Learning Forum

[rinaldi]

Hi There,

I am trying to use Discovery Live for an external air flow around a bike wheel. 

I am having problems and doubting about the solution that Discovery Live calculates because it looks very dependent on the size of the "wind tunnel" created (that is always way to small, almost touching the wheel in all the directions, that is why I started to enlarge it of about 200mm for front and sides, 1000mm on the back). 

The Fx,Fy,Fz I calculate with different enclosures reach 40% difference.

 

Could someone explain why this happens? Increase the enclosure size is decreasing the mesh quality? How can I know when the mesh is accurate enough to give reliable results? 

 

Btw, are the boundary layers meshed at all? 

 

Thanks

Alberto

[Naresh Patre]

Hello rinaldialberto

The domain for external flow simulation should be such that the flow around the object is properly captured. Having a domain almost touching the object i.e. wheel is not recommended as the domain is not capturing the flow field around the wheel. Please check out the Tutorials section of External fluid flow under Discovery Live tutorials for more details.

[rinaldi]

Hi Naresh Patre ,

what I wrote is that Discovery Live generates a small domain. I always enlarge it in order to capture the flow. 

What I meant is that depending on how large the domain is I get different results (even if both are large enough to capture the flow behavior).

 

Cheer

Alberto

[ERab]

Hi @rinaldialberto

Please review this post. Discovery Live: “Thin” geometric features and relative GPU adjustments

While it is focused on structures, it talks about the relative size of geometry which also applies to fluids.

While the accuracy of the solver doesn’t change with size, Discovery Live’s ability to fully capture geometry detail is.  The more RAM you have on your GPU, and the higher you set “Fidelity” above the Play/Pause button at the bottom of your screen, the more geometry detail is captured.  The smaller you make your fluid domain relative to the solids inside, the more detail is captured.

When the geometry is not properly captured like in the above link, the results appear less accurate.  The solver operates equally of all geometry, but not all geometry is captured depending on the overall size of the model, and the relative size of the detail.

[rinaldi]

Hi Ethan ,

 

thanks for your reply. 

The post you linked explains what I was worried about. 

How can I understand if the details are not captured? Is there any parameter or file (a log file maybe) that notes down these characteristic? 

Thanks,

Alberto