Crowd dynamics under emergency conditions: using non-human organisms in the development of a pedestrian crowd model
thesisposted on 15.05.2017, 04:21 authored by Shiwakoti, Nirajan
The rapid movement of large numbers of people is critical in emergency and/or panic situations, such as during the evacuation of buildings, stadiums, theatres, and public transport stations; outdoor events such as public assemblies, open concerts, and religious gatherings; and community evacuations following natural disasters or terrorist attacks. Perhaps the most critical reason for studying collective pedestrian dynamics under emergency/panic conditions is the lack of complementary data to develop and validate an explanatory model. That lack of data is likely to explain why very few models focus on panic situations. The bulk of the literature is restricted to the study of normal evacuation processes. Even the researchers responsible for developing the few existing models of crowd panic have identified the need for more rigorous modelling frameworks and the development of approaches to assess the reliability of model predictions. The broad aim of this dissertation is to use empirical data from non-human organisms in the development of a pedestrian traffic model under emergency conditions. Experiments undertaken with non-human organisms under panic conditions are a crucial component of the research reported here. Those experiments are found to be a promising and feasible means of circumventing the limitations posed by the scarcity of complementary human data under panic conditions. Argentine ants (Linepithema humile) were used as test organisms in the experiments reported here because they are abundant and simple to maintain in the laboratory. The experiments reported in this thesis reflect an original attempt to study the effects of structural features, that is, the layout of the escape area, on the collective movement patterns of non-human entities during rapid egress and to translate those results to the study of human panic. Large potential effects from the adjustments of small structural features of the escape area have been demonstrated via experiments with panicking Argentine ants. Insights from the experiments with panicking Argentine ants, along with previous studies on animal dynamics and pedestrian dynamics, have been used in the development of a simulation model called EmSim (short for Emergency Simulation). The formulation for the model recognises the role of both attractive and repulsive forces in maintaining the coherence of collective dynamics under panic conditions. To date, consideration of both repulsive and attractive forces has received limited attention in studies of crowd panic reported in the literature. Also the granular forces for pushing behaviour were modified to consider the case of discontinuity when the relative velocity is zero or near to zero. A first attempt has also been made to scale the model parameters for collective pedestrian traffic via ant traffic, based on a scaling concept commonly used in biology. With this innovative framework combining insights from biology and traffic engineering, there is scope to compare the collective movement patterns of non-human biological entities and pedestrians in order to devise sound strategies to aid evacuation. The proposed model also provides insight into the minimal interactions or physical mechanisms required for the emergence of collective dynamics. The nature of those underlying mechanisms was investigated through experiments with panicking ants. The proposed model is first calibrated and validated (with independent data) through simulation of panicking ant traffic as observed from the experiment and then scaled up for the human panic situation. Since data does not exist for direct measurement of model parameter values appropriate for panicking humans, the parameter values in the model were allometrically scaled up from the ant values to human values. The model predictions for collective pedestrian traffic were consistent with observations of collective traffic for ants. This consistency suggests that there are fundamental features of crowd behavior that transcend the biological idiosyncrasies of the organisms involved. The effectiveness of the proposed modelling framework is also validated through the comparison of the simulation results for the pedestrian traffic with the observed data from the experiment (under non-panic conditions). For normal (non-panic) conditions, the model was validated with experimental data on pedestrian traffic; specifically through comparisons of: •headway distributions in uni-directional traffic, •speed distributions and lane formation in bi-directional traffic, and, •outflow from bottlenecks of various widths. The results provide reassurance of the robustness of the model in explaining the collective dynamics of the panicking individuals despite the differences in speed, size and other biological details between ants and humans. The results also demonstrate the capability of the EmSim model to represent both non-panic and panic conditions within the same modelling framework. The model organism approach is commonplace in medical research but not in engineering, yet it is shown in this dissertation that it has enormous potential to provide insight and theoretical understanding of crowd panic. It will enhance understanding about what properties of panic are inherent to the physical nature of crowds, and what properties depend on idiosyncratic details. Also in biology, little attention has been given to the study of the effect of nest design elements on collective movements of social insects. The experiments that are reported here address those gaps in the study of alarm traffic in social insects by focussing attention to the relationship between nest architecture and internal traffic under alarm conditions. It is expected that the experimental studies and modelling framework presented in this dissertation will appeal to a broad audience, including researchers interested in social insects and nest architecture, self-organization, evacuation and traffic dynamics and engineering.