You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@@ -97,3 +100,61 @@ The reference values $$\rho_{ref}, T_{ref}, v_{ref}$$ equal the initial state va
97
100
98
101
**Note:** The initial state is also used as boundary conditions for `MARKER_FAR`.
99
102
103
+
## Turbulence Models ##
104
+
105
+
| Solver | Version |
106
+
| --- | --- |
107
+
|`*_RANS`| 7.4.0 |
108
+
109
+
This section describes how to setup turbulence models for RANS simulations. Turbulence is activated using the option `KIND_SOLVER= RANS`, or `KIND_SOLVER= INC_RANS`
110
+
A turbulence model can then be selected via the option `KIND_TURB_MODEL`
111
+
Different submodels and parameters are specified via the different options listed below.
112
+
The turbulent Prandtl number can be modified with the option `PRANDTL_TURB` (the default is 0.9).
113
+
114
+
### Spalart-Allmaras (SA) ###
115
+
116
+
SU2 implements several versions and corrections of the SA model.
117
+
The model is selected using `KIND_TURB_MODEL= SA` and the modifications via the `SA_OPTIONS` list. If this list is empty, then SU2 defaults to `SA-noft2`.
118
+
The freestream and inlet conditions are specified via the option `FREESTREAM_NU_FACTOR= 3` (ratio of SA variable to freestream kinematic viscosity).
119
+
120
+
The following modifications are allowed (refer to [NASA's TMR](https://turbmodels.larc.nasa.gov/spalart.html) for further info):
121
+
- Versions:
122
+
-`NEGATIVE` - Negative SA model.
123
+
-`EDWARDS` - Edwards modification.
124
+
-`BCM` - BCM transitional model.
125
+
-`WITHFT2` - SA model **with** ft2 term, note that by default we omit this term.
126
+
- Corrections:
127
+
-`QCR2000` - Quadratic contitutive relation used in the stress tensor.
-`ROTATION` - Dacles-Mariani et al. rotation correction.
130
+
131
+
All the modifications can be combined with each other expect `NEGATIVE` and `EDWARDS`.
132
+
For example, to specify `SA-neg-R-comp-QCR2000` use `SA_OPTIONS= NEGATIVE, WITHFT2, ROTATION, COMPRESSIBILITY, QCR2000`.
133
+
**However, some combinations are not considered standard**, e.g. `SA-neg` should have the ft2 term, whereas `SA-noft2-Edwards` and `SA-noft2-BCM` should not have the ft2 term, and they are usually not combined with other corrections (see TMR for more details). To use non-standard combinations it is necessary to add `EXPERIMENTAL` to the option list, e.g. `SA_OPTIONS= NEGATIVE, BCM, EXPERIMENTAL`.
134
+
135
+
The rough wall correction is implicitly turned on by specifying roughness values for wall markers via the `WALL_ROUGHNESS` option.
136
+
137
+
### Shear Stress Transport (SST) ###
138
+
139
+
SU2 implements the "Standard" (1994) and 2003 versions of the SST model along with several modifications.
140
+
141
+
**Note:** Currently all versions are "modified" i.e. the turbulence kinetic energy (TKE) is not included in the viscous stress tensor.
142
+
143
+
The main model is selected using `KIND_TURB_MODEL= SST` and the version and modifications via the `SST_OPTIONS` list. If this list is empty SU2 defaults to the baseline 1994 model, `V1994m` (see warning below). The options allow for a version and a set of modifiers to the version.
144
+
The freestream and inlet conditions are specified via the options `FREESTREAM_TURBULENCEINTENSITY= 0.05` (5%) and `FREESTREAM_TURB2LAMVISCRATIO= 10` (ratio of turbulent to laminar viscosity).
145
+
146
+
**Note:** The default values for these options are suitable for internal flows but may be too high for external aerodynamics problems.
147
+
148
+
The following modifications are allowed:
149
+
- Versions:
150
+
-`V1994m` - SSTm **WARNING:** Our implementation has a small [inconsistency with the literature](https://github.com/su2code/SU2/issues/1551), which will be resolved in the next major SU2 update (i.e. version 8).
