BiGlobal Stability Analysis of 3D Channel Flow case at Re15000
Dear Nektarers, I am running a 3D direct stability analysis test case, the Re15000 Channel Flow from the tutorial (PPF_R15000_3D, ARPACK as the solver). As I understand, this is a BiGlobal stability analysis with 2D base flow and the spanwise wavenumber Beta. It means that for a given Beta, the Eigenvalues can be calculated. What I am seeking is the Eigenvalues under different spanwise wavenumber Beta, so I turn "Beta Set To Zero" as false (as in the PPF_R15000_3D). And the solver converged showing 4 Eigenvalues with Real and Imaginary part of each, with no information of Beta. How can I find the Eigenvalues under different spanwise wavenumber Beta? Thank you very much. Best Regards, Xu
Hi Xu, You need to manipulate the Lz parameter which specifies the distance over which the expansion is performed (Lz=2Pi / Beta). Best, Stan On czw, 2018-11-08 at 17:54 +0100, Xu Chu wrote:
Dear Nektarers, I am running a 3D direct stability analysis test case, the Re15000 Channel Flow from the tutorial (PPF_R15000_3D, ARPACK as the solver). As I understand, this is a BiGlobal stability analysis with 2D base flow and the spanwise wavenumber Beta. It means that for a given Beta, the Eigenvalues can be calculated. What I am seeking is the Eigenvalues under different spanwise wavenumber Beta, so I turn "Beta Set To Zero" as false (as in the PPF_R15000_3D). And the solver converged showing 4 Eigenvalues with Real and Imaginary part of each, with no information of Beta. How can I find the Eigenvalues under different spanwise wavenumber Beta? Thank you very much.
Best Regards, Xu
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
Hi Stan, thank you very much. I tried it, it works! I have another question about the ARPACK and modified Arnoldi. I am now using ARPACK for BiGlobal LST (2D base flow), which is serial. I tried to use modified Arnoldi, which should be parallel. But it runs parallel only with up to 4 cores. If more cores are given, it just stopped there without results output. Does it have anything to do with the limited mode (single mode) in the spanwise direction? I already try direct solver, multigrid and XXT. Best, Xu On 11/09/2018 09:52 AM, Stanisław Gepner wrote:
Hi Xu,
You need to manipulate the Lz parameter which specifies the distance over which the expansion is performed (Lz=2Pi / Beta).
Best, Stan
On czw, 2018-11-08 at 17:54 +0100, Xu Chu wrote:
Dear Nektarers, I am running a 3D direct stability analysis test case, the Re15000 Channel Flow from the tutorial (PPF_R15000_3D, ARPACK as the solver). As I understand, this is a BiGlobal stability analysis with 2D base flow and the spanwise wavenumber Beta. It means that for a given Beta, the Eigenvalues can be calculated. What I am seeking is the Eigenvalues under different spanwise wavenumber Beta, so I turn "Beta Set To Zero" as false (as in the PPF_R15000_3D). And the solver converged showing 4 Eigenvalues with Real and Imaginary part of each, with no information of Beta. How can I find the Eigenvalues under different spanwise wavenumber Beta? Thank you very much.
Best Regards, Xu
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
Hi Xu, It has been a while since I had fun with using the stability solver, so I might be mixing something up. I am not sure as to why your simulation stops, but could you possibly try with a larger mesh? Could it be that the mesh you are using, when partitioned, result in a small number of elements per core? Also, have you tried running the pure 2D case without the homogeneous expansion in parallel, and did it behave the same - i.e. just stops? Best, Stan On czw, 2018-11-15 at 16:57 +0100, Xu Chu wrote:
Hi Stan,
thank you very much. I tried it, it works!
I have another question about the ARPACK and modified Arnoldi. I am now using ARPACK for BiGlobal LST (2D base flow), which is serial. I tried to use modified Arnoldi, which should be parallel. But it runs parallel only with up to 4 cores. If more cores are given, it just stopped there without results output. Does it have anything to do with the limited mode (single mode) in the spanwise direction? I already try direct solver, multigrid and XXT.
Best,
Xu
On 11/09/2018 09:52 AM, Stanisław Gepner wrote:
Hi Xu,
You need to manipulate the Lz parameter which specifies the distance over which the expansion is performed (Lz=2Pi / Beta).
Best, Stan
On czw, 2018-11-08 at 17:54 +0100, Xu Chu wrote:
Dear Nektarers, I am running a 3D direct stability analysis test case, the Re15000 Channel Flow from the tutorial (PPF_R15000_3D, ARPACK as the solver). As I understand, this is a BiGlobal stability analysis with 2D base flow and the spanwise wavenumber Beta. It means that for a given Beta, the Eigenvalues can be calculated. What I am seeking is the Eigenvalues under different spanwise wavenumber Beta, so I turn "Beta Set To Zero" as false (as in the PPF_R15000_3D). And the solver converged showing 4 Eigenvalues with Real and Imaginary part of each, with no information of Beta. How can I find the Eigenvalues under different spanwise wavenumber Beta? Thank you very much.
Best Regards, Xu
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
Hallo Stan, thank you for your kind response. After some tryouts, I figured it out. It seems Nektar in general is very sensitive to the mesh load pro core. Small number of elements pro core or large number both will not work. The same thing happens in the DNS- the simulation (both direct solver and XXT) just hang in there without giving any results. In this case, I just increased the number of partitions, and the modified Arnoldi solver runs. Best, Xu Chu On 11/16/2018 10:24 AM, Stanisław Gepner wrote:
Hi Xu,
It has been a while since I had fun with using the stability solver, so I might be mixing something up.
I am not sure as to why your simulation stops, but could you possibly try with a larger mesh? Could it be that the mesh you are using, when partitioned, result in a small number of elements per core? Also, have you tried running the pure 2D case without the homogeneous expansion in parallel, and did it behave the same - i.e. just stops?
Best, Stan
On czw, 2018-11-15 at 16:57 +0100, Xu Chu wrote:
Hi Stan,
thank you very much. I tried it, it works!
I have another question about the ARPACK and modified Arnoldi. I am now using ARPACK for BiGlobal LST (2D base flow), which is serial. I tried to use modified Arnoldi, which should be parallel. But it runs parallel only with up to 4 cores. If more cores are given, it just stopped there without results output. Does it have anything to do with the limited mode (single mode) in the spanwise direction? I already try direct solver, multigrid and XXT.
Best,
Xu
On 11/09/2018 09:52 AM, Stanisław Gepner wrote:
Hi Xu,
You need to manipulate the Lz parameter which specifies the distance over which the expansion is performed (Lz=2Pi / Beta).
Best, Stan
On czw, 2018-11-08 at 17:54 +0100, Xu Chu wrote:
Dear Nektarers, I am running a 3D direct stability analysis test case, the Re15000 Channel Flow from the tutorial (PPF_R15000_3D, ARPACK as the solver). As I understand, this is a BiGlobal stability analysis with 2D base flow and the spanwise wavenumber Beta. It means that for a given Beta, the Eigenvalues can be calculated. What I am seeking is the Eigenvalues under different spanwise wavenumber Beta, so I turn "Beta Set To Zero" as false (as in the PPF_R15000_3D). And the solver converged showing 4 Eigenvalues with Real and Imaginary part of each, with no information of Beta. How can I find the Eigenvalues under different spanwise wavenumber Beta? Thank you very much.
Best Regards, Xu
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
participants (2)
-
Stanisław Gepner
-
Xu Chu