Evacuation planning using multiobjective evolutionary optimization approach

In an emergency situation, evacuation is conducted in order to displace people from a dangerous place to a safer place, and it usually needs to be done in a hurry. It is necessary to prepare evacuation plans in order to have a good response in an emergency situation. A central challenge in developing an evacuation plan is in determining the distribution of evacuees into the safe areas, that is, deciding where and from which road each evacuee should go. To achieve this aim, several objective functions should be brought into consideration and need to be satisfied simultaneously, though these objective functions may often conflict with each other.     

An extended model for describing pedestrian evacuation under the threat of artificial attack

An extended floor field model was proposed to investigate evacuation behaviors of pedestrians under the threat of artificial attack. In this model, pedestrian movement governed by the static and dynamic floor field, and the motion and assault of artificial attacker were involved simultaneously. Further, injuries with lower velocity and deaths of pedestrians caused by the attacker during evacuation were considered. And a new parameter $k_t$ was introduced. It is the sensitivity coefficient of attack threat floor field and could reflect quantitatively the extent of effect of attack threat on the decision-making of the individual. Moreover, effects of several key parameters such as the sensitivity coefficient, assault intensity and pedestrian density on evacuation dynamics were studied. Results show that pedestrian evacuation would display interesting phenomena transiting from rolling behavior to along-the-wall motion with aggravating extent of the impact of attackers on pedestrians, which refers $k_t$ in the model varying from 0.5 to 0.8. As assault intensity increases, more casualties would be caused and the available evacuation time would decrease, which means people have to flee the room in a shorter time period for survival. When the pedestrian density increases, more clogging at the exit would be generated and pedestrians would be more difficult to evacuate due to the limited capacity of egress and the reduction in the average speed of pedestrian flow caused by the injured. And the injured with limited motion capacity could hardly complete the evacuation owing to that they need more evacuation time and would retard the speed of the pedestrian flow.     

Macroscopic modeling and simulations of room evacuation

We analyze numerically two macroscopic models of crowd dynamics: the classical Hughes model and the second order model being an extension to pedestrian motion of the Payne–Whitham vehicular traffic model. The desired direction of motion is determined by solving an eikonal equation with density dependent running cost, which results in minimization of the travel time and avoidance of congested areas. We apply a mixed finite volume-finite element method to solve the problems and present error analysis for the eikonal solver, gradient computation and the second order model yielding a first order convergence. We show that Hughes’ model is incapable of reproducing complex crowd dynamics such as stop-and-go waves and clogging at bottlenecks. Finally, using the second order model, we study numerically the evacuation of pedestrians from a room through a narrow exit.     

Pattern-based evacuation planning for urban areas

The population of an urban area may be in danger due to disasters like floods, hurricanes, chemical or nuclear accidents. This requires decisions to protect the affected population. One decision may be to evacuate the affected area. For the exceptional case of an evacuation an approach to reorganize the traffic routing of the endangered area is developed. In this paper a two-stage heuristic solution approach for a pattern-based mixed integer dynamic network flow model is presented. The model restructures the traffic routing such that the evacuees leave the evacuation area as safe as possible and as early as possible within the considered time horizon.     

Microscopic information processing and communication in crowd dynamics

Due, perhaps, to the historical division of crowd dynamics research into psychological and engineering approaches, microscopic crowd models have tended toward modelling simple interchangeable particles with an emphasis on the simulation of physical factors. Despite the fact that people have complex (non-panic) behaviours in crowd disasters, important human factors in crowd dynamics such as information discovery and processing, changing goals and communication have not yet been well integrated at the microscopic level. We use our Microscopic Human Factors methodology to fuse a microscopic simulation of these human factors with a popular microscopic crowd model. By tightly integrating human factors with the existing model we can study the effects on the physical domain (movement, force and crowd safety) when human behaviour (information processing and communication) is introduced.In a large-room egress scenario with ample exits, information discovery and processing yields a crowd of non-interchangeable individuals who, despite close proximity, have different goals due to their different beliefs. This crowd heterogeneity leads to complex inter-particle interactions such as jamming transitions in open space; at high crowd energies, we found a freezing by heating effect (reminiscent of the disaster at Central Lenin Stadium in 1982) in which a barrier formation of naïve individuals trying to reach blocked exits prevented knowledgeable ones from exiting. Communication, when introduced, reduced this barrier formation, increasing both exit rates and crowd safety.     

Study of bottleneck effect at an emergency evacuation exit using cellular automata model, mean field approximation analysis, and game theory

An improved cellular automaton model for pedestrian dynamics was established, where both static floor field and collision effect derived from game theory were considered. Several model parameters were carefully determined by previous studies. Results obtained through model-based simulation and analytical approach (derived from mean field approximation) proved that outflow rate from an evacuation exit, which is usually estimated using outflow coefficient in building codes in Japan, can be improved by placing an appropriate obstacle in front of the exit. This can reduce collision probability at the exit by increasing collisions around the obstacles ahead of the exit.     

An efficient algorithm for the evacuation problem in a certain class of networks with uniform path-lengths

In this paper, we consider the evacuation problem in a network which consists of a directed graph with capacities and transit times on its arcs. This problem can be solved by the algorithm of Hoppe and Tardos [B. Hoppe, É. Tardos, The quickest transshipment problem, Math. Oper. Res. 25(1) (2000) 36–62] in polynomial time. However their running time is high-order polynomial, and hence is not practical in general. Thus, it is necessary to devise a faster algorithm for a tractable and practically useful subclass of this problem. In this paper, we consider a network with a sink s such that (i) for each vertex v≠s the sum of the transit times of arcs on any path from v to s takes the same value, and (ii) for each vertex v≠s the minimum v-s cut is determined by the arcs incident to s whose tails are reachable from v. This class of networks is a generalization of grid networks studied in the paper [N. Kamiyama, N. Katoh, A. Takizawa, An efficient algorithm for evacuation problem in dynamic network flows with uniform arc capacity, IEICE Trans. Infrom. Syst. E89-D (8) (2006) 2372–2379]. We propose an efficient algorithm for this network problem.     

Simulation of pedestrian evacuation strategies under limited visibility

Evacuation process under limited visibility is different from that under good visibility. To investigate the evacuation efficiency under limited visibility, different evacuation strategies including walking along the wall (S1), following the average movement direction (S2) and following the average position (S3) are proposed in this paper. Performances of these strategies under different visibilities, densities and exit widths are compared based on a revised social force model. Simulation results show that strategy S1 is more effective at low densities while strategy S2 and S3 are more efficient at high densities. It is noted that strategy S2 outperforms S3 under the same condition. In addition, strategy S1 is not sensitive to the change of exit width due to the movement mode of walking along the wall, while the following strategies (S2 and S3) have a better performance under the wide exit condition. The evacuation time for different proportions of pedestrians adopting one strategy in the mixed strategy situation is investigated and the optimal mixture proportion is also discussed. This study provides a new insight into the effect of different evacuation strategies on pedestrian evacuation, which is helpful for evacuees or organizers of public events to make an efficient evacuation plan under limited visibility.     

An optimization-based overtaking model for unidirectional pedestrian flow

We propose an optimization-based model for simulating the overtaking behaviour in the unidirectional pedestrian flow. A ‘visual area’ is introduced so that agents could receive the information regarding their surroundings and react by choosing one of three options: to move straight on, to dodge to the left, or to dodge to the right. And a side preference of each pedestrian for evading and overtaking is implemented based on traffic ‘social norms’. The model was validated by reproducing the experimentally obtained pedestrian flow patterns. The effects of the initial pedestrian formation on overtaking behaviour and the evacuation time have been analysed in different geometries. The results show that pedestrian flow patterns after overtaking are obviously influenced by both the initial positions and density of the slow pedestrians in the front. Phase changes of pedestrian formation are observed in both experiment and simulations. On the other hand, for sparse pedestrian crowds, the egress time of the fast individuals is mainly impacted by the horizontal distance between the initial positions of the slow pedestrians in the front, especially in the geometry with a bottleneck.