[nhowlett]

I'm investigating a flow scenario related to public health, in particular possible innovations around PPE.

Basically it simulates the scenario of an virus-infected person transmitting bioaerosols when 'speaking' while another person is 'listening' at 1 metre away. This 2nd person is wearing a face shield, while breathing normally thru only their nose.

The full OpenFOAM simulation files for a 3D LES computation is available on my GitHub for reuse and collaboration [1]! Although we're using some serious computing power for the transient (time-stepping) approach (20 CPU cores per simulation), you could possibly get away using significantly less cores as we can do a single simulation in ~ 5 hours (we deemed the flow behaviour is asymptotic at ~ 45 flow-time secs) if you relax the convergence criteria away from standard values (we're interested in innovation applications primarily, not in extremely high accuracy of the fluid dynamics).

[1] https://github.com/TessellateDataSci...dOptimisations

If you have any difficulty understanding anything please reply in this thread so we can do some better instructional design. Thanks!

[nhowlett]

Originally Posted by nhowlett

Although we're using some serious computing power for the transient (time-stepping) approach (20 CPU cores per simulation), you could possibly get away using significantly less cores as we can do a single simulation in ~ 5 hours.

Seems I was quite wrong about scaling across processors. I checked and you can probably do a simulation for the flow-time of 45 secs in ~ 12 hours using only 4 cores.