[alligngr8]

Hi Hou

I have sent you a mail.

Thanks

[prasanthnitt]

I have used SIMPLE long back and so I am not familiar with the details. The pressure outflow boundary conditions can be tricky. Incase of divergence of the poisson solver I would recommend that you validate your code for the simple lic driven cavity case where the velocity and pressure boundary conditions are well defined. Your pressure boundary conditions seems to be okay, there could be a typo in your code.
Does it blow up in the very first time step? You could add the mass deficit term to the velocity boundary and set dp/dn=0 at the out flow boundary. This is what is done in fractional step method. And dont set your v velocity outflow as zero, set it as dv/dn=0. v=0 should be an outcome and not something that you impose at the outflow.

[alligngr8]

Originally Posted by prasanthnitt

I have used SIMPLE long back and so I am not familiar with the details. The pressure outflow boundary conditions can be tricky. Incase of divergence of the poisson solver I would recommend that you validate your code for the simple lic driven cavity case where the velocity and pressure boundary conditions are well defined. Your pressure boundary conditions seems to be okay, there could be a typo in your code.
Does it blow up in the very first time step? You could add the mass deficit term to the velocity boundary and set dp/dn=0 at the out flow boundary. This is what is done in fractional step method. And dont set your v velocity outflow as zero, set it as dv/dn=0. v=0 should be an outcome and not something that you impose at the outflow.

Hi
if I set dv/dn =0 , this will imply dv/dx =0 for 2D flow. But from where this condition arises.? If I am not wrong then du/dx=0 from no stress and then I will get dv/dy=0 from continuity. But from where I will get dv/dx=0.

when solving for lid driven cavity, do I need to extend velocity and pressure domain outside boundary.?

[shahrooz.omd]

Originally Posted by alligngr8

Hi
if I set dv/dn =0 , this will imply dv/dx =0 for 2D flow. But from where this condition arises.? If I am not wrong then du/dx=0 from no stress and then I will get dv/dy=0 from continuity. But from where I will get dv/dx=0.

when solving for lid driven cavity, do I need to extend velocity and pressure domain outside boundary.?

dv/dx + du/dy = 0, if du/dy = 0 then dv/dx = 0. in this case u = 0 and du/dy = 0

however i think I didn't get what you really are asking.

[alligngr8]

Hello
I was asking about how dv/dn = 0 at outlet.

[shahrooz.omd]

Originally Posted by alligngr8

Hello
I was asking about how dv/dn = 0 at outlet.

You need to do a scale analysis on continuity equation. Scale of v will be DU/L when L is large enough v reduces to zero and continuity equation will be du/dx= 0. You can find more detailed explanation in adrian bejan's book titled convection heat transfer.

[alligngr8]

Originally Posted by shahrooz.omd

You need to do a scale analysis on continuity equation. Scale of v will be DU/L when L is large enough v reduces to zero and continuity equation will be du/dx= 0. You can find more detailed explanation in adrian bejan's book titled convection heat transfer.

Hi,
Since I am solving flow between parallel plates, then for fully developed case du/dx wil be zero.

How does scale of v gives dv/dx=0 at outlet this is what I was asking. I think it should be either dv/dy=0 at outlet or v=0.

[shahrooz.omd]

Originally Posted by alligngr8

Hi,
Since I am solving flow between parallel plates, then for fully developed case du/dx wil be zero.

How does scale of v gives dv/dx=0 at outlet this is what I was asking. I think it should be either dv/dy=0 at outlet or v=0.

when v = 0 at that region then dv/dx=0, it's obvious. but for boundary condition v=0 is more proper imo.

[prasanthnitt]

Originally Posted by alligngr8

Hi
if I set dv/dn =0 , this will imply dv/dx =0 for 2D flow. But from where this condition arises.? If I am not wrong then du/dx=0 from no stress and then I will get dv/dy=0 from continuity. But from where I will get dv/dx=0.

when solving for lid driven cavity, do I need to extend velocity and pressure domain outside boundary.?

If your channel is long enough you can set either dv/dn=0 or v=0, both should give you the same answer. ONce the flow is fully developed the flow quantities dont change in the streamwise direction. If your channel is not long enough then you will face problems specifying the outflow bcn.
For lid driven cavity, anything you extend beyond the physical domain size would be ghost cells and would depend on how you implement the bondary conditions. Since you are doing FVM you can directly set the fluxes to be zero at the boundaries and so extension beyond the boundary can be avoided. I hope this helps.

[nazia]

I need a matlab code for solving incompressible flow in a tube can anybody help me?

[sisy]

Hello I need a matlab code for Roe linearization for 1D shallow water equation with entropy fix.
Is there any one who can help me?

[Mass]

Hi everybody,

I am using the Dr. Seibold code for solving the N.S. in 2D.
I am going to solve the Navier-Stokes for a 2D rectangular domain where the Dirichlet boundary conditions are applied at north, south and west sides; but on the east-side two boundary conditions should be applied. Part of the east-side has no-slip condition (Dirichlet B.C.) and for the rest of it (dv/dx)=f(y) is held, where v is the vertical (in y-direction) component of the velocity vector.
I know how to modify the "boundary conditions" section of the code, where "vE" is specified, to have part of the east-side to follow no slip condition; but I need help about how to apply the Neumann B.C. for the rest of its length.

Thanks

[umair_tariq]

Originally Posted by cici

Hi All,

I have written a MATLAB code for a 2-D lid driven cavity problem, and it works fast and well, the results are consistent with experimental data. But this code does not work for the duct flow, which is also a classic example in many references.

I wonder it is due to the change of the definition of boundary conditions or the scheme itself. What is the main issue behind this.

Thanks.

where is the code you have mentioned ?