Hi, i cant get Nektar++ to work with 3D meshes containing tetrahedrons or any other 3D elements with triangular faces. The problem occurs when the local base vertex is the vertex with the largest angle. I first suspected a bug in the APESolver but noticed that the problem occurs with the NavierStokesCFE system as well. The ADRSolver is not affected as far as i can tell. I appended four test cases to demonstrate the problem. Two APE cases with two prismatic elements with the local base vertex in the right angle vertex (APE_3DChannel_1LPrismX_laxFri.xml) and in the 45° angle vertex (APE_3DChannel_1SPrismX_laxFri.xml). The two ther cases use the NavierStokesCFE equations on hexahedrons and prisms. Is there something wrong with my cases setup or is this something that isnt working in nektar++? Cheers, Kilian -- Kilian Lackhove, M.Sc. Fachgebiet für Energie- und Kraftwerkstechnik L1|08 114 Technische Universität Darmstadt Jovanka-Bontschits-Straße 2 D-64287 Darmstadt Germany Tel.: +49 6151 16 - 5186 Fax: +49 6151 16 - 6555 e-mail: lackhove@ekt.tu-darmstadt.de
Hi Killian, I have just tested your APE files and they do not seem to give a good answer in a fix/GetFacePhys branch with error s of about 200 000 in p. However it does seem to give a more reasonable answer in the feature/substepping branch. So I do not know what the issue is but I have undertaking various developments in the substepping branch to optimise for my use of Tets and Prism in the Incompressibe Navier Stokes solver. The output for one of your cases is given below. Is this correct? Cheers, Spencer. Spencers-MacBook-Pro:APESolver ssherw$ ./APESolver-g APE_3DChannel_1LPrismX_laxFri.xml ======================================================================= EquationType: APE Session Name: APE_3DChannel_1LPrismX_laxFri Spatial Dim.: 3 Max SEM Exp. Order: 5 Expansion Dim.: 3 Riemann Solver: LaxFriedrichs Advection Type: Projection Type: Discontinuous Galerkin Advection: explicit AdvectionType: WeakDG Diffusion: explicit Time Step: 2.93258e-07 No. of Steps: 1000 Checkpoints (steps): 100 Integration Type: ClassicalRungeKutta4 ======================================================================= Initial Conditions: - Field p: 100*(0.5*sin(k*x-PI/2)+0.5) - Field u: 100*(0.5*sin(k*x-PI/2)+0.5) / (Rho0 * c) - Field v: 0 - Field w: 0 Writing: "APE_3DChannel_1LPrismX_laxFri_0.chk" Steps: 100 Time: 2.93258e-05 CPU Time: 0.377132s Writing: "APE_3DChannel_1LPrismX_laxFri_1.chk" Steps: 200 Time: 5.86515e-05 CPU Time: 0.375962s Writing: "APE_3DChannel_1LPrismX_laxFri_2.chk" Steps: 300 Time: 8.79773e-05 CPU Time: 0.378889s Writing: "APE_3DChannel_1LPrismX_laxFri_3.chk" Steps: 400 Time: 0.000117303 CPU Time: 0.372519s Writing: "APE_3DChannel_1LPrismX_laxFri_4.chk" Steps: 500 Time: 0.000146629 CPU Time: 0.375492s Writing: "APE_3DChannel_1LPrismX_laxFri_5.chk" Steps: 600 Time: 0.000175955 CPU Time: 0.374156s Writing: "APE_3DChannel_1LPrismX_laxFri_6.chk" Steps: 700 Time: 0.00020528 CPU Time: 0.372089s Writing: "APE_3DChannel_1LPrismX_laxFri_7.chk" Steps: 800 Time: 0.000234606 CPU Time: 0.374489s Writing: "APE_3DChannel_1LPrismX_laxFri_8.chk" Steps: 900 Time: 0.000263932 CPU Time: 0.379276s Writing: "APE_3DChannel_1LPrismX_laxFri_9.chk" Steps: 1000 Time: 0.000293258 CPU Time: 0.402799s Writing: "APE_3DChannel_1LPrismX_laxFri_10.chk" Time-integration : 3.7828s Writing: "APE_3DChannel_1LPrismX_laxFri.fld" ------------------------------------------- Total Computation Time = 4s ------------------------------------------- L 2 error (variable p) : 6.94608 L inf error (variable p) : 24.6149 L 2 error (variable u) : 0.0169185 L inf error (variable u) : 0.0599543 L 2 error (variable v) : 1.14782e-15 L inf error (variable v) : 7.35552e-15 L 2 error (variable w) : 6.57571e-16 L inf