151
+
-`V2003m` - SST-2003m (no known inconsistencies).
152
+
- Production modifications:
153
+
-`VORTICITY` - Uses vorticity to compute the source term instead of strain-rate magnitude.
154
+
-`KATO_LAUNDER` - Uses the Kato-Launder modification (vorticity times strain-rate).
155
+
-`UQ` - Production is computed using a modified stress tensor for [uncertainty quantification](https://su2code.github.io/tutorials/UQ_NACA0012/). **Note** with this modification TKE is always included in the stress tensor.
156
+
- Corrections:
157
+
-`SUSTAINING` - SST with controlled decay.
158
+
- Curvature corrections are currently not implemented.
159
+
160
+
Modifications from each of these three groups can be combined, for example `SST_OPTIONS= V2003m, VORTICITY, SUSTAINING`
Copy file name to clipboardExpand all lines: _docs_v7/Theory.md
+3-2Lines changed: 3 additions & 2 deletions
Original file line number
Diff line number
Diff line change
@@ -168,9 +168,10 @@ Within the `INC_EULER` solver, we discretize the equations in space using a fini
168
168
169
169
Available for `RANS`, `INC_RANS`.
170
170
171
-
The Shear Stress Transport (SST) model of Menter and the Spalart-Allmaras (S-A) model are two of the most common and widely used turbulence models. The S-A and SST standard models, along with several variants, are implemented in SU2. The reader is referred to the [NASA Turbulence Modeling Resource](https://turbmodels.larc.nasa.gov/index.html) (TMR) for the details of each specific model, as the versions in SU2 are implemented according to the well-described formulations found there.
171
+
SU2 implements several variants of the SST and SA turbulence models, for specifics of the models please see the [NASA Turbulence Modeling Resource](https://turbmodels.larc.nasa.gov/index.html) (TMR).
172
+
For information on how to use turbulence models in SU2 see the [users guide](https://su2code.github.io/docs_v7/Physical-Definition/).
172
173
173
-
Within the turbulence solvers, we discretize the equations in space using a finite volume method (FVM) with a standard edge-based data structure on a dual grid with vertex-based schemes. The convective and viscous fluxes are evaluated at the midpoint of an edge.
174
+
The edge-based finite volume discretization of flow solvers is also used in turbulence solvers. Convective fluxes are evaluated using a scalar upwind scheme (1st or 2nd order).
Copy file name to clipboardExpand all lines: _tutorials/compressible_flow/Transitional_Flat_Plate/Transitional_Flat_Plate.md
+7-7Lines changed: 7 additions & 7 deletions
Original file line number
Diff line number
Diff line change
@@ -42,7 +42,7 @@ For verification, we will be comparing SU2 results against the results of natura
42
42
43
43
### Problem Setup
44
44
45
-
The length of the flat plate is 1.5 meters, and it is represented by an adiabatic no-slip wall boundary condition. There is a symmetry plane located before the leading edge of the flat plate. Inlet boundary condition is used on the left boundary of the domain, and outlet boundary condition is applied to the top and right boundaries of the domain. The freestream velocity, density, viscosity and turbulence intensity (%) is specified as 50.1 m/s, 1.2 kg/m^3, 1.8e-05 and 0.18%, respectively. Since the Mach number is about 0.15, compressibility effects are negligible; therefore, the incompressible flow solver can be employed.
45
+
The length of the flat plate is 1.5 meters, and it is represented by an adiabatic no-slip wall boundary condition. There is a symmetry plane located before the leading edge of the flat plate. Inlet boundary condition is used on the left boundary of the domain, and outlet boundary condition is applied to the top and right boundaries of the domain. The freestream velocity, density, viscosity and turbulence intensity (%) is specified as 50.1 m/s, 1.2 kg/m^3, 1.8e-05 and 0.18% (u'/U=0.0018), respectively. Since the Mach number is about 0.15, compressibility effects are negligible; therefore, the incompressible flow solver can be employed.
