Deboarding simulation · A320, 30 rows 3-3, door L1 · top view to scale

Who yields the aisle?

After landing, people already standing in the aisle usually stop to let each row out. The hypothesis: it would be faster if the aisle never stopped and those still seated waited for a real gap, using the wait to get their things ready. Both planes carry exactly the same passengers; only that rule changes.

See both cabins in 3D → Contribute data from a real video

Courtesy

the aisle always yields (p_yield = 1)

0:00 0 out – pax/min

Aisle first

nobody in the aisle stops (p_yield = 0)

0:00 0 out – pax/min

Seated, getting ready Ready, waiting for their row Standing, wants to step out Stepping into the aisle Taking the bag down (blocks) Walking back for their bag Walking (with bag) Stopped to let a row out Stopped on the way Travelling together

Flight parameters

Changing any of them regenerates both planes with the same passengers and reruns the Monte Carlo.

Courtesy of the top plane Probability that someone walking down the aisle stops to let a row out.

Rows

Occupancy

Have a bag in the overhead bin Taking it down takes 1.5 to 30 s; half take less than 7.3 s.

Take it down standing in their row, if in reach Everyone else takes it down in the aisle, blocking it.

Bags stored away from their row If the bag is behind and out of reach, they wait for the aisle to clear and walk against the flow.

Travel with a seat neighbour Share of half-rows where neighbours leave together: whoever is in the aisle waits for the others, whatever the rule.

Stop once on the way out A second item, a coat, looking back: they block the aisle for 1.5 to 12 s.

Door: seconds between passengers If the door is the bottleneck, the aisle rule stops mattering.

Passengers off the plane

Cumulative over time, for this particular plane.

Courtesy Aisle first

Many planes, not just one

One plane can go well or badly by chance. Here 40 different planes are simulated with the parameters above, under both rules and intermediate degrees of courtesy. Differences are paired: the same passengers are compared.

Planes per point Computing…

Courtesy, mean total time –

Aisle first, mean total time –

Time saved by aisle first –

Total time by degree of courtesy

Mean over all planes, with a 95% confidence interval.

Show the data as a table

Probability of yielding	Mean total time	95% CI	Mean exit per passenger	Mean rate

When does it stop holding?

Sweep with 100 planes per value (node sweep.js, the same model as this page and the Python simulator): one parameter changes at a time and the rest keep their defaults. The last column is how much longer courtesy takes than aisle first.

Parameter	Value	Courtesy	Aisle first	Difference (95% CI)

How the model works

Every number is a parameter with its source; those without data yet are estimates waiting for video measurements (see DATA.md).

Continuous aisle. Each person has a body depth (28-45 cm), a free walking speed (0.5-1.1 m/s; median 0.8 m/s, measured by Schultz) and limited acceleration. They keep 15 cm from the person ahead when stopped and a 0.6 s headway when moving; once stopped, they wait for 25 cm more before starting again, so queues restart as a wave.
Inside the row. Everyone gets ready seated at the same time (0-10 s). The aisle seat leaves first; the others stand up and shuffle sideways once there is room.
Bags. 65% have a bag overhead. If they can reach it from their row, 80% take it down while they wait; otherwise they take it down in the aisle, blocking it. Carrying a bag, they take up 25 cm more and walk 15% slower.
Bins. 28% of bags are not above their row. If the bag is ahead, they stop to pick it up on the way; if it is behind and out of reach, they wait until the aisle up to it is empty, reserve it and walk against the flow.

Groups and stops. In a third of the half-rows, neighbours travel together: whoever is in the aisle steps a little ahead and waits for the others to get out behind them, whatever the rule. They give up if someone else gets in between or after 30 s. And 15% stop once on the way out (a second item, a coat) for 1.5 to 12 s.
The rule being compared. When approaching a row where someone is getting up, a walker decides once whether to stop and let the whole row out (probability p_yield).
Gaps. Under any rule, someone at the row edge may step out without being let in if there is room and nobody is closer than 1 m plus 1.5 s per m/s of speed.
Calibration. With courtesy, a full A320 takes 13.2 min (about 14 pax/min), a little slower than observed: 15-17 pax/min, and 6:27 to 13 min with at least 80% occupancy. Groups and stops are still estimates waiting for video data.
Same code. This page runs an exact port of the Python simulator, with the same random number generator. Check against Python: pending