Density-driven queue formation and synchronized movement in stairwell evacuation: A hybrid experimental and simulation study

Existing stairwell evacuation studies face critical bottlenecks: insufficient validation of simulation models on inclined surfaces, disconnection between experiments and simulations, ambiguous formation mechanisms of high-density evacuation bottlenecks, and lack of quantitative analysis on the role of synchronized and queuing behaviors. To address these issues, this study integrates experimental and simulation methods to investigate density-driven queue formation and cooperative movement in vertical evacuation. Single-file evacuation experiments were conducted to collect microscopic movement data (e.g., step length, speed difference), which was used to calibrate and validate a Social Force Model (SFM) tailored for inclined stair scenarios. The validated model was then extended to multi-lane evacuation simulations to replicate real-world conditions. Key findings include: (1) The SFM, after experimental calibration, reliably predicts macroscopic evacuation trends on stairs, filling the gap of inclined-plane model validation; (2) Three prerequisite conditions for triggering the synchronization mechanism are identified: approximately linear movement trajectory, avoidance of directional change areas, and speed difference within ±0.5 m/s between adjacent pedestrians; (3) Self-organized queuing behavior significantly alleviates bottlenecks at stair ascents, reducing evacuation time by at least 38% under high-density conditions (ρ≥1.39 people/m²); (4) An optimal speed range (1.0–1.2 m/s) is quantified for high-density evacuation, balancing efficiency and safety. This research establishes an "experiment-simulation" closed-loop framework, clarifies the intrinsic link between microscopic behaviors and macroscopic efficiency, and provides actionable insights for optimizing stairwell design (e.g., width setting, queue guidance) and emergency evacuation strategies (e.g., speed control, 2-second departure interval).