Using Analytics to Safely Manage Large-Scale Crowds


The Hajj – the great Islamic pilgrimage to Makkah in Saudi Arabia – is known to be the largest annually occurring pedestrian problem in the world. Each year up to 4 million pilgrims approach the holy sites at the region of Makkah to perform their religious duty. The Hajj comprises several rituals on consecutive days at different sites. The Jamarat Bridge, where the stoning-of-the-devil ritual is performed, is one of the focal sites. Over 4 days the pilgrims throw pebbles against the so-called Jamarah pillars (representing the devil) which are embedded in the Jamarat Bridge. This ritual is known to be particularly crowded. Until 2006, several sad crowd disasters (stampedes) with thousands of casualties occurred due to overcrowding of the scarce infrastructures. In the aftermath of the crowd disaster in 2006, the Ministry of Municipal and Rural Affairs of the Kingdom of Saudi Arabia (MOMRA) launched projects of total $7 billion to prevent crowd disasters in the future.

In particular, MOMRA started to develop an effective crowd management system that employs a wide range of tools from operations research, analytics, and crowd dynamics. At the heart of this trailblazing decision support system a Pilgrim Scheduler is implemented that assigns stoning times to the pilgrims. Further, pilgrims are assigned to paths that lead from their accommodation to the Jamarat complex in a strict one way manner. Hence, counter or intersecting flows are avoided.  The Pilgrim Scheduler is a mixed integer program that minimizes the deviation of the assigned stoning times from the preferred stoning times. A set of constraints ensure smooth utilization of infrastructures such as roads, tunnels, and the 5 story Jamarat Bridge. Due to the problem size (more than 3 million variables) a multi-stage fix-and-optimize heuristic has been developed yielding high quality solutions in minutes. Additionally to the Pilgrim Scheduler MOMRA employs sophisticated operations research tools to assign pilgrims to stations and to trips of the AlMashaaer Makkah Metro that moves pilgrims between the holy sites, and for layout planning of the tent city that accommodates the pilgrims. Further essential elements of the decision support system are tools for simulation and controlling: state-of-the-art analytical models of pedestrian dynamics simulate the flow of pilgrims through infrastructures (metro stations, for example) and the street network. MOMRA uses latest radio-frequency identification and video technologies to control pilgrim flows. Together, simulation and controlling systems allow for real time predictions of high crowd densities. The technology is thus a base of an effective alert-and-response strategy for the operation.

The decision support system provides solutions to MOMRA and operation agencies that enable uncongested and smooth flows of the pilgrims as well as maximum satisfaction of the pilgrims concerning their scheduled stoning time, i.e. the time to perform the religious ritual. Indeed, operations research helped to reduce operation and maintenance costs. For example, the smoothed and well balanced flows enabled a dramatically reduction in security personnel (saving more than 20% personnel cost). But most importantly: operations research significantly contributed to saving lives. Since 2007 no crowd disaster has happened. Lessons learned from Hajj crowd management and the integration of operations research can help to manage and operate mass gatherings all over the world.