CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   SU2 (https://www.cfd-online.com/Forums/su2/)
-   -   Turbulent flow at low Mach number (https://www.cfd-online.com/Forums/su2/113906-turbulent-flow-low-mach-number.html)

robyTKD February 28, 2013 16:17
Turbulent flow at low Mach number
 
Hello,

I am trying to simulate the turbulent flow around airfoil NACA0012 at low Mach number (say M=0.15) and Re=6e6 in order to perform a shape optimization, then I also need adjoint simulation (impossible with the incompressible formulation, is it right?).
Up to now, I don't have a satisfying enough solution in terms of pressure distribution. I have some questions:
  • Do you have any advice to build a good mesh? My mesh: http://www.4shared.com/zip/8Zgpycw1/...12_40ksu2.html.
  • From your twitter account, I notice that you are validating low Mach cases. Do you have any suggestion to run my case? (space and time discretization, multigrid…)

Thank you in advance,
Roberto

robyTKD March 4, 2013 10:42
1 Attachment(s)
Referring to your last tweet, can you please give me some information about the settings used to have that results?
I am trying to reproduce the same case, but the pressure distribution has a strange behavior near leading edge, as you can see from the Cp plot I attach.

fpalacios March 6, 2013 11:22
Quote:
Originally Posted by robyTKD (Post 411393)
Referring to your last tweet, can you please give me some information about the settings used to have that results?
I am trying to reproduce the same case, but the pressure distribution has a strange behavior near leading edge, as you can see from the Cp plot I attach.
Thanks Roberto,
Could you please download the last version of the code (it contains some relevant improvements with respect the previous one)
http://su2.stanford.edu/download/svn/SU2_Rev1206.zip

after unzipping the code you will find a folder called NACA0012_RANS with the files that we have used for the compressible and incompressible validation.

Best,
Francisco

samiam1000 March 7, 2013 09:34
Hi Francisco and thanks for the new version of the solver.

I ran it and both the compressible and incompressible cases work well.

The point is that I try to solve the adjoint problem (either compressible or incompressible), I can not run it and I get this error:

Code:
zampini@pc-zampini:~/SU2_Rev1206/SU2_Rev1206/NACA0012_RANS/comp$ SU2_CFD default_comp.cfg

-------------------------------------------------------------------------
|            SU2 Suite (Computational Fluid Dynamics Code)            |
-------------------------------------------------------------------------

------------------------ Physical case definition -----------------------
Continuous Navier-Stokes adjoint equations with frozen viscosity.
Mach number: 0.15.
Angle of attack (AoA): 0 deg, and angle of sideslip (AoS): 0 deg.
Reynolds number: 6e+06.
No restart solution, use the values at infinity (freestream).
Read flow solution from: solution_flow.dat.
Surface(s) where the force coefficients are to be evaluated: airfoil.
The reference length/area (force coefficient) is 1.
The reference length (moment computation) is 1.
Reference origin (moment computation) is (0.25, 0, 0).
Input mesh file name: mesh_NACA0012_turb_897x257.su2

----------------------- Design problem definition -----------------------
Drag objective function.
Primitive variables gradient threshold: 100.

---------------------- Space numerical integration ----------------------
Jameson-Schmidt-Turkel scheme for the adjoint inviscid terms.
JST viscous coefficients (1st, 2nd, & 4th): 0.15, 0, 0.02.
The method includes a grid stretching correction (p = 0.3).
Average of gradients with correction (viscous adjoint terms).
Piecewise constant integration of the Navier-Stokes eq. source terms.
Gradient Computation using weighted Least-Squares method.

---------------------- Time numerical integration -----------------------
Local time stepping (steady state simulation).
Euler implicit method for the adjoint equations.
No CFL ramp.
Courant-Friedrichs-Lewy number:        8

------------------------- Convergence criteria --------------------------
Maximum number of iterations: 99999999999.
Reduce the adjoint density residual 6 orders of magnitude.
The minimum value for the adjoint density residual is 10^(-10).

-------------------------- Output information ---------------------------
Writing a flow solution every 5000 iterations.
Writing the convergence history every 1 iterations.
The output file format is Paraview (.vtk).
Convergence history file name: history.
Adjoint solution file name: solution_adj.dat.
Restart adjoint file name: restart_adj.dat.
Adjoint variables file name: adjoint.
Surface adjoint coefficients file name: surface_adjoint.
Surface(s) to be plotted: airfoil.

------------------- Config file boundary information --------------------
Navier-Stokes wall boundary marker(s): airfoil.
Far-field boundary marker(s): farfield.