46
46
47
47
### Mesh Description
48
48
@@ -62,17 +62,17 @@ Several of the key configuration file options for this simulation are highlighte
62
62
% LINEAR_ELASTICITY, POISSON_EQUATION)
63
63
SOLVER= INC_RANS
64
64
%
65
-
% Specify turbulent model (NONE, SA, SA_NEG, SST)
65
+
% Specify turbulent model (NONE, SA, SST)
66
66
KIND_TURB_MODEL= SA
67
67
%
68
-
% Specify transition model (NONE, BC)
69
-
KIND_TRANS_MODEL= BC
68
+
% Specify transition model
69
+
SA_OPTIONS= BCM
70
70
%
71
-
% Specify Turbulence Intensity (%)
72
-
FREESTREAM_TURBULENCEINTENSITY = 0.18
71
+
% Specify Turbulence Intensity (u'/U)
72
+
FREESTREAM_TURBULENCEINTENSITY = 0.0018
73
73
```
74
74
75
-
The governing equations are RANS with the Spalart-Allmaras (`SA`) turbulence model. By entering `KIND_TRANS_MODEL= BC`, the Bas-Cakmakcioglu Algebraic Transition Model is activated. This model requires freestream turbulence intensity that is to be used in the transition correlation, thus the `FREESTREAM_TURBULENCEINTENSITY` option is also used. The BC model achieves its purpose by modifying the production term of the 1-equation SA turbulence model. The production term of the SA model is damped until a considerable amount of turbulent viscosity is generated, and after that point, the damping effect on the transition model is disabled. Thus, a transition from laminar to turbulent flow is obtained.
75
+
The governing equations are RANS with the Spalart-Allmaras (`SA`) turbulence model. By entering `SA_OPTIONS= BCM`, the Bas-Cakmakcioglu Algebraic Transition Model is activated. This model requires freestream turbulence intensity that is to be used in the transition correlation, thus the `FREESTREAM_TURBULENCEINTENSITY` option is also used. The BC model achieves its purpose by modifying the production term of the 1-equation SA turbulence model. The production term of the SA model is damped until a considerable amount of turbulent viscosity is generated, and after that point, the damping effect on the transition model is disabled. Thus, a transition from laminar to turbulent flow is obtained.
76
76
77
77
The incompressible freestream properties are specified as follows. (Please see "Notes" for freestream properties of other transitional flat plate test cases).
The details of the Zero Pressure Gradient Flat Plate case are taken from the [NASA TMR website](https://turbmodels.larc.nasa.gov/flatplate.html).
15
11
16
12
By comparing the SU2 results of the flat plate case against CFL3D and FUN3D on a sequence of refined grids and seeing agreement of key quantities, we can build a high degree of confidence that the SA and SST models are implemented correctly. Therefore, the goal of this case is to verify the implementations of the SA and SST models in SU2.
@@ -50,20 +46,27 @@ We will compare the convergence of the drag coefficient on the flat plate with g
50
46
51
47
Both the SA and SST models exhibit excellent agreement in the figures below. With grid refinement, we see that both drag and skin friction values asymptote very close to those of FUN3D and CFL3D (and additional codes not shown here but displayed on the NASA TMR), which builds high confidence in the implementations of these two turbulence models in SU2.
The two main SST models, 1994m and 2003m, are compared against FUN3D and CFL3D. Note that for FUN3D and CFL3D, for this case only the 1994m results are available.
Copy file name to clipboardExpand all lines: _vandv/index.md
+2Lines changed: 2 additions & 0 deletions
Original file line number
Diff line number
Diff line change
@@ -27,3 +27,5 @@ Code-to-code comparisons of drag and skin friction on a turbulent flat plate is
27
27
Code-to-code comparisons for a bump in a channel, which results in pressure gradients, is presented using data from the NASA Turbulence Modeling Resource.
Results for the 30p30n airfoil, mesh independence study at low angle-of-attack, and determination of maximum lift, both comparing different numerical schemes.
The details of the Mach 5 SWBLI validation case are taken from the [NPARC Alliance Validation Archive](https://www.grc.nasa.gov/www/wind/valid/m5swbli/m5swbli.html).
