[claco]

Hi All,

I tried to use the "type fan" boundary condition in OpenFOAM.

In order to investigate this b.c., I created a wind tunnel, with an inlet and an outlet. Then I created a 2D disk simulating the rotor, onto which I applied the fan boundary condition.

I set it in such a way to specify a constant pressure jump. I used the rhoSimpleFoam solver, in OF1.7.1.

Simulation residuals are good, and residuals convergence is clearly reached.

Then I compared the results, in terms of totalPressure distribution, with those gained with a commercial code run (the mesh, the pressure jump and the other boundary conditions being the same).

As regards OF results, it seems that the wake due to the fan weakens nearly immediately, preventing me from studying the interference of the latter with possible bodies located downstream (not present in this model).

I used like-second-order discretization schemes in OF.

I ask You what can I do in order to "augment" the fan effects, such those resulting from the commercial code run.

In fact, if the fan effects are only local, I cannot investigate the aerodynamic interference resulting from the wake impinging on the downstream bodies, which is the aim of this work.

I hope someone can help me, since I don't know what it can be due to.

I can accept some differences between the two models, but these ones seem to me unacceptable.


Yours sincerely,


Claudio

P.S. the totalPressure ranges in the two figures are different, but it can be seen that the rotor effects are qualitatively different.

[Sylvain]

Hi Claudio

In OF the fan BC only prescribes a jump of the pressure trough the interface. Unlike fluent (which is your commercial code I guess) it doesn't generate the rotational velocity induced by the rotor nor the increase of turbulence due to the fan. The solution should be to write specific cyclic BCs for k, epsilon and U which represent the behaviour of the variables though the fan. Lot of work.

I'm currently working in modeling the fan of a closed loop WT, and the classic fan BC wasn't good enough for my case.

If OF 2.0.x you have an new BC called fanPressure which is a BC of type patch. I made a post last week about it.

http://www.cfd-online.com/Forums/ope...essure-bc.html

With such a BC you can prescribe you pressure jump according to you fan law, but you can also prescribe BCs of type patch for k and epsilon which represent the increase of turbulence through the fan. (like in the pimpleFoam/TJunction tutorial).

Might be enough to represent the "wake effect" of you fan.

Hoping it is useful

Sylvain

[claco]

Thank You Sylvain,

after neglecting the increase of turbulence due to the rotor, I would like to reproduce the Fluent increase of totalPressure downstream of the rotor. Moreover I'm worried about the fact that the totalPressure decreases too much quickly when compared to that of Fluent. This fact constitutes my real concern.

This quick dissipation of total pressure when compared to that featuring Fluent simulation, makes me scared.

Thank You,

Claudio

P.S.: I built up the mesh with Tgrid. Then I specified a type fan for the rotor, just before exporting the mesh in msh format. After that I imported the mesh via the fluent3DMeshToFoam utility. The rotor patch is automatically recognized as a cyclic patch. I did not make use of createBaffles or other utilities, and I imposed directly a type fan for the rotor, in file p, whereas into the other files I imposed a type cyclic. Is this procedure correct, or I have to use createBaffles or other utilities? The residuals are converged, but the results seems to be wrong, as highlighted in #1. On the contrary, in tutorial TJunctionFan the topoSet and the createBaffles commands are used, and results make sense. So I think this is the key point of my problems, because I do not make use of this utilities. I kindly ask you your opinion.