---------------- Flow & Non-dimensionalization information ---------------
Viscous flow: Computing pressure using the ideal gas law
based on the freestream temperature and a density computed
from the Reynolds number.
--Input conditions:
Grid conversion factor to meters: 1
Ratio of specific heats: 1.4
Specific gas constant (J/(kg.K)): 287.87
Freestream pressure (N/m^2): 184090
Freestream temperature (K): 300
Freestream density (kg/m^3): 2.13163
Freestream velocity (m/s): (52.1572,0) -> Modulus: 52.1572
Freestream energy (kg.m/s^2): 217263
Freestream viscosity (N.s/m^2): 1.853e-05
--Reference values:
Reference pressure (N/m^2): 184090
Reference temperature (K): 300
Reference energy (kg.m/s^2): 86361.1
Reference density (kg/m^3): 2.13163
Reference velocity (m/s): 293.873
Reference viscosity (N.s/m^2): 626.428
--Resulting non-dimensional state:
Mach number (non-dimensional): 0.15
Reynolds number (non-dimensional): 6e+06
Reynolds length (m): 1
Froude number (non-dimensional): 16.6554
Specific gas constant (non-dimensional): 0.999998
Freestream temperature (non-dimensional): 1
Freestream pressure (non-dimensional): 1
Freestream density (non-dimensional): 1
Freestream velocity (non-dimensional): (0.177482,0) -> Modulus: 0.177482
Freestream energy (non-dimensional): 2.51575
Freestream viscosity (non-dimensional): 2.95804e-08
Force coefficients computed using freestream values.

---------------------- Read grid file information -----------------------
Two dimensional problem.
57824 points.
57344 interior elements. 2 surface markers.
704 boundary elements in index 0 (Marker = farfield).
256 boundary elements in index 1 (Marker = airfoil).

------------------------- Geometry preprocessing ------------------------
Setting local point and element connectivity.
Checking the numerical grid orientation.
Identifying edges and vertices.
Computing centers of gravity.
Setting the control volume structure.
Area of the computational grid: 875555.
Searching for closest normal neighbor on the surface.
Searching for sharp corners on the geometry.

------------------------- Solution preprocessing ------------------------
Initialize jacobian structure (Navier-Stokes' equations). MG level: 0.
Initialize jacobian structure (SA model).

------------------ Integration and solver preprocessing -----------------
Area projection in the y-plane = 0.998982.
Set Near-Field boundary conditions (if any).
Set Interface boundary conditions (if any).

------------------------------ Begin solver -----------------------------
Single iteration of the direct solver to store flow data.
[pc-zampini:08664] *** Process received signal ***
[pc-zampini:08664] Signal: Segmentation fault (11)
[pc-zampini:08664] Signal code: Address not mapped (1)
[pc-zampini:08664] Failing at address: (nil)
[pc-zampini:08664] [ 0] /lib/libpthread.so.0(+0xeff0) [0x7fc62c635ff0]
[pc-zampini:08664] [ 1] SU2_CFD(_ZN11CNSSolution13PreprocessingEP9CGeometryPP9CSolutionPP9CNumericsP7CConfigt+0x199) [0x60ff19]
[pc-zampini:08664] [ 2] SU2_CFD(_ZN21CMultiGridIntegration13FAS_MultigridEPPP9CGeometryPPPP9CSolutionPPPPP9CNumericsPP7CConfigtttmt+0x226) [0x493256]
[pc-zampini:08664] [ 3] SU2_CFD(_ZN21CMultiGridIntegration19SetMultiGrid_SolverEPPP9CGeometryPPPP9CSolutionPPPPP9CNumericsPP7CConfigtmt+0x258) [0x491cb8]
[pc-zampini:08664] [ 4] SU2_CFD(_Z20AdjMeanFlowIterationP7COutputPPP12CIntegrationPPP9CGeometryPPPP9CSolutionPPPPP9CNumericsPP7CConfigPP16CSurfaceMovementPP19CVolumetricMovementPPP14CFreeFormChunkm+0x2f9) [0x4a1e79]
[pc-zampini:08664] [ 5] SU2_CFD(main+0x11a1) [0x6ad191]
[pc-zampini:08664] [ 6] /lib/libc.so.6(__libc_start_main+0xfd) [0x7fc62c2e3c8d]
[pc-zampini:08664] [ 7] SU2_CFD() [0x459109]
[pc-zampini:08664] *** End of error message ***
Segmentation fault
zampini@pc-zampini:~/SU2_Rev1206/SU2_Rev1206/NACA0012_RANS/comp$
Any idea about how to solve this?

Thanks a lot,
Samuele

economon March 12, 2013 00:56
Hi Samuele,

Could you please check that you are using the same spatial discretization for both the flow and adjoint problems in the config file (i.e. use either JST or ROE-2ND_ORDER for both CONV_NUM_METHOD_FLOW and CONV_NUM_METHOD_ADJ)? Could you also please check that they are both using the same type of time integration (i.e. both EULER_IMPLICIT for instance)?

In general, you are not required to use the same scheme, however we recently discovered a small issue in the config options which will be fixed in the next release. For now, this should fix your problem, I think.

Cheers,
Tom

Caroline January 3, 2014 03:18
I write a post here but then I searched the forum and its already been solved. so can anyone delete this post for me pls? thanks.


All times are GMT -4. The time now is 23:46.