Lessons in crowd computing

A sytems approach can help prevent the frequent stampedes during religious festivals

Chaos: "Human movement is naturally turbulent." Picture shows crowds of devotees during the Kumbh Mela in Nashik.

On July 14 this year, at least 27 people were crushed to death on the banks of the river Godavari in Andhra Pradesh when a crowd surged towards the ghat, causing a stampede. This is yet another example of a tragedy that was waiting to happen, given the lax crowd management systems in India and our unwillingness and inability to learn from similar incidents in the past.

To understand the situation, we can use a systems approach: even a passing familiarity with networks, fluid mechanics and operations research (queuing principles and assembly line balancing) are enough to start with. Network and flow principles used by irrigation engineers to control floods and dam bursts, or by the police to manage traffic, or by sanitary engineers to control water/ sewage flow, or by electric engineers to prevent tripping and excess loads, or by IT specialists to ensure data flow are all equally applicable in crowd management and control. Of course, the caveat is that we are dealing with people.

To take an example, if in-flow exceeds out-flow over a period of time, the volume of the material inside the system will continue to rise till it exceeds the designed capacity of the confined space at some point. Two things will then occur: either the flow channel will collapse under pressure or the space will become unfit for survival. The result will be a disaster of the kind that happened at the Godavari Pushkaram. Careful planning, monitoring and regulation could have prevented the disaster, as evident from what was put in place after the disaster happened.

Dimensions of planning

The first step is to plan. Three dimensions need to be considered: the spatial, temporal and the event-related; or, ‘a confluence of place, time and event’. If any space, say, a cricket ground, is required to host a crowd, then it first needs to be studied. Next, if at any point of time a sudden surge of activity is anticipated, then additional planning is needed. Third, if there is a specific event happening in that space-time context, this will imply extra attention.

So, for instance, we know that a temple attracts crowds during festivals, and all these people aim to visit the

garbhagriha or sanctum sanctorum. Therefore, it is this space especially that needs to be carefully monitored. Second, during any event, festivities will reach a peak at some point: in marriages it is when the

muhurat starts, for example. Last, some events can precipitate uncontrollable crowd behaviour: for instance, distribution of

prasad , which can render crowds unruly.

Of course, sudden, unforeseen emergencies (like a fire) can also trigger unfortunate incidents. Even if such incidents can’t be predicted, they should always be anticipated and prepared for.

Administrators should understand all three dimensions of crowd management. Solutions include enlarging or creating additional spaces for the event, counters (ghats in this case), extending the time for the event, planning for large crowds and, finally, anticipating whether any particular event can trigger sudden, unusual movement. The confluence of event, space and time can then be either avoided by repeating the event at other locations, or by creating more regulated space.

The second aspect of crowd management is ‘flow regulation’. On normal days, one would expect crowds to be distributed randomly, peaking perhaps twice or thrice. However, during the confluence of event, space and time, there may be a sudden, sharp peak; even if arrivals follow the famous bell or normal distribution curve, the curve may have a narrow base and a sharp peak (sudden increase in arrivals). Therefore, flattening this curve by regulating the rate at which the crowd builds up can be achieved through controlling crowd flow at the entry and exit points. If, in a continuous flow process, monitoring the flow and keeping check on how it’s being exceeded is not possible, a system can be created whereby people can be allowed in batches of fixed numbers. Again, like in an assembly line, balancing is important: every stage of flow should roughly take an equal amount of time to negotiate, or we will have queues building up at some stages, which will become bottlenecks.

Third, it is easier to manage steady rather than turbulent flows, but human movements will naturally be turbulent. To ensure steady flows, apart from manned entry and exit points, it might be useful to make all movements unidirectional, segregating the flows in different directions. Signage, announcements and volunteers to assist should be available. Buffer zones to cater to additional flows are essential. Decongestion is possible only if movement regulation starts from the peripheries. There is no point imposing order at the location itself, since managing huge masses of people within a limited space is difficult; spreading out to the peripheries is a better option.

Monitoring the crowd

Last, even if planning is done well, systematic and meticulous monitoring is important. The feedback received through monitoring can help regulate the system. If, at some point, outflow doesn’t happen at the planned rate, the control room should, through some mechanism (say CCTV), check the reason and immediately act to decongest and restore the flow. Those who are part of the crowd (or a traffic jam) will never know where the bottleneck is; only someone outside the system, but monitoring it closely, will be able to recognise this and act immediately. Regulation will include controlling flows at entry and exit points, decongestion and removing bottlenecks along flow paths and resorting to transfers to temporary buffer spaces, if necessary.

All this may require additional resources. But it is better to spend more than lament later. It is also important to create awareness. Lack of information often confuses crowds: rumours and panic can lead to tragedies. The benefits of queuing up should be explained to children early. A contingency plan should be in place including evacuation and approach channels, first aid, ambulance, fire services, water supply and so on.

A systems approach where lessons from one field are applied to others is usually successful and productive. The sooner we learn to do this in the case of crowd management, the better. Only then can we call ourselves a developed country.

(T.K. Ramachandran is with the IAS. Views are personal.)