CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (https://www.cfd-online.com/Forums/openfoam-solving/)
-   -   AMI simulation unstable within first few timesteps (https://www.cfd-online.com/Forums/openfoam-solving/130334-ami-simulation-unstable-within-first-few-timesteps.html)

craig0a February 24, 2014 13:09

AMI simulation unstable within first few timesteps
 
Hello everyone,

There are a couple of threads on this topic already, but after looking through them carefully, I am still not sure how to go about fixing my particular problem, since my simulation seems to be breaking down in a slightly different way .

Short: With my AMI set to not be rotating, my patch source/target weights are close to 1 when running, but my simulation is going unstable within a few timesteps. In Paraview, it looks like there is a build up of pressure/high velocity at the upstream surface of my AMI.

Long:

I am simulating a vertical axis wind turbine in 3D (again, like so many on this forum!). The geometry was made in Solidworks, modified in Blender to ensure the mesh was closed and to refine the triangulation, and exported as an ascii .obj file. In my case, I have three blades, a tower, and an "innerCylinderSmall" which encompasses just the blades, brushing the top surface of the tower.

I then followed the propellor tutorial as closely as possible. So my command flow is:

blockMesh
surfaceFeatureExtract
snappyHexMesh
topoSet -dict system/removeRedundantZones.topoSetDict
topoSet -dict system/createInletOutletSets.topoSetDict
topoSet -dict system/createAMIFaces.topoSetDict
createPatch -overwrite
createBaffles -overwrite
mergeOrSplitBaffles -split -overwrite
transformPoints -scale "(0.00508 0.00508 0.00508)"

If I am understanding correctly, and this seems to be supported by my results, the "innerCylinderSmall" should be used in snappyHexMesh to create a more refined grid around my blades, but then it is removed and replaced with AMI surfaces. (So my AMI surface is a closed cylinder.)

Checking my results in ParaView, the mesh looks okay (I need to switch to an unstructured grid near the blades, but that can come later). The geometries seem to match up alright; ParaView recognizes both AMI1 and AMI2 as patches and they overlap exactly.

Running checkMesh, I get "Mesh OK". The cell sizes are about two orders of magnitude smaller than that used in the propeller tutorial. Again, AMI1 and AMI2 are recognized, each with 110444 faces, albeit one has 110682 points while the other has 110666 points.

When I try to run the simulation, setting the rotation to zero, running in water (like the propeller) at a Reynolds number of 600, for the first ~100 steps residuals, Courant number, patch/source target weights, etc. seem to be doing fine, but then in about 5 steps, everything explodes. (The number of steps to blow up seems to be the same regardless of initial time step size and I have been trying two orders of magnitude smaller than the propeller tutorial to match my smaller cell size.) Looking at the time sequence in Paraview, I am getting effectively unbounded growth in velocity and pressure at the upstream edge of my AMI surface (actually between blades of my geometry.)

I am guessing that this has something to do with how I am defining/meshing/setting boundary conditions on my AMI surface, but I don't know how to go about debugging this.

If anyone has the time to help me through this, I would greatly appreciate it. At present, I don't know what files/screenshots may be pertinent, so if you can help, please let me know what you would like to see.

Thank you!

Best,
A. Craig


All times are GMT -4. The time now is 10:04.