[carl jung]

Hi,

I have a question about the terminology.
Considering no slip condition, there is a huge difference in velocity between the wall and the fluid. If the grid is not fine enough near the surface, the gradient of the velocity will be a spike rather than a smooth peak. I think it's a kind of numerical error or discretization error, but is there any particular terminology about this?

Thank you,

Carl

[mprinkey]

It is called "unphysical." If the flow is viscous and viscous effects are important to the flow, ignoring boundary layer behavior means the model fails to capture the flow dynamics. Pressure drops will be completely wrong, heat transfer rates will be completely wrong, etc. Failing to resolve diffusion layers in diffusion-domained processes ignores the principle physical effects.

Free slip conditions do make sense, say for external aero, when one is only calculating lift and drag results are not a consideration.

[FMDenaro]

I think that a correct terminology in CFD (and numerical analysis) is related to the concept of "insufficient resolution" in the representation of the solution. When the gradient is high, you can think to the function as represented by a large numbers of wavenumber components. Your large grid size cuts all wavenumbers greater than the Nyquist wavenumber. As mprinkey wrote, that leads to an unphysical smoothing of the velocity profile.

[carl jung]

Thank you all for the replies. There might not be an exact term for that. Is 'numerical diffusion' related to this?

[FMDenaro]

Originally Posted by carl jung

Thank you all for the replies. There might not be an exact term for that. Is 'numerical diffusion' related to this?

Well, numerical diffusion is related to the type of the scheme that produces a term in the local truncation error that is similar to a diffusion. Considerino only the low resolution of a grid you cannot say it will produce a numerical diffusion.

[carl jung]

Thank you for the clarification!