2. AB²C modeling framework
2.1. Activity-based transportation model
Time-space modeling of activities and trips is nowadays preferably done using activity-based transportation
models. These models predict diaries for each synthetic agent in a predefined population. Single parameters that are
predicted are start and end time of an activity, day of week, type of activity, transport mode, location, household
members involved, etc.. The FEATHERS activity-based model was developed for Flanders [18]; this model is
largely based on the earlier developed ALBATROSS, an activity-based model for the Netherlands [19]. These
activity-based models lean on rules derived from decision trees (based on revealed preference data) and are
stochastic in nature. Weeklong diaries for all adult agents (older than 18) in the population are generated and hourly
OD-matrices are constructed. The trips are assigned to the road network resulting in traffic flows. In traffic assignment, capacity of the roads and congestion effects are taken into account (equilibrium assignment using
TransCAD software). Dynamic population density is also calculated: not based on static addresses, but based on the
locations that agents actually visit during each hour. Individual diaries of agents with specific characteristics (i.e.
low or high SEC) are produced to calculate their exposure to BC in the AB²C model.
2.2. Air pollution modelling
To model the minute-to-minute exposure of individuals to BC, three (groups of) submodels were implemented,
each to model a specific part of the exposure.