Using Nektar for solving compressible flow case
Hi All, I am trying using Nektar for solving the compressible flow over a cavity at Re - 10^7 per meter. I tried to set the parameters as per the user guide but the solution terminates saying "NaN error found during time integration". Initially, I used structured grid with fine mesh near the walls, but it didn't work. And when I use a coarse structured grid it runs for a while and then terminates saying the same issue. When I tried with a lesser Reynolds no. (0.3x10^7) and a coarse unstructured grid, I worked but had a very bad residual. The tutorials provided by Nektar are all cases of incompressible flow, which works fine. I can't find any examples of using compressible flow solver in the tutorials. Can anyone please give me an example of how to set the parameters and solver info for using compressible flow solver. And what changes in the parameters and solverinfo should be done to solve steady cases of compressible flow in Nektar. Thanking you. Regards, Dev
Hi All, Hi Dev, I have similar problems by running a simulation of an airfoil at high Re (Re=408000) with the compressible flow solver. It terminates every time with the same report "Nan found during time integration". Actually I still have no solution for that, but it seems to be the same problem. Maybe somebody can help us? Best regards Fabian ________________________________ Von: nektar-users-bounces@imperial.ac.uk [nektar-users-bounces@imperial.ac.uk]" im Auftrag von "devabratasahoo@campus.technion.ac.il [devabratasahoo@campus.technion.ac.il] Gesendet: Montag, 5. Dezember 2016 09:32 An: nektar-users@imperial.ac.uk Betreff: [Nektar-users] Using Nektar for solving compressible flow case Hi All, I am trying using Nektar for solving the compressible flow over a cavity at Re - 10^7 per meter. I tried to set the parameters as per the user guide but the solution terminates saying "NaN error found during time integration". Initially, I used structured grid with fine mesh near the walls, but it didn't work. And when I use a coarse structured grid it runs for a while and then terminates saying the same issue. When I tried with a lesser Reynolds no. (0.3x10^7) and a coarse unstructured grid, I worked but had a very bad residual. The tutorials provided by Nektar are all cases of incompressible flow, which works fine. I can't find any examples of using compressible flow solver in the tutorials. Can anyone please give me an example of how to set the parameters and solver info for using compressible flow solver. And what changes in the parameters and solverinfo should be done to solve steady cases of compressible flow in Nektar. Thanking you. Regards, Dev
Hi both, Happy to help you with this but I have to give a fairly pessimistic warning, which is that these cases are likely to be very challenging with the compressible solver, due to the explicit time-stepping in use in this code. This might be somewhat mitigated by your goals -- i.e. do you want to do a full DNS or under resolved (uDNS/ILES) simulation? In terms of why things are blowing up, I obviously don't have your meshes, but for very high-Re cases, I can guess that you almost certainly have significant under-resolution even if you're attempting an under resolved simulation, particularly in high-shear regions (e.g. cavity lid/corners and aerofoil surface). The second problem is that if this isn't the case, or if it's been corrected, you will then have quite severe timestep restrictions due to the explicit time stepping. I think the first step in any case is to try a lower Re simulation, perhaps O(10^3), to establish the simulation. You can then try to increase the Reynolds number incrementally, using this flow field as an initial condition, to see if this helps setting the simulation up. Cheers, Dave
On 5 Dec 2016, at 10:16, Selbach, Fabian <fabian.selbach@student.uni-siegen.de> wrote:
Hi All, Hi Dev,
I have similar problems by running a simulation of an airfoil at high Re (Re=408000) with the compressible flow solver. It terminates every time with the same report "Nan found during time integration". Actually I still have no solution for that, but it seems to be the same problem.
Maybe somebody can help us?
Best regards Fabian Von: nektar-users-bounces@imperial.ac.uk [nektar-users-bounces@imperial.ac.uk]" im Auftrag von "devabratasahoo@campus.technion.ac.il [devabratasahoo@campus.technion.ac.il] Gesendet: Montag, 5. Dezember 2016 09:32 An: nektar-users@imperial.ac.uk Betreff: [Nektar-users] Using Nektar for solving compressible flow case
Hi All,
I am trying using Nektar for solving the compressible flow over a cavity at Re - 10^7 per meter. I tried to set the parameters as per the user guide but the solution terminates saying "NaN error found during time integration". Initially, I used structured grid with fine mesh near the walls, but it didn't work. And when I use a coarse structured grid it runs for a while and then terminates saying the same issue. When I tried with a lesser Reynolds no. (0.3x10^7) and a coarse unstructured grid, I worked but had a very bad residual. The tutorials provided by Nektar are all cases of incompressible flow, which works fine. I can't find any examples of using compressible flow solver in the tutorials. Can anyone please give me an example of how to set the parameters and solver info for using compressible flow solver. And what changes in the parameters and solverinfo should be done to solve steady cases of compressible flow in Nektar.
