Hi,
I have removed the "3d only" restriction from the Navier Stokes and related terms. There is also a new example: examples/navier_stokes/navier_stokes2d.py.
r.
Now it remains to implement a robust flow solver. Even this small example shows, that the solution is not obtained easily - try decreasing the viscosity, and/or increase the Dirichlet velocity - the solver would not converge.
Cheers, r. PS: As mentioned in Terri Oda's e-mail, you should blog about your work so far ASAP!
On 07/01/2013 06:12 PM, Ankit Mahato wrote:
awesome :)
On Monday, 1 July 2013 15:05:16 UTC+5:30, Robert Cimrman wrote:
Hi,
I have removed the "3d only" restriction from the Navier Stokes and related terms. There is also a new example: examples/navier_stokes/navier_stokes2d.py.
r.
R,
I was getting an error while running the example. ./simple.py examples/navier_stokes/navier_stokes2d.py sfepy: left over: ['data_dir', 'verbose', '_filename', '__builtins__', '__package__', '__doc__', '__name__', '__file__'] sfepy: reading mesh (/home/ankit/sfepy/meshes/2d/rectangle_fine_quad.mesh)... sfepy: ...done in 0.02 s sfepy: creating regions... sfepy: Right sfepy: Bottom sfepy: Top sfepy: Surface sfepy: Omega sfepy: Walls sfepy: Driven sfepy: ...done in 0.03 s sfepy: equation "balance": sfepy: + dw_div_grad.5.Omega(fluid.viscosity, v, u) + dw_convect.5.Omega(v, u) - dw_stokes.5.Omega(v, p) = 0 sfepy: equation "incompressibility": sfepy: dw_stokes.5.Omega(u, q) = 0 sfepy: setting up dof connectivities... sfepy: ...done in 0.00 s sfepy: using solvers: ts: no ts nls: newton ls: ls sfepy: updating variables... sfepy: ...done sfepy: matrix shape: (44557, 44557) sfepy: assembling matrix graph... sfepy: ...done in 0.13 s sfepy: matrix structural nonzeros: 1728264 (8.71e-04% fill) sfepy: updating materials... sfepy: fluid sfepy: ...done in 0.01 s sfepy: nls: iter: 0, residual: 2.023986e-02 (rel: 1.000000e+00) convect_build_vtg(): ERR_Switch mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! Traceback (most recent call last): File "./simple.py", line 146, in <module> main() File "./simple.py", line 143, in main app() File "/home/ankit/sfepy/sfepy/applications/application.py", line 29, in call_basic return self.call(**kwargs) File "/home/ankit/sfepy/sfepy/applications/pde_solver_app.py", line 213, in call nls_status=nls_status) File "/home/ankit/sfepy/sfepy/solvers/ts_solvers.py", line 29, in __call__ state = problem.solve(state0=state0, nls_status=nls_status) File "/home/ankit/sfepy/sfepy/fem/problemDef.py", line 933, in solve vec = solvers.nls(vec0) File "/home/ankit/sfepy/sfepy/solvers/nls.py", line 345, in __call__ mtx_a = fun_grad(vec_x) File "/home/ankit/sfepy/sfepy/fem/evaluate.py", line 66, in eval_tangent_matrix mtx = pb.equations.eval_tangent_matrices(vec, mtx) File "/home/ankit/sfepy/sfepy/fem/equations.py", line 640, in eval_tangent_matrices self.evaluate(mode='weak', dw_mode='matrix', asm_obj=tangent_matrix) File "/home/ankit/sfepy/sfepy/fem/equations.py", line 526, in evaluate asm_obj=asm_obj) File "/home/ankit/sfepy/sfepy/fem/equations.py", line 766, in evaluate ret_status=True) File "/home/ankit/sfepy/sfepy/terms/terms.py", line 1473, in evaluate diff_var, **kwargs) File "/home/ankit/sfepy/sfepy/terms/terms.py", line 1311, in eval_real status = self.call_function(out, fargs) File "/home/ankit/sfepy/sfepy/terms/terms.py", line 1296, in call_function raise ValueError('term evaluation failed! (%s)' % self.name) ValueError: term evaluation failed! (dw_convect)
Regards, Ankit
PS: I was not aware that we have to blog weekly. I will blog for week 1 and 2 asap.
On Tuesday, 2 July 2013 13:58:20 UTC+5:30, Robert Cimrman wrote:
Now it remains to implement a robust flow solver. Even this small example shows, that the solution is not obtained easily - try decreasing the viscosity, and/or increase the Dirichlet velocity - the solver would not converge.
