With growing city density and mass gatherings held all over the world in urban spaces, crowd disasters have been happening each year. In considering the avoidance of crowd disasters and the reduction of fatalities, it is important to analyze the efficient spatial layout of the public space in situations of high crowd density. Compared with traditional empirical design methods, computational approaches have better abilities for quantitative analysis and are gradually being adopted in the planning and management of the urban public space. In this paper, we investigated the official documents, publicly available videos, and materials of the Shanghai waterfront crowd disaster which happened on December 31, 2014. Based on the investigation, a detailed site survey was conducted and pedestrian flow data were acquired. To test the influence of different spatial layouts, an agent-based simulator is built, following the ASPFver4.0 (Agent Simulator of Pedestrian Flow) pedestrian walking rules. With the surveyed pedestrian flow data, the original spatial layout of the Shanghai Bund waterfront together with five other comparison scenarios are tested, including both space design and crowd management improvements. In the simulation results, the efficiencies of different space design and crowd management solutions are compared. The results show that even simple crowd control measures such as capacity reserve and more proper route planning will allow for a positive improvement in crowd safety. The results also compare the efficiency of different spatial operations and give general suggestions to the problems urban public space designers should consider in high-density environments.

Objective: The potential for deadly human stampedes to occur at any mass gathering event highlights this unique form of crowd disaster as deserving of special attention from both scientific and planning perspectives. Improved understanding of human stampedes is indispensable in the mitigation of this type of mass casualty. With relatively few peer-reviewed reports on deadly human stampedes, information from news reports and the Internet is essential to increased collective understanding. Without incorporating nontraditional sources, no other way to reasonably acquire sufficient data is available. This study analyzed human stampede events from 1980 to 2007 to identify epidemiological characteristics associated with increased mortality.

Methods: A LexisNexis search was followed by sequential searches of multiple Internet-based English-language news agencies. Date, country, geographical region, time of occurrence, type of event, location, mechanism, number of participants, number injured, and number of deaths were recorded. Bivariate analyses of number of deaths or injuries were conducted using a nonparametric Wilcoxon rank test. Multivariate regression was performed to determine the factors associated with increased number of fatalities during stampede events.

Results: A total of 215 human stampede events were reported from 1980 to 2007, resulting in 7069 deaths and at least 14,078 injuries from 213 events with available fatality information and 179 events with injury information. In bivariate analysis, stampedes occurring in the Middle East, in developing countries, outdoors, or associated with religious events had the highest median number of deaths. In multivariate analysis, events that occurred in developing countries and outdoors were associated with increased number of fatalities. Stampedes that occurred in the context of sports, religious, music, and political events, or that had a unidirectional mechanism, also increased the relative number of deaths.

Conclusions: Several epidemiological features of human stampedes associated with increased mortality are identified. Standardized collection of epidemiological data pertaining to human stampedes is strongly recommended, and further study of this recurrent, distinctive disaster is warranted. (Disaster Med Public Health Preparedness. 2009;3:217–223)