The alternative to RANS models are models that resolve at least a portion of the turbulence for at least a portion of the flow domain. Such models are generally termed ‘Scale-Resolving’.
The most widely known SRS modeling concept is Large Eddy Simulation (LES). It is based on the approach of resolving large turbulent structures in space and time down to the grid limit everywhere in the flow. However, while widely used in the academic community, LES had very limited impact on industrial simulations. The reason lies in the excessively high resolution requirements for wall boundary layers. Near the wall, the largest scales in the turbulent spectrum are nevertheless geometrically very small and require a very find grid and a small time step. In addition, unlike RANS, the grid cannot only be refined in the wall normal direction, but also must resolve turbulence in the wall parallel plane. This can only be achieved for flows at very low Reynolds number and on very small geometric scales (the extent of the LES domain cannot be much larger than 10-100 times the boundary layer thickness parallel to the wall). For this reason the use of LES is only recommended for flows where wall boundary layers are not relevant and need not be resolved or for flows where the boundary layers are laminar due to the low Reynolds number. In such cases, the most balanced LES model is the WALE model (see Wall-Adapting Local Eddy-Viscosity (WALE) Model in the Theory Guide). It offers a good compromise between model complexity and generality. It also allows computing laminar shear (boundary) layers without any model impact.
An extension to LES is Wall-Modeled LES (WMLES). It allows the
LES computation of wall bounded flows at higher Reynolds number without
the large increase in grid resolution required for conventional LES
at high Reynolds numbers. For WMLES, the grid resolution is largely
independent of the Reynolds number with respect to the grid cells
required per boundary layer volume. An enhanced version of WMLES called
WMLES -
is also available. WMLES does not provide
zero eddy-viscosity for flows with constant shear. Therefore, it does
not allow the computation of transitional effects, and can produce
overly large eddy-viscosities in separating shear layers. The enhanced
WMLES
-
formulation overcomes these deficiencies.
See Algebraic Wall-Modeled LES Model (WMLES) in the Theory Guide for further details.
In order to avoid the high resolution requirements of LES, numerous hybrid models have been developed in recent years. These are models that combine certain elements of RANS and LES approaches in a way that allows for the simulation of high Reynolds number flows. With hybrid models, the attached wall boundary layers are typically covered by the RANS part of the model, while large detached regions are handled in ‘LES’ mode, meaning with a partial resolution of the turbulent spectrum in space and time. Hybrid models rely on a strong enough flow instability to generate turbulent structures in the separated zone. This is typically the case for flows behind bluff bodies, where URANS models predict single-mode periodic vortex shedding. Hybrid models allow these vortices to generate smaller eddies down to the available grid limit. Figure 11.1: Illustration of SST-URANS vs. SST-SAS Models shows a typical scenario: While the application of a standard RANS model in unsteady mode results in a single frequency vortex shedding (left), the application of hybrid models allows a break-up of the large structures into smaller scales. This is beneficial for predicting the correct mixing behind the body or to extract spectral information (for example, for acoustic simulations). At the same time, the wall boundary layers are covered by the RANS part of the hybrid model avoiding the excessive resolution requirements of LES.
Figure 11.1: Illustration of SST-URANS vs. SST-SAS Models shows a circular
cylinder in a cross flow at .
The left hand side illustrates the SST-URANS model, while the right-hand
side illustrates the SST-SAS model. The isosurface of
is colored according to the eddy viscosity ratio
(note that the scale in the right-hand
side figure is smaller by a factor of 14).
The SAS modeling approach (see Scale-Adaptive Simulation (SAS) Model in the Theory Guide) as proposed
by Menter et al. ([69]
[70]) is based on the introduction of the
von Karman length scale, , into the turbulence equations (for the BSL and
SST models, it enters into the
-equation).
is defined as the ratio of the first divided by
the second derivative of the velocity vector (times the von Karman
constant
=0.41):
(11–1) |
The inclusion of this term allows the model to adjust its length scale to already resolved scales in the flow and thereby provide a low enough eddy viscosity to allow the model to operate in ‘LES’ mode.
The SAS approach has the advantage that the RANS part of the model is unaffected by the grid spacing and will therefore not allow a deterioration of model accuracy as seen in DES in regions of refined grid but insufficient flow instability. However, in cases where the flow instability is not strong enough, the SAS will remain in RANS mode and will not produce unsteady structures. While this is often a sign that the RANS model is still reasonably capable of handling the flow, it is a limitation if unsteady information is required (for example, in acoustics). In such cases, the internal interface option of the ELES implementation can be applied to convert modeled turbulence into resolved turbulence (see Internal Interface Without LES Zone in the Theory Guide).
