Non-physical smoke layering in a large multi-compartment simulation

[clarahickel-eng]

Hi everyone,

I am currently simulating a fire in a large industrial-style building on a single floor measuring 39 meters long, 30 meters deep, and 2.7 meters high. To account for smoke propagation time and oxygen distribution, I have divided the main area into six compartments connected by large vertical wall vents (designed to simulate an open space).

Technical details of my setup :

• Fire : 3MW (t-squared growth) located in comp 1
• Internal Vents : 14.8m wide, 2.5m high, with 0.1m offset from the floor
• External Openings : 5 windows (1m*1m) at a height of 1.5m and one door at floor level in comp 6
• Leakage : Wall Leak Area Ratio = 1.7E-04 and Floor öeak Area Ratio = 5.2E-05 m2/m2
• Software : CFAST version 7.7.5.

The issue :
The behavior of the smoke layer appears unrealistic: the smoke layer in the fire compartment (comp 1) remains very thin (almost nonexistent, while the farthest compartment (comp 6) fills up much faster and shows a significantly lower smoke layer interface (see attached image).

My questions :

• How can the smoke "bypass" the fire compartment and preferentially accumulate at the far end of the floor ?
• Does the 10cm soffit between compartments provide insufficient flow resistance to allow for a stable upper layer to form in the room of origin ?
• Are there any known limitations or specific settings for &VENT parameters when trying to model a large open floor as multiple zones ?

Best regards

[drjfloyd]

CFAST is not a computational fluid dynamics model. Dividing up a large room like this is not in line with the fundamental assumptions of a zone model.

[clarahickel-eng]

Following your previous advice, I understand the limitations of zone modeling for large open spaces.
However, I now need to model a physical mezzanine for another project. Since this creates a real structural separation, what is the best practice in CFAST?
Should I model the area under the mezzanine as a separate compartment connected via a Horizontal Vent (HVENT) to the main hall, or is there a better way to capture the plume entrainment as it spills from under the mezzanine?
Would this 9-compartment model be feasible ?

[mcgratta]

The reason why FDS (Fire Dynamics Simulator) was developed was because some fire scenarios do not fit neatly into the zone model assumptions.