Abstract
Understanding choice behavior during building evacuations is crucial for predicting pedestrian dynamics, optimizing emergency management strategies, and improving the safety of buildings. However, due to the limited ability of individuals to perceive and combine multiple attributes, they often rely on preferences rather than making strategic and rational decisions. The decision-making process involves cognitive mechanisms that are influenced by internal preferences and option contexts.In this study, a series of virtual evacuation scenarios were designed, and an online Stated Preference (SP) survey was conducted to investigate the influence of internal preferences on exit choice behavior during building evacuations. The results revealed a systematic and patterned preference in exit choice regardless of the decisions made by experienced evacuees. The weighting function between subjective and objective entities exhibited an S shape, and the function was estimated using an empirical equation. The results of the experiments conducted in a virtual environment agreed well with the Cumulative Prospect Theory (CPT), a famous decision-making behavior theory which was proposed to describe decision-makers' preferences over different outcomes.
To simulate the above decision-maker’s preference in the evacuation model, CPT was integrated with the Cellular Automata (CA) model (CPT-CA) to mimic exit and route choice behaviors. Moreover, several reported value functions, risky probability weighting functions, and relevant parameters were used to simulate evacuation movements, and the study found that the CPT-CA model can provide near-realistic evacuation patterns and evacuation times, and the Tversky-Kahneman function form and Hensher et al.'s function form provided a robust explanation for experimental data.
Furthermore, the choice context can facilitate or hinder the evaluation of attributes, and thus preference reversals may occur in the same choice set but in different contexts. Researchers have long known that the context effect – i.e. preference reversals depending on the availability of other options – plays an important role in decision-making behavior. Thus, a series of virtual reality (VR) experiments were conducted in this study to examine the influence of the context effect on exit choice. The results provide novel and clear empirical evidence that preference reversals involving three major effects, the compromise effect, similarity effect, and attraction effect, significantly impact exit choice during evacuation. Notably, the Accentuation of Differences model (AOD) outperformed other discrete choice models, emphasizing evacuees' psychological perspectives on option attributes and the pivotal role of context effects in exit choice.
To model these context effects in building evacuation, the Context Effects in a Social Force (SF) exit choice model (CE-SF) was implemented and outperformed the traditional Utility Function SF model (UF-SF) in terms of evacuation trajectories and exit utilities. Substantial evidence of three context effects during evacuations, regardless of levels of urgency, was provided in this study. The successful integration of both CPT-CA and CE-SF modules within a unified framework enables the comprehensive simulation of evacuation processes. This approach not only facilitates the identification of bottlenecks and dynamic flow patterns but also yields precise estimations of evacuation times across a spectrum of architectural layouts.
| Date of Award | 29 Apr 2024 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Wai Ming LEE (Supervisor) |