Fire accident seriously threatens the stable operation and personnel security in the subway station, and thermal stratification is an important factor used to determine smoke settlement. Therefore, this paper carried out a series of model experiments by altering the heat release rate (HRR), fire locations, and ventilation conditions. The vertical temperature of the central longitudinal axis was measured, and the S value representing that thermal stratification characteristic was calculated and compared under various fire scenarios. The results showed that the temperature changing curve in the vertical direction was significantly affected by ventilation volume, but it had little to do with the HRR and fire location. The increase of HRR could influence the value of S by enhancing the upper temperature for the whole space and heating the lower temperature near the fire. In addition, the S value decreased with the distance under different longitudinal fire source locations, whereas the variety of S was not monotonical with the deviating transverse distance. And the larger ventilation volume was conducive to restricting smoke diffusion and improving the value of S. This work could provide data support and theoretical reference for controlling smoke stratification under subway station fire.

With the development of urban transportation, metros have become an important means of travel for residents. However, casualty and economic loss might occur in metro systems due to various emergencies. Social media has gradually become the main way to express people’s needs, which provides a new analysis perspective for risk management in metros. This study takes the Zhengzhou Metro flood as an example and collects relevant social media data. Then, the analysis method of social media data evolution characteristics in metro emergencies is proposed. Finally, the evolution characteristics of social media data are analyzed from three aspects: spatiotemporal distribution, emotional distribution, and hot topics classification. The results show that: The temporal distribution of social media data is affected by the emergency process and official media; the spatial distribution of social media data reflects the distribution of stations affected by emergency and temporary shelters; timely and appropriate official media reports are conducive to guiding public emotions toward positive; and the key hot topics can be divided into disaster environment (DE), disaster impact (DI), disaster carrier (DC), emergency management (EM), positive comment (PC), and negative comment (NC). The proposed method can provide support for public opinion analysis and risk management in metro emergencies.

Large metro transfer stations have been widely constructed in China, among which the double-island station faces the serious fire safety issues owing to its large passenger flow. In this paper, simulation cases were carried out to investigate the effectiveness of different ventilation modes by jointly operating tunnel ventilation fan (TVF) and platform screen doors (PSD) under two typical fire scenarios in the platform. The numerical model was established by Fire Dynamics Simulator software and verified via reduced-scale model experiments. The results indicate that the TVF mode of supplying at the end near fire and exhausting at the other end is superior to that of exhausting at both ends. Besides, activating more PSD and TVF on the both sides of platform will restrict smoke in one end to the greater extent. During a fire in the middle of the platform, opening all PSD near tunnel-2 and TVF in tunnel-2 and tunnel-3 is the most appropriate mode. While during a fire at the left end of the platform, activating all PSD and TVF on both sides is the optimal operation mode. The conclusions can provide guidance for smoke control design and on-site emergency ventilation operation in double-island platform fire.