skewness and convergence
 

Hi all,

I really want to know one thing. I have seen a reply by Xaar that it is becuase of the highly skewed elements, the continuity doesn't fall below the imposed limits. Can anybody tell quantitatively that what is the limiting % of high skewness beyond which the continuity or other parameters donot converge properly, preferably in internal flows? (sorry for asking this, as "quantitatively" is meaningless when you are dealing with Fluent and CFD!!) Sometimes we cannot avoid to have highly skewed elements.

Thanks

Rajesh

Re: skewness and convergence
 

let me explain a little about this issue when you apply continuity, what you need is values of variable (u,v,w) at face center. This value can be calculated from values of cells sharing this face. from these values you can get fluxes at face and sum of fluxes equal to zero is continuity. To couple pressure-velocity, you get one more parameter, that is p. so we have fluxes in terms of (u,v,w,p), but these fluxes do not sum to zero, so we calculate pressure correction. from this pressure correction we correct the velocity and thus fluxes.

now comes the cretical part, the way pressure correction equation is formed, it becomes poisson equation in terms of pressure correction variable. This poisson equation has, terms that require you to calculate gradients of (u,v,w,p) at the face center and normal to face , in a orthogonal cell system, these values can be calculated accurately, and thus satisfying the continuity is easy. But when the non-orthogonality increases, these terms introduce errors, and thus it becomes increasing difficult to get correct value as non-orthogonolity increases.

i hope you get the point,

now this also means one more thing, that if the cells are non orthogonal but the gradients are zero, or scalar value does not change, then, the errors will also be very very less, thus if you have cells in the portion where flow does not change much, they will not create problems. But if it is in sensetive portion where flow changes too much, it creates lot of problems in satisfying continuity.

i hope it clears.

Re: skewness and convergence
 

Thanx Zxaar

Your reply gave me some light into how we should modify our geometry according to the nature of the flow...Thanx a lot

Rajesh