Joint modeling of arrivals and parking durations for freight loading zones: Potential applications to improving urban logistics

This paper analyzes truck parking patterns in urban freight loading zones by jointly modeling the vehicle arrival rates and the parking durations. Three models were explored: 1) Count data (Negative Binomial) for vehicle arrivals, 2) Survival (Weibull) model for parking duration and 3) A joint model for arrivals and duration. The count data model estimates the parking demand i.e., the rate of truck arrival, while the survival model estimates the probability that a truck is parked for one more minute. The joint model is compared with separate models for predictability and performance. The dataset used in this research is obtained using a mobile phone parking application, at eight loading zones in the city Vic, Spain over an 18-month period from July 2018 to December 2019, comprised of vehicle parking durations, date, time of arrival and departure, professional activity, and vehicle type (weight). The parking activity data are complemented with built in environment variables of the loading zones, such as the number of establishments in a certain radius, the average walking distance to establishments, the presence of pedestrian pavement, the number of traffic lanes, among others. The joint model outperforms the models estimating the arrival rates and durations separately in goodness of fit and predictability. The model results showed that truck arrival rates vary significantly across days of the week, months, and arrival times. The parking durations are highly dependent on professional activity, vehicle type, and size. Tuesdays and Wednesdays have higher arrival rates compared to other days of a week (except Sundays). Among activities, the transport and parcels require longer parking durations. Among the vehicle types, trucks with gross weight larger than 3.5 tons park longer. This paper concludes by explaining the potential of these modeling approaches in improving urban freight operations, evaluation of various policy implications, limitations, and future research