error (variable w) : 4.27839e-15 Hi, i cant get Nektar++ to work with 3D meshes containing tetrahedrons or any other 3D elements with triangular faces. The problem occurs when the local base vertex is the vertex with the largest angle. I first suspected a bug in the APESolver but noticed that the problem occurs with the NavierStokesCFE system as well. The ADRSolver is not affected as far as i can tell. I appended four test cases to demonstrate the problem. Two APE cases with two prismatic elements with the local base vertex in the right angle vertex (APE_3DChannel_1LPrismX_laxFri.xml) and in the 45° angle vertex (APE_3DChannel_1SPrismX_laxFri.xml). The two ther cases use the NavierStokesCFE equations on hexahedrons and prisms. Is there something wrong with my cases setup or is this something that isnt working in nektar++? Cheers, Kilian -- Kilian Lackhove, M.Sc. Fachgebiet für Energie- und Kraftwerkstechnik L1|08 114 Technische Universität Darmstadt Jovanka-Bontschits-Straße 2 D-64287 Darmstadt Germany Tel.: +49 6151 16 - 5186 Fax: +49 6151 16 - 6555 e-mail: lackhove@ekt.tu-darmstadt.de<mailto:lackhove@ekt.tu-darmstadt.de><APE_3DChannel_1LPrismX_laxFri.xml><APE_3DChannel_1SPrismX_laxFri.xml><Couette_WeakDG_LDG_SEM_3D.xml><Couette_WeakDG_LDG_SEM_3D_prismX.xml>_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk<mailto:Nektar-users@imperial.ac.uk> https://mailman.ic.ac.uk/mailman/listinfo/nektar-users Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ s.sherwin@imperial.ac.uk<mailto:s.sherwin@imperial.ac.uk> +44 (0) 20 759 45052
Hi Killian, I think the branch feature/DGopt may also work and so I guess the issue is related to these updates? Cheers, Spencer. On 19 Dec 2014, at 16:36, Sherwin, Spencer J <s.sherwin@imperial.ac.uk<mailto:s.sherwin@imperial.ac.uk>> wrote: Hi Killian, I have just tested your APE files and they do not seem to give a good answer in a fix/GetFacePhys branch with error s of about 200 000 in p. However it does seem to give a more reasonable answer in the feature/substepping branch. So I do not know what the issue is but I have undertaking various developments in the substepping branch to optimise for my use of Tets and Prism in the Incompressibe Navier Stokes solver. The output for one of your cases is given below. Is this correct? Cheers, Spencer. Spencers-MacBook-Pro:APESolver ssherw$ ./APESolver-g APE_3DChannel_1LPrismX_laxFri.xml ======================================================================= EquationType: APE Session Name: APE_3DChannel_1LPrismX_laxFri Spatial Dim.: 3 Max SEM Exp. Order: 5 Expansion Dim.: 3 Riemann Solver: LaxFriedrichs Advection Type: Projection Type: Discontinuous Galerkin Advection: explicit AdvectionType: WeakDG Diffusion: explicit Time Step: 2.93258e-07 No. of Steps: 1000 Checkpoints (steps): 100 Integration Type: ClassicalRungeKutta4 ======================================================================= Initial Conditions: - Field p: 100*(0.5*sin(k*x-PI/2)+0.5) - Field u: 100*(0.5*sin(k*x-PI/2)+0.5) / (Rho0 * c) - Field v: 0 - Field w: 0 Writing: "APE_3DChannel_1LPrismX_laxFri_0.chk" Steps: 100 Time: 2.93258e-05 CPU Time: 0.377132s Writing: "APE_3DChannel_1LPrismX_laxFri_1.chk" Steps: 200 Time: 5.86515e-05 CPU Time: 0.375962s Writing: "APE_3DChannel_1LPrismX_laxFri_2.chk" Steps: 300 Time: 8.79773e-05 CPU Time: 0.378889s Writing: "APE_3DChannel_1LPrismX_laxFri_3.chk" Steps: 400 Time: 0.000117303 CPU Time: 0.372519s Writing: "APE_3DChannel_1LPrismX_laxFri_4.chk" Steps: 500 Time: 0.000146629 CPU Time: 0.375492s Writing: "APE_3DChannel_1LPrismX_laxFri_5.chk" Steps: 600 Time: 0.000175955 CPU Time: 0.374156s Writing: "APE_3DChannel_1LPrismX_laxFri_6.chk" Steps: 700 Time: 0.00020528 CPU Time: 0.372089s Writing: "APE_3DChannel_1LPrismX_laxFri_7.chk" Steps: 800 