By continuing to use the site you agree to our Privacy & Cookies policy

Your browser seems to have cookies disabled. For the best experience of this website, please enable cookies in your browser.


Your browser is no longer supported

For the best possible experience using our website we recommend you upgrade to a newer version or another browser.


Go with the flow

technical & practice A new method of analysing how people really move around buildings should take the guesswork out of public spaces

Architects have always considered themselves to be adept at knowing how people will move around their buildings. Ask them how they know, and they will either tell you it's because they have designed it in such a way as to 'encourage' particular flow patterns; or that they just know, okay.

After all, it is something many of them are supposed to have been trained in.

But, unfortunately, the experience of a lot of designed environments indicates that theory and practice do not always combine satisfactorily.

The invention of the specialist 'wayfinding' discipline was supposed to fill the gap, but these, too, are similarly vague about what they can really do to influence behaviour. In the end, many architects simply farm off a critical aspect of design to a wayfinding consultant armed with a psychology A-level and a signage contractor's catalogue. So how do people really move around the built environment, whether public spaces or private buildings? How can we identify aspects of design that conflict with the real-world experience of public movements to enable us to improve on designs and avoid mistakes?

Examples of flow problems that require a designer's attention range from the criss-crossing of weekday commuter traffic on London's Underground network, to the way visitors negotiate an exhibition. Very often along existing urban streets there are certain features that create bottlenecks disturbing urban flow - intentional or otherwise. At shopping centres emergency evacuation is sometimes not as smooth as planned because displays have been badly placed or because people react differently in distressing circumstances.At football grounds the turnstile is a weak point - a bottleneck and location for potential disasters as the crowd build-up increases. How well routing and pinch-points have been considered is something that can only really be assessed after the event.

But all too often finding fault-lines means waiting for a catastrophic failure and an enquiry into some accident or other.

Simulating hazard Legion, a relatively new company based in London's Waterloo, has now launched a program and an advisory assistance package so that more scientifically based predictive knowledge can be fed into the equation. Just before Christmas, I went down to its offices along the Embankment, overlooking the Houses of Parliament, to see what it had to offer.

In a quite unassuming office block I was seated among an array (almost a legion) of besuited execs and saw the dreaded slick Powerpoint presentation loom into view. However, as the meeting went on, I became pleasantly engaged in a discussion about development opportunities, applications and research methods, and realised that this package contained more than a fair share of common sense, insight and scientific rigour.What we used to call R&D. Given the fact that there is a lot of US expertise and money in this venture, I was also pleased to realise that this wasn't exactly a hard sell; the team seemed confident and yet relatively self-effacing about its provenance.

Getting ahead The new Legion version 1.7 software predicts crowd behaviour and quantifies individual experiences by creating a virtual model of an environment to test (or replicate) its design, effectiveness and operation in near-real situations. As Legion kept telling me, its programs are based on years of research into pedestrian behaviour and 'millions of detailed measurements of real people moving and interacting'.

Essentially, the company has used filmed footage of actual locations showing how people act and react in particular circumstances. Even though it says most movements are not really country-specific (crowds in the US move and flow, over time, in similar ways to crowds in Australia, for example), Legion has still taken the time and effort to create models from local and national data. The company has worked with Singapore's MTR and the Kowloon-Canton Railway, and its European clients include London's Crossrail, London Underground Limited (LUL) and French network SNCF.

It also provided pedestrian planning software and services to Sydney's and Athens'Olympics.

Legion uses client-operated CCTV camera footage from football stadia, shopping malls or public transport interchanges, or from its own cameras, to build up a base of filmed evidence of crowd behaviour. Each sequence is then analysed to develop a generic database of information points. Each head is dotted on the frame and a reference of his or her individual characteristics - sex, (assumed) age, etc - are logged relative to that dot. The laborious task of inputting these vast amounts of data results in an easy-to-read plan of people movements. The plan layout of the building is then overlaid and when the data video is run, the dots move as shown on the live-action video. (Obviously, care has to be taken when translating the CCTV video footage, which tends to photograph people at an oblique angle, and turning this into the plan, as the dots have to represent the central point of a person relative to the plan. However, Legion can do this with accuracy. ) Admittedly, this system simply gives a record of how people move about a real concrete environment. In order to assess how people would flow around the same environment if changes were made to the layout, it requires a predictive capacity. This is the key to Legion's knowledge base:

having analysed pedestrian behaviour from a wide variety of sources, and over many years, its detailed measurements have been built up into a believable predictive mapping tool.

And so the company can create virtual models of various environments to assess their design and operational effectiveness, in terms of the way people negotiate the spaces.

The key to Legion's success, and confidence, is that its predictive models have been tested in real environments - blind trials - and its predictions have been borne out by real flow patterns. Previously, movement models used in this fledgling industry have approached the question by drawing comparisons between the way humans move and the way water flows. But this has been shown to be an inadequate and misrepresentational view of human behaviour.

Get in line One of Legion's more interesting simulations shows a circular corridor (doughnut-shaped on plan), with dots (representing people) within it.

The simulation - based on expert evidence, appraisals of