Cheers, r. PS: As mentioned in Terri Oda's e-mail, you should blog about your work so far ASAP!
On 07/01/2013 06:12 PM, Ankit Mahato wrote:
awesome :)
On Monday, 1 July 2013 15:05:16 UTC+5:30, Robert Cimrman wrote:
Hi,
I have removed the "3d only" restriction from the Navier Stokes and related terms. There is also a new example: examples/navier_stokes/navier_stokes2d.py.
r.
On 07/02/2013 01:58 PM, Ankit Mahato wrote:
R,
I was getting an error while running the example. ./simple.py examples/navier_stokes/navier_stokes2d.py sfepy: left over: ['data_dir', 'verbose', '_filename', '__builtins__', '__package__', '__doc__', '__name__', '__file__'] sfepy: reading mesh (/home/ankit/sfepy/meshes/2d/rectangle_fine_quad.mesh)... sfepy: ...done in 0.02 s sfepy: creating regions... sfepy: Right sfepy: Bottom sfepy: Top sfepy: Surface sfepy: Omega sfepy: Walls sfepy: Driven sfepy: ...done in 0.03 s sfepy: equation "balance": sfepy: + dw_div_grad.5.Omega(fluid.viscosity, v, u) + dw_convect.5.Omega(v, u) - dw_stokes.5.Omega(v, p) = 0 sfepy: equation "incompressibility": sfepy: dw_stokes.5.Omega(u, q) = 0 sfepy: setting up dof connectivities... sfepy: ...done in 0.00 s sfepy: using solvers: ts: no ts nls: newton ls: ls sfepy: updating variables... sfepy: ...done sfepy: matrix shape: (44557, 44557) sfepy: assembling matrix graph... sfepy: ...done in 0.13 s sfepy: matrix structural nonzeros: 1728264 (8.71e-04% fill) sfepy: updating materials... sfepy: fluid sfepy: ...done in 0.01 s sfepy: nls: iter: 0, residual: 2.023986e-02 (rel: 1.000000e+00) convect_build_vtg(): ERR_Switch mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! Traceback (most recent call last): File "./simple.py", line 146, in <module> main() File "./simple.py", line 143, in main app() File "/home/ankit/sfepy/sfepy/applications/application.py", line 29, in call_basic return self.call(**kwargs) File "/home/ankit/sfepy/sfepy/applications/pde_solver_app.py", line 213, in call nls_status=nls_status) File "/home/ankit/sfepy/sfepy/solvers/ts_solvers.py", line 29, in __call__ state = problem.solve(state0=state0, nls_status=nls_status) File "/home/ankit/sfepy/sfepy/fem/problemDef.py", line 933, in solve vec = solvers.nls(vec0) File "/home/ankit/sfepy/sfepy/solvers/nls.py", line 345, in __call__ mtx_a = fun_grad(vec_x) File "/home/ankit/sfepy/sfepy/fem/evaluate.py", line 66, in eval_tangent_matrix mtx = pb.equations.eval_tangent_matrices(vec, mtx) File "/home/ankit/sfepy/sfepy/fem/equations.py", line 640, in eval_tangent_matrices self.evaluate(mode='weak', dw_mode='matrix', asm_obj=tangent_matrix) File "/home/ankit/sfepy/sfepy/fem/equations.py", line 526, in evaluate asm_obj=asm_obj) File "/home/ankit/sfepy/sfepy/fem/equations.py", line 766, in evaluate ret_status=True) File "/home/ankit/sfepy/sfepy/terms/terms.py", line 1473, in evaluate diff_var, **kwargs) File "/home/ankit/sfepy/sfepy/terms/terms.py", line 1311, in eval_real status = self.call_function(out, fargs) File "/home/ankit/sfepy/sfepy/terms/terms.py", line 1296, in call_function raise ValueError('term evaluation failed! (%s)' % self.name) ValueError: term evaluation failed! (dw_convect)
Did you rebuild the sources?
Regards, Ankit
PS: I was not aware that we have to blog weekly. I will blog for week 1 and 2 asap.
It's either weekly or bi-weekly...
r.
On Tuesday, 2 July 2013 13:58:20 UTC+5:30, Robert Cimrman wrote:
Now it remains to implement a robust flow solver. Even this small example shows, that the solution is not obtained easily - try decreasing the viscosity, and/or increase the Dirichlet velocity - the solver would not converge.
Cheers, r. PS: As mentioned in Terri Oda's e-mail, you should blog about your work so far ASAP!
