MODELING DEMAND FOR A MOBILE APPLICATION-BASED CARPOOLING PROGRAM AT THE AMERICAN UNIVERSITY OF BEIRUT

Abstract

Carpooling is a potential sustainable mode of transportation because of its promising benefits. These benefits include reducing the financial burden on commuters by allowing them to share the cost of a trip with each other. It also has several environmental benefits, including reduced energy use and emission of greenhouse gases. Moreover, carpooling reduces the demand for parking, road congestion, and vehicle miles traveled. Today, carpooling programs are being implemented at different companies and institutions, including universities, to alleviate commuting challenges and establish more efficient travel habits. The purpose of this research is to evaluate the market demand potential of a mobile application-based carpooling program at the American University of Beirut and users’ trade-offs among different attributes of carpooling options. This research involves the design and implementation of a stated preferences survey targeted to all AUB students and staff members in order to help understand the probable impacts of the app-based carpooling on modal choices so as to design an efficient app that accounts for users’ needs. Using the survey data, a discrete mixed logit choice model representing the joint choice of switching from the current mode to carpooling and arrival time choice at AUB is estimated. The model consists of two sub-models: (i) the decision to consider using the app, and (ii) the decision to use carpooling and choice of arrival time conditional on using the app. The results of Model 1 suggest that individuals who live outside Greater Beirut are more likely to consider carpooling. The results show that the primary factors that affect the decision to choose a carpooling offer are time, cost, schedule delay early and late, presence of at least one female other than the participant being a female, and the average rating of all carpool members. The demand models can be used when designing an organization-based app to inform the customization of ride-matching offers for the AUB community based on their preferences. The estimated model is then applied in a dynamic simulation context to test the performance of a practical dynamic carpooling system that involves matching passengers and drivers using a simple greedy algorithm. The results under the base scenario, where the actual market share in the proposed carpooling program is not available, show the market share of those who consider using the app as drivers, passengers, or both is, on average, 68.82%. The average cost savings per person was found to be 9.529 km. The findings show that as the ratio of drivers to passengers (L_d/L_p) increases, the total system 〖CO〗_2 saved increases. We conclude that an app-based carpooling program has good potential for the AUB community. Further studies are needed to determine the practical implementation-level requirements for a successful and self-sustained program.