Abstract Manager

Microsimulation modelling is often conducted for a ‘typical day’. Due to historical data limitations, this typical day is a construct formed from data averaged across numerous days. Model calibration, therefore, relies on a synthesized set of hypothetical conditions rather than observed ones. In the United States, the Federal Highway Administration (FHWA) maintains the Traffic Analysis Toolbox, a set of guidelines intended to provide a consistent approach to using transportation tools across the country.  Volume III (Guidelines for Applying Traffic Microsimulation Modeling Software) was updated in 2019 to include a new process for selecting base travel conditions against which to calibrate a microsimulation model. This new process utilizes clustering algorithms to group observation days into clusters of days with similar travel characteristics. Microsimulation models are then calibrated against specific days selected to be most representative of their given clusters. Parsons applied the new FHWA guidance to a project on the I-80/I-94 corridor in Indiana, just east of Chicago, in what was described by our FHWA project partners as the first non-academic application of the process. Our presentation will first delve into the data requirements and general process of the cluster analysis method, also discussing open-source software available to conduct the analysis. Next, we will present lessons learned from Parsons’ experience applying the method in Indiana, including a discussion on difficulties with data sources, scope definition and more. Lastly, we will connect the new clustering method to the Canadian context, addressing questions such as “what Canadian data sources are available” and “where and under what conditions might this method be appropriate?”