On 07/01/2013 06:12 PM, Ankit Mahato wrote:
awesome :)
On Monday, 1 July 2013 15:05:16 UTC+5:30, Robert Cimrman wrote:
Hi,
I have removed the "3d only" restriction from the Navier Stokes and related terms. There is also a new example: examples/navier_stokes/navier_stokes2d.py.
On 07/02/2013 02:06 PM, Robert Cimrman wrote:
On 07/02/2013 01:58 PM, Ankit Mahato wrote:
R,
I was getting an error while running the example. ./simple.py examples/navier_stokes/navier_stokes2d.py sfepy: left over: ['data_dir', 'verbose', '_filename', '__builtins__', '__package__', '__doc__', '__name__', '__file__'] sfepy: reading mesh (/home/ankit/sfepy/meshes/2d/rectangle_fine_quad.mesh)... sfepy: ...done in 0.02 s sfepy: creating regions... sfepy: Right sfepy: Bottom sfepy: Top sfepy: Surface sfepy: Omega sfepy: Walls sfepy: Driven sfepy: ...done in 0.03 s sfepy: equation "balance": sfepy: + dw_div_grad.5.Omega(fluid.viscosity, v, u) + dw_convect.5.Omega(v, u) - dw_stokes.5.Omega(v, p) = 0 sfepy: equation "incompressibility": sfepy: dw_stokes.5.Omega(u, q) = 0 sfepy: setting up dof connectivities... sfepy: ...done in 0.00 s sfepy: using solvers: ts: no ts nls: newton ls: ls sfepy: updating variables... sfepy: ...done sfepy: matrix shape: (44557, 44557) sfepy: assembling matrix graph... sfepy: ...done in 0.13 s sfepy: matrix structural nonzeros: 1728264 (8.71e-04% fill) sfepy: updating materials... sfepy: fluid sfepy: ...done in 0.01 s sfepy: nls: iter: 0, residual: 2.023986e-02 (rel: 1.000000e+00) convect_build_vtg(): ERR_Switch mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! mem_free_mem(): error exit! Traceback (most recent call last): File "./simple.py", line 146, in <module> main() File "./simple.py", line 143, in main app() File "/home/ankit/sfepy/sfepy/applications/application.py", line 29, in call_basic return self.call(**kwargs) File "/home/ankit/sfepy/sfepy/applications/pde_solver_app.py", line 213, in call nls_status=nls_status) File "/home/ankit/sfepy/sfepy/solvers/ts_solvers.py", line 29, in __call__ state = problem.solve(state0=state0, nls_status=nls_status) File "/home/ankit/sfepy/sfepy/fem/problemDef.py", line 933, in solve vec = solvers.nls(vec0) File "/home/ankit/sfepy/sfepy/solvers/nls.py", line 345, in __call__ mtx_a = fun_grad(vec_x) File "/home/ankit/sfepy/sfepy/fem/evaluate.py", line 66, in eval_tangent_matrix mtx = pb.equations.eval_tangent_matrices(vec, mtx) File "/home/ankit/sfepy/sfepy/fem/equations.py", line 640, in eval_tangent_matrices self.evaluate(mode='weak', dw_mode='matrix', asm_obj=tangent_matrix) File "/home/ankit/sfepy/sfepy/fem/equations.py", line 526, in evaluate asm_obj=asm_obj) File "/home/ankit/sfepy/sfepy/fem/equations.py", line 766, in evaluate ret_status=True) File "/home/ankit/sfepy/sfepy/terms/terms.py", line 1473, in evaluate diff_var, **kwargs) File "/home/ankit/sfepy/sfepy/terms/terms.py", line 1311, in eval_real status = self.call_function(out, fargs) File "/home/ankit/sfepy/sfepy/terms/terms.py", line 1296, in call_function raise ValueError('term evaluation failed! (%s)' % self.name) ValueError: term evaluation failed! (dw_convect)
Did you rebuild the sources?
I have tried it now on a 32 bit box to be sure, and found no problems...
Regards, Ankit
PS: I was not aware that we have to blog weekly. I will blog for week 1 and 2 asap.
It's either weekly or bi-weekly...
But as it is week no. 3, it should be up ASAP :)
r.
r.
On Tuesday, 2 July 2013 13:58:20 UTC+5:30, Robert Cimrman wrote:
Now it remains to implement a robust flow solver. Even this small example shows, that the solution is not obtained easily - try decreasing the viscosity, and/or increase the Dirichlet velocity - the solver would not converge.
Cheers, r. PS: As mentioned in Terri Oda's e-mail, you should blog about your work so far ASAP!
