Atrium Detection on a Beam Grid and Hotel Notification
A six-storey hotel is being renovated. Guest-room corridors are 2.4 m wide. The lobby is a two-storey atrium with beam-and-slab construction: 500 mm deep beams on a 4 m x 4 m grid form pockets spanning the atrium ceiling. The contractor proposes a single row of spot-type photoelectric smoke detectors on a 9.1 m x 9.1 m grid across the whole atrium ceiling, and notification by corridor sounders only, measured at 75 dBA at 3 m in the corridor. Guest-room doors are rated, tight-fitting and gasketed. A transfer fan moves air from the corridor into the guest rooms through a return plenum. Mechanical-room ambient noise is 95 dBA. The designer must review the detection and notification concept.
1. Evaluate the proposed detector grid given the beam construction. Will it detect a fire in the atrium in time to support the smoke control concept?
Read the worked answer
The beams change the ceiling
500 mm is 19.7 in, which exceeds the 18 in threshold at which beam pockets are treated as separate compartments for detector spacing. Each 4 m x 4 m pocket behaves as its own small ceiling: smoke has to fill the pocket and bank under the beam soffit before a detector anywhere else responds. The 9.1 m smooth-ceiling spacing credit simply does not apply across the beams.
What the proposal actually provides
- A 9.1 m x 9.1 m grid is roughly one detector per 83 m2.
- A 4 m x 4 m pocket is 16 m2, so most pockets contain no detector at all.
- A fire in an empty pocket must fill that pocket, spill over the beam, and fill adjacent pockets before it reaches the nearest detector - minutes of delay in a two-storey volume, and far too late for the smoke control system to hold the layer above the design height.
Correction
- At least one detector centered in every pocket, mounted on the pocket ceiling and set back from the beam face (not on the beam soffit).
- Keep detectors clear of supply diffusers (typically 3 ft) so the airflow does not blow smoke away from the head.
- This is roughly a fivefold increase in device count, and the contractor needs to price it.
Higher-level judgment
Spot-type photoelectric detection in a tall atrium is often the wrong technology regardless of spacing. Projected-beam or air-sampling detection at several levels follows the smoke plume through stratification and responds far earlier. If the atrium has a smoke control system, the detection layout has to be tied to the smoke-filling analysis that sized the exhaust - a device grid chosen by rule of thumb and not coordinated with that analysis is a design failure even if every pocket has a detector.
2. Evaluate the notification concept: 75 dBA in the corridor, sounders in the corridor only. Does it work for sleeping guests? What about the mechanical room?
Read the worked answer
Sleeping areas are the governing case
- General public-mode criterion: 15 dBA above ambient at the listener.
- Sleeping areas are stricter: the alarm must produce 75 dBA at the pillow, or 15 dBA above ambient if that is greater.
- 75 dBA in the corridor, attenuated by a rated, gasketed door (realistically 20-30 dB), gives roughly 45-55 dBA at the pillow. That is 20-30 dB short.
- Trying to close the gap from the corridor would require 95-105 dBA in the corridor, which is not a design - it is a hearing hazard and an annoyance complaint generator.
Conclusion: corridor-only notification is not an acceptable substitute for in-room notification. Provide audible appliances in each guest room, synchronized on the same circuit, and use 520 Hz low-frequency sounders - low frequency wakes sleeping occupants far more reliably, particularly children, the hearing-impaired and people impaired by alcohol or sleep aids. Verify at the pillow of the worst-case (largest, best-sealed) room, not at a sample door.
Mechanical room
- Ambient 95 dBA requires 110 dBA. That is at the practical ceiling for notification appliances; audible output is normally capped around 110 dBA at minimum hearing distance.
- Provide a documented ambient measurement (sound levels vary with equipment running), specify appliances that can achieve 110 dBA at all required points, and provide visual notification so the alarm is unambiguous. Where ambient exceeds 105 dBA, visual-only notification is the intended answer.
Separate issue the concept misses: detection in the corridor does not satisfy the requirement for smoke alarms in each guest sleeping room, and the transfer fan moving corridor air into rooms through a return plenum needs a duct detector if its airflow exceeds the threshold and needs its smoke spread path reviewed either way.