Hi,
I tried to run my (steady flow) simulation with a timestep of 0.03, which gave me the results as can be found in attachment (large timestep), with mesh 2 being a (slightly) finer mesh than mesh 1 (being the estimate mesh from SV). Mesh 1 appears to give me a good results when I visualize the velocity at the inlet (parabolic profile). Mesh 2 however does not give me back the parabolic profile at the inlet.
When I chance the timestep to 0.005 (small timestep in attached pictures), mesh 2 appears to give me the correct profile while mesh 1 now looks very bad. For the other parameters I used exactly the same values for four cases.
What can be the problem for this since it looks like SV don't apply my boundary conditions correctly?
Thanks in advance!
Best wishes,
Charlotte
Strange results after simulation
-
Charlotte Thyssen
- Posts: 3
- Joined: Fri Sep 29, 2017 2:19 am
Strange results after simulation
- Attachments
-
- mesh2_smallTimestep.png (166.45 KiB) Viewed 455 times
-
- mesh2_LargeTimestep.png (270.15 KiB) Viewed 455 times
-
- mesh1_smallTimestep.png (235.24 KiB) Viewed 455 times
-
- mesh1_largeTimestep.png (159.15 KiB) Viewed 455 times
-
Justin Tran
- Posts: 109
- Joined: Mon Sep 30, 2013 4:10 pm
Re: Strange results after simulation
Hi Charlotte,
I am a bit surprised that the smaller timestep on mesh1 (the courser mesh) seems to not work. The mesh2 x timestep1 errors look like a diverging simulation to me, so it makes sense that lowering the timestep allowed this to converge.
A few clarifying questions:
1. Can you check the residual values for all the simulations you tried? The residual can be found in the file "histor.dat". This file shows information about the solver at every timestep. The first column of histor.dat shows the timestep number, the second column shows the wall time since the simulation started in seconds, and the third column shows the average residual. The residual shows how much the current solution for velocities and pressures in the domain satisfy the Navier-Stokes equations. Ideally, we would like to see this residual drop to below 0.01 for a converged simulation. Can you check the residual at the last timestep for each of your steady simulations to see if they converge or not?
2. The velocity magnitudes which you show in your figures is extremely high (though the parabola of the velocity itself is very blue). What is the inflow rate which you specify in your model? Also, do you know what measurement system your image data was taken in (usually this is mm or cm) and what units you used for the boundary conditions? The boundary conditions should be in the same units as the image data.
3. What is the outlet boundary condition which you are specifying in your model?
Thank you!
I am a bit surprised that the smaller timestep on mesh1 (the courser mesh) seems to not work. The mesh2 x timestep1 errors look like a diverging simulation to me, so it makes sense that lowering the timestep allowed this to converge.
A few clarifying questions:
1. Can you check the residual values for all the simulations you tried? The residual can be found in the file "histor.dat". This file shows information about the solver at every timestep. The first column of histor.dat shows the timestep number, the second column shows the wall time since the simulation started in seconds, and the third column shows the average residual. The residual shows how much the current solution for velocities and pressures in the domain satisfy the Navier-Stokes equations. Ideally, we would like to see this residual drop to below 0.01 for a converged simulation. Can you check the residual at the last timestep for each of your steady simulations to see if they converge or not?
2. The velocity magnitudes which you show in your figures is extremely high (though the parabola of the velocity itself is very blue). What is the inflow rate which you specify in your model? Also, do you know what measurement system your image data was taken in (usually this is mm or cm) and what units you used for the boundary conditions? The boundary conditions should be in the same units as the image data.
3. What is the outlet boundary condition which you are specifying in your model?
Thank you!