Emergency Evacuation in Classrooms: A Review

Abstract

Classroom evacuation is constrained by relatively fixed occupant positions, dense desk-and-chair arrangements, restricted aisles, and limited exits, and is jointly influenced by occupant behavior, spatial conditions, and emergency scenarios. This review proposes a four-stage framework comprising scenario and data, experiments and modeling, analysis and evaluation, and optimization and validation, and synthesizes studies on evacuation methods, evaluation indicators, scenario-specific characteristics, and optimization strategies. Existing approaches include evacuation drills and experiments, simulation software, cellular automata, social force models, and agent-based models, with increasing applications of machine learning, reinforcement learning, and virtual reality. Evaluation mainly focuses on time, movement, density and flow, paths, and exits. Fire, low-visibility conditions, earthquakes, and violent attacks exhibit distinct evacuation characteristics, while the effectiveness of exit configuration, desk-and-chair layout, organizational guidance, safety education, special-needs support, and dynamic route guidance varies across occupant groups and scenarios. Overall, research is shifting from single-factor analysis toward integrated analysis of occupants, space, and hazards. However, gaps remain in empirical data, model calibration and validation, multi-factor optimization, and continuous evacuation analysis from classrooms to building exits. Future studies should strengthen multi-scenario data support, experimental–model cross-validation, coordinated optimization, and intelligent decision-making and emergency training.