9 minute read

Unveiling the Dynamics of Crowds: Insights from the Latest Research

BY WILLIAM O’TOOLE, CFEE

The rapid growth of events and festivals worldwide is reflected in the rise of specialized companies, including those in security, crowd management, and research. During my time in Dubai, I witnessed an exponential increase in the number of supplier companies within five years. For example, Saudi Arabia transitioned from hosting only one event, the religious Hajj, to organizing billion-dollar festivals and concerts across the entire country. Alongside this trend, there has been a parallel growth in the fields of risk management and crowd management, both as services and sciences.

Recognizing the need and opportunity, universities applied for research funding in this high-risk field of crowd management. Input from event operations and research is crucial. As a result, Dr. Milad Haghani from the School of Civil and Environmental Engineering at the University of New South Wales convened the Crowd Safety Summit in May 2023, an international conference that brought together researchers and practitioners from the UK, Belgium, Germany, Australia, Japan, and France.

Figure 1 Volunteers tracked while exiting. The data is analyzed to determine the factors influencing the rate of flow and patterns of behavior (photo courtesy of Dr Haghani)

Attending the two-day conference provided me with a glimpse of the latest advancements in this field. Unfortunately, progress in risk management disciplines is often driven by disasters. Many presenters at the conference mentioned the recent crowd crush incidents in Korea and El Salvador. These tragic events underscore the need for data, information, and analysis. Fortunately, there is now an abundance of data available from around the world. It is now the responsibility of researchers to make sense of this data and contribute to operational planning and response for festivals and events.

The conference covered three main areas: simulation, experimentation, and experience. Essentially, university researchers utilized data collected through experiments to develop models of crowd dynamics. For instance, Dr. Paul Geoerg, a researcher with the Association of Fire Protection Engineering in Germany, conducted experiments where subjects were tracked via overhead cameras as they exited through doorways. The researcher then examined the impact of including wheelchairs in the crowd. Similarly, Dr. Haghani employed a tracking method to study how people make choices about which exit to use and how they adapt their choices while in motion. Figure 1 shows a typical photo of these experiments, with subjects moving towards various exits.

Of particular interest to me was Ramon Spaaj’s presentation on violence at sports events. He made a distinction between retrospective risk, which focuses on what we should have done, and pre-emptive risk, which concentrates on what we can do to prevent violence. Professor Spaaj, from Victoria University, described how some gangs use these sports events as a cover for their violent activities. Moreover, he highlighted that the presence of police and security personnel can actually contribute to an increase in violence. However, he concluded that certain segments of the crowd have the ability to deescalate problems before they escalate further. As a solution, he proposed exploring ways to develop relationships with these groups, rather than calling for more external security. I’m sure we have all witnessed situations where members of the crowd have managed to calm others and prevent incidents from rapidly growing into disasters.

The old models initially regarded a crowd as similar to a fluid or traffic flow. The next approach was to view a crowd as a collection of individual ‘particles’ that are influenced by forces such as social and internal drives. Crowd behavior is determined by the interaction and summation of these forces. In both models, common terms used were ‘herding,’ ‘irrational,’ and ‘panic.’ According to Dr. Haghani, the conclusion from this perspective was that the crowd needed to be controlled with security. The solution to any issues was to call in more security.

The new model regards the crowd as an ally in its management, raising the question: How can we enhance this aspect of crowd management? This model considers crowds as complex entities with emergent behaviors. This means that using extra force does not mean extra compliance. Such a model needs validation through experiments and real-life experience, necessitating conducting experiments with volunteers to create and refine these models. Based on the preliminary results of their work, it appears that the research is still in its early stages. However, I anticipate that it will gain momentum as more experience and data are gathered from events worldwide.

During the presentations, various aspects of crowds were illustrated with simulations, including pedestrian movement, avoidance maneuvers, and bidirectional flow. Dr. Alexandre Nicolas from the French National Centre for Research mentioned a new safety concern called ‘smartphone walking,’ where moving crowd members are distracted by looking at their phones.

However, not every presentation focused on modeling. Steve Laws, an Event Safety and Security consultant at Taylor Bridges, shared a familiar story about a 28-day event in Qatar. The event promoter, realizing that there weren’t enough attendees on the first day, announced free entry for people wearing the local football club shirt. This decision increased attendance but also led to an increase in drunken behavior. The promoter’s subsequent response was to restrict free entry to couples and females only. Consequently, large numbers of young men arrived wearing their football shirts and expecting free entry, and you can guess the result.

