CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Main CFD Forum (https://www.cfd-online.com/Forums/main/)
-   -   problem in DNS in a pipe (https://www.cfd-online.com/Forums/main/107904-problem-dns-pipe.html)

 hnemati October 9, 2012 05:51

problem in DNS in a pipe

1 Attachment(s)
Hi guys
I am running a DNS code in a pipe. For laminar flow I have a problem.
as you see in 2d plot of radial velocity at inlet, there are some strange number of velocity. it should be zero but it is not.
I will appreciate any help.

 FMDenaro October 9, 2012 06:20

Quote:
 Originally Posted by hnemati (Post 385670) Hi guys I am running a DNS code in a pipe. For laminar flow I have a problem. as you see in 2d plot of radial velocity at inlet, there are some strange number of velocity. it should be zero but it is not. I will appreciate any help.
You should address more details ... Re number, number of cells, inflow type ...
are you solving incompressible flows? did you check the divergence-free constraint in each cell? Are you sure of the numerical stability contraints?

From the figure it seems that the radial velocity accounts for the wall and the incipient boundary layer in a very low Re flow with plug velocity inlet, but I may be wrog ...

 hnemati October 9, 2012 07:58

Quote:
 Originally Posted by FMDenaro (Post 385673) You should address more details ... Re number, number of cells, inflow type ... are you solving incompressible flows? did you check the divergence-free constraint in each cell? Are you sure of the numerical stability contraints? From the figure it seems that the radial velocity accounts for the wall and the incipient boundary layer in a very low Re flow with plug velocity inlet, but I may be wrog ...
Hi
the Re=60 based on friction velocity. number of mesh is: r=30 theta=16, z=96.
Yes this is for incompressible flow. the divergence is Ok. I am using the profile ((Re/4.)*(1.- 4.*(rp(i))**2.)) for inlet which is analytitical solution. I put radial and tangential velocity 0 at z=0 .

 FMDenaro October 9, 2012 09:46

Quote:
 Originally Posted by hnemati (Post 385705) Hi I am so grateful for your answer. the Re=60 based on friction velocity. number of mesh is: r=30 theta=16, z=96. Yes this is for incompressible flow. the divergence is Ok. I am using the profile ((Re/4.)*(1.- 4.*(rp(i))**2.)) for inlet which is analytitical solution. I put radial and tangential velocity 0 at z=0 .
Ok, are you sure you get a full steady state?
After you have to refine the grid and see what happens, usually the small value in the normal-to-wall velocity component can be controlled. See also your threshold value in the pressure solver, can be to high.
After that, you can check further by changing the inflow/outflow condition in a periodic condition, this way you can check the inner solution.

 kuihe October 22, 2012 12:10

how strange the velocity is? Can you email me and we can discuss furtherly.

Quote:
 Originally Posted by hnemati (Post 385670) Hi guys I am running a DNS code in a pipe. For laminar flow I have a problem. as you see in 2d plot of radial velocity at inlet, there are some strange number of velocity. it should be zero but it is not. I will appreciate any help.