On 07/01/2013 06:12 PM, Ankit Mahato wrote:
awesome :)
On Monday, 1 July 2013 15:05:16 UTC+5:30, Robert Cimrman wrote:
Hi,
I have removed the "3d only" restriction from the Navier Stokes and related terms. There is also a new example: examples/navier_stokes/navier_stokes2d.py.
On Tuesday, 2 July 2013 13:58:20 UTC+5:30, Robert Cimrman wrote:
Now it remains to implement a robust flow solver. Even this small example shows, that the solution is not obtained easily - try decreasing the viscosity, and/or increase the Dirichlet velocity - the solver would not converge.
Yes R,
The solution is not obtained easily. I am looking into it.
PS: Here are blog posts for week 1 & 2 Kindly tell me if this will do before I send it to terri oda:
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc... http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
Cheers,
r. PS: As mentioned in Terri Oda's e-mail, you should blog about your work so far ASAP!
On 07/01/2013 06:12 PM, Ankit Mahato wrote:
awesome :)
On Monday, 1 July 2013 15:05:16 UTC+5:30, Robert Cimrman wrote:
Hi,
I have removed the "3d only" restriction from the Navier Stokes and related terms. There is also a new example: examples/navier_stokes/navier_stokes2d.py.
r.
On Tuesday, 2 July 2013 19:06:16 UTC+5:30, Ankit Mahato wrote:
On Tuesday, 2 July 2013 13:58:20 UTC+5:30, Robert Cimrman wrote:
Now it remains to implement a robust flow solver. Even this small example shows, that the solution is not obtained easily - try decreasing the viscosity, and/or increase the Dirichlet velocity - the solver would not converge.
Yes R,
The solution is not obtained easily. I am looking into it.
PS: Here are blog posts for week 1 & 2 Kindly tell me if this will do before I send it to terri oda:
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
Hi R,
Just wanted your views. Does the problem of Navier-Strokes solver lies with the implementation or the algorithm which is used.
Cheers,
r. PS: As mentioned in Terri Oda's e-mail, you should blog about your work so far ASAP!
On 07/01/2013 06:12 PM, Ankit Mahato wrote:
awesome :)
On Monday, 1 July 2013 15:05:16 UTC+5:30, Robert Cimrman wrote:
Hi,
I have removed the "3d only" restriction from the Navier Stokes and related terms. There is also a new example: examples/navier_stokes/navier_stokes2d.py.
r.
On 07/02/2013 03:46 PM, Ankit Mahato wrote:
On Tuesday, 2 July 2013 19:06:16 UTC+5:30, Ankit Mahato wrote:
On Tuesday, 2 July 2013 13:58:20 UTC+5:30, Robert Cimrman wrote:
Now it remains to implement a robust flow solver. Even this small example shows, that the solution is not obtained easily - try decreasing the viscosity, and/or increase the Dirichlet velocity - the solver would not converge.
Yes R,
The solution is not obtained easily. I am looking into it.
PS: Here are blog posts for week 1 & 2 Kindly tell me if this will do before I send it to terri oda:
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
Hi R,
Just wanted your views. Does the problem of Navier-Strokes solver lies with the implementation or the algorithm which is used.
Mostly the algorithm, but it might be also the formulation. I am far from CFD, but people there seem to be preferring a dimensionless form of the incompressible NS equations. It also depends on the discretization/FE spaces used. It's really a broad subject, and there is still no a silver bullet solver. Maybe ask your thesis advisor/colleagues doing CFD? Searching the net is really of no help here, as it returns so many things... Expert advice is needed :)
r.
Cheers,
r. PS: As mentioned in Terri Oda's e-mail, you should blog about your work so far ASAP!
On 07/01/2013 06:12 PM, Ankit Mahato wrote:
awesome :)
On Monday, 1 July 2013 15:05:16 UTC+5:30, Robert Cimrman wrote:
Hi,
I have removed the "3d only" restriction from the Navier Stokes and related terms. There is also a new example: examples/navier_stokes/navier_stokes2d.py.
r.
On Tuesday, 2 July 2013 19:23:05 UTC+5:30, Robert Cimrman wrote:
On 07/02/2013 03:46 PM, Ankit Mahato wrote:
On Tuesday, 2 July 2013 19:06:16 UTC+5:30, Ankit Mahato wrote:
On Tuesday, 2 July 2013 13:58:20 UTC+5:30, Robert Cimrman wrote:
Now it remains to implement a robust flow solver. Even this small
example
shows, that the solution is not obtained easily - try decreasing the viscosity, and/or increase the Dirichlet velocity - the solver would not converge.
