Adaptive routing for subway passenger evacuation

Bashynskyi Oleh

Lviv State University of Life Safety

https://orcid.org/0000-0002-0243-7519

Peleshko Marta

Lviv State University of Life Safety

https://orcid.org/0000-0002-9315-1590

Beseda Andrii

Lviv State University of Life Safety

https://orcid.org/0009-0004-6252-1431

Ketsmur Rostyslav

Lviv State University of Life Safety

https://orcid.org/0009-0000-5602-8000

DOI: https://doi.org/10.52363/2524-0226-2026-43-8

Keywords: subway, evacuation, emergency, routing, artificial intelligence, graph modeling, passenger flow

Аnnotation

The article explores approaches to modeling the evacuation of people from the subway in com-bined emergency situations and demonstrates an example of using artificial intelligence support to select an evacuation route. The relevance of the topic is due to the fact that the subway combines high passenger density, a closed underground space, and dependence on narrow elements of the evac-uation infrastructure. For Ukraine, this problem acquires additional importance, since metro stations are used not only as transport facilities, but also as shelters during military threats. The purpose of the study is to generalize modern approaches to modeling the evacuation of subway passengers in com-bined emergency situations and demonstrate a scenario in which the artificial intelligence approach to route selection is compared with a static evacuation scheme. Methodologically, the study combines the analysis of scientific papers with demonstration graph modeling in the Google Colab environment using Python. The metro station is presented as a weighted graph, in which the length, conditional hazard index, and throughput are given for each section. Two scenarios were implemented: evacua-tion without artificial intelligence, where the route was determined by the shortest path length, and a scenario with artificial intelligence support, where the route selection was carried out taking into ac-count the risk and current load on the sections. As a result, it was found that under a static scheme, passenger flow is concentrated on a shorter but riskier route, while the adaptive approach allows you to redirect traffic to bypass the dangerous zone. In the scenario with artificial intelligence support, the average risk index decreased from 9.0 to 1.2, the number of passengers in the risk zone decreased from 60 to 0, and the overload factor of the critical section decreased from 5.0 to 2.4. At the same time, the average route length increased from 3.0 to 4.4. It is concluded that even simplified AI-supported modeling reveals the benefits of adaptive routing in the context of combined threats in the metro.

References

1. Mandal, T., Rao, K. R., Tiwari, G. (2023). Evacuation of metro stations: A review. Tunnelling and Underground Space Technology, 140, 105304. doi: 10.1016/j.tust.2023.105304
2. Qin, J., Liu, C., Huang, Q. (2020). Simulation on fire emergency evacuation in special subway station based on Pathfinder. Case Studies in Thermal Engineering, 21, 100677. doi: 10.1016/j.csite.2020.100677
3. Zhang, N., Liang, Y., Zhou, C., Niu, M., Wan, F. (2022). Study on fire smoke distribution and safety evacuation of subway station based on BIM. Applied Sciences, 12(24), 12808. doi: 10.3390/app122412808
4. Zuo, S., Mao, Z., Fan, C., Chen, X., Gong, M., Ren, J. (2024). Dynamic planning of crowd evacuation path for metro station based on Dynamic Avoid Smoke A-Star algorithm. Tunnelling and Underground Space Technology, 154, 106145. doi: 10.1016/j.tust.2024.106145
5. Huang, P., Lin, X., Liu, C., Fu, L., Yu, L. (2024). A real-time automatic fire emergency evacuation route selection model based on decision-making processes of pedestrians. Safety Science, 169, 106332. doi: 10.1016/j.ssci.2023.106332
6. Bunkus, S. V., Ohnieva, O. (2024). Doslidzhennia metodiv modeliuvannia stsenariiv evakuatsii v umovakh nadzvychainykh sytuatsii. Visnyk Khersonskoho Natsionalnoho Tekhnichnoho Universytetu, 4(91), 226–235. doi: 10.35546/kntu2078-4481.2024.4.29
7. Kurylo A. H., Kustov M. V., Zimin S. I., Hubenko A. O. (2025). Modelyuvannya ryzykiv kaskadnykh avariy na zaliznychnomu transporti v umovakh viyny. Problemy nadzvychaynykh sytuatsiy, 2(42), 123–137. doi: 10.52363/2524-0226-2025-42-9
8. Mitina, N. B., Plys, M. M., Rohalov, M. V., Malynovska, N. V. (2018). Metodychno-praktychni aspekty planuvannia ta orhanizatsii evakuatsii v umovakh nadzvychainykh sytuatsii. Budivnytstvo, Materialoznavstvo, Mashynobuduvannia. Seriia: Bezpeka Zhyttiediialnosti, 105, 148–153. doi: 10.30838/P.CMM.2415.250918.151.144
9. Popko, S. O., Kolokolnikov, V. O., Cherepnov, I. A. (2025). Mozhlyvist vy-korystannia shtuchnoho intelektu dlia poperedzhennia ta likvidatsii nadzvychainykh sytuatsii v metropoliteni. In S. O. Vambol (Ed.), Bezpeka liudyny u suchasnykh umovakh: zbirnyk dopovidei 17-i Mizhnarodnoi naukovo-metodychnoi konferentsii, 143–145. NTU “KhPI”. Available at: https://repository.kpi.kharkov.ua/handle/KhPI-Press/96596
10. Poteriaiko, S. P., Bielikova, K. H., Tverdokhlib, O. S. (2022). Problemni pytannia evakuatsii naselennia v bezpechni mistsia v umovakh voiennoho stanu. Ukrainskyi keis [Problematic issues of evacuating the population to safe places under martial law: The Ukrainian case]. Publichne Uriaduvannia, 5(33), 58–63. doi: 10.32689/2617-2224-2022-5(33)-8
11. Poteriaiko, S. P., Bielikova, K. H., Tverdokhlib, O. S., Orlova, N. S. (2022). Ekonomiko-matematychne modeliuvannia prohnoznoho otsiniuvannia diievosti funktsionuvannia yedynoi derzhavnoi systemy tsyvilnoho zakhystu. Fi-nansovo-Kredytna Diialnist: Problemy Teorii ta Praktyky, 1(42), 293–303. doi: 10.55643/fcaptp.1.42.2022.3676
12. Kodeks tsyvilnoho zakhystu Ukrainy [Code of Civil Protection of Ukraine], № 5403-VI (2012). Available at: https://zakon.rada.gov.ua/laws/show/5403-17#Text

Received by the editorial board: 10.03.2026

Accepted for publication: 13.04.2026

Date of publication (release): 30.05.2026