The Cost of the Omniscience Assumption: Quantifying Bias in Fire Evacuation Predictions Under Incomplete Information

Desheng Cao; Xin Liu; Jingjing Xia; Qingxin Xia

Authors

Desheng Cao School of Computer and Information Security, University of Emergency Management, Beijing, China Author https://orcid.org/0009-0004-9430-9996
Xin Liu School of Computer and Information Security, University of Emergency Management, Beijing, China Author https://orcid.org/0009-0008-8063-7429
Jingjing Xia School of Computer and Information Security, University of Emergency Management, Beijing, China Author https://orcid.org/0009-0006-8154-254X
Qingxin Xia School of Computer and Information Security, University of Emergency Management, Beijing, China Author https://orcid.org/0000-0002-3108-3326

Keywords:

Omniscience assumption, Occupant evacuation, Prediction bias, Multi-agent systems, Fire simulation

Abstract

Evacuation simulation is the main basis for determining safety margins in performance-based fire design, and its reliability depends on assumptions about the state of crowd information. Existing microscopic evacuation models generally imply an omniscient assumption: individuals know the locations of all exits and fire sources at any time, and information is received instantly. This contradicts behavioral evidence from real fires, where perception is limited, cognition is insufficient, and information is delayed. The prediction errors caused by this assumption haven't been systematically quantified yet. This paper breaks down the omniscience assumption into three orthogonal information channels: perception, cognition, and delay, represented respectively by the visual radius that diminishes with smoke density, the completeness of knowledge about exits and fire sources, and the update interval for decision information. On a multi-agent platform coupling fire, smoke, and humans, we designed gradient ablation experiments: first downgrading each channel separately to isolate their individual contributions, then combining downgrades to identify channel interactions. Deviations are measured across four aspects: evacuation time T95, number of casualties, safety margin distribution, and balance in exit usage. Experiments show that the cognitive channel is the most stable source of independent bias (dominating 16 out of 18 scenarios), while the perception and delay channels only appear under high-density conditions with few exits and smoke spread, and the channels work together only when there's a cognitive gap. This suggests that an all-knowing model is a pretty good approximation for normal real-world scenarios, and its cost is mainly in the extreme cases where information is completely missing, providing a quantitative basis for prioritizing prediction corrections and information system investments.

Downloads

Download data is not yet available.

References

Gwynne, S., Galea, E. R., Owen, M., Lawrence, P. J., & Filippidis, L. (1999). A review of the methodologies used in evacuation modelling. Fire and Materials, 23(6), 383-388. https://doi.org/10.1002/(SICI)1099-1018(199911/12)23:6<383::AID-FAM715>3.0.CO;2-2

Wang, Y., Kyriakidis, M., & Dang, V. N. (2021). Incorporating human factors in emergency evacuation: An overview of behavioral factors and models. International Journal of Disaster Risk Reduction, 60, 102254. https://doi.org/10.1016/j.ijdrr.2021.102254

Helbing, D., Farkas, I., & Vicsek, T. (2000). Simulating dynamical features of escape panic. Nature, 407(6803), 487-490. https://doi.org/10.1038/35035023

Blue, V. J., & Adler, J. L. (1999). Emergent fundamental pedestrian flows from cellular automata microsimulation. Transportation Research Record, 1644, 29-36. https://doi.org/10.3141/1644-04

Kirchner, A., & Schadschneider, A. (2002). Simulation of evacuation processes using a bionics-inspired cellular automaton model for pedestrian dynamics. Physica A: statistical mechanics and its applications, 312(1-2), 260-276. https://doi.org/10.1016/S0378-4371(02)00857-9

Burstedde, C., Klauck, K., Schadschneider, A., & Zittartz, J. (2001). Simulation of pedestrian dynamics using a two-dimensional cellular automaton. Physica A: Statistical Mechanics and its Applications, 295(3-4), 507-525. https://doi.org/10.1016/S0378-4371(01)00141-8

Hartzell, G. E., & Emmons, H. W. (1988). The fractional effective dose model for assessment of toxic hazards in fires. Journal of Fire Sciences, 6(5), 356-362. https://doi.org/10.1177/073490418800600504

Zhai, Y., Han, Z. L., Qu, L., & Han, C. (2025). Research on the coupled simulation of fire and pedestrian-vehicle evacuation in underground parking lot. Chinese Journal of Underground Space and Engineering, 21(6), 2217-2226. https://doi.org/10.20174/j.JUSE.2025.06.36

