Abstract: During the propagation of smoke generated by coal mine underground fires within tunnels, its flow state is high-ly susceptible to the influence of internal obstacles, often leading to the formation of localized smoke stagnation zones. This increases the risk of toxic gas accumulation and seriously threatens personnel evacuation and emer-gency decision-making. To systematically reveal the influence mechanisms of obstacles on smoke flow character-istics, this paper is based on the Particle Image Velocimetry (PIV) experimental method and conducted a study in a 1:10 scaled arched tunnel model. Burning sandalwood was used as the source of smoke and tracer particles, fo-cusing on three key factors: obstacle shape, height, and distance from the fire source, and their effects on the evo-lution of recirculation vortices and velocity distribution. Experiments revealed that regarding obstacle shape, the fixed geometric leading edge of a square obstacle forces the smoke to separate strongly at the sharp corners, forming a structurally stable and relatively large recirculation vortex. The maximum reverse smoke velocity reached -0.027 m/s, and the recirculation vortex height was 184 mm, both significantly higher than those of the circular obstacle. Therefore, compared to circular obstacles, square obstacles pose a higher risk due to their stronger smoke retention capacity and require greater attention in ventilation and evacuation design. Regarding obstacle height, a 10-mm high obstacle mainly formed a small-scale, high-velocity recirculation vortex, while a 20-mm high obstacle resulted in a vortex height extending to 233 mm, but the maximum reverse smoke velocity decreased to -0.015 m/s, forming a large-scale, low-velocity recirculation vortex. Thus, although the vortex flow velocity is weaker with taller obstacles, the large-scale smoke stagnation zone they create increases the accumula-tion space for toxic gases and the risk of personnel being trapped. Regarding the distance between the fire source and the obstacle, at 10-20 mm, the smoke impact was intense, and the vortex structure failed to fully develop, with a maximum reverse smoke velocity of approximately -0.020 m/s. At 20-30 mm, the vortex structure was most stable, and the velocity peaked at approximately -0.028 m/s. As the distance further increased to 30-40 mm, the vortex was pushed downstream, and the velocity weakened to about -0.023 m/s. At 40-50 mm, the smoke ve-locity further decreased to approximately -0.021 m/s. Therefore, there exists a critical distance of 20-30 mm be-tween the obstacle and the fire source, where the velocity peaks and the vortex structure is most stable. When the distance is too close, the intense flow impact prevents the vortex from fully developing; when the distance is too far, the smoke velocity decays, and the vortex flow weakens significantly. In practical fire risk assessment, target-ed analysis combined with obstacle position is necessary. The above research results provide theoretical support for the understanding of smoke spread patterns in coal mine tunnels.