[Schnauzer]

Hi guys!

I'm doing a 2D VIV study over a cylinder in Fluent . First , I perfomed analysis of the flow around a static cylinder, and the results were consistent and did not have many problems. For the VIV analysis , I wrote a UDF to simulate the oscilation of the cylinder in cros-flow direction using dynamic mesh and here the problems started . I am using a structured mesh and I'm having the problem of " negative cell volume detected" . Initially, I tried to use the layering method , but it presented many problems. Then I tried using diffusion smoothing method , but the problem persisted . I've thought about changing the mesh to triangles , but I avoid it because the structured mesh shows better convergence.
Does anyone have any suggestions of how to deform structured mesh without having this problem of negative volume?

Thx!

[Far]

Originally Posted by Schnauzer

Hi guys!

I'm doing a 2D VIV study over a cylinder in Fluent . First , I perfomed analysis of the flow around a static cylinder, and the results were consistent and did not have many problems. For the VIV analysis , I wrote a UDF to simulate the oscilation of the cylinder in cros-flow direction using dynamic mesh and here the problems started . I am using a structured mesh and I'm having the problem of " negative cell volume detected" . Initially, I tried to use the layering method , but it presented many problems. Then I tried using diffusion smoothing method , but the problem persisted . I've thought about changing the mesh to triangles , but I avoid it because the structured mesh shows better convergence.
Does anyone have any suggestions of how to deform structured mesh without having this problem of negative volume?

Thx!

there are few steps you can follow:

1. Use concept of two domains : inner and outer.

2. give motion to whole inner domain

3. Use re meshing option with appropriate settings. this is very important

Alternatively you can use

1. 6DOF rigid body solver or

2. Two way FSI

[Schnauzer]

Hi!
Thanks for reply.

I'm using 6DOF rigid body solver.
The suggestion is to create two domains. Assign movement to the domain closest to cylinder using the same UDF i'm using for the 6DOF moviment of the cylinder? So, the inner domain moves together with the cylinder?

The outer domain assign deforming using smoothing and remeshing methods?

Here is the mesh that i'm using.

Thanks!

mesh 1.jpg

[Far]

yes correct. only thing to be taken care is to treat the interface carefully.

[Schnauzer]

I tried again to perform the simulation and the problem persists.
The mesh is deforming near the inner domain, even using smoothing diffusion method with diffusion parameter = 3.

I read in the Fluent Users Guide that higher values ​​of diffusion parameter preserve larger regions of the mesh near the moving body and cause regions away of the moving body to absorb more of the motion, but this is not happening.
I'm doing something wrong, but I dont know exactly what.

[Far]

Originally Posted by Schnauzer

I tried again to perform the simulation and the problem persists.
The mesh is deforming near the inner domain, even using smoothing diffusion method with diffusion parameter = 3.

I read in the Fluent Users Guide that higher values ​​of diffusion parameter preserve larger regions of the mesh near the moving body and cause regions away of the moving body to absorb more of the motion, but this is not happening.
I'm doing something wrong, but I dont know exactly what.

Make diffusion factor = 0.3 and also enable remeshing.

[Daban]

i am doing similar problem and i had your problem i resolved it with small "max itteration/timestep" and a diffusion value of 1 for remeshing.
how did u write your UDF file can you share it?