[huangxianbei]

Hi, all
I have just installed OF-2.1.1 which contains channelFoam solver. I'm doing a channel simulation. So firstly I want to clarify the accuracy of this solver. I ran the Allrun script in the channel395 tutorial and noticed something strange.

1. As for the case, Re_tau=395, nu=2e-5m^2/s, so u_tau=7.9e-3m/s and thus the pressure gradient can be calculated, gradP=u_tau^2/h=6.24e-5m/s^2, h is the half-width of the channel,h=1m. However, when monitoring the pressure gradient, the pressure gradient varies all the time and maintains a tendency to fall down and down. I'm confused that how can we identify the solution to reach a statical turbulent flow?

2. In the tutorial case, there are 2 folders, "0" and "0.org", take a look at into the "0" folder, it can be seen that all the initial fields are specified by nonuniform values. The only utility I know is the perturbU which can generate a parabolic velocity field for channel flow, but I don't know how to generate a proper "p" or "nusgs" field. Can someone figure it out?

[huangxianbei]

After a period of searching , I continue my calculation for a much longer time, the velocity at the domain center oscillating around 0.15m/s(the Ubar=0.1335m/s), about 1.1times. So this is exactly what it should be. The first problem is solved.

However, I'm still not quite understand the second problem. How can we initialize the p ,nusgs fields? Or my misunderstanding, just perturbU required for the velocity field?

[huangxianbei]

In fact, if I use perturbU utility without change as this

// Expected turbulent Re(tau) Reynolds number with respect to utau
Retau 395;

// Streamwise component of flow. 0=x, 1=y, 2=z
streamwise 0;

// Spanwise component of flow. 0=x, 1=y, 2=z
spanwise 2;

// Halfheight of channel. This is the direction normal to both streamwise and
// spanwise directions.
h 1.0;

// Set (overwrite) velocity to laminar profile
setBulk false;

// Perturb velocity with some cosine like perturbations
perturb true;

// Perturbation properties
//wall normal circulation as a fraction of Ubar/utau
duplus 0.25;
//spanwise perturbation spacing in wall units
betaPlus 200;
//streamwise perturbation spacing in wall units
alphaPlus 500;
//transverse decay
sigma 0.00055;
//linear perturbation amplitude as a fraction of Ubar
epsilon 0.005;

The result U filed is quite different from the U in the "0" folder of the tutorial. I don't know how the U field comes from in the tutorial. If anyone knows , please tell me about that

[rob3rt 0ng]

Hi Huang,

Have you got the answer on the second part of your question?

I'm very interested to know the answer.

Thanks,
Robert

[huangxianbei]

Originally Posted by rob3rt 0ng

Hi Huang,

Have you got the answer on the second part of your question?

I'm very interested to know the answer.

Thanks,
Robert

Hi, Robert
It has been a long time after I investigated the channel flow case. In my opinion, the fields are generated by a fully developed case so that the user can quickly has an idea of the right flow field.

Xianbei

[rob3rt 0ng]

Hi Huang,

So, if the 0 folder has been defined a posteriori then what is the purpose for defining parameters in the 0.org folder?

Thanks,
Robert

[huangxianbei]

Originally Posted by rob3rt 0ng

Hi Huang,

So, if the 0 folder has been defined a posteriori then what is the purpose for defining parameters in the 0.org folder?

Thanks,
Robert

I would regard it as a guidence on setting up the boundary conditions for a case without posteriori.