[aryan]

Hi
I have written a 2D code for the SIMPLE method on the co-located grid. I want to study the fluid flow around an obstacle. For this purpose, I alter the source terms appeared in the momentum equation such that the velocity in the obstacle region is zero. But I have problem with the pressure equation. Since the velocities in the solid zone are zero, all of the coefficient s of aW, aE, aS, aN, Su, Sp and aP are zero for the pressure equation (actually very small values because of the numerical and round off errors). This causes a wrong solution for the pressure and consequently the momentum equations. I was wondering if someone could help me to solve this problem.
Regards

[nickna]

Hi,
You need to deal with the boundary cells specially to ensure the correct boundary conditions of pressure (to solve the Poission eq.). It may not be feasible to treat all of the cells in a general way.

[aryan]

Hi nickna

But I think this is practical: "Computational Flow Modeling for Chemical Reactor Engineering" By Vivek V. Ranade, Page 172.

[andy_]

Originally Posted by aryan

Hi
I have written a 2D code for the SIMPLE method on the co-located grid. I want to study the fluid flow around an obstacle. For this purpose, I alter the source terms appeared in the momentum equation such that the velocity in the obstacle region is zero. But I have problem with the pressure equation. Since the velocities in the solid zone are zero, all of the coefficient s of aW, aE, aS, aN, Su, Sp and aP are zero for the pressure equation (actually very small values because of the numerical and round off errors). This causes a wrong solution for the pressure and consequently the momentum equations. I was wondering if someone could help me to solve this problem.
Regards

The SIMPLE method staggers the velocity components and so this problem does not arise.

A co-located arrangement requires the introduction of "pressure smoothing" in order to prevent pressure-velocity decoupling. It looks as if you are doing this in an implicit manner by fiddling/averaging coefficients rather than with explicit pressure gradient terms. A first step could be to work out what your smoothing fluxes actually are and whether you want to apply them at the boundaries. For example, swirling flow in collocated schemes tends to jump badly at boundaries when the large pressure smoothing term at the boundary is not included.

For incompressible flows a pressure equation can be derived from the momentum equations and then constrained to be divergence free. This can be used to determine good treatments for pressure on the boundary given your particular numerical scheme. There is plenty in the literature on this topic.

[aryan]

Hi andy

Thanks for your guidance but I still think that the problem remains, as it occures in my code apparently. When the velocity in the block zone is zero the fluxes are zero and consequently the pressure equation coefficients are also zero. Correct?

[aryan]

Originally Posted by andy_

The SIMPLE method staggers the velocity components and so this problem does not arise.

Hi andy

Thanks for your guidance but I still think that the problem remains, as it occures in my code apparently. When the velocity in the block zone is zero the fluxes are zero and consequently the pressure equation coefficients are also zero. Correct?

[arjun]

Originally Posted by aryan

Hi andy

Thanks for your guidance but I still think that the problem remains, as it occures in my code apparently. When the velocity in the block zone is zero the fluxes are zero and consequently the pressure equation coefficients are also zero. Correct?

Depends.

If you decided to treat the solid part as fluid then it has viscosity and thus coefficients are not 0.

If you decided to treat solid part as solid then you do not need to solve pressure correction equation in this part and thus this does not go into matrix so coefficients have no meaning.