[PopeySailor]

Hello, can I ask anyone for advice with CFD simulation of fan in Helyx OS? My aim is to simulate blades rotation and I want to achieve volumetric flow rate on outlet as value given by manufacturer (980 rpm, 40 m3/s). Till now, I achieved only very small values, and solution did not converged well. I am begginer so I am not sure about what settings to use. Previously I use:
• Transient simulation
• Inlet: Zero pressure gradient, Outlet: total pressure = 0
• Blades: moving wall (rad/s, axis in centre)
• Sim. Time: 10 s, time step: 0,01 s
If you have any recommendations, I will be very grateful.

Fan02.png

Fan01.png

Blades.PNG

[spenceme]

I'm not familar with the helyx os workflow, but typically to get advice on this you will need to post up some of your input files. specifically your boundary conditions probably here. also details about your timestep as 0.01s is probably much to large to start with. and your mesh...if the image of your rotor is the meshed model then your mesh needs work.

[Fouch]

Agree.
Firstly check your mesh and refine.
Secondly decrease your time step to 1e-6

[PopeySailor]

Thanks for response, file with BC and input geometry is attached here. Also I tried finer mesh and time step as adviced, but still got very small values of volumetric flow rate. Sory if some information is missing, I will supply as soon as possible. OF CFD is very new to me.

boundary.txt
Geometry.zip

[Fouch]

You didn't send the values of the boundary conditions for U.
Could you share it ?

[PopeySailor]

Of course, sory

U.txt
p.txt

[PopeySailor]

SOLVED, co for future Foamers with similar problem! Problem was not in not small enough time step or mesh, but opposite. It takes some time to velocity profile to occure (around 40 s), so it was neccessary to set simulation time larger. With time step 0,5 results occur in scale of expected ones (smaller is better but for scale of results and fast preview...).

Still got another question. In transient simulation, it takes some timo to velocity profile to fully develop. Is there some trick to estimate scale of time which has to be watched out ( for example to know that it is smaller than 1 s, around 10 s etc.)? In this case, if I set 1s of simulation with time step of 0,00001 s, it will count for long time, but no reasonable result will be achieved.

Thanks a lot.

[fxzf]

Do you need a mixing plane？

[PopeySailor]

Have no experience with this feature, so probably not. Have to study it first, sory