upwind v.s. chequerboard
 

Hi! eveybody

Does anyone know how the upwind schemes eliminate the chequerboard oscillation on pressure and velocity field? Does it mean that use upwind scheme to determine the velocity on contral volume surface?

Best regard

Darcy

Re: upwind v.s. chequerboard
 

An upwind discretization of the advection term as nothing to do with the elimination of checkerboard board pressure solutions.

1) An upwind discretization of the advection term stabilized the solution for advection driven transport of a scalar quantity.

2). The checkerboard type solution for pressure as to do with incorrect space functions for velocity and pressure.

Re: upwind v.s. chequerboard
 

Hi,

Sebastien is definitively true!!

To avoid checkerboard pressure solutions, the easiest way is to use staggered grid (see Patankar) for pressure and velocity, which is equivalent to use Raviart-Thomas element for velocity and Q0 element for pressure.

Bye

Re: upwind v.s. chequerboard
 

Or of course if you opt for the "better" more suitable Co-located grid, you may use the so called "Rhie and Chow" interpolation. Co located grids is the obvious choice as unstructured grids may be handled without too much complications.

Re: upwind v.s. chequerboard
 

hi,dear Sebastien,

I have a problem about the upwind or hybrid scheme,I want to extend my code(central difference for all space derivatives) to fluid field including separation,now,I use Adams-bashforth for the convective terms[i.e.,du/dx(t+1)=3/2(du/dx(t))-1/2(du/dx(t-1))],I wonder,if I can use upwind or hybrid scheme for du/dx(t) and du/dx(t-1) to evaluate du/dx(t+1)?here,t is the time step.

thanks