11
+
12
+
<palign="center">
13
+
<imgsrc="/vandv_files/swbli/mach.png"alt="Mach number contours (SST-2003m)" />
14
+
</p>
15
+
**Figure 1** - Mach number contours (SST-2003m).
16
+
17
+
We validate our implementation of the SA and SST models by comparing the SU2 numerical results on a sequence of grids against experimental results.
18
+
19
+
## Problem Setup
20
+
21
+
The main geometry features and flow conditions are according to the [main reference](https://www.grc.nasa.gov/www/wind/valid/m5swbli/m5swbli.html). In this study, the inlet was extended 10mm to avoid the intersection of the supersonic inlet with a no-slip wall (that extension is modelled with a slip wall).
22
+
The SU2 configuration files used in this study are [available here](https://github.com/su2code/SU2/blob/develop/TestCases/vandv/rans/swbli/).
23
+
These are applicable to all grid levels, however note that simulations on finer grids were restarted from the results on the previous (coarser) level.
24
+
Mean flow convective fluxes were computed with Roe's scheme and a limited MUSCL reconstruction (Green-Gauss gradients and Venkatakrishnan's limiter).
25
+
The SA-neg and SST-2003m turbulence models were used with first order advection.
26
+
SU2 was run with "freestream equal Mach" non-dimensionalization for all configurations.
27
+
28
+
## Mesh Description
29
+
30
+
Quad-dominant meshes of increasing density were used to perform a grid convergence study.
31
+
The meshes were generated using GMSH where a refinement factor was applied for all sizes and counts.
32
+
Particular attention was given to the y+ on the bottom plate (smaller than 1 on the coarsest level), the main shock, and the separation region.
33
+
The GMSH script can be downloaded from the [SU2 V&V GitHub repository](https://github.com/su2code/VandV/tree/master/rans/swbli).
34
+
The mesh designations and approximate sizes are:
35
+
36
+
- L1 "coarse" (2 x "fine") - 37k quadrilaterals
37
+
- L2 "medium" (1.41 x "fine") - 76k quadrilaterals
38
+
- L3 "fine" - 146k quadrilaterals
39
+
40
+
## Results
41
+
42
+
Given the focus of this validation case (interaction between a shock wave and a boundary layer) it is of particular interest to analyze how well CFD predicts the skin friction coefficient on the bottom plate and the separation (caused by the shock wave) and re-attachment locations.
43
+
Figure 2 compares the skin friction coefficient for the two turbulence models and three mesh levels used, with the experimental values.
44
+
45
+
<palign="center">
46
+
<imgsrc="/vandv_files/swbli/cf.png"alt="Comparison of skin friction coefficient." />
47
+
</p>
48
+
**Figure 2** - Comparison of skin friction coefficient.
49
+
50
+
The results do not change significantly between meshes L2 and L3, they were also not sensitive to other perturbations such as refining the mesh around the main shock, or global refinement (i.e. what would be an L4 mesh). The results for these other tests are omitted for simplicity.
51
+
52
+
Both turbulence models predict that separation takes place (negative skin friction). The start of the separation region is better predicted with SST and it is close to the experiment. However, both models underestimate the re-attachment location.
53
+
54
+
These results with SU2 are further from the experiment than the numerical results presented in the [main reference](https://www.grc.nasa.gov/www/wind/valid/m5swbli/m5swbli.html), which is likely due to compressibility effects not being included in the SA and SST-2003m models.
55
+
This hypothesis was tested by including the SA mixing-layer compressibility correction, i.e. using SA-neg-comp (by adding `COMPRESSIBILITY` to `SA_OPTIONS`).
56
+
Figure 3 compares the results obtained with this model on meshes L2 and L3 with the experimental results, the SA-neg results on the L3 mesh are also shown for reference.
57
+
With compressibility effects the size of the separation region is closer to the experiment.
58
+
This V&V case will be updated with results from the SST-2003 (not modified) model once it is implemented.
59
+
60
+
<palign="center">
61
+
<imgsrc="/vandv_files/swbli/cf_comp.png"alt="Skin friction coefficient with SA compressibility correction." />
62
+
</p>
63
+
**Figure 3** - Skin friction coefficient with SA compressibility correction.
0 commit comments