Crowd Safety

#1. The Root Cause of Crowd Disasters

Íse Murphy

Dec 2, 2021 — 7 min read

Part one of Íse’s dissertation into the impact of regulation and legislation on crowd safety at outdoor events.

Crowd disasters occur when people are given insufficient information and space (Sime, 1999). It is rarely the case that only one factor causes a disaster (Helbing and Mukerji, 2012). In particular, increase or decrease in density is the difference between crowds moving from a ‘jamming’ to a ‘non-jamming’ state, accumulating over time resulting in a sudden shift of crowd dynamic state (Smith, 1995; Zheng et al, 2010). Research suggests that when density increases above 5 people per metre squared (ppm²) threshold, members of the crowd lose individual movement and behave as one entity (Fruin, 1993; Helbing and Mukerji, 2012). When density reaches this critical level, physical movement is almost impossible (Alkhadim et al, 2018) causing “crowd turbulence” (Helbing et al, 2007). The high pressures that develop within the crowd are powerful enough to bend steel barriers and push down brick walls (Lee et al, 2006). People are unable to control their bodies due to pressure, as each breath is exhaled and pressure prevents inhalation causing slow death by asphyxiation.

Historic Crowd Disasters

The UK in particular has experienced notable crowd disasters, mostly related to football stadia (Melrose et al, 2011). The table below outlines the key crowd disasters in recent history that have impacted legislation relating to crowd safety, noting that there have been no stadium disasters since 1989.

In contrast, the following table outlines notable reported crowd incidents in outdoor events. Note how incidents are still continuing to today and that there have been no public report or inquiry into any of the incidents.

Identifying Trends

Crowd scientists have compiled disaster databases over the last century in order to understand causality (Still, 2020; Asgary, 2018). I also compiled a database of crowd disasters spanning from 1988 to 2018 as summarised in the table below. Entertainment events, including music shows, festivals and cultural events account for 44% more incidents compared to sport events. This illustrates that potentially double the amount of crowd incidents happen in outdoor events than in stadia, yet in the UK there is specific legislation surrounding the safety of spectators in sports stadia but none for outdoor events. Furthermore, academic research in events tends to focus on economic and financial impact (Carlsen et al, 2010; Backman, 2018) rather than safety.

Predictable, Preventable

Identifying the cause of crowd disasters is crucial in understanding if a disaster was foreseeable and therefore preventable (Helbing et al, 2007; Still et al, 2020). In court, causality is often a key topic of debate when determining duty of care to an event organiser or venue owner, and breach of care to an injured plaintiff (Pearl, 2015). Possible issues could arise long before the event that directly or indirectly impacts the likelihood of the disaster occuring. Turner (1978) discusses an “incubation period” whereby a chain of events develop and accumulate unnoticed, possibly long before a disaster happens.

Global situations outside of an event organiser's control, such as the Coronavirus Pandemic, will have an impact on the safety of local events, and we are seeing the affect of it in the security and stewarding staffing and skills shortage crisis. This indirect issue has a direct impact on the safety of our events. Not only are we challenged by shortage of staff and their experience in dealing with crowd safety, but we are also faced with what appears to be an increase in deviant behaviour (Astroworld, France's Ligue 1, London O2, Wembley), which brings its own set of challenges to consider.

Identifying the reasons why disasters happen helps to understand if the disaster was foreseeable. The HSE (2004) classes causality into the following categories:

Immediate (proximate) causes: the agent of damage, injury or ill health (e.g., stadium stand collapses)
Underlying (distal) causes: unsafe acts and unsafe conditions (e.g. untested stadium design for safe crowd movement)
Root (distal) causes: the failure from which all failures grow (e.g. lack of management safety culture, prioritising finance over safety etc.)

Identifying proximate and distal causes helps to understand the sequence of actions or inactions that allowed the situation to develop and a disaster to occur (Still, 2000). As discussed in the last article, historically the crowd were blamed as the cause of the disaster, owing to “panic” (Drury and Stott, 2011). However when facts are analysed, the proximate cause of the disaster is usually owed to errors in space design and management (Fruin, 1993; Still et al, 2020). Proximate cause is then usually the subject of inquiry (Elliott and McGuinness, 2002), not the underlying distal cause "failure from which all failures grow".