I would like to highlight some key points that caught my attention. Ben Crabb from the Cabinet Office Emergency Planning College in the UK emphasized the importance of risk resilience and effective communication. To illustrate his point, he described an incident of fake ticketing at a recent festival. In this case, the scanners would have detected the fake tickets, so the seller of the forged tickets cut the fiber optic cable connected to the event site. As a result, the scanners stopped working, allowing an additional 1000 people to enter the venue.

Most presenters referred to the Swiss Cheese Model, which emphasizes the concept of layering. According to this model, errors are bound to happen, so it is crucial to have multiple levels of protection in place for events. In event security, this approach is commonly seen, with measures such as searches, behavior analysis, stand-off areas, CCTV, security staff, police presence, and more. Further information about this can be found in the paper linked below.

Figure 2 Screen shot of instantaneous crowd analysis, detecting and analyzing density, flow and mood in six areas of the festival (courtesy of Dr Andrew Tatrai, Dynamic Crowd Measurement)

Regarding terrorist attacks, several presenters mentioned the UK government’s response to the bombing at the Ariana Grande concert in Manchester. This response, currently in draft form, is known as Martyn’s Law: Terrorism (Protection of Premises) Bill, which mandates changes in almost all venues. Additionally, the company Vardogyir presented their software on data-driven terrorism risk assessment.

Several practitioners noted the significant shift in crowd mood and psychology post-COVID compared to pre-COVID. They all expressed the attendees’ expectation for more engagement and a noticeably rowdier atmosphere at events.

Part of the conference theme, driven by the practitioners, emphasizes that planning is necessary but not sufficient. This notion is exemplified by the work of Dynamic Crowd Management. I had the opportunity to work on the mathematical analysis to develop the original model, and witnessing it in action was truly enlightening. The model was developed based on the experiences of crowd managers working on the ground. They shared their current project, which involves managing a free event attended by 1.3 million people over 20 days at a festival on Sydney harbor. Drawing from their expertise, they identified indicators that precede a crisis. By promptly detecting these indicators and relaying the solutions to the on-ground staff, crises can be prevented. Time is of the essence in such situations. They developed AI software that utilizes existing cameras to measure density, flow, and mood. These three variables serve as indicators for potential issues. Travis Semmens demonstrated this with an example from a previous event. The software detected a change in mood in a specific area, which went unnoticed by the on-ground security due to it being within the crowd. Thanks to the software, the security staff was able to intervene immediately and relocate parts of the crowd to alleviate the pressure. Figure 2 shows the screen at an event.

There was significantly more information and modeling discussed during the Summit than what I have described here. Topics such as crowd density measurement using hexagonal design, software developments like CrowdScan, Olympic pedestrian planning, and crowds at live screenings were all covered in fascinating detail. As the duty holder for your events, it is important to know that crowd safety is actively being researched. If you would like to receive more information, I can easily send it to you. Please feel free to email me at events@epms.net.

Dr. Haghani has graciously granted us access to a highly readable paper on crowd safety research titled “Contemporary challenges in crowd safety research and practice, and a roadmap for the future: The Swiss Cheese Model of Crowd Safety and the need for a Vision Zero target.” You can find the paper here.

To wrap up, let me share a little anecdote. Top architects who design new event venues take crowd movement very seriously. Last year, I was asked to conduct a risk assessment for a new venue that can accommodate over 14,000 people. The venue received a high rating on my comprehensive checklist, thanks to excellent planning regarding factors such as signage, lighting, queuing, width of aisles, and exits.

As the audience arrived and exited the venue, I observed them closely. On exiting, they had to pass through a narrow tunnel to reach the transportation area. This aspect was not included in the builder’s plans, as it was outside their so-called ‘red line’. This situation highlights the limitations of theory and emphasizes the significance of operational experience in integrating crowd management. The excellent flow rates of this venue plan efficiently fed the audience into a pinch point, posing a potential disaster.

William (Bill) O’Toole, CFEE is an events development specialist consultant who has authored several textbooks on events, risk, and crowds. His recent publications include “Crowd Management: Risk Security Health” (2020), “Safe and Healthy Crowded Places” (Government manual, available for free online), and “Risk Management for Events” (2022). He has trained event teams around the world in project and risk management. He is a member of the IFEA World Board and is currently researching the application of advanced risk tools listed in ISO31010 to the management of events and festivals.

This article is from: