Skip to content

Commit 35ae620

Browse files
pcarruscagpcarruscag
committed
revise table, list incompressible options
1 parent cbe97a6 commit 35ae620

File tree

1 file changed

+42
-28
lines changed

1 file changed

+42
-28
lines changed

‎_docs_v7/Convective-Schemes.md‎

Lines changed: 42 additions & 28 deletions
Original file line numberDiff line numberDiff line change
@@ -28,6 +28,8 @@ They are selected via option `CONV_NUM_METHOD_FLOW` and fall under the two broad
2828
Central schemes tend to be more robust whereas second order upwind schemes can be more accurate (i.e. less dissipative).
2929
To achieve second order upwind schemes need to be used with MUSCL reconstruction (`MUSCL_FLOW = YES`), see the "gradients and limiters" page for the MUSCL-related options.
3030

31+
**Note:** MUSCL options have no effect on central schemes or on coarse multigrid levels in general.
32+
3133
## Compressible Flow ##
3234

3335
| Solver | Version |
@@ -36,36 +38,48 @@ To achieve second order upwind schemes need to be used with MUSCL reconstruction
3638

3739
### Central Schemes ###
3840

39-
- `JST`: Jameson-Schmidt-Turkel scheme with scalar dissipation defined by the second and fourth order dissipation coefficients in option `JST_SENSOR_COEFF = (2nd, 4th)` the default values are 0.5 and 0.02 respectively;
40-
- `JST-KE`: Equivalent to `JST` with 0 fourth order coefficient (computational effort is reduced as solution Laplacians no longer need to be computed);
41-
- `LAX-FRIEDRICH`: The simplest of central schemes with a first order dissipation term specified via `LAX_SENSOR_COEFF` (the default is 0.15), this scheme is the most stable and least accurate due to its very dissipative nature.
41+
- `JST` - Jameson-Schmidt-Turkel scheme with scalar dissipation defined by the second and fourth order dissipation coefficients in option `JST_SENSOR_COEFF = (2nd, 4th)` the default values are 0.5 and 0.02 respectively. This scheme offers a good compromise between accuracy and robustness but it will over predict viscous drag contributions in low-Re meshes.
42+
- `JST-KE` - Equivalent to `JST` with 0 fourth order coefficient (the computational effort is slightly reduced as solution Laplacians no longer need to be computed);
43+
- `LAX-FRIEDRICH` - The simplest of central schemes with a first order dissipation term specified via `LAX_SENSOR_COEFF` (the default is 0.15), this scheme is the most stable and least accurate due to its very dissipative nature.
44+
45+
The option `CENTRAL_JACOBIAN_FIX_FACTOR` (default value 4.0) affects all central schemes.
46+
In implicit time marching it improves the numerical properties of the Jacobian matrix so that higher CFL values can be used.
47+
To maintain CFL at lower-than-default values of dissipation coefficients, a higher factor should be used.
4248

43-
Option `CENTRAL_JACOBIAN_FIX_FACTOR` (default value 4.0) affects all central schemes, in implicit time marching it improves the numerical properties of the Jacobian matrix so that higher CFL values can be used.
49+
**Note:** The Lax-Friedrich scheme is always used on coarse multigrid levels when any central scheme is selected.
4450