Thanking you. Regards, Dev
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
-- David Moxey (Research and Teaching Fellow) d.moxey@imperial.ac.uk | www.imperial.ac.uk/people/d.moxey Room 364, Department of Aeronautics, Imperial College London, London, SW7 2AZ, UK.
Hi Dave, Thanks for the reply. I will try as suggested. I am attaching the meshes tried so far. I could get the results only for coarse structured mesh and unstructured mesh but with really bad residuals that too only on lower Re No. As soon as I try using fine or mesh things are blowing up. [cid:2153f66c-2c3d-4234-a99c-cac2b920115f] FINE [cid:da687011-9911-4886-88fc-c780634f1f3d] MEDIUM[cid:1e2cd113-b436-4868-9e7a-f722e56447f1] COARSE[cid:2c6bf011-c586-4761-a8e4-f8d8978e7ae8] UNSTRUCTURED COARSE Also you can see in the picture below the Parameter and solver info when its terminating. [cid:508336a7-bca7-4735-8c08-ee826e4d6669] And below you can see the result when the solver works for coarse structured aswell as unstructured grid [cid:7270ce63-86ff-4aca-a492-9a28a7a64a15] Maybe now you can give some more suggestions on how to sort out the issues. Thanking you. Regards, Dev ________________________________ From: David Moxey <d.moxey@imperial.ac.uk> Sent: 05 December 2016 12:42:41 To: Selbach, Fabian Cc: Devabrata Sahoo; nektar-users@imperial.ac.uk Subject: Re: [Nektar-users] Using Nektar for solving compressible flow case Hi both, Happy to help you with this but I have to give a fairly pessimistic warning, which is that these cases are likely to be very challenging with the compressible solver, due to the explicit time-stepping in use in this code. This might be somewhat mitigated by your goals -- i.e. do you want to do a full DNS or under resolved (uDNS/ILES) simulation? In terms of why things are blowing up, I obviously don't have your meshes, but for very high-Re cases, I can guess that you almost certainly have significant under-resolution even if you're attempting an under resolved simulation, particularly in high-shear regions (e.g. cavity lid/corners and aerofoil surface). The second problem is that if this isn't the case, or if it's been corrected, you will then have quite severe timestep restrictions due to the explicit time stepping. I think the first step in any case is to try a lower Re simulation, perhaps O(10^3), to establish the simulation. You can then try to increase the Reynolds number incrementally, using this flow field as an initial condition, to see if this helps setting the simulation up. Cheers, Dave
On 5 Dec 2016, at 10:16, Selbach, Fabian <fabian.selbach@student.uni-siegen.de> wrote:
Hi All, Hi Dev,
I have similar problems by running a simulation of an airfoil at high Re (Re=408000) with the compressible flow solver. It terminates every time with the same report "Nan found during time integration". Actually I still have no solution for that, but it seems to be the same problem.
Maybe somebody can help us?
Best regards Fabian Von: nektar-users-bounces@imperial.ac.uk [nektar-users-bounces@imperial.ac.uk]" im Auftrag von "devabratasahoo@campus.technion.ac.il [devabratasahoo@campus.technion.ac.il] Gesendet: Montag, 5. Dezember 2016 09:32 An: nektar-users@imperial.ac.uk Betreff: [Nektar-users] Using Nektar for solving compressible flow case
Hi All,
I am trying using Nektar for solving the compressible flow over a cavity at Re - 10^7 per meter. I tried to set the parameters as per the user guide but the solution terminates saying "NaN error found during time integration". Initially, I used structured grid with fine mesh near the walls, but it didn't work. And when I use a coarse structured grid it runs for a while and then terminates saying the same issue. When I tried with a lesser Reynolds no. (0.3x10^7) and a coarse unstructured grid, I worked but had a very bad residual. The tutorials provided by Nektar are all cases of incompressible flow, which works fine. I can't find any examples of using compressible flow solver in the tutorials. Can anyone please give me an example of how to set the parameters and solver info for using compressible flow solver. And what changes in the parameters and solverinfo should be done to solve steady cases of compressible flow in Nektar.
Thanking you. Regards, Dev
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
-- David Moxey (Research and Teaching Fellow) d.moxey@imperial.ac.uk | www.imperial.ac.uk/people/d.moxey<http://www.imperial.ac.uk/people/d.moxey> Room 364, Department of Aeronautics, Imperial College London, London, SW7 2AZ, UK.
participants (3)
-
David Moxey
-
devabratasahoo＠campus.technion.ac.il
-
Selbach, Fabian