PENG Yuqi, LI Chao, YANG Ruihang, WANG Dachuan, ZHOU Tiejun
[Objective] Deep underground spaces present unique challenges owing to their strong closure, long evacuation distances, and high evacuation difficulty. In disaster situations, evacuees must navigate long, confined horizontal channels, increasing the risk of overcrowding and trampling, which can severely compromise safety. Existing national standards, industry norms, and research results involving underground space safe evacuation primarily target shallow underground spaces, rail transit, and civil defense areas, which are difficult to adapt to the evacuation requirements of deep underground spaces. This highlights the urgent need for specific research into the evacuation design of deep underground spaces. Given the scarcity of relevant guidelines and studies and the potential for congestion during large-scale evacuation in these horizontal channels of deep underground spaces, there is a critical need to optimize the design of horizontal channel widths. [Methods] A lattice-like deep underground space model was developed based on the underground space planning at Nanyang Technological University in Singapore. This model connects chambers in series through horizontal channels. In the first layer of the model, chamber exits and channels are identified, with internal chamber channels, evacuation channel intersections, and vertical evacuation facilities being abstracted as source, intersecting, and terminating nodes, respectively. The shortest evacuation paths from the source to the terminating nodes are considered as directed sides in the network. Then, the topology network model is established. The betweenness centrality index from complex network analysis is used to assess the importance of nodes. Based on these calculations, horizontal channels are divided into three levels: first, second, and third. This classification helps construct a graded road network model. Current specifications guide the assignment of basic widths for these channels at each level. The study uses GoAhead, a self-developed pedestrian evacuation dynamics simulation software, to simulate evacuation scenarios. By setting different evacuation widths for various working conditions, the simulation evaluates evacuation density and time as people move through key node areas in real time, providing insights into evacuation effectiveness. [Results] The evacuation simulation results showed the following: (1) When the width of the third-level horizontal channel was 6.0 m or less, increasing its width effectively reduced evacuee density, alleviated congestion, and shortened the evacuation time. However, beyond 6.0 m, further width increases did not affect the evacuation time. (2) When the width of the second-level horizontal channel was 9.0 m or less, the maximum evacuation density and time were reduced as width increases, thus effectively improving the evacuation efficiency. Between 9.0 and 11.0 m, the density tended to rise as wider channels increased the horizontal walking distance, resulting in longer evacuation times. Beyond 11.0 m, the density decreased again. (3) When the width of the first-level horizontal channel was 15.0 m or less, the maximum evacuation density and time decreased as width increased, thus effectively improving evacuation efficiency. Beyond 15.0 m, the maximum evacuation density continued to decrease, whereas the evacuation time remained stable. A width of 17.0 m was optimal for minimizing crowding and maintaining safety, though exceeding this brought unnecessary economic costs. (4) The suggested horizontal channel widths in deep underground spaces were 6.0 m for third-level channels, 9.0 m (left) and 13.0 m (right) for second-level channels, and 17.0 m for first-level channels. [Conclusion] By comparing the results of evacuation time and density across different widths, this study establishes reasonable horizontal channel widths at each level, providing both theoretical and technical support for the safe evacuation design of horizontal channels in deep underground spaces and helping establish a complete underground safety evacuation system. However, this study focuses on the preliminary design of safety evacuation in deep underground space. Future studies should incorporate the psychological and behavioral factors of pedestrians in deep underground spaces. Testing and calibrating theoretical results with practical engineering cases and actual evacuation data is crucial to improving the safety evacuation model for these deep underground spaces.
