An ABM lets you look at all modes of transport
#1 in our “What can you do with an ABM?” series
TLDR: Real people use multiple modes of transport, and we need to understand this when planning transport. An Agent-Based Model (ABM) is the ideal tool for this.
City Modelling Lab is all about simulating the transport choices of virtual people (the agents in agent-based modelling). These virtual people try to achieve a set of activities over the course of a simulated day, and they can choose their mode of transport, when they travel, and the route(s) they take. This means we don’t prescribe what an individual person might choose, but let them explore their options and find out what works for them.
We simulate all modes of transport (so far we have the following in our models: private cars, taxis, walking, cycling, buses, trains, ferries, trams, cable cars, e-bikes, and short-haul flights), and can create modes that currently don’t exist (we could implement drone taxis or new types of vehicle if we wanted).
This means we can look at the detail of trips; both multimodally (trips across all modes) and intermodally (how people combine multiple modes for a single trip). This is what gets us past traditional pitfalls that lead to highway models recommending that the answer is more highways.
The graphic below shows how we can look at these complex trips, both by distance and by time. Traditionally, planners consider a trip to be by its ‘dominant’ mode, usually the mode that goes furthest or is used longest.
However, most public transport trips use multiple modes, and how you classify ‘main mode’ can be complex. Picking the dominant mode by distance paints a very different picture than if you pick by duration. We avoid this problem by preserving information about each mode used as part of a single trip.
One model, all modes, including trips that use multiple modes
In simulation, agents experiment with changing the mode, route, or timing of their trip and evaluate how these options improve or reduce their utility for the day. At the end of hundreds of iterations of exploring options, an individual agent may well be choosing a completely different mode.
This is why we don’t talk about rail demand or road demand in the model, but travel demand across all of our agents’ plans. Preserving the full detail of the modes they chose means we can get more insight into the true use of the network and the impacts of these choices on carbon, equity, and more.
This is the first post in the series “What can you do with an Agent-Based Model?”
