Local timescale to start a steady-state simulation?
 

Hi guys, I've searched this forum for a thread about starting a steady-state simulation using local timescale but unfortunately, I didn't find any (or maybe my search key is just off). Also, I found the FAQ topic regarding using the local timescale factor for simulations needing tighter convergence/oscillating residuals, but this is for when the simulation is in progress.

So my question really is: is it ok to start a steady-state simulation with the set initial conditions using a local timescale factor? I already tried this and got a pretty good monotonic convergence (though the rate of convergence slows down later on and the wall scale does not disappear). What are the implications in the numerics when I use the local timescale factor at the beginning, since what I usually read about local timescale factor is "only use them when the flow is complex, and with uniform elements and moderate aspect ratio"?

Thank you in advance to those who will answer my question! :)

In practice, I never use it and stay away from it as much as I can.

For large aspect ratio meshes, typical of boundary layers flows (CFX is a viscous code), the local timescale factor may show false convergence. It seems smooth convergence, but sometimes if you change to a physical timescale the flow continues converging; therefore, the previous one was not.

What kind of flow are you modeling that you need to resort to it?

Advice:
Set Output Equation Residuals
Write a backup file
Find where the maximum residual is located
Determine the reason why the residual may be stuck in such region, mesh quality, unresolved recirculation zone, too close to an outlet, etc

Hi! Thanks for replying. I'm just modelling a solar panel using an inclined plate with supports. I did run some iterations using local timescale and when I shifted to physical timescale, it went chaotic. Is this a sign of "false convergence"? Also, do you know why local timescale behaves as such?

Thanks!