Dynamic Structured mesh around cylinder problem!!!

ludosbiz · August 14, 2015, 06:13

Hi guyz,

first explanation, than comes the question!

i carried on analysis on stationary 2D cylinder with Fluent, with a mesh made in ICEM and all resulted as expected, no problem and appreciable physically results have been obtained.
Then i moved to oscillating cylinder in cross flow direction, and here came the problems with dynamic mesh, since the layering method creates eccessive distortion on the lateral sides of the cylinder, whereas on the upper side and below the boundary, it works fine. I thought indeed to change the mesh in triangles to use remeshing option, but results take more time to converge(a lot more) and are not so strictly right as before, comparing the stationary(remeshed) grid with the results obtained before. It seems that some amount of diffusion in generated from this newly made mesh.
Now the question: any suggestion on how to make the stuctured mesh deform properly,avoiding these lateral side effects? i mean, anybody knows how to use the layering method, avoiding the lateral mesh to distort till almost collapse? thx for the help to anyone, if u need screenshot just ask!

ludosbiz · January 27, 2016, 04:10

i'll post here a reply to my own question, in order to help those out there facing the same problems. Hope to be clear

i solved the issues related to the dynamic remeshing after a while, it has required the use of the Diffusion Smoothing method with the Diffusion Parameter set from 1 to 10.

Well, remeshing itself without layering is successful if u are using a structured mesh as i did, and yes, it is enough to reach convergence and stability of the iterations.

The diffusion parameter should be chosen as a function of the width of the domain:
if u chose a D<1, all of the deformation of the mesh needed to account for the oscillation, will be concentrated in the near wake of the moving wall thus generating highly skewed mesh in that zone, whereas affecting few or none the rest of the domain.
If u chose a D>10 the motion of the cylinder brings the whole domain to collaborate in the deformation to account for the oscillation, thus preserving the wall boundary from excessive skewness and allowing a better integration of the drag and lift forces.
I will link here an image with D=10 as i used in my work to let u understand properly what i mean.
The elements near the wall rest practically unvaried.

The link: http://i68.tinypic.com/k3xavq.jpg

Best wishes for you work dudes!

Schnauzer · March 23, 2016, 16:15

Hi!

I am performing vortex induced vibration simulations and I am having trouble with dynamic mesh . I tried to change the mesh by triangular elements, but another kind of problems apperead like convergence problems. My structured mesh showed excellent results for the stationary case. How could I set the dynamic mesh to avoid problems of negative cell volumes? I tried to use layering, but I still have with the same problem. Did you have problems using remeshing with structured mesh?

Thx for the help!

August 14, 2015, 06:13	Dynamic Structured mesh around cylinder problem!!!	#1
ludosbiz New Member Ludo Join Date: Jul 2015 Posts: 14 Rep Power: 11	Hi guyz, first explanation, than comes the question! i carried on analysis on stationary 2D cylinder with Fluent, with a mesh made in ICEM and all resulted as expected, no problem and appreciable physically results have been obtained. Then i moved to oscillating cylinder in cross flow direction, and here came the problems with dynamic mesh, since the layering method creates eccessive distortion on the lateral sides of the cylinder, whereas on the upper side and below the boundary, it works fine. I thought indeed to change the mesh in triangles to use remeshing option, but results take more time to converge(a lot more) and are not so strictly right as before, comparing the stationary(remeshed) grid with the results obtained before. It seems that some amount of diffusion in generated from this newly made mesh. Now the question: any suggestion on how to make the stuctured mesh deform properly,avoiding these lateral side effects? i mean, anybody knows how to use the layering method, avoiding the lateral mesh to distort till almost collapse? thx for the help to anyone, if u need screenshot just ask!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[ICEM] surface mesh merging problem	everest	ANSYS Meshing & Geometry	44	April 14, 2016 07:41
Update of the variables after dynamic mesh motion.	gtg258f	OpenFOAM Programming & Development	9	January 18, 2014 11:08
[ICEM] ICEM Structured Mesh Problem	OMJT	ANSYS Meshing & Geometry	3	March 22, 2013 11:06
Can CFX solve problem with structured mesh?	Peted	CFX	2	June 1, 2004 00:45
unstructured vs. structured grids	Frank Muldoon	Main CFD Forum	1	January 5, 1999 11:09

January 27, 2016, 04:10		#2
ludosbiz New Member Ludo Join Date: Jul 2015 Posts: 14 Rep Power: 11	i'll post here a reply to my own question, in order to help those out there facing the same problems. Hope to be clear i solved the issues related to the dynamic remeshing after a while, it has required the use of the Diffusion Smoothing method with the Diffusion Parameter set from 1 to 10. Well, remeshing itself without layering is successful if u are using a structured mesh as i did, and yes, it is enough to reach convergence and stability of the iterations. The diffusion parameter should be chosen as a function of the width of the domain: if u chose a D<1, all of the deformation of the mesh needed to account for the oscillation, will be concentrated in the near wake of the moving wall thus generating highly skewed mesh in that zone, whereas affecting few or none the rest of the domain. If u chose a D>10 the motion of the cylinder brings the whole domain to collaborate in the deformation to account for the oscillation, thus preserving the wall boundary from excessive skewness and allowing a better integration of the drag and lift forces. I will link here an image with D=10 as i used in my work to let u understand properly what i mean. The elements near the wall rest practically unvaried. The link: http://i68.tinypic.com/k3xavq.jpg Best wishes for you work dudes!

March 23, 2016, 16:15		#3
Schnauzer New Member LRubino Join Date: Feb 2016 Location: Brazil Posts: 6 Rep Power: 10	Hi! I am performing vortex induced vibration simulations and I am having trouble with dynamic mesh . I tried to change the mesh by triangular elements, but another kind of problems apperead like convergence problems. My structured mesh showed excellent results for the stationary case. How could I set the dynamic mesh to avoid problems of negative cell volumes? I tried to use layering, but I still have with the same problem. Did you have problems using remeshing with structured mesh? Thx for the help!