Yes R,
The solution is not obtained easily. I am looking into it.
PS: Here are blog posts for week 1 & 2 Kindly tell me if this will do before I send it to terri oda:
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
Hi R,
Just wanted your views. Does the problem of Navier-Strokes solver lies with the implementation or the algorithm which is used.
Mostly the algorithm, but it might be also the formulation. I am far from CFD, but people there seem to be preferring a dimensionless form of the incompressible NS equations. It also depends on the discretization/FE spaces used. It's really a broad subject, and there is still no a silver bullet solver. Maybe ask your thesis advisor/colleagues doing CFD? Searching the net is really of no help here, as it returns so many things... Expert advice is needed :)
Yes R dimensionless form of the equations are preferred. I have implemented Patankar's SIMPLE algorithm previously (http://en.wikipedia.org/wiki/SIMPLE_algorithm) [also SIMPLER and SIMPLEC] when I did CFD course but using Finite Difference Method. I will look into it if I find FEM approach. Also I wanted to ask you if we need to stick to FEM for CFD because people use Finite Volume Method [FVM] for CFD.
r.
Cheers,
r. PS: As mentioned in Terri Oda's e-mail, you should blog about your work so far ASAP!
On 07/01/2013 06:12 PM, Ankit Mahato wrote:
awesome :)
On Monday, 1 July 2013 15:05:16 UTC+5:30, Robert Cimrman wrote:
Hi,
I have removed the "3d only" restriction from the Navier Stokes and related terms. There is also a new example: examples/navier_stokes/navier_stokes2d.py.
r.
On 07/02/2013 04:07 PM, Ankit Mahato wrote:
On Tuesday, 2 July 2013 19:23:05 UTC+5:30, Robert Cimrman wrote:
On 07/02/2013 03:46 PM, Ankit Mahato wrote:
On Tuesday, 2 July 2013 19:06:16 UTC+5:30, Ankit Mahato wrote:
On Tuesday, 2 July 2013 13:58:20 UTC+5:30, Robert Cimrman wrote:
Now it remains to implement a robust flow solver. Even this small
example
shows, that the solution is not obtained easily - try decreasing the viscosity, and/or increase the Dirichlet velocity - the solver would not converge.
Yes R,
The solution is not obtained easily. I am looking into it.
PS: Here are blog posts for week 1 & 2 Kindly tell me if this will do before I send it to terri oda:
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
Hi R,
Just wanted your views. Does the problem of Navier-Strokes solver lies with the implementation or the algorithm which is used.
Mostly the algorithm, but it might be also the formulation. I am far from CFD, but people there seem to be preferring a dimensionless form of the incompressible NS equations. It also depends on the discretization/FE spaces used. It's really a broad subject, and there is still no a silver bullet solver. Maybe ask your thesis advisor/colleagues doing CFD? Searching the net is really of no help here, as it returns so many things... Expert advice is needed :)
Yes R dimensionless form of the equations are preferred. I have implemented Patankar's SIMPLE algorithm previously (http://en.wikipedia.org/wiki/SIMPLE_algorithm) [also SIMPLER and SIMPLEC] when I did CFD course but using Finite Difference Method. I will look into it if I find FEM approach. Also I wanted to ask you if we need to stick to FEM for CFD because people use Finite Volume Method [FVM] for CFD.
Yes, we need it in a FEM context, sfepy cannot do FVM. But (big)IMHO those two approaches have things in common, and once you have the matrices and algorithm... It should even be possible to compute fluxes and similar quantities, if needed, though I never tried that and think there is no time to implement FVM in sfepy within your project.
r.
Hi R,
For our Navier-Stokes currently we use the Newton method with backtracking line-search. in OpenFoam and most of the CFD code the linearization approach is based on Patankar's SIMPLE algorithm.1 I talked to my professor who told me that SIMPLE is used in commercial softwares like FLUENT too.