Marlow, F., Jacob, J., & Sagaut, P. (2022). A multidisciplinary model coupling lattice-Boltzmann-based CFD and a social force model for the simulation of pollutant dispersion in evacuation situations. Building and Environment, 205, 108212. https://doi.org/10.1016/j.buildenv.2021.108212

Mirahadi, F., McCabe, B., & Shahi, A. (2019). IFC-centric performance-based evaluation of building evacuations using fire dynamics simulation and agent-based modeling. Automation in Construction, 101, 1-16. https://doi.org/10.1016/j.autcon.2019.01.007

Xie, J. B., Chen, K. C., Kwan, T. H., & Yao, Q. H. (2021). Numerical simulation of the fire emergency evacuation for a metro platform accident. Simulation, 97(1), 19-32. https://doi.org/10.1177/0037549720961433

Cao, R. F., Lee, E. W. M., Xie, W., Gao, D. L., Chen, Q., Yuen, A. C. Y., Yeoh, G. H., & Yuen, R.-K.-K. (2023). Development of an agent-based indoor evacuation model for local fire risks analysis. Journal of Safety Science and Resilience, 4(1), 75-92. https://doi.org/10.1016/j.jnlssr.2022.09.006

McGrattan, K. B., Baum, H. R., Rehm, R. G., Hamins, A., Forney, G. P., Floyd, J. E., ... & Prasad, K. (2000). Fire dynamics simulator--Technical reference guide. Gaithersburg: National Institute of Standards and Technology, Building and Fire Research Laboratory. https://doi.org/10.6028/NIST.SP.1018e6

Liu, M. T., Zhu, W., Wang, Y. F., & Zheng, J. C. (2021). Modeling and simulation of exit selection behavior in pedestrian evacuation based on information perception and transmission. Sustainability, 13, 13194. https://doi.org/10.3390/su132313194

Wang, X., Chraibi, M., Chen, J., Li, R. Y., & Ma, J. (2022). Modeling boundedly rational route choice in crowd evacuation processes. Safety Science, 147, 105590. https://doi.org/10.1016/j.ssci.2021.105590

Yen, H. H., Lin, C. H., & Tsao, H. W. (2022). Novel smoke-aware individual evacuation and congestion-aware group evacuation algorithms in IoT-enabled multi-story multi-exit buildings. IEEE Access, 10, 119402-119418. https://doi.org/10.1109/ACCESS.2022.3221757

Ji, Y. P., Wang, W. S., Zheng, M. Y., & Chen, S. (2022). Real time building evacuation modeling with an improved cellular automata method and corresponding IoT system implementation. Buildings, 12(6), 718. https://doi.org/10.3390/buildings12060718

Mendoza, M. G., Waldburger, L., Lee, J., & Sastry, S. (2026). Hierarchical Generative Agents for Simulating Sequential Human Behavior. arXiv preprint arXiv:2606.14989. https://doi.org/10.48550/arXiv.2606.14989

Ronchi, E., Reneke, P. A., & Peacock, R. D. (2014). A method for the analysis of behavioural uncertainty in evacuation modelling. Fire Technology, 50(6), 1545-1571. https://doi.org/10.1007/s10694-013-0352-7

Jin, T. (1972). Visibility through fire smoke (III). Bulletin of Japan Association for Fire Science and Engineering, 22(1_2), 11-15. https://doi.org/10.11196/kasai.22.11

Isobe, M., Helbing, D., & Nagatani, T. (2004). Experiment, theory, and simulation of the evacuation of a room without visibility. Physical Review E, 69(6), 066132. https://doi.org/10.1103/PhysRevE.69.066132

Zhiming, F., Xingpeng, X., Lixue, J., Xiaolian, L., & Nan, H. (2021). Study on the exit-selecting behavior in underground indoor space with fire using a virtual experiment. Tunnelling and Underground Space Technology, 112, 103936. https://doi.org/10.1016/j.tust.2021.103936

Bonabeau, E. (2002). Agent-based modeling: Methods and techniques for simulating human systems. Proceedings of the National Academy of Sciences, 99(Suppl. 3), 7280-7287. https://doi.org/10.1073/pnas.082080899

Proulx, G. (1995). Evacuation time and movement in apartment buildings. Fire Safety Journal, 24(3), 229-246. https://doi.org/10.1016/0379-7112(95)00023-M