Investigating the interaction between crowd dynamics and train operations through agent-based modeling

• Main Author: Dela Torre, Alexczar
• Format: text
• Language: English
• Published: Animo Repository 2021
• Subjects: Local transit; Railroad trains; Rail passengers; Computer Sciences
• Online Access: https://animorepository.dlsu.edu.ph/etdm_comsci/11; https://animorepository.dlsu.edu.ph/cgi/viewcontent.cgi?article=1011&context=etdm_comsci
• Institution: De La Salle University

Rapid transit systems (RTS) are essential in supporting the mobility of the population in growing urban centers. In the pursuit of understanding the dynamics of such complex systems, simulation models have been developed for and used by transportation planners to aid with the management of such transportation networks. Agent-based modeling, wherein a system is viewed in terms of agents and their interactions with each other and their environment, is a simulation modeling paradigm that is most suited for the modeling of complex systems such as RTS. While several agent-based models of RTS have been previously developed, the simultaneous and integrated modeling of their train operations with their passenger crowd dynamics have been inadequate. This research uses agent-based modeling to study the dynamics between train operations and passenger crowds in an RTS. As a case study, the RTS of Metro Manila were investigated through the developed simulation model. The model was validated through the use of smart card data and station video recordings. Through face validation using the video recordings as references, the crowd dynamics aspect of the simulation was found to capture most behaviors seen in actual train station environments. As for the validation of the models through the smart card data, the LRT-2 model was found to capture the behavior of its real-world counterpart the best, compared to the models of the other train systems. As the model which captures real-world dynamics the best, scenarios were run through the LRT-2 model to determine the consequences. It was observed that deploying less than 4 trains results in a considerable increase in the mean passenger travel time. Furthermore, when the ridership for the LRT-2 is scaled up alongside the latest Philippine population growth rate of 1.63%, by the year 2023, a 0.78% increase in mean passenger travel time, 2.16% increase in average total time spent, 4.7% increase in the average load factor of all trains, and a 0.47% increase in the time the trains stop at the platforms were observed. In the future, a more refined crowd dynamics model is recommended to be integrated with the developed agent-based model. Moreover, optimizations and interventions are strongly urged to be implemented so that the model could scale better. Lastly, the application of more informed and accurate station layout data, train deployment procedures, and crowd management policies is suggested when performing future experimentation using the model.