I found few papers which tells us some other approaches. Do have a look at them and lend your views:
- http://www.google.co.in/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&ved=0CC0QFjAA&url=http%3A%2F%2Fwww.wias-berlin.de%2Fpeople%2Fjohn%2FPP99_13.ps&ei=mW3VUZqzOMmzrgeKuYD4DA&usg=AFQjCNEp9_rShrLSjkYdax6bOimSrkD-KQ&sig2=8S654V-zz2vd4mFZOilZCw&bvm=bv.48705608,d.bmk
- http://numerik.iwr.uni-heidelberg.de/Oberwolfach-Seminar/CFD-Course.pdf
- http://dspace.uta.edu/bitstream/handle/10106/5144/JIAJAN_uta_2502M_10764.pdf
- http://www.reaction-eng.com/downloads/nksolver_pernice.pdf
- http://aero-comlab.stanford.edu/Papers/birkenjamesonproceedings09.pdf
- https://cs.uwaterloo.ca/research/tr/1993/02/CS-93-02.pdf
- http://www.cs.sandia.gov/~rstumin/backtrack.pdf
- http://repository.cmu.edu/cgi/viewcontent.cgi?article=1032&context=math
- http://www8.cs.umu.se/kurser/5DA001/HT07/lectures/newton-handouts.pdf
PS: For the python 3 fix which I had forgotten earlier :( . While going through the codes I came across that we use output() in base.py to print. You have already called if sys.version[0] < '3': basestr = basestring else: basestr = str So basically we know the python version and call the print function according to the python version. If I am correct it is quite easy to fix then, am I?
On Tuesday, 2 July 2013 19:57:11 UTC+5:30, Robert Cimrman wrote:
On 07/02/2013 04:07 PM, Ankit Mahato wrote:
On Tuesday, 2 July 2013 19:23:05 UTC+5:30, Robert Cimrman wrote:
On 07/02/2013 03:46 PM, Ankit Mahato wrote:
On Tuesday, 2 July 2013 19:06:16 UTC+5:30, Ankit Mahato wrote:
On Tuesday, 2 July 2013 13:58:20 UTC+5:30, Robert Cimrman wrote:
Now it remains to implement a robust flow solver. Even this small
example
shows, that the solution is not obtained easily - try decreasing the viscosity, and/or increase the Dirichlet velocity - the solver would not converge.
Yes R,
The solution is not obtained easily. I am looking into it.
PS: Here are blog posts for week 1 & 2 Kindly tell me if this will do before I send it to terri oda:
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
Hi R,
Just wanted your views. Does the problem of Navier-Strokes solver lies with the implementation or the algorithm which is used.
Mostly the algorithm, but it might be also the formulation. I am far from CFD, but people there seem to be preferring a dimensionless form of the incompressible NS equations. It also depends on the discretization/FE spaces used. It's really a broad subject, and there is still no a silver bullet solver. Maybe ask your thesis advisor/colleagues doing CFD? Searching
net is really of no help here, as it returns so many things... Expert advice is needed :)
Yes R dimensionless form of the equations are preferred. I have implemented Patankar's SIMPLE algorithm previously (http://en.wikipedia.org/wiki/SIMPLE_algorithm) [also SIMPLER and SIMPLEC] when I did CFD course but using Finite Difference Method. I will look into it if I find FEM approach. Also I wanted to ask you if we need to stick to FEM for CFD because
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc... the people
use Finite Volume Method [FVM] for CFD.
Yes, we need it in a FEM context, sfepy cannot do FVM. But (big)IMHO those two approaches have things in common, and once you have the matrices and algorithm... It should even be possible to compute fluxes and similar quantities, if needed, though I never tried that and think there is no time to implement FVM in sfepy within your project.
r.
On 07/04/2013 03:02 PM, Ankit Mahato wrote:
I found few papers which tells us some other approaches. Do have a look at them and lend your views:
- http://www.google.co.in/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&ved=0CC0QFjAA&url=http%3A%2F%2Fwww.wias-berlin.de%2Fpeople%2Fjohn%2FPP99_13.ps&ei=mW3VUZqzOMmzrgeKuYD4DA&usg=AFQjCNEp9_rShrLSjkYdax6bOimSrkD-KQ&sig2=8S654V-zz2vd4mFZOilZCw&bvm=bv.48705608,d.bmk - http://numerik.iwr.uni-heidelberg.de/Oberwolfach-Seminar/CFD-Course.pdf - http://dspace.uta.edu/bitstream/handle/10106/5144/JIAJAN_uta_2502M_10764.pdf - http://www.reaction-eng.com/downloads/nksolver_pernice.pdf - http://aero-comlab.stanford.edu/Papers/birkenjamesonproceedings09.pdf - https://cs.uwaterloo.ca/research/tr/1993/02/CS-93-02.pdf - http://www.cs.sandia.gov/~rstumin/backtrack.pdf - http://repository.cmu.edu/cgi/viewcontent.cgi?article=1032&context=math - http://www8.cs.umu.se/kurser/5DA001/HT07/lectures/newton-handouts.pdf
From those links, I think only the first two links are interesting for us (other texts describe FV, or are 2D only etc.). In CFD-Course.pdf, check especially section 5, where some approaches to solving the (non)linear system are given (e.g. a Schur complement approach). Unfortunately (from the complexity/time constraint point of view), most people seem to agree that multigrid is the way to go.
