Information-driven behavioural dynamics in indoor gas-leak evacuation

Indoor gas-leak emergencies pose severe evacuation challenges due to rapid hazard diffusion, confined spaces, and strong information asymmetries within crowds. Existing evacuation models largely emphasise physical risk fields or aggregate communication rates, while insufficiently representing how warning information is actually received, accepted, and acted upon by individuals. This study develops an information-driven evacuation model by extending the social force framework to explicitly incorporate information reception as a core behavioural mechanism. Pedestrians transition dynamically among uninformed, directly aware, and informed states based on hazard exposure, interpersonal information transmission, and social influence, with information reception quantified through Effective Information Reception (EIR) that integrates cognition, credibility, and transmission attenuation. Simulation experiments in a simplified indoor gas-leak environment systematically examine the effects of EIR, information propagation range, and crowd density on evacuation dynamics. Results reveal clear threshold and saturation effects in information-driven evacuation: higher EIR and larger propagation ranges substantially accelerate evacuation under low-to-medium densities, while their marginal benefits diminish in dense crowds where congestion dominates. Crowd density modulates whether evacuation dynamics are primarily constrained by information availability or by physical congestion, while persistent spatial edge effects hinder complete evacuation even under favourable information conditions. These findings highlight the central role of information reception in shaping evacuation performance and provide guidance for designing effective emergency communication strategies in enclosed hazardous environments.