The Proximate Cause

When designing event sites, the capacity of the space must be so that it can safely accommodate our crowds (and staff); allowing them to enter safely, move around the space safely and exit safely, in good time and with no restriction in the event of an emergency. As previously discussed, density over 5 ppm² changes the dynamic of the crowd to start moving as one entity, and so as planners we are looking to consider the validity of our site long before we welcome crowds into it. Indoor spaces are governed by building regulations with strict capacity calculating guidelines, but outdoor spaces do not have the same regulations.

The proximate cause can be associated to the critical point in a crowd whereby a catastrophic ‘jump’ occurs that instantly changes the crowd dynamic, resulting in a crowd disturbance. Models such as the Cusp Catastrophe Model offer a three dimensional visual representation on how a situation develops, reaches a critical point in which going past that point creates a sudden, irreversible change.

A cusp-catastrophe model of a crowd (Zheng et al, 2010)

In particular, increase or decrease in density is the difference between crowds moving from a ‘jamming’ to a ‘non-jamming’ state (Zheng et al, 2010) (Smith, 1995). In this state, the density is so that physical movement is almost impossible. The Hillsborough Disaster (Hillsborough Independent Panel, 2012) is an example of how an increase in density can cause catastrophic failure of crowd dynamics. Spectators had already filled central pens in the west stand and with high fences to prevent pitch invasions, there was no way out except via the entrance. A crowd built up at the Leppings Lane entrance and to relieve the pressure, management opened the gates, letting the crowd flow in. This crowd moved forward into the same central pens, which were already full. This action increased density, causing a crowd crush. Spectators could not escape forward onto the pitch because of the high fencing.

Another example is the crowd disaster at Love Parade, Duisburg in 2010 where overcrowding occurred on the only shared ingress and egress ramp into the site. The study into the disaster by Helbing and Mukerji (2012) identified as a result of the design of the space and lack of action by management on warning signs of a crowd disaster, density increased above a critical point and a crowd crush occurred.

These disasters are examples that when density increases above 5 ppm² the crowd begin to behave like a fluid mass (Fruin, 2002; Lee and Hughes, 2005; Pearl, 2015). This shift in crowd behaviour from individual independent movements to moving as one mass can be described as a catastrophic shift from one state to another on the cusp-catastrophe model. If density is regulated, then risk to the crowd is regulated (Still, 2014). However, if we consider the actions/inactions that took place long before the disaster occured, we may be able to erradicate the errors that over time accumulated to the catastrophic moment.

The Distal/Root Cause

Turner (1994) identified man-made disasters as “socio-technical” events; whereby in order for a system to work successfully, both the social and technical elements need to be focused on and optimised (Cherns, 1987). Using this model, Turner (1994) recognised the majority of disasters are caused by social, managerial or administrative errors. Challenger and Clegg (2011) identified through the Socio-Technical Model that there needs to be a number of failings for a disaster to occur. This model includes; goals, people, buildings/infrastructure, technology, culture and processes/procedures. They identified the distal cause of the Hillsborough Disaster to be connected to social, managerial or administrative errors (Challenger and Clegg, 2011) which was similarly highlighted by Lea et al (1998). Broadening disaster research outside of the scope of crowds, the distal cause appears to remain the same. For example, the cause of the Stardust Fire in Dublin 1981 was a failure of management to "adhere to legislation, regulation and guidelines" (Keane, 1982); the cause of the Piper Alpha Disaster in 1988 that killed 165 people, was attributed to “an accumulation of management errors” (Paté-Cornell, 1992); the cause of the Chernobyl disaster was deemed to be due to “poor safety culture” within the organisation (Pidgeon, 1997); and the cause of the Aberfan Disaster in 1966 was lack of action taken by those in responsible positions with knowledge of relevant factors (Couto, 1989).

Disaster and The Law

Understanding the causes of disasters helps us to learn how to prevent them so they hopefully never occur again. Investigating the history of crowd disasters highlights the parallel timeline of legislation and regulation change. With every disaster, there is often a corresponding change to legislation and regulation. The next chapter will delve into the evolution of legislation and regulation in the UK relating to crowd disasters and the impacrt it has on crowd safety.