resolution and simulation time

[Radon_Autodesk]

Hi , Im a new to CFD and have several questions,

1. to start I need to decrease my voxel size as much as I can to test a grille design with openings around 20-40 cm, I can go down to Voxel size 6-10 cm but there is no reproducibility from one model to another; even setting the same tunnel parameters the voxel size comes out different, also I should not make the tunnel too small to avoid wall effects, any ideas?

2. I am running the simulations at 250%, but it takes for ever, even after the Cd is constant for hours the simulation stays in "transient", what can I do to make it "stable"?

3. why after running the simulation for hours in the "velocity" mode, if I change to "pressure" I get nothing but a whole green screen? so to obtain the pressure data I have to run it again.. 

4. My computer has 8 GB ram and 3.1 Ghz processor, would this be a limitation?

thanks a lot in advace

[Radon_Autodesk]

Hi , I solved question #3, all I had to do is to move the plane cut of the image down as it was completely above the model...

[hartogj · Accepted solution]

Hi

1. Flow Design is intended for early stage external flow simulation. If you are modeling something like a grill, that probably involves airflow through a fairly complex set of geometry. You can do some of that work in Flow Design but it is really intended for flow around bulky objects like vehicles, buildings, etc. The reason I mention this is that is sounds like you are having to use a really fine mesh to capture flow behavior in the grill. For this kind of application you may be better off with a full featured simulation tool like Simulation CFD (or others)

2. The solution will report "stable" after sufficient iterations have completed. If you are all the way up at 250% on a 3D model, timesteps may be small and each iteration will take relatively long to complete, yieding a longer period of time before you hit "stabilized." Just a rule of thumb to keep in mind - results will tend to stabilize after 1 fluid residence time has been simulated. In other words, a 10m long tunnel with 10m/s wind will usually require 1 sec of simulated time before results are stable. The time may be a bit higher with higher mesh resolution, but this is a rough rule of thumb.

3. Sounds like you have this figured out.

4. Those specs sound okay but again, if you are running a model that is requiring mesh resolution all the way up to 250% on a 3D model, it will take a while. If you crank up the resolution beyond what the machine can handle memory-wise, it is designed to back off automatically to avoid consuming too much memory.

Hope this helps

Jon