To proceed with the gsoc, maybe it would be good if you, in parallel to this, tried to create an example with all the equations coupled, to have something to play with. It could be small and use non-realistic viscosity to make the solution easier. Use a 2D mesh, as that could be made reasonably fine.
Then you could develop/try some iterative schemes to solve the system. (e.g. solve flow i -> solve energy i -> solve flow i+1 ...).
What do you think? r.
On Wednesday, 10 July 2013 22:19:47 UTC+5:30, Robert Cimrman wrote:
On 07/04/2013 03:02 PM, Ankit Mahato wrote:
I found few papers which tells us some other approaches. Do have a look at them and lend your views:
--http://numerik.iwr.uni-heidelberg.de/Oberwolfach-Seminar/CFD-Course.pdf
-http://dspace.uta.edu/bitstream/handle/10106/5144/JIAJAN_uta_2502M_10764.pdf
- http://www.reaction-eng.com/downloads/nksolver_pernice.pdf -http://aero-comlab.stanford.edu/Papers/birkenjamesonproceedings09.pdf
- https://cs.uwaterloo.ca/research/tr/1993/02/CS-93-02.pdf - http://www.cs.sandia.gov/~rstumin/backtrack.pdf -http://repository.cmu.edu/cgi/viewcontent.cgi?article=1032&context=math
-http://www8.cs.umu.se/kurser/5DA001/HT07/lectures/newton-handouts.pdf
From those links, I think only the first two links are interesting for us (other texts describe FV, or are 2D only etc.). In CFD-Course.pdf, check especially section 5, where some approaches to solving the (non)linear system are given (e.g. a Schur complement approach). Unfortunately (from the complexity/time constraint point of view), most people seem to agree that multigrid is the way to go.
okey
To proceed with the gsoc, maybe it would be good if you, in parallel to this, tried to create an example with all the equations coupled, to have something to play with. It could be small and use non-realistic viscosity to make the solution easier. Use a 2D mesh, as that could be made reasonably fine.
okey
Then you could develop/try some iterative schemes to solve the system. (e.g. solve flow i -> solve energy i -> solve flow i+1 ...).
What do you think?
Yes this is the iterative scheme i have to apply. But currently everything is steady state. The final problem has to be unsteady as then phase change effect will be seen evolving with time.
r.
PS: By the way I recieved a mail from the professor that he has caught viral fever due to changing weather so he would not be able to meet me this week. I also caught fever yesterday and i just got out of bed a few minutes back so couldn't reply to the thread earlier.
On 07/02/2013 03:36 PM, Ankit Mahato wrote:
On Tuesday, 2 July 2013 13:58:20 UTC+5:30, Robert Cimrman wrote:
Now it remains to implement a robust flow solver. Even this small example shows, that the solution is not obtained easily - try decreasing the viscosity, and/or increase the Dirichlet velocity - the solver would not converge.
Yes R,
The solution is not obtained easily. I am looking into it.
PS: Here are blog posts for week 1 & 2 Kindly tell me if this will do before I send it to terri oda:
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc... http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
It seems ok, just correct the following typo: Navier-Strokes -> Navier-Stokes :)
It would be interesting to see the Peclet number graphs. Also, did you try some other, more interesting, geometries?
r.
Cheers,
r. PS: As mentioned in Terri Oda's e-mail, you should blog about your work so far ASAP!
On 07/01/2013 06:12 PM, Ankit Mahato wrote:
awesome :)
On Monday, 1 July 2013 15:05:16 UTC+5:30, Robert Cimrman wrote:
Hi,
I have removed the "3d only" restriction from the Navier Stokes and related terms. There is also a new example: examples/navier_stokes/navier_stokes2d.py.
r.
On Tuesday, 2 July 2013 19:16:37 UTC+5:30, Robert Cimrman wrote:
On 07/02/2013 03:36 PM, Ankit Mahato wrote:
On Tuesday, 2 July 2013 13:58:20 UTC+5:30, Robert Cimrman wrote:
Now it remains to implement a robust flow solver. Even this small
example
shows, that the solution is not obtained easily - try decreasing the viscosity, and/or increase the Dirichlet velocity - the solver would not converge.
Yes R,
The solution is not obtained easily. I am looking into it.
PS: Here are blog posts for week 1 & 2 Kindly tell me if this will do before I send it to terri oda:
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
It seems ok, just correct the following typo: Navier-Strokes -> Navier-Stokes :)
Done.
