[jmjohansen]

Dear all,

I'm usually working with large systems at high Re and large dimensions, i.e. not paying to much attention to the boundary layer behavior.


Now I have to do some calculations on laminar flows. Both laminar flows in pipes and along flat plates.

I can easily estimate the boundary layer thickness. But what how well should I discretisize the boundary layer? Any rule of thumbs?
And since the boundary layer in a laminar flow will continously grow, is there a clever way to take this into account in FVM calculations? (I have quit a long plate).

BR,

Joakim

[FMDenaro]

Originally Posted by jmjohansen

Dear all,

I'm usually working with large systems at high Re and large dimensions, i.e. not paying to much attention to the boundary layer behavior.


Now I have to do some calculations on laminar flows. Both laminar flows in pipes and along flat plates.

I can easily estimate the boundary layer thickness. But what how well should I discretisize the boundary layer? Any rule of thumbs?
And since the boundary layer in a laminar flow will continously grow, is there a clever way to take this into account in FVM calculations? (I have quit a long plate).

BR,

Joakim

an easy way is to ensure a good quality is that the Reynolds cell number is O(1)

[cfdivan]

Originally Posted by jmjohansen

Dear all,

I'm usually working with large systems at high Re and large dimensions, i.e. not paying to much attention to the boundary layer behavior.


Now I have to do some calculations on laminar flows. Both laminar flows in pipes and along flat plates.

I can easily estimate the boundary layer thickness. But what how well should I discretisize the boundary layer? Any rule of thumbs?
And since the boundary layer in a laminar flow will continously grow, is there a clever way to take this into account in FVM calculations? (I have quit a long plate).

BR,

Joakim

Hi Joakim,

As mentioned by FMDenaro you better force your Reynolds cell number to O(1)...O(3) should provide you decent results as well. I guess that there are so different ways to treat the wall in commercial codes, you better check it in your code.
Based on boundary layer development equations, u can predict the size of the wall closest cell. By checking the eddy viscosity ratio (for example) in near wall you can evaluate if your approximation was reasonable or not and iterate on mesh refinement.

Regards,