The DES model (see Detached Eddy Simulation (DES) in the Theory Guide) achieves
the switch between RANS and LES by a comparison of the turbulent length
scale with the grid spacing
. The model selects the minimum
of both and thereby switches between RANS and LES mode by replacing
in the
-equation by:
(11–2) |
Once the model selects the grid spacing as the minimum, the model is operating in ‘LES’ mode.
The grid spacing enters explicitly into the DES model. This
can affect the RANS solution in regions, where the grid is between
RANS and LES resolution (so-called ‘gray zones' in DES) and/or
where the flow instability is not strong enough to generate LES structures.
Another issue to consider with DES is the problem of ‘grid-induced-separation’
(GIS). It occurs if the grid for an attached wall boundary layer flow
is refined to a point where the DES limiter becomes active and affects
the RANS solution. However, in such situations, the flow instability
is not strong enough to balance the reduced RANS content by resolved
turbulence. This will typically result in an artificial flow separation
at the location of grid refinement. It typically happens if (
being the boundary
layer thickness). Remedies for this situation have been proposed by
Menter et al. who recommended using the F1 blending function of the
SST-DES model to shield the boundary layers from the DES limiter.
Later, alternative blending functions for the same purpose have been
proposed by Spalart et al. ([99])—resulting
in the terminology Delayed DES (DDES). The DDES model as originally
proposed for the Spalart-Allmaras model provided limited protection
against GIS for two-equation models such as BSL, SST, and
-
. Therefore, the DDES function
has been re-calibrated for the BSL, SST, and
-
models and is now the recommended
choice and the default setting when using these models.
A further refinement is provided by the Improved DDES (IDDES) formulation of Strelets et al. ([105]), which extends the LES zone of the model to the outer part of wall boundary layers. This allows the simulation of wall boundary layers in Wall-Modeled LES (WMLES) mode. In this model, the IDDES model is applied like a LES model, typically with the specification of unsteady inlet conditions. The grid resolution requirements for WMLES are much less stringent than for LES.
All of the above shielding function variants are available in ANSYS Fluent.
For the Spalart-Allmaras model, the original DDES shielding function
is used. For the -
, BSL, and SST models, the DDES function
has been re-calibrated for better boundary layer protection. The re-calibrated
DDES function is the default selection and is recommended over the
use of the F1 or F2 functions for the BSL / SST model. Despite the
potential difficulties in the application of hybrid methods, they
have the potential to greatly expand the usage of Scale-Resolving
Simulation models for engineering applications, as they avoid the
excessive resolution required by LES for wall boundary layers. IDDES
in WMLES mode is an advanced option and should only be used if you
are familiar with the original literature and the grid requirements
for this model.
The DES and DDES models in ANSYS Fluent have been superseded by the SDES and SBES models described in Shielded Detached Eddy Simulation (SDES) and Stress-Blended Eddy Simulation (SBES).
The Shielded Detached Eddy Simulation (SDES) model and the Stress-Blended Eddy Simulation (SBES) model are hybrid RANS-LES models with an improved shielding function compared to DDES / IDDES. They provide strong shielding of the RANS boundary layer, faster “transition” from RANS to LES in separating shear layers, and (for SBES) explicit selection of the LES model. For implementation details, see Shielded Detached Eddy Simulation (SDES) and Stress-Blended Eddy Simulation (SBES) in the Theory Guide; for setup details, see Including the SDES or SBES Model with BSL, SST, and Transition SST Models.
These models are recommended over the older DES / DDES model formulations. The SBES model is considered the optimal selection, as it offers the clearest distinction between RANS and LES zones.
As pointed out in the previous sections, hybrid models rely on flow instabilities to generate turbulent structures in large separated regions without the explicit introduction of unsteadiness through the boundary conditions. However, there are situations, where such instabilities are not present or are not reliable to serve this purpose. In such cases, it is desirable to apply RANS and the LES models in predefined zones and provide clearly defined interfaces between them. At these interfaces, the modeled turbulent kinetic energy from the upstream RANS model is converted explicitly to resolved scales at an internal boundary to the LES zone. The LES zone can then be limited to the region of interest where unsteady results are required. ELES is available in ANSYS Fluent and allows the combination of most RANS model with classical LES models. It is important to emphasize that in this mode, a full LES resolution is required within the LES zone. In the LES zone, using the WALE model is recommended (see Wall-Adapting Local Eddy-Viscosity (WALE) Model in the Theory Guide).