Time: 0.000234606 CPU Time: 0.374489s Writing: "APE_3DChannel_1LPrismX_laxFri_8.chk" Steps: 900 Time: 0.000263932 CPU Time: 0.379276s Writing: "APE_3DChannel_1LPrismX_laxFri_9.chk" Steps: 1000 Time: 0.000293258 CPU Time: 0.402799s Writing: "APE_3DChannel_1LPrismX_laxFri_10.chk" Time-integration : 3.7828s Writing: "APE_3DChannel_1LPrismX_laxFri.fld" ------------------------------------------- Total Computation Time = 4s ------------------------------------------- L 2 error (variable p) : 6.94608 L inf error (variable p) : 24.6149 L 2 error (variable u) : 0.0169185 L inf error (variable u) : 0.0599543 L 2 error (variable v) : 1.14782e-15 L inf error (variable v) : 7.35552e-15 L 2 error (variable w) : 6.57571e-16 L inf error (variable w) : 4.27839e-15 Hi, i cant get Nektar++ to work with 3D meshes containing tetrahedrons or any other 3D elements with triangular faces. The problem occurs when the local base vertex is the vertex with the largest angle. I first suspected a bug in the APESolver but noticed that the problem occurs with the NavierStokesCFE system as well. The ADRSolver is not affected as far as i can tell. I appended four test cases to demonstrate the problem. Two APE cases with two prismatic elements with the local base vertex in the right angle vertex (APE_3DChannel_1LPrismX_laxFri.xml) and in the 45° angle vertex (APE_3DChannel_1SPrismX_laxFri.xml). The two ther cases use the NavierStokesCFE equations on hexahedrons and prisms. Is there something wrong with my cases setup or is this something that isnt working in nektar++? Cheers, Kilian -- Kilian Lackhove, M.Sc. Fachgebiet für Energie- und Kraftwerkstechnik L1|08 114 Technische Universität Darmstadt Jovanka-Bontschits-Straße 2 D-64287 Darmstadt Germany Tel.: +49 6151 16 - 5186 Fax: +49 6151 16 - 6555 e-mail: lackhove@ekt.tu-darmstadt.de<mailto:lackhove@ekt.tu-darmstadt.de><APE_3DChannel_1LPrismX_laxFri.xml><APE_3DChannel_1SPrismX_laxFri.xml><Couette_WeakDG_LDG_SEM_3D.xml><Couette_WeakDG_LDG_SEM_3D_prismX.xml>_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk<mailto:Nektar-users@imperial.ac.uk> https://mailman.ic.ac.uk/mailman/listinfo/nektar-users Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ s.sherwin@imperial.ac.uk<mailto:s.sherwin@imperial.ac.uk> +44 (0) 20 759 45052 _______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk<mailto:Nektar-users@imperial.ac.uk> https://mailman.ic.ac.uk/mailman/listinfo/nektar-users Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ s.sherwin@imperial.ac.uk<mailto:s.sherwin@imperial.ac.uk> +44 (0) 20 759 45052
Hi Spencer, i tested the fix/GetFacePhys and the feature/DGopt branches and bothwork fine for the testcases from my last email. I am going to run some additional tests next week and will keep you updated if they worked, too. I also want to identify the changes that fix the problem, so i can cherry-pick them into my APE branches until the feature/DGopt branch is merged into master. Many Thanks, Kilian Am Fr, 19. 12 2014, 16:48:49 schrieb Sherwin, Spencer J:
Hi Killian,
I think the branch feature/DGopt may also work and so I guess the issue is related to these updates?
Cheers, Spencer.
On 19 Dec 2014, at 16:36, Sherwin, Spencer J <s.sherwin@imperial.ac.uk<mailto:s.sherwin@imperial.ac.uk>> wrote:
Hi Killian,
I have just tested your APE files and they do not seem to give a good answer in a fix/GetFacePhys branch with error s of about 200 000 in p. However it does seem to give a more reasonable answer in the feature/substepping branch. So I do not know what the issue is but I have undertaking various developments in the substepping branch to optimise for my use of Tets and Prism in the Incompressibe Navier Stokes solver. The output for one of your cases is given below. Is this correct?
Cheers, Spencer.
