The aircraft#
I left at a gate in week 2#
SLOP2710 · Week 10 · Tomas Rehak
The server and the queue
occupy the same space.
Two blocking events#
The stow — variable duration, and the variance is the cost
The seat shuffle — pure self-inflicted damage, caused entirely by the order people were let on
Four strategies#
Back to front — what many airlines use
Random — scatters people along the aisle
Outside-in (WILMA) — kills the seat shuffle
Steffen — alternating rows, outside-in. Consecutive boarders two rows apart
Rank them.#
Where does back-to-front land?#
- boarding this run
- —
- mean over 40 runs
- —
- vs back to front
- —
- worst row queue held up
- —
- Steffen mean
- —
Calling rows 30 to 36#
puts every bag-stowing passenger into the same five metres of aisle at the same moment.
You have carefully concentrated the interference.
And it was tested with real people#
Proposed 2008. Field experiment with a mock cabin, volunteers and luggage broadly confirmed it.
Roughly half the time of back-to-front.
So why does nobody use it?#
Three reasons, none of them technical#
It fights the revenue model. Zone 1 is a product.
It separates people travelling together.
It is hard to explain at a gate, in a second language, over a poor PA.
Rank your options by how they perform
when half-followed,
not by how they perform when obeyed.
The gate is a network#
deplane → clean → board
fuel · cater · load, partly in parallel
You speed up boarding by four minutes.#
How much shorter is the turnaround?#
Bottlenecks move#
Check-in → security. Same journey time, smaller room.
Braess, 1968. Adding a road can make every journey longer.
Emergency departments. The bottleneck is discharge, upstairs.
Bufferbloat. More buffer, more latency, no more throughput.
Back-to-front does not create interference.
It arranges for the interference to happen
where the queue is standing.