Computer modelling of aircraft boarding

I love this article I came across on the BBC website this weekend.

As someone who’s travelled on a lot of planes, I can fully understand the motivation to study methods of boarding a plane. Traditionally, boarding is done in this sequence:

1. Those needing special assistance (e.g. those for whom walking is difficult)

2. Families with young children

3. Then, by row number, in blocks (smaller or larger depending on the mood of the person with the microphone), from the back of the plane forward.

And business and first class passengers can board ‘at their leisure…’ (I hate that phrase).

The ‘back-to-forward’ makes some sense, in that you shouldn’t have people obstructing the aisle at the front of the plane while they throw things in the overhead lockers, while there is a queue of people behind them wanting to get to the back.

But is it the best way of doing it?   According to Jason Steffen, an astrophysicist and computer modeller, the answer is no. He’s used computer simulations of different boarding patterns (e.g. window seats first, then middle seats, then aisle seats) to see what might work better. You can read the original article here. One interesting result is that a completely random way of boarding (which seems to happen anyway with half the airlines – passengers just ignore the instructions given to them to wait until their row is called) is better than a back-to-front method!

Now, some of the various methods have been trialled with volunteer ‘passengers’ in a plane mock-up in Hollywood for a TV programme. It sounds like a great piece of entertainment, but I have a few questions about this from a scientific perspective (and so do those posting comments on the BBC article).

Computer modelling is a really great tool. It enables you to study problems that could otherwise be unreachable. For example, construct computer models of the brain, and it enables you to study phenomena without having to go to the expense of or grappling with the ethics of sticking things into animals. Build a climate model, and you can see how changing the carbon dioxide output of the world would influence the earth’s climate in 50 years’ time. A computer model of the movement of pesticide droplets over a crop field can tell help determine under what weather conditions a farmer should be spraying his crop and what kind of nozzles to use, etc.

But computer models on their own don’t mean a lot. They need validation. That is, they need to be tested against real situations. That can be done in a number of ways – e.g.test each part of the model separately, or test the whole lot at once – and for some models it’s harder than others. Any computer model is only as good as the processes it accounts for, and, in practice, there will be many, many other processes going on that the computer model ignores. The art of getting a good model is in capturing the most important processes, and ignoring the insignificant ones.

I reckon boarding a plane has many processes going on that make it a very complicated thing to study and validate well. A Hollywood mock-up probably won’t get it right. Does it account for factors such as passenger compliance (e.g. those who don’t understand the English in which the instructions are given)? Passengers trying to board with three oversize suitcases and clogging everything up? Families wanting to board together, not in separate waves? Stressed passengers who just want to get on board? Cabin crew running up and down the aisle to fetch that seat-belt extension for the man in 32A? The guy in the window seat who turns up late? And so forth.   I would hazard a guess that if this ‘validation’ exercise were done at a real airport, with real passengers on a real plane with real cabin staff, there might be a different outcome.

Maybe some airline will pick this up, try it, and some other way of boarding (perhaps the budget airline pile-yourself-on method) will become the norm.  But it will need a proper validation study first.

Leave a Reply