Velocity tolerance

[Matthew Bilson]

Hi - 

Running a tunnel case here (12 m long, 2.5 m wide, 1.25 m high) and having problems with numerical stability. I wanted to flag something I saw. When I reduced the velocity tolerance the models still went unstable, and it appeared like the velocity tolerance was having no impact.

Looking at the User Guide Section 9.1, it states that:

If you do not want FDS to suspend the pressure iteration, set SUSPEND_PRESSURE_ITERATIONS to F on the PRES line. This will be done automatically if you change the default values of VELOCITY_TOLERANCE or MAX_PRESSURE_ITERATIONS.

Grid resolution in my model was around 0.05 m, making velocity tolerance 0.025 m/s (dx/2) (model had three grids, but all had same resolution). I had the following line set for CASE A (*.out file attached):

&PRES PRESSURE_TOLERANCE=100.0,MAX_PRESSURE_ITERATIONS=200,VELOCITY_TOLERANCE=0.001,TUNNEL_PRECONDITIONER=.TRUE. /

For CASE B I added extra instructions (out file also attached):

&PRES PRESSURE_TOLERANCE=100.0,MAX_PRESSURE_ITERATIONS=200,VELOCITY_TOLERANCE=0.001,TUNNEL_PRECONDITIONER=.TRUE.,CHECK_POISSON=.TRUE.,SUSPEND_PRESSURE_ITERATIONS=.FALSE. /

CASE A went unstable and the velocity tolerance was not met just prior to this. For CASE B - it appears to be holding stable. Could it be that "SUSPEND_PRESSURE_ITERATIONS=.FALSE." is not automatically changing when we change the velocity tolerance, thereby meaning that even though we reduce the velocity tolerance, the instability keeps happening because it is not driving the errors low enough?

[Kevin McGrattan]

Bad wording, I suspect. If you set SUSPEND_PRESSURE_ITERATIONS=F, it will remain that way. It means that FDS will continue to iterate the pressure even if little progress is being made. The iterations will stop only when the max is reached, or the tolerance is satisfied. 

Can you send me the buggy case.