Crowd control: the engineering lessons learned from the Hillsborough disaster

Finally, after 27 years, this week’s verdict on the 1989 Hillsborough disaster – confirming that the 96 fans who lost their lives were unlawfully killed – provides something approaching closure for the families, friends and loved ones left behind.

Screen shot of the Greenwich University developed EXODUS crowd simulation tool

There are still some exceptionally tough times ahead. The expected legal proceedings against those responsible will dredge up more unbearable memories.

But those affected by the tragedy can perhaps take some small comfort from the fact that, although it’s taken nearly three decades to bring those accountable to justice, Hillsborough did at least trigger a major rethink in the design of stadiums which has reduced the chances of such a tragedy happening again. And for that, the engineering community perhaps deserves some credit.

In the wake of Hillsborough, following the Taylor report (both in 1989), the layout of football stadiums changed dramatically, most notably through the introduction of all-seater stadiums, which limit the number of supporters able to access a ground. “We have seen massive improvements in stadium safety over the past 50 years”, said Mike Robinson, chief executive of the British Safety Council, in response to this week’s events.

But much more significantly, Hillsborough was one of a number of incidents that triggered research into an entirely new field: the study of crowd flows.

One of the pioneers of this field is Greenwich University’s Professor Ed Galea, whose Fire Safety Engineering Group has played a major role in the development of simulation tools that are now used around the world to plan crowd control strategies and refine building designs.

The tragic events at Hillsborough were, Galea told The Engineer, something of a catalyst for the development of these approaches. “My research in this area started with the Manchester Boeing fire in 1985, then came Piper Alpha and then Hillsborough: it was one thing after another, and you could see how important it was to understand crowd dynamics. Incidents like this really kickstarted the research.”

Today, these kinds of tools are widely used. “You can’t design a crowded place now without using some sort of crowd modeling tool,” he said. “We don’t have to rely on so-called ‘magic numbers’; we can simulate events and use it for planning the normal management of crowds and also how to manage crowds when things go wrong.”

Although each situation is unique, simulation has taught us some valuable lessons, and highlighted some common risks: such as the hazards posed by what Galea terms “Unsegregated bi-directional flows”, i.e. crowds entering and leaving in the same space. “This is a bad thing to do,” he said, “and also you don’t want flows crossing it each other if you can avoid it.”

If you understand these risks, he said, you can develop strategies for moving people around a venue that prevents such situations from occurring in the first place.

Galea developed the first version of his so-called EXODUS software back in 1986. This was a relatively simple tool – not dissimilar to the CFD software used by engineers to model fluid flows.

Since then, thanks to advances in computing power and an improved understanding of crowds, its capabilities have grown considerably, and today’s version of the software is able to simulate the often unpredictable behaviour of individuals. This, said Galea, represents a major advance in the utility of crowd modelling. “You have to understand the behaviour of people because it’s the behaviour that’s driving where people go and what they do.”

The next step, he said, is to look at using the tools not just for planning, but for real-time input during an incident. This is a capability that’s already been demonstrated by Galea’s team through the EU-funded GETAWAY project. Through this project, the team developed an intelligent “way finding system”, that examined where people are in a building, where the hazard is, and calculates in real time the best way to evacuate them.

There’s little doubt, said Galea, that had such tools been around at the time of the Hillsborough disaster, things would have been very different.

“One of the things that the tools would have been able to do in advance of something like Hillsborough, would have been that you could use them to develop procedure for what happens if crowds are reaching a particular critical density of people. What if I close this gate? What if I opened that gate? These are things that could have been done in advance of Hillsborough. To develop a strategy rather than guessing.”

None of this will be of much comfort to those affected by the appalling events of 15th April 1989, but it does at least reduce the chances of such a disaster happening again, and demonstrates that while the long process of bringing those responsible to account has only just begun, some of the technical lessons, at least, were learned a long time ago.