Has anyone made a success with the 2D FFD optimization?
Dear Everyone,
I tried to run the optimization with 2D FFD several times but not succeeded.Can anyone give me an example? I would be grateful for any help you could provide. Thanks! |
Quote:
- Look in the DESIGNS/ folder - output from the CFD and gradient calculations at each step is stored there, and there may be helpful error messages. - Check to see if SU2_CFD on the same config file will run. Any error messages from that may be helpful - Check the result of running either finite_difference or continuous_adjoint (whichever one you were using for the optimization - see shape_optimization.py -h). If it can't get a gradient, it won't work. - Check to see if it works with other DV types - Set a nonzero value for one of your design variables, and run SU2_DEF with VISUALIZE_DEFORMATION set to YES. Check the output volume file - if no deformation has occurred, something is wrong with the DV parameters. (And the output of SU2_DEF will say what deformation it was attempting) |
Thanks for the answer!
4 Attachment(s)
Thanks for the answer!
Here are several files I got when computing the gradient with FFD. The FFd_box.png : Attachment 41909 The config file inv_NACA0012_adv_CFD.txt : Attachment 41910 The surface_adjoint.xls : Attachment 41911 The of_grad_cd.txt : Attachment 41912 I can't find why the gradient is such high number. Cheers |
Quote:
As I stated earlier, I would encourage you to apply test deformations on your DVs to see what deformation is actually being applied. |
Thanks for the answer!
3 Attachment(s)
Thanks for the answer!
I changed the DV like this DV_KIND= FFD_CONTROL_POINT_2D DV_MARKER= ( airfoil ) DV_PARAM= ( wing, 12.0, 1.0, 0.0, 1.0) DV_VALUE= 0.001 as you see,a small change with one ffd point.However,the output mesh is such a disordered mesh. The su2 I run is the exe download form http://su2.stanford.edu/download.html ,which is the single thread v4 on windows. The FFd_box : http://www.cfd-online.com/Forums/att...1&d=1441617847 The config file : http://www.cfd-online.com/Forums/att...1&d=1441617969 The output mesh: http://www.cfd-online.com/Forums/att...1&d=1441617932 Cheers! |
Can anyone help?
|
Quote:
I changed the DV like this DV_KIND= FFD_CONTROL_POINT_2D DV_MARKER= ( airfoil ) DV_PARAM= ( wing, 12.0, 1.0, 0.0, 1.0) DV_VALUE= 0.001 as you see,a small change with one ffd point.However,the output mesh is such a disordered mesh. The su2 I run is the exe download form http://su2.stanford.edu/download.html ,which is the single thread v4 on windows. The FFd_box : http://www.cfd-online.com/Forums/att...1&d=1441617847 The config file : http://www.cfd-online.com/Forums/att...1&d=1441617969 The output mesh: http://www.cfd-online.com/Forums/att...1&d=1441617932 Cheers! |
Quote:
Settings that can effect the deformation (which were not included in your config file, for some reason) - DEFORM_STIFFNESS_TYPE usually INVERSE_VOLUME, but you might also try CONSTANT_STIFFNESS or WALL_DISTANCE. - DEFORM_TOL_FACTOR usually 0.001, controls how far the solver tries to converge - DEFORM_LINEAR_SOLVER usually FGMRES - MARKER_MOVING should get set by DV_MARKER in optimization, and it looks like something is being moved, but just to be extra sure why not. |
5 Attachment(s)
Quote:
Thanks for the answer! It doesn't work. The terminal output : Attachment 42101 The config file : Attachment 42102 The original surface : Attachment 42103 The deformed surface : Attachment 42104 The deformed mesh : Attachment 42106 If you have got a successful test case with FFD control point 2d ,can you give me an example? Sorry to bother you again! Many thanks! |
Quote:
|
5 Attachment(s)
Quote:
I changed the FFD_DEFINITION,but it doesn't work. inv_NACA0012_adv.txt Attachment 42141 orginal_box_and_mesh.png Attachment 42142 deformed_box_and_mesh.png Attachment 42143 origial_surface.png Attachment 42144 deformed_surface.png Attachment 42145 As you can see in the deformed surface , the node id is jumped from 98 to 42 then to 95. I'm wondering there are some bugs in the program. Sorry to bother you again! Many thanks! |
Quote:
|
Quote:
FFD_DEFINITION= ( wing, -0.025, -0.08, 0.0, 1.025, -0.08, 0.0, 1.025, 0.08, 0.0, -0.025, 0.08, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0)) The new setting FFD_DEFINITION= ( wing, -0.025, -0.1, 0.0, 1.025, -0.1, 0.0, 1.025, 0.1, 0.0, -0.025, 0.1, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0)) I changed the y coordinate from 0.08 to 0.1,change DV_KIND to FFD_SETTING, run SU2_DEF, and then try the 2D FFD deformation using the resulting mesh_out. As you can see in the deformed surface , the node id is jumped from 98 to 42 then to 95.I can't find why the id node of the surface is disordered. Sorry to bother you again! Many thanks! |
How do you view the FFD control points along with the mesh?
I know how to view the mesh, just not the FFD control points. |
Quote:
Just run your .cfg file with SU2_DEF and it creates FFD box mesh file. You can append FFD mesh into your original mesh in paraview. |
looking for reply
hey xiongdi
you are a zhongguo ren duiba, i am using SU2 too, ruguo you wanna some jiaoliu, lianxi me! yuezehua@buaa.edu.cn |
All times are GMT -4. The time now is 10:50. |