Issue 007 - Safety engineering - Evacuation rate
How many usable exits does a crowded pub need during a fast fire?
A July 2026 fire at the Rong Beer Na Lat Phrao pub in Bangkok killed dozens of people. Reuters reported that police were investigating possible negligence, including obstructed or locked exits, while AP later reported at least 28 deaths, 73 injuries, and possible blocked exits or flammable ceiling materials.
The problem
Estimate the minimum exit flow rate needed to evacuate a crowded pub before smoke conditions became life-threatening.
For this calculation, use the scenario from the original prompt: assume roughly 300 to 500 people are inside the venue. How many usable normal-doorway equivalents would be needed to evacuate everyone in about 2 to 3 minutes?
Because Fermi problems target an order of magnitude, I normally use no more than two significant digits and write most calculations in scientific notation; the Fermi reference explains both conventions.
Before checking sources
Matt's first pass
This was an especially simple calculation as far as the math goes, but I think the assumptions that go into it are questionable.
The problem suggests 300 to 500 people in the venue. Assuming it is a 500-person-capacity venue, I chose a middle value of 400 people, which is 100 less than full capacity.
The problem also suggests the fire could build fast enough that smoke conditions would be life-threatening in about 2 to 3 minutes. That will depend a lot on the size and layout of the venue, and perhaps other conditions such as ventilation and ceiling height.
fast evacuation case ~= 400 people / 120 seconds
~= 3.3 people/second
slower evacuation case ~= 400 people / 180 seconds
~= 2.2 people/secondIf an orderly egress allows about 1 person per second per exit, then a venue that size needs at least 3 to 4 usable exits.
doorway-equivalents needed ~= 2.2 to 3.3 people/second / 1 person/second/door
~= 3 to 4 usable doorway-equivalentsIf there are fewer than that many, I would not expect all of the occupants to exit before smoke conditions became life-threatening.
But I think the per-exit rate would be worse than this. People might panic and aggregate at the exits, slowing the flow rate. I also expect panicked people would not distribute themselves evenly across all exits, so the total time to get everyone out would increase further. In an ideal world, the number of emergency exits should be some multiple of the calculated minimum, maybe double or triple.
Calibration Score
Matt's Calibration Score: 90 / 100
Higher is better: earn points for accurate pegs, sound models, correct math, and a result close to the sourced answer. The image shows percent full of it: 100 minus the Calibration Score.
Pegs: 20/30. The attendance and exit-flow pegs were close enough for the problem.
Model: 30/30. Occupants divided by escape time is the right first-order egress model.
Math: 10/10. The arithmetic was clean.
Result: 30/30. The corrected answer was well aligned with the first pass.
Grounding facts
A single ordinary doorway at 1 person per second clears only about 120 to 180 people in 2 to 3 minutes. That is not enough for a crowded 300-to-500-person venue. Three usable doorway-equivalents move roughly 360 to 540 people in the same window under idealized conditions; blocked exits, poor distribution, smoke, and panic can erase that margin quickly.
After checking sources
Check and recalibrate
Matt's calculation is a good first-order model. The required total flow rate is just occupants divided by available time:
required flow ~= 300 to 500 people / 120 to 180 seconds
~= 1.7 to 4.2 people/secondA normal doorway can plausibly move about 1 person per second under stressed but still moving conditions. NIST's review of U.S. egress-code history notes that observed door behavior can limit flow to about 50 to 60 people per minute, or roughly 0.8 to 1.0 people per second. A Pathfinder verification example based on an IMO door-flow test uses an upper bound of 1.33 people per second for a 1-meter exit, while SFPE-style simulated flows are lower.
doorway-equivalents ~= (1.7 to 4.2 people/second) / (0.8 to 1.3 people/second/door)
~= about 2 to 5 doorway-equivalentsThe practical answer is therefore about 3 to 5 fully usable doorway-equivalents for a 300-to-500-person venue if the survival window is only 2 to 3 minutes. Matt's 3-to-4 result is right in the middle of the calibrated range.
The important caveat is that "doorway-equivalent" does not mean a door merely exists somewhere in the building. It has to be unlocked, unobstructed, visible, reachable, and connected to a path people can actually use under smoke, darkness, stress, and crowd pressure.
Post-check reflection
Matt's reflection
None of my assumptions were wrong or especially poor. I pretty much nailed them. The corrected number was well aligned with what I calculated, and mathematically this one was straightforward.
The scale of the true value does not surprise me, but it really does emphasize the importance of emergency exit availability and clear markings. The calculations seem to support the news items. It is a matter I rarely think about, but it is clearly of critical importance when emergencies arise.
Recommended memory peg
Remember that a normal doorway under stress is roughly 1 person per second, or about 60 people per minute. For evacuation Fermi problems, the core equation is required flow = people / available seconds.
Reader results
Bars show how submitted estimates sort into the answer choices from the gut-check prompt.