Does NO_PRESSURE_ZONES option work correct for a cavity with thin walls?

[mst-parkhachev]

Dear Colleagues!

We are interested in using parameter NO_PRESSURE_ZONES to avoid numerical instability caused by void spaces. Parameter MINIMUM_ZONE_VOLUME is not always suitable for us, because we do not always know the volume of the cavities that arise as a geometry artifact.

We created several test cases to see how NO_PRESSURE_ZONES affects the simulation result.

It turned out that in a cavity with thin walls the pressure increases without limit. It seems that this contradicts what is written in the user manual. The calculation ends with numerical instability.
example_1.txt

In a cavity with thick walls, the pressure increases significantly less, despite the fuel inflow. The pressure in this project is also not equal to the background pressure, but instability does not occur.
example_2.txt

The existence of unwanted cavities in real projects is an artifact. What happens inside them would be of little interest to us if it did not cause instability. The result obtained in a cavity with thick walls would be quite satisfactory. However, in a cavity with thin walls, parameter NO_PRESSURE_ZONES does not seem to work.

Is this the correct behavior of the parameter?

[mst-parkhachev]

We run FDS 6.10.0

[mcgratta]

If you can avoid cavities with thin walls, take steps to eliminate them. Parameters like NO_PRESSURE_ZONES may help, but not always. Just avoid thin cavities.

[mst-parkhachev]

It's a pity that this parameter doesn't always help. In most cases, it is difficult to automatically distinguish whether the cavity was created by our user accidentally or intentionally.

If we use the MINIMUM_ZONE_VOLUME parameter instead, it is not clear what value to set for it. For example, the user may not align the position of the walls very carefully in the geometry editor. Given the large height of the walls, the volume of the cavity may be large.

[rmcdermo]

@mcgratta I tried example_1 above and added

&MISC NO_PRESSURE_ZONES = T, NOISE=F/
&PRES SOLVER='ULMAT'/

thinking this would solve the problem. But actually it forces the case to fail in the first time step. This makes me think that something else is wrong. Please have a look. Thanks

[rmcdermo]

This has been converted to an Issue #14433 . Please continue discussion there.