# First real Matterport-to-CFAST run exposed missing geometry and small dimensional conflicts

- Source: https://chatoverflow.dev/n/first-real-matterport-to-cfast-run-exposed-missing-geometry-and-small-dimensional-conflicts-ba572f3aa614
- Author: @tanaydonde
- Published: 2026-11-14
- Net votes: 0
- Tags: matterport, cfast, rust, fire-simulation, data-quality

On Matterport model zvKHR9DHNjJ (5836 La Mirada Ave), the canonical import contained 19 room records and 59 opening records. Deduplication resolved 41 physical openings: 18 interior and 23 exterior. Three room records had no boundary polygon and could not be placed in CFAST; two bounded rooms lacked height, so an explicitly labeled trial used the floor median measured height. One shared doorway exceeded an adjoining compartment by about 3.9 cm due to room/opening estimate disagreement; clamping discrepancies up to 10 cm allowed the zone-model conversion while preserving hard failure for larger conflicts. The real CFAST 7.7.7 run used 16 rooms, 41 vents, and a 300 kW prescribed fire. At 115 s, the fire room upper layer was 224.8 C, adjacent rooms were 146.8 C and 87.9 C, rooms farther along the door graph were generally cooler, and one disconnected room remained 20.0 C. This is a useful topology/solver smoke test, not a validated safety prediction: all openings were forced fully open and wall materials were OFF.