4551
### Upwind Schemes ###
4652

47-
The following table lists the available upwind schemes for compressible flow and what secondary options apply to each one.
48-
49-
| Scheme \ Option | `ROE_KAPPA` | `ENTROPY_FIX_COEFF` | `ROE_LOW_DISSIPATION` | `USE_ACCURATE_FLUX_JACOBIANS` | `MIN/MAX_ROE_TURKEL_PREC` |
50-
| --- | --- | --- | --- | --- | --- |
51-
| `ROE` - Classic Roe scheme | X | X | X | | |
52-
| `L2ROE` - Low dissipation Low Mach Roe | X | X | | | |
53-
| `LMROE` - Rieper's Low Mach Roe scheme | X | X | | | |
54-
| `TURKEL_PREC` - Row with Turkel preconditioning | | | | | X |
55-
| `AUSM` | | | | | |
56-
| `AUSMPLUSUP` - AUSM+up | | | | X | |
57-
| `AUSMPLUSUP2` - AUSM+up2 | | | | X | |
58-
| `SLAU` | | | X | X | |
59-
| `SLAU2` | | | X | X | |
60-
| `HLLC` | X | | | | |
61-
| `CUSP` | | X | | | |
62-
| `MSW` - Modified Steger-Warming scheme | | | | | |
63-
64-
- `ROE_KAPPA`, default 0.5, constant that multiplies the left and right state averages;
65-
- `ENTROPY_FIX_COEFF`, default 0.001, puts a lower bound on dissipation by limiting the minimum convective Eigenvalue to a fraction of the speed of sound;
66-
- `ROE_LOW_DISSIPATION`, default `NONE`, methods to reduce dissipation where certain conditions are verified, `FD` (wall distance based), `NTS` (Travin and Shur), `FD_DUCROS` and `NTS_DUCROS` as before plus Ducros' shock sensor;
67-
- `USE_ACCURATE_FLUX_JACOBIANS`, default `NO`, if set to `YES` accurate flux Jacobians are used instead of Roe approximates, slower on a per iteration basis but in some cases allows much higher CFL values to be used;
68-
- `MIN/MAX_ROE_TURKEL_PREC`, defaults 0.01 and 0.2, reference Mach numbers for Turkel preconditioning.
53+
- `ROE` - Classic Roe scheme;
54+
- `L2ROE` - Low dissipation Low Mach Roe (L^2 Roe);
55+
- `LMROE` - Rieper's Low Mach Roe;
56+
- `TURKEL_PREC` - Roe scheme with Turkel preconditioning;
57+
- `AUSM` - Advection Upstream Splitting Method;
58+
- `AUSMPLUSUP` - AUSM+up, revised Mach and pressure splittings;
59+
- `AUSMPLUSUP2` - AUSM+up2, uses an alternative pressure flux formulation;
60+
- `SLAU` - Simple Low dissipation AUSM scheme;
61+
- `SLAU2` - SLAU with the alternative pressure flux formulation;
62+
- `HLLC` - Harten-Lax-van Leer-Contact;
63+
- `CUSP` - Convective Upwind Split Pressure;
64+
- `MSW` - Modified Steger-Warming.
65+
66+
Some of the schemes above have tunning parameters or accept extra options, the following table lists those options and indicates to which schemes they apply (if a scheme does not appear on the table, no options apply to it).
67+
68+
| Option \ Scheme | `ROE` | `L2ROE` | `TURKEL_PREC` | `AUSMPLUSUP[2]` | `SLAU[2]` | `HLLC` | `CUSP` |
69+
| --------------------------------- | ----- | ------- | ------------- | --------------- | --------- | ------ | ------ |
70+
| **`ROE_KAPPA`** | X | X | X | | | X | |
71+
| **`ENTROPY_FIX_COEFF`** | X | X | X | | | | X |
72+
| **`ROE_LOW_DISSIPATION`** | X | | | | X | | |
73+
| **`USE_ACCURATE_FLUX_JACOBIANS`** | | | | X | X | | |
74+
| **`MIN/MAX_ROE_TURKEL_PREC`** | | | X | | | | |
75+
76+
- `ROE_KAPPA`, default 0.5, constant that multiplies the left and right state sum;
77+
- `ENTROPY_FIX_COEFF`, default 0.001, puts a lower bound on dissipation by limiting the minimum convective Eigenvalue to a fraction of the speed of sound. Increasing it may help overcome convergence issues, at the expense of making the solution sensitive to this parameter.
78+
- `ROE_LOW_DISSIPATION`, default `NONE`, methods to reduce dissipation in regions where certain conditions are verified, `FD` (wall distance based), `NTS` (Travin and Shur), `FD_DUCROS` and `NTS_DUCROS` as before plus Ducros' shock sensor;
79+
- `USE_ACCURATE_FLUX_JACOBIANS`, default `NO`, if set to `YES` accurate flux Jacobians are used instead of Roe approximates, slower on a per iteration basis but in some cases allows much higher CFL values to be used and therefore faster overall convergence;
80+
- `MIN_ROE_TURKEL_PREC` and `MAX_ROE_TURKEL_PREC`, defaults 0.01 and 0.2 respectively, reference Mach numbers for Turkel preconditioning.
81+
82+
**Note:** Some schemes are not compatible with all other features of SU2, the AUSM family and CUSP are not compatible with unsteady simulations of moving grids, non-ideal gases are only compatible with the standard Roe and HLLC schemes.
6983

7084
## Incompressible Flow ##
7185

@@ -75,11 +89,11 @@ The following table lists the available upwind schemes for compressible flow and
7589

7690
### Central Schemes ###
7791

78-
92+
`JST` and `LAX-FRIEDRICH` are available with low speed preconditioning, the afforementioned 1st, 2nd, and 4th order dissipation coefficients apply to these schemes but the `CENTRAL_JACOBIAN_FIX_FACTOR` option does not.
7993

8094
### Upwind Schemes ###
8195

82-
96+
`FDS` - Flux Difference Splitting with low speed preconditioning, this scheme does not have tunning parameters.
8397

8498
## Turbulence Equations ##
8599

0 commit comments

Comments
 (0)