[florian_krause]

Dear Foamers!

currently I am trying to set-up and solve a case of a pipe flow with cylindrical cross section. The length of the pipe is L=5D with D beeing the pipe diameter. I am using pisoFoam with the k-epsilon RAS model in OF-1.6.x.

Since I want to couple the inlet and the outlet with a cyclic boundary condition, I have set-up my U as following:
---
dimensions [0 1 -1 0 0 0 0];

internalField uniform (0 0 1.4);

boundaryField
{
newCyclicInOutlet
{
type cyclic;
}

pipe_wall
{
type fixedValue;
value uniform (0 0 0);
}
}
---

for k and eps I just use some reasonable values for a turbulent intensity of about 10%

My first grid point normal from the wall is located, such that should be in the region y+ > 20.

The final; goal is to obtain a fully developed turbulent flow of a Reynolds number of 7000 based on the pipe diamter and the centreline velocity.

The PROBLEM when I perform the calculation until convergence of U.z is that the tends to become zero. So it seems that it either diffuses heavily or there is something wrong with the combination of my cyclic boundary conditions and my field initialization.

Would appreciate any suggestions!!! (just let me know if you need more info!)

thanks a lot!
Florian

[florian_krause]

Ok "solved" it. I have choosen a too large endtime for the set pipe length and initial velocity, so the diffusion makes sense.

Got a nice velocity profile - compared to some exp. and DNS data - for the Re_centreline = 7000 flow, when I look at the flow at ealier times.

Will try to use now some forcing term to lower the diffusion.

have a nice evening!

Florian

[hhuang84]

Hey,
Can you just put the input files here? I am trying to do such a similar job in simpleFoam. but it fail to get converged. Thanks.

[Sunxing]

Hi Florian,

I want to know how do you set the 0/p document. May I have a look at it please?

Regards,

Sunxing