Spencers-MacBook-Pro:APESolver ssherw$ ./APESolver-g APE_3DChannel_1LPrismX_laxFri.xml ======================================================================= EquationType: APE Session Name: APE_3DChannel_1LPrismX_laxFri Spatial Dim.: 3 Max SEM Exp. Order: 5 Expansion Dim.: 3 Riemann Solver: LaxFriedrichs Advection Type: Projection Type: Discontinuous Galerkin Advection: explicit AdvectionType: WeakDG Diffusion: explicit Time Step: 2.93258e-07 No. of Steps: 1000 Checkpoints (steps): 100 Integration Type: ClassicalRungeKutta4 ======================================================================= Initial Conditions: - Field p: 100*(0.5*sin(k*x-PI/2)+0.5) - Field u: 100*(0.5*sin(k*x-PI/2)+0.5) / (Rho0 * c) - Field v: 0 - Field w: 0 Writing: "APE_3DChannel_1LPrismX_laxFri_0.chk" Steps: 100 Time: 2.93258e-05 CPU Time: 0.377132s Writing: "APE_3DChannel_1LPrismX_laxFri_1.chk" Steps: 200 Time: 5.86515e-05 CPU Time: 0.375962s Writing: "APE_3DChannel_1LPrismX_laxFri_2.chk" Steps: 300 Time: 8.79773e-05 CPU Time: 0.378889s Writing: "APE_3DChannel_1LPrismX_laxFri_3.chk" Steps: 400 Time: 0.000117303 CPU Time: 0.372519s Writing: "APE_3DChannel_1LPrismX_laxFri_4.chk" Steps: 500 Time: 0.000146629 CPU Time: 0.375492s Writing: "APE_3DChannel_1LPrismX_laxFri_5.chk" Steps: 600 Time: 0.000175955 CPU Time: 0.374156s Writing: "APE_3DChannel_1LPrismX_laxFri_6.chk" Steps: 700 Time: 0.00020528 CPU Time: 0.372089s Writing: "APE_3DChannel_1LPrismX_laxFri_7.chk" Steps: 800 Time: 0.000234606 CPU Time: 0.374489s Writing: "APE_3DChannel_1LPrismX_laxFri_8.chk" Steps: 900 Time: 0.000263932 CPU Time: 0.379276s Writing: "APE_3DChannel_1LPrismX_laxFri_9.chk" Steps: 1000 Time: 0.000293258 CPU Time: 0.402799s Writing: "APE_3DChannel_1LPrismX_laxFri_10.chk" Time-integration : 3.7828s Writing: "APE_3DChannel_1LPrismX_laxFri.fld" ------------------------------------------- Total Computation Time = 4s ------------------------------------------- L 2 error (variable p) : 6.94608 L inf error (variable p) : 24.6149 L 2 error (variable u) : 0.0169185 L inf error (variable u) : 0.0599543 L 2 error (variable v) : 1.14782e-15 L inf error (variable v) : 7.35552e-15 L 2 error (variable w) : 6.57571e-16 L inf error (variable w) : 4.27839e-15
Hi,
i cant get Nektar++ to work with 3D meshes containing tetrahedrons or any other 3D elements with triangular faces. The problem occurs when the local base vertex is the vertex with the largest angle.
I first suspected a bug in the APESolver but noticed that the problem occurs with the NavierStokesCFE system as well. The ADRSolver is not affected as far as i can tell.
I appended four test cases to demonstrate the problem. Two APE cases with two prismatic elements with the local base vertex in the right angle vertex (APE_3DChannel_1LPrismX_laxFri.xml) and in the 45° angle vertex (APE_3DChannel_1SPrismX_laxFri.xml). The two ther cases use the NavierStokesCFE equations on hexahedrons and prisms.
Is there something wrong with my cases setup or is this something that isnt working in nektar++?
Cheers,
Kilian
-- Kilian Lackhove, M.Sc. Fachgebiet für Energie- und Kraftwerkstechnik L1|08 114 Technische Universität Darmstadt Jovanka-Bontschits-Straße 2 D-64287 Darmstadt Germany
Tel.: +49 6151 16 - 5186 Fax: +49 6151 16 - 6555 e-mail: lackhove@ekt.tu-darmstadt.de<mailto:lackhove@ekt.tu-darmstadt.de><APE_3DCha nnel_1LPrismX_laxFri.xml><APE_3DChannel_1SPrismX_laxFri.xml><Couette_WeakDG_ LDG_SEM_3D.xml><Couette_WeakDG_LDG_SEM_3D_prismX.xml>_______________________ ________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk<mailto:Nektar-users@imperial.ac.uk> https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ
s.sherwin@imperial.ac.uk<mailto:s.sherwin@imperial.ac.uk> +44 (0) 20 759 45052
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk<mailto:Nektar-users@imperial.ac.uk> https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ
s.sherwin@imperial.ac.uk<mailto:s.sherwin@imperial.ac.uk> +44 (0) 20 759 45052
participants (2)
-
Kilian Lackhove
-
Sherwin, Spencer J