique solutions).
The formulation produces “pseudo-transitional” results (that is, unphysical laminar zones).
Because of these problems, the model formulation of choice in today’s industrial
codes for the ε-equation is a so-called two-layer formulation. It avoids the solution of
the -equation in the viscous sublayer and overwrites it with an algebraic
formulation based on a simple mixing length model.
In Fluent, the two-layer model is the basis of the Enhanced Wall Treatment
(EWT-), which is a
-insensitive formulation for all ε-equation based models. In the
two-layer approach, the fluid domain is subdivided into a viscosity-affected region and a fully
turbulent region. The blending of the two regions is determined by a turbulent Reynolds number
(see Enhanced Wall Treatment ε-Equation (EWT-ε) for more details).
However, using a turbulent Reynolds number for the demarcation of the flow regime has some drawbacks:
Regions with very low values of turbulence kinetic energy might easily have a turbulent Reynolds number smaller than 200. These regions will therefore be treated with a near-wall formulation, even though they might be far away from the wall (for example, regions with a very low turbulence level).
The Wolfstein model is not consistent with the -equation for non-equilibrium (pressure gradient) flows. The solution of the
combination therefore depends on the switching location.
For a mesh that is coarse with a of the first cell near the switching location, the model has a tendency to
oscillate, as it switches back and forth between time steps. This oscillation prevents
convergence.
In order to avoid such drawbacks, the Menter-Lechner near-wall treatment has been developed
as an alternative formulation that is not based on the two-layer approach. It also uses a new
low-Re formulation that is designed to avoid the previously listed deficiencies of existing
-
low-Re formulations.
The goal of a -insensitive near-wall treatment is the
-independent prediction of the wall shear stress and wall heat flux (assuming a
sufficient resolution of the boundary layer). The formulation should switch gradually from wall
functions to a low-Re formulation when the mesh is refined. This also requires a blending of
various quantities between the viscous sublayer and the logarithmic region.
The wall shear stress is needed as a boundary condition and is calculated as:
(4–359) |
where denotes the density. Both of the friction velocities (
and
) are blended between the viscous sublayer and the logarithmic region. The
following formulation is used for the blending of
:
(4–360) |
For the friction velocity , the following formulation is used:
(4–361) |
The main idea of the Menter-Lechner near-wall treatment is to add a source term to the
transport equation of the turbulence kinetic energy that accounts for near-wall effects. The standard
-
model is modified as shown in the following equations (for the sake of
simplicity, buoyancy effects are not included):
(4–362) |
(4–363) |
(4–364) |
where ,
,
,
, and
.
The additional source term is active only in the viscous sublayer and accounts for low-Reynolds number
effects. It automatically becomes zero in the logarithmic region. The exact formulation of this
source term is at this point proprietary and is therefore not provided here.
In combination with the Menter-Lechner near-wall treatment, the iterative treatment and
linearization of the -
two-equation model has been improved. This modification is activated by
default when this near-wall treatment is used.
The Menter-Lechner near-wall treatment can be used together with the standard, realizable,
and RNG -
turbulence models.