Can anyone explain why using the local timescale factor option is not advisable for obtaining final results? I know it has been mentioned somewhere but cannot find the reference.

Regards

Adam

The local timescale factor will advance the solution within each control volume at a multiple of the local timescale. This means that as a whole, the solution is advancing at different rates throughout the solution and can make the RMS and MAX residuals appear to be low when a solution may not actaully be acheived.

If the solution is truly converged, changing to a Physical Timescale will have no effect, whereas if it hasn't, the residuals will jump back up. It is best to think of this as a sanity check.

Personally, I never use the Local Timescale Factor. It used to be necessary in some cases to get things going, but the solver is much more robust now. Generally, if the solver blows with a Physical Timescale and you need to resort to this, it is only masking other problems such a grid or setup issues.

Regards, Robin

Hi Adam,

this techique is useful when the problem has a wide range of time scales but since the local time step can be very small in some cells it is possible that you run what seems to be a long time but by the global scale is not enough to bring it to steady state.

If you can running a physical time step is advisable since a steady solution should run for all time steps...that is the theory anyhow. In practice especially with complex physics this is not so. If you run problems with all one characteristic time scale like a blade passage then the physical time step usually works fine...that is why CFX makes this strong recommendation.

Hope this helps,

Regards,

Bak_Flow