Mesh movement during runtime
1 Attachment(s)
Hello there,
although I already was searching here and there, I am still not sure how to use the diffusivity of the velocityLaplacian solver in the dynamicMeshDict. I want to move my mesh during runtime depending on the pressure on the patch boundary "airfoil" (optimization of a 2D-airfoil). But somehow the diffusivity (uniform, quadratic, motionDirectional etc.) does not change much the results and I always get some overlapping cells (see attachment). If I choose different values for uniform I also don't see any change in the result. Here is the dynamicMeshDict I am using: HTML Code:
dynamicFvMesh dynamicMotionSolverFvMesh; Kind regards, Matthias |
I'm not sure you can get the behaviour you want with the standard motion solvers. I'm not sure how you are coupling the pressure on the patch with the motionSolver. velocityLaplacian means that the displacement of the points is computed with velocity specified as initial and boundary conditions of the laplacian equation.
The diffusivity for the points mainly affects how "stiff" the mesh is. The different options are there to modify how diffusivity is computed based on several criteria. I know there is one called inverseDistance that has the diffusivity be proportional to the inverse distance from certain patches (so its highest near the patches and then falls off as you move further). I think you will need some kind of distance dependant diffusivity to help things move along. |
Thanks, mturcios.
The coupling of pressure and motionSolver I am doing via the pointMotionU. The diffusivity inverseDistance I also tried, but the results are just slightly different and I still get overlapping cells. But is it normal that different values, e.g. in motionDirectional, give no difference in the result? So, is my dynamicMeshDict correct in principle? |
Quote:
|
Dear foamer
I use dynamic mesh for "Flow-induced deformation of a flexible thin structure as manifestation of heat transfer enhancement" article. the geometry is same as turek-hron FSI benchmark. the quadratic inverseDistance uses 1/L^2 relation of laplacian equation of mesh motion, but when deflection amplitude increases the mesh deform near the moving boundary and make it distorted as below. http://uupload.ir/files/sf1r_1.png I looking for a better diffusion parameter for this case. I think some mesh layers of the moving boundary should move with boundary (for example 3layers) and then diffusion spread in mesh domain to prevent distortion of mesh in first mesh layer of moving boundary. do you have any suggestion for any formulation to change the code base it? Thanks. |
Hi Hgholami,
I am experiencing the same issue as you when using the displacementLaplacian solver with a cantilever foil. Have you been able to find a fix for this? Regards, Conor |
Hi Conor
I used velocity solver prefer than displacement solver. At now, the refVelocityLaplacian is prefer than other. Also "diffusivity quadratic inverseDistance" is good. But still it can't support large deflection. I also modified "quadratic" in motionDiffusivity for compatibility with geometry but still it is not sufficient. May you can use overset in foam-extend4.1. the solids4Foam for FSI problems use it. Quote:
|
All times are GMT -4. The time now is 22:50. |