CFAST simulation speed

[Jianjun Hu]

Hello,

I have run several different cases with the fire at different locations in one building configuration. It turns out that in some cases, the simulation can be done very fast, while in other cases, the simulation would become frozen at certain simulation steps for a long time and then more forward to following steps. The simulation would take very long time to finish. 

Would it possible to understand what it is happening when the simulation becomes frozen? Is it just because of numerical iteration? I wonder if there is any way to avoid the situation so to speed up the simulation?

Best regards,

Jay

[Richard Peacock]

The slow down is purely because the model decides to take smaller time steps to resolve the problem numerically.  There are a number of things that can cause this and typically very small changes in the input file can speed things up.  Most often, it's because a wall vent goes all the way to the ceiling so that an upper layer cannot form in one or more compartments.  Adding a tiny soffit to the vent can usually resolve this.

[fde]

I am observing a slowing down in cases where the fire is underventilated. Is that normal behaviour?

[dr_jfloyd]

Yes, the equations are much more stiff when the fire is underventilated.

When a fire is well ventilated, changes in the vent flow rate (provided they do not cause underventilation) do not affect the fire size in the compartment. When a fire is underventilated, at 13000 kJ/kg of O2, small changes in vent flow can result in large changes in the HRR which changes the pressure which changesthe vent flow. 

[fde]

Thank you for the explanation and the confirmation.