Abstract Manager

The proliferation of GPS and cell-network connected devices has revolutionized location data collection. Even in developing countries, smartphones and other connected devices are affordable and widely used. This presents a substantial opportunity for transportation modelling applications, as it permits the accurate tracking of a large proportion of an area’s population without having to rely on expensive and, at times, biased surveys. At the same time, techniques used to expand the data collected by a single cell carrier to be representative of an entire population are not well established. 

Arcadis was retained by a private interest to conduct a feasibility study for a prospective toll highway in Central America. The study involved estimating the size of the travel market, the potential time savings offered by the new highway, the local value of travel time savings, optimal toll rates, and traffic and revenue over a 15-year horizon period. The project used innovative and non-traditional data sources like cell phone location data and the Google Maps API to supplement conventional traffic counts to inform conclusions and recommendations.  

A partnership with a local cell carrier to obtain zone-based location records for all users that were observed in the study area over a two-year period was central to the analysis. The data—amounting to over 500 GB—were fused with road-network data sourced from Open Street Map to impute trip starts and ends and distinguish them from trips passing through a zone. Traffic counts around the study area were used to iteratively adjust the cell data to better represent the area’s complete travel market, inclusive of all modes of travel and trip makers that were not customers of the local cell carrier, resulting in a complete trip table. The Google Maps API was used to determine zone pairs that would save travel time if the highway was used. This, in combination with a locally-calculated value of travel time savings and the trip table described above, was then applied to estimate toll traffic and revenue for various toll scenarios.  

The presentation will provide an overview of the process used to deliver long term traffic and revenue forecasts using a combination of novel and established data. The focus will be on informing the audience on how “big-data” sources can replace expensive travel surveys in early-stage studies for revenue-generating infrastructure projects, a topic that is highly relevant given uncertain funding arrangements many infrastructure owners and operators face.