From cellular automata to agent-based models : addressing urban issues

• Main Author: Quek, Wei Liang
• Other Authors: Chew Lock Yue
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Science::Physics
• Online Access: https://hdl.handle.net/10356/136755
• Institution: Nanyang Technological University

The continual urbanization of our modern world brings forth problems which are increas- ingly difficult to solve. This difficulty is mainly due to the complexity of the systems involved. Comprehensive strategies to solve these problems have to be based on rigor- ous treatments to understand the key interactions governing the systems’ complexities. Using various forms of microscopic modelling, we seek to address two issues in urban systems: congestion when processing a large number of vehicles, and the persistence of bus bunching. By means of the Nagel-Schreckenberg model, we investigated the maximum vehicular flow rate achieved by different solution strategies undertaken to process large amount of traffic. The evaluated analytical form of this flow rate found that for large-scale expansion, parallel expansion - or lane-expansion - is less efficient than serial expansion - which takes the form of a multi-point tollbooth - in the absence of human driving behavior. However when considering human reaction time, it diminishes the efficacy of the serial expansion such that it is no longer tenable for traffic processing. By proposing a novel combination of serial and parallel expansions, the analytical flow rate shows that optimal efficiencies are achieved via configurations with few (many) lanes of a large (small) number of serial units when the processing time is short (long). The problem of bus bunching can be found in every bus system. Using a discrete model, the quantitative dynamics of bus bunching was evaluated analytically. It was found that passenger arrival rate and the difference in buses’ velocities are essential factors in the dynamics of bus bunching. Based on this, a bus-bunching agent-based model known as the Empirically-based Monte-Carlo Bus-network (EMB) model is proposed. A case study on NTU’s Shuttle Bus System demonstrates that the EMB model accurately captures the bus bunching dynamics of the empirical system. Using the EMB model, three classes of intervention strategies - holding, no-boarding and centralized-pulsing - were studied. It was found that the holding and no-boarding strategies are only effective at specific scenarios, whereas the centralized-pulsing seems to be a potentially comprehensive strategy. Through that, we suggest a series of future works to ascertain its practicability as an implementable strategy to solve bus bunching.