Learning Forum

[Milne Ando]

G'day,

I wonder that what the number means in "skewness correction" field. Correction times per time step? Does increasing it make calculation more stable?

Hope someone can help me. Thank you.

[Rob]

If you stick skewness correction into the documentation it'll give you more details. Assuming you have a decent mesh it'll not do much, if you don't have a decent mesh I recommend fixing it rather than rely on the numerics. I don't think I've ever altered the default.

[Milne Ando]

Hello,

okay, i'll keep both corrections default, and my mesh is fine( lowest orthogonal quality is 0.3). I have tried PRESTO! and body force weighted, however calculation still went unstable. In a 3D FSI problem, is body force weighted scheme recemmended?

[Rob]

What fluid physics and models are you looking at? The solver discretisation isn't linked to FSI, it's purely for the CFD part. 

[Milne Ando]

Hello,

I want to simulate a solid cylinder swinging like a pendulum (a kind of vortex-induced vibration), and you've suggested me to do one-way simulation in Fluent with dynamic mesh and 6DOF solver.

[Rob]

So it's not FSI as it's not bending.  It's purely 6DOF motion, so probably Body Force Weighted as the flow probably doesn't do a great deal. 

[Milne Ando]

Okay, I'll quit body force weighted scheme. And as what I know is that FSI also includes moving structure without deforming, so hmm it has a generalised definition. Well let's think about six DOF Properties; Another subscriber and I have discussed what the possible problem might be. We suspect that maybe Moment of Inertia should be filled in moment of inertia about the axis of rotation instead of its CG. Since the user guide only claims that in Inertia Tensor field those components (Ixx, Iyy,.etc.) are defined relative to the CG. What do you think?

[Rob]

I suspect it's the standard definition of a moment of inertia. Does the video in the Help system give you any pointers? 

[Milne Ando]

I don't know how I can access that video, hmm maybe we can ask your programmers; he/she must know the answer 'cause they can check CFD codes. If we still can't solve this mystery, I may turn to transient structural (two-way coupling) as a fallback method, since it can cover that part. But assigning a dummy UDF to the inflation mesh moving with structure will be another problem. There's a sample UDF in the user guide p.3323, but I don't have the file called udf.h, so how can #include "udf.h" be valid?

[Rob]

udf.h is one of the headers that are in the Fluent install. 

Click on Help in Fluent, then you'll find both the documentation and videos.