It would be interesting to see the Peclet number graphs. Also, did you try some other, more interesting, geometries?
Actually I varied the convective velocity and c to observe the variation as I had the graph in my book alongside. It was in accordance but I did not plot it [ grave mistake :( ] and moved ahead at looking into the 3d navier stokes code.
r.
Cheers,
r. PS: As mentioned in Terri Oda's e-mail, you should blog about your work so far ASAP!
On 07/01/2013 06:12 PM, Ankit Mahato wrote:
awesome :)
On Monday, 1 July 2013 15:05:16 UTC+5:30, Robert Cimrman wrote:
Hi,
I have removed the "3d only" restriction from the Navier Stokes and related terms. There is also a new example: examples/navier_stokes/navier_stokes2d.py.
r.
On 07/02/2013 04:02 PM, Ankit Mahato wrote:
On Tuesday, 2 July 2013 19:16:37 UTC+5:30, Robert Cimrman wrote:
On 07/02/2013 03:36 PM, Ankit Mahato wrote:
On Tuesday, 2 July 2013 13:58:20 UTC+5:30, Robert Cimrman wrote:
Now it remains to implement a robust flow solver. Even this small
example
shows, that the solution is not obtained easily - try decreasing the viscosity, and/or increase the Dirichlet velocity - the solver would not converge.
Yes R,
The solution is not obtained easily. I am looking into it.
PS: Here are blog posts for week 1 & 2 Kindly tell me if this will do before I send it to terri oda:
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
It seems ok, just correct the following typo: Navier-Strokes -> Navier-Stokes :)
Done.
It would be interesting to see the Peclet number graphs. Also, did you try some other, more interesting, geometries?
Actually I varied the convective velocity and c to observe the variation as I had the graph in my book alongside. It was in accordance but I did not plot it [ grave mistake :( ] and moved ahead at looking into the 3d navier stokes code.
Maybe you could start using the ipython notebooks for that kind of plots/exploration? It might be handy both for your thesis, and the blog.
r.
r.
Cheers,
r. PS: As mentioned in Terri Oda's e-mail, you should blog about your work so far ASAP!
On 07/01/2013 06:12 PM, Ankit Mahato wrote:
awesome :)
On Monday, 1 July 2013 15:05:16 UTC+5:30, Robert Cimrman wrote:
Hi,
I have removed the "3d only" restriction from the Navier Stokes and related terms. There is also a new example: examples/navier_stokes/navier_stokes2d.py.
r.
On Tuesday, 2 July 2013 19:35:30 UTC+5:30, Robert Cimrman wrote:
On 07/02/2013 04:02 PM, Ankit Mahato wrote:
On Tuesday, 2 July 2013 19:16:37 UTC+5:30, Robert Cimrman wrote:
On 07/02/2013 03:36 PM, Ankit Mahato wrote:
On Tuesday, 2 July 2013 13:58:20 UTC+5:30, Robert Cimrman wrote:
Now it remains to implement a robust flow solver. Even this small
example
shows, that the solution is not obtained easily - try decreasing the viscosity, and/or increase the Dirichlet velocity - the solver would not
Yes R,
The solution is not obtained easily. I am looking into it.
PS: Here are blog posts for week 1 & 2 Kindly tell me if this will do before I send it to terri oda:
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
http://ankitmahato.blogspot.in/2013/07/python-software-foundation-sfepy-gsoc...
It seems ok, just correct the following typo: Navier-Strokes -> Navier-Stokes :)
Done.
It would be interesting to see the Peclet number graphs. Also, did you
some other, more interesting, geometries?
Actually I varied the convective velocity and c to observe the variation as I had the graph in my book alongside. It was in accordance but I did not plot it [ grave mistake :( ] and moved ahead at looking into the 3d
converge. try navier
stokes code.
Maybe you could start using the ipython notebooks for that kind of plots/exploration? It might be handy both for your thesis, and the blog.
Okie R. Will do so form now onwards.
r.
r.
Cheers,
r. PS: As mentioned in Terri Oda's e-mail, you should blog about your work so far ASAP!
On 07/01/2013 06:12 PM, Ankit Mahato wrote:
awesome :)
On Monday, 1 July 2013 15:05:16 UTC+5:30, Robert Cimrman wrote: > > Hi, > > I have removed the "3d only" restriction from the Navier Stokes and > related > terms. There is also a new example: > examples/navier_stokes/navier_stokes2d.py. > > r. >
participants (2)
-
Ankit Mahato -
Robert Cimrman