Home News Modelling Experience Resources About

Simulating One Million People

Following is a summary of the findings of a modelling exercise simulating air travel by one million people over a period of two thousand days with FLAP in place. Travellers were split into six bands to repesent the real-world balance between frequent and infrequent flyers discussed here and here. In the top band travellers fly six to twelve times a year. In the bottom band flyers fly zero to one time a year.

The initial daily allowance was calculated as the average total daily distance travelled during the first 100 days. This was then enforced, and reduced by 0.05% each day, from Day 101 onwards.

The source code for the FLAP implementation and modelling tool used is available here to download and use.

Global Air Travel is Reduced

This chart shows how the daily allowance and the total daily distance travelled varies over time. Air travel is reduced by 40% over the course of five years. The yearly cycle of variation in distance travelled is due to the modelling of seasonal traffic flows. You can explore a running model in action my checking out the bot accounts in the FLyingIsAPrivilege Web App.

Frequent Flyers Can Still Fly More

This chart shows how the average distance travelled every 100 days by the most and least frequent flyers compares over time. Frequent traveller flying is reduced whilst infrequent traveller flying remains steady, but despite this frequent flyers fly more throughout the 2000 day period.

Infrequent Flyers are Impacted Less

This chart shows the percentage of trips that the most frequent and infrequent flyers are forced to cancel at the planning stage (see Experience for an explanation of how this works in practice). Infrequent flyers suffer far less disruption to their travel plans than frequent flyers.

Opportunities for Improvement

The model used to generate these results is basic. Whilst it does simulate global air traffic on geniune routes between genuine airports it diverges from real-world flying patterns in many ways, including:

We don't believe any of these omssions negate the validity of the results but nevertheless plan to address some in future iterations of the modelling tool.

For more details on this and the modelling exercise as a whole please refer to this white paper.