[scoop_them_mids]

Hello, a generic question more about physics than CFD software.

I used to simulate an axial gas flow, with typically v = 3m/s and L = 1m. Averaging was done over 8 seconds of simulated time. So, it took 24 passages of a fluid particle through the axis to get a reliable average.

Now I have a shear layer axis with, say, v = 10m/s, L = 20m. To travel that distance 20 times at that speed takes 40 seconds. So far I ran the case for 20 seconds max, which didn't give great results, so the idea above seems to make sense, but it could be completely coincidental.

Any suggestions?

[FMDenaro]

Originally Posted by scoop_them_mids

Hello, a generic question more about physics than CFD software.

I used to simulate an axial gas flow, with typically v = 3m/s and L = 1m. Averaging was done over 8 seconds of simulated time. So, it took 24 passages of a fluid particle through the axis to get a reliable average.

Now I have a shear layer axis with, say, v = 10m/s, L = 20m. To travel that distance 20 times at that speed takes 40 seconds. So far I ran the case for 20 seconds max, which didn't give great results, so the idea above seems to make sense, but it could be completely coincidental.

Any suggestions?

The convective time is one of the characteristic times you can have in a flow problem. You can think also to a diffusive time, an acoustic time...

[scoop_them_mids]

Hi FMDenaro,

Thanks for your reply. By convective time I assume you mean the large eddy timescale, which is indeed what I base my estimation on. Basically I want to know if one can scale the time averaging duration of one problem to another similar problem but of a different size.

In other words, "in problem A the averaging duration allows for 20 eddy passages through a probe. If this yields a good average velocity, can I adjust the averaging duration in problem B to 20 eddy passages and hope to get an equally good average?"

[FMDenaro]

Originally Posted by scoop_them_mids

Hi FMDenaro,

Thanks for your reply. By convective time I assume you mean the large eddy timescale, which is indeed what I base my estimation on. Basically I want to know if one can scale the time averaging duration of one problem to another similar problem but of a different size.

In other words, "in problem A the averaging duration allows for 20 eddy passages through a probe. If this yields a good average velocity, can I adjust the averaging duration in problem B to 20 eddy passages and hope to get an equally good average?"

I suggest to work using the non dimensional Numbers. You can do any combination of velocity and lenght scales that drive to the same Re number.
Of course assuming that the Mach number has no relevance in your problem.