Sprinkler credit versus flashover in a self-storage conversion
A 1960s godown in Kwun Tong is being converted to a self-storage facility. The fire engineering team is asked whether the existing sprinkler system, designed to ordinary hazard, can be credited in a performance-based fire safety strategy for a 4.5 m × 6.0 m × 2.7 m storage compartment. The compartment has one 1.0 m × 2.1 m doorway, assumed open. The proposed storage is packed polyurethane foam and cardboard with a potential peak heat release rate of 8 MW, growing at a fast t-squared rate (α = 0.047 kW/s²). The sprinkler heads are 68 °C, quick response, 5 mm bulb, spaced at 3.5 m, with the nearest head 1.8 m above the fuel. The client argues that sprinklers will always control the fire before flashover. The approving authority asks for a quantitative check and a clear statement of the model's limitations. A colleague suggests using the McCaffrey/Quintiere flashover correlation with A√H.
1. Using the A√H flashover correlation, estimate the heat release rate at flashover and the time to reach it for the assumed fast t-squared fire. State whether flashover could occur before sprinkler activation and explain what the calculation does and does not prove.
Read the worked answer
Flashover threshold. Use Q_FO ≈ 750 × A√H (kW). Here A = 1.0 × 2.1 = 2.1 m² and H = 2.1 m, so A√H = 2.1 × √2.1 = 2.1 × 1.449 = 3.04 m^2.5. Therefore Q_FO ≈ 750 × 3.04 ≈ 2.28 MW. A common alternative coefficient of 610 kW gives 1.86 MW, so the defensible statement is a threshold of roughly 2 MW, with real uncertainty.
Time. For Q = αt² with α = 0.047 kW/s²: t = √(2,280/0.047) = √48,510 ≈ 220 s, about 3.7 minutes. Using 1.86 MW gives about 199 s. The fire therefore reaches the estimated flashover threshold in roughly 3–4 minutes.
Could flashover precede sprinkler activation? It could. A quick-response 68 °C head 1.8 m above a fast fire in a small compartment may operate in the same time band, but the result is not determined by this calculation. It depends on the ceiling jet temperature and velocity, the RTI, the distance from the fire axis, spray cooling and shielding by the storage. The client's assertion that sprinklers will 'always' control before flashover is not supported.
What the calculation does not prove. It does not prove that flashover occurred, that the sprinkler failed, or that the fire followed the assumed growth curve. It assumes a single opening, a well-mixed upper layer and a known heat release rate. It is not valid for a closed, under-ventilated compartment, and it does not model sprinkler spray or storage geometry. To accept a sprinkler credit, require a zone or CFD model with the actual fuel layout, a sprinkler activation analysis, water-supply reliability, and sensitivity cases for the door open and closed.
2. The client now proposes to treat the compartment as closed and argue that flashover cannot occur. Explain why that argument is technically unsafe and set out the additional analyses and evidence you would require before accepting any sprinkler credit.
Read the worked answer
Closing the doorway does not make the compartment safe. It changes the fire from a ventilation-controlled to a fuel-rich, under-ventilated condition. The A√H correlation is not valid for a closed compartment, and a closed room can still reach flashover because heat accumulates and the linings and fuel receive increasing radiation. When an occupant or firefighter opens the door, the inflow of air can produce a backdraught or a rapid increase in burning. The argument is therefore technically unsafe and does not represent a realistic worst case.
Additional work required:
- Fire model. Use a zone model or CFD with the actual compartment geometry, the storage racks, the door open/closed cases and the ceiling vent conditions. Model the growth phase, not just a steady fire.
- Sprinkler response. Calculate the first head operating time using the actual RTI, spacing, ceiling height and fire location. Check spray coverage, storage shielding and water supply duration and pressure.
- Sensitivity. Run cases for a 5 MW fire, a fire at the rack face, a fire shielded by packaging, and a door opened at 5 minutes. Report the flashover margin in each case.
- Reliability. Provide evidence of maintenance, valve supervision, pump reliability and water supply. A performance-based credit cannot rest on an ideal sprinkler that is not assured in service.
- Acceptance criteria. State the tenability or flashover criterion, the modelling uncertainty and the peer-review process. If the margin is small, do not grant the credit; require additional passive protection or a redesign of the storage.