A 2019 full-scale field experiment in the Tengmieshan road tunnel in Yunnan, China, followed 20 pedestrians evacuating through a 334 m long, 9-degree upward-sloping, 2.0 m wide emergency path toward the surface during a 0.5 MW tunnel fire at about 1025 m elevation, and proposed a four-stage phenomenological model of evacuee performance (adaption, stabilization, fatigue, sprint). The study itself left four issues unresolved: the mechanism behind the right-side running preference that was more pronounced among female participants; whether the four-stage model generalizes to populations not acclimatized to the reduced-oxygen environment; how smoke obscuration in the path would interact with the information signs that participants rated highly under clear conditions; and the unmeasured quantitative effect of continuous handrails on evacuation speed and fatigue. Adjacent literature reports gender speed differences in a long inclined tunnel, altitude-dependent walking speeds, slope-movement speed models, and guidance effectiveness under no-visibility conditions - but none of the four questions has been answered as stated; all remain open.

Background

Extra-long road tunnels in mountainous western China increasingly rely on dedicated pedestrian evacuation paths: enclosed passages connecting the tunnel bore to the surface that can themselves be hundreds of meters long and steeply inclined. When a fire blocks the tunnel, occupants must walk (usually upward) through such a path while exposed to a combination of stressors that ordinary evacuation research does not capture jointly: reduced ambient oxygen at mountainous elevation, a long ascent, a steep grade, and a confined, monotonous environment. Because evacuation-safety assessment compares the time available before conditions become untenable with the time occupants need to reach safety, walking speed along the path - and how it deteriorates with distance, fatigue, and personal characteristics - is the decisive design parameter.

The primary source is a full-scale field experiment by Zhang et al. (2019) in the newly built Tengmieshan road tunnel in Yunnan, China (tunnel elevation about 1025 m, ambient oxygen content 19.31%; the paper frames this as a high-altitude setting). Twenty volunteers (10 male, 10 female, aged 17-52) evacuated through the tunnel's 334 m long, 9-degree upward-sloping, 2.0 m wide evacuation path while a 0.5 MW diesel-gasoline pool fire burned in the tunnel 220 m away; the tunnel lights were on and longitudinal jet fans kept the path itself free of smoke, so visibility remained clear throughout. Nineteen of the 20 participants completed the ascent unaided within 15 minutes. The study recorded response time (the interval from ignition to the start of movement; no significant male-female difference, means 8.2 s and 8.3 s; correlation 0.313 between age and response time), continuous walking speeds from video (mean 1.40 m/s for men versus 1.20 m/s for women, with faster speed decay for women and for participants with weaker athletic ability), lateral position, and questionnaire ratings of the technical installations. From speed profiles and interviews the authors proposed a four-stage phenomenological model of evacuee performance on the path: an adaption stage on entry, a stabilization stage, a fatigue stage in which speed decays, and a final sprint stage (not reached by all participants; stage boundaries were defined only qualitatively). Participants preferred to run along the right side of the path; this bias intensified over the ascent and was more pronounced among women, which the authors could only hypothesize to reflect greater physical exhaustion and wall-seeking for balance, noting that no existing study explains the gender difference. Questionnaires rated the pre-recorded voice alarm and the information signs (way-finding signage) favorably, and 25% of participants suggested installing continuous handrails along the path. All participants were local residents accustomed to the elevation, and the sample of 20 was too small for demographic-subgroup statistics.

The surrounding literature documents pieces of the problem but not its resolution. Gender differences in evacuation walking speeds are repeatedly observed, and were confirmed for a long inclined tunnel by Storm and Celander (2022); field tests by Yan et al. (2022) show that altitude measurably reduces walking speed, with physiological markers (blood-oxygen saturation, heart-rate variability, perceived exertion) tracking the effect; pedestrian speed on slopes varies non-linearly with grade and distance travelled (Nong et al., 2025); pedestrians on building stairs prefer lanes adjacent to the wall and handrail for balance (Xu et al., 2026); and guidance effectiveness under compromised visibility is an active thread (An et al., 2022; Zeng et al., 2024; Chu et al., 2026). None of these addresses the four questions below in the combined setting of an extra-long, steep, reduced-oxygen evacuation path.

Problem Statement

Determine, quantitatively, how pedestrian evacuation along an extra-long steep-slope evacuation path in a reduced-oxygen (mountainous-altitude) tunnel fire depends on the factors that the Tengmieshan field experiment of Zhang et al. (2019) could not resolve. The four questions share one scientific objective - a design-usable account of evacuation performance in this setting, sufficient to inform path width and lateral usage, handrail provision, signage design, and egress-time calculation. The four sub-questions below are jointly posed toward that objective:

(1) Lateral-bias mechanism: identify the mechanism or combination of mechanisms - physiological (fatigue- and balance-driven wall-seeking), biomechanical, or behavioral and cultural (handedness, right-side traffic conventions, following behavior) - that produces the right-side running preference in such paths, and explain why it is stronger among women. The candidate mechanisms are separated only by evidence (quantitative or experimental) that discriminates among them, and the hypothesised fatigue- and balance-driven wall-seeking explanation, on its own, is not such discriminating evidence.

(2) Model generalizability: test whether the four-stage performance model (adaption - stabilization - fatigue - sprint), which was derived from 19 acclimatized local participants on a single 334 m, 9-degree path at about 1025 m elevation, describes the staged speed evolution of other populations; the required contrast is occupants not acclimatized to reduced-oxygen environments. Because the source fixed the stage boundaries only qualitatively, the four stages need quantitative bounds - the speed ranges and durations that bound each stage - which qualitative stage labels do not supply. The non-acclimatized contrast is itself fixed by an explicit membership criterion (for example residence or exposure history relative to the ~1025 m source site, and/or a measured physiological marker).

