Investigating Microsimulation Error in Activity-Based Travel Demand Forecasting Using Confidence Intervals

Activity-based models of travel demand using micro-simulation approach inevitably include stochastic error that is caused by the statistical distributions of random components. As a result, running a traffic micro-simulation model several times with the same inputs will obtain different outputs. In order to take the variation of outputs in each model run into account, a common approach is to run the model multiple times and to use the average value of the results. The question then becomes: what is the minimum number of model runs required to reach a stable result (i.e., with a certain level of confidence that the obtained average value can only vary within an acceptable interval). In this study, systematic experiments are carried out by using the FEATHERS framework, an agent-based micro-simulation model particularly developed for Flanders, Belgium. Six levels of geographic detail are taken into account, which are Building block level, Subzone level, Zone level, Superzone level, Province level, and the whole Flanders. Three travel indices, i.e., the average daily number of trips per person, the average daily distance traveled per person, and the average daily number of activities per person, as well as their corresponding segmentations, are estimated by running the model 100 times. The results show that the more detailed geographical level is considered, the larger the number of model runs is needed to ensure confidence of a certain percentile of zones at this level to be stable. In addition, based on the time-dependent origin-destination table derived from the model output, traffic assignment is performed by loading it onto the Flemish road network, and the total vehicle kilometers traveled in the whole Flanders are computed subsequently. The stable results at the Flanders level provides model users with confidence that application of the FEATHERS at an aggregated level only requires limited model runs.