(3) Smoke and signage interaction: quantify how smoke obscuration in the evacuation path changes the effectiveness of the information signs that participants rated highly under clear visibility, and of the other guidance installations (for example the voice alarm).

(4) Handrail effect: measure the quantitative effect of continuous handrails on safe evacuation speed and fatigue along the steep path, comparing speed and physiological and subjective fatigue with and without handrails under otherwise comparable conditions.

References

  1. Yuxin Zhang, Zhiguo Yan, Hehua Zhu, Yi Shen, Qinghua Guo, Qingchao Guo (2019). Experimental investigation of pedestrian evacuation using an extra-long steep-slope evacuation path in a high altitude tunnel fire. Sustainable Cities and Society 46:101423. DOI 10.1016/j.scs.2019.101423, https://doi.org/10.1016/j.scs.2019.101423 (source paper).
  2. Artur Storm, Eva-Sara Celander (2022). Field evacuation experiment in a long inclined tunnel. Fire Safety Journal 132:103640. DOI 10.1016/j.firesaf.2022.103640, https://doi.org/10.1016/j.firesaf.2022.103640.
  3. Guanfeng Yan, Mingnian Wang, Tao Yan, Pengcheng Qin (2022). Evacuation speed of human beings in road tunnels at different altitudes. Tunnelling and Underground Space Technology 128:104651. DOI 10.1016/j.tust.2022.104651, https://doi.org/10.1016/j.tust.2022.104651.
  4. Tingting Nong, Zhichao Zhang, Tao Wang, Wenke Zhang, Jingyu Tan, Eric Wai Ming Lee, Meng Shi (2025). Dynamics analysis of pedestrian movement on slopes: Modelling, simulations and experiments. Physica A: Statistical Mechanics and its Applications 668:130589. DOI 10.1016/j.physa.2025.130589, https://doi.org/10.1016/j.physa.2025.130589.
  5. Yuxin Zhang, Hehua Zhu, Qinghua Guo, Ricky Carvel, Zhiguo Yan (2021). The effect of technical installations on evacuation performance in urban road tunnel fires. Tunnelling and Underground Space Technology 107:103608. DOI 10.1016/j.tust.2020.103608, https://doi.org/10.1016/j.tust.2020.103608.
  6. Yuxin Zhang, Xinyan Huang (2024). A Review of Tunnel Fire Evacuation Strategies and State-of-the-Art Research in China. Fire Technology 60(2):859-892 (published online December 2022). DOI 10.1007/s10694-022-01357-5, https://doi.org/10.1007/s10694-022-01357-5.
  7. Wenjuan An, Mintao Hu, Jianping Gao, Yaohua Xie, Shanchuan Yu (2022). Experiment on Evacuation Behavior in Full-Scale Tunnel with Different Guiding Pattern. International Journal of Intelligent Transportation Systems Research 20(3):793-804. DOI 10.1007/s13177-022-00327-z, https://doi.org/10.1007/s13177-022-00327-z.
  8. Guang Zeng, Zhiyuan Li, Rui Ye, Shuchao Cao (2024). Evacuation performance of individuals and social groups in a tunnel under no visibility. Tunnelling and Underground Space Technology 146:105661. DOI 10.1016/j.tust.2024.105661, https://doi.org/10.1016/j.tust.2024.105661.
  9. Yaxin Li, Xuepeng Jiang, Qirui Wang, Anzhe Jiang (2025). Full-Scale Experimental Study on Evacuation Behavior of Underwater Shield Tunnel with Evacuation Stairs. Fire Technology 61(5):3875-3898. DOI 10.1007/s10694-025-01760-8, https://doi.org/10.1007/s10694-025-01760-8.
  10. Kunkun Chu, Dan Zhou, Hongyi Wang (2026). Full-scale experimental study on evacuation behavior in a smoke-filled high-speed train carriage with multimodal evacuation guidance strategies. Tunnelling and Underground Space Technology 176:107884. DOI 10.1016/j.tust.2026.107884, https://doi.org/10.1016/j.tust.2026.107884.
  11. Chunhua Xu, Ning Ding, Erhao Zhang, Qinan Xu (2026). Pedestrian Decision-Making Behavior During Stair Evacuation: An Experiment Study on Stair Lane-Selection Preferences. Fire 9(2):64. DOI 10.3390/fire9020064, https://doi.org/10.3390/fire9020064.
  12. Kongjin Zhu, Lizhi Peng, Ning Guo, Bangyu Wang, Xiang Ling, Peng Mei (2023). Individual Ascent and Descent Evacuation Performance on Long Stairs Considering the Effects of Stepping Pattern and Light Conditions. Fire Technology 59(2):903-923. DOI 10.1007/s10694-023-01380-0, https://doi.org/10.1007/s10694-023-01380-0.
  13. Yusuke Miyazaki, Kohei Shoda, Koji Kitamura, Yoshifumi Nishida (2023). Assessing Handrail-Use Behavior during Stair Ascent or Descent Using Ambient Sensing Technology. Sensors 23(4):2236. DOI 10.3390/s23042236, https://doi.org/10.3390/s23042236.
  14. Yuxin Zhang, Max Kinateder, Xinyan Huang, William H. Warren (2025). Modeling competing guidance on evacuation choices under time pressure using virtual reality and machine learning. Expert Systems with Applications 262:125582. DOI 10.1016/j.eswa.2024.125582, https://doi.org/10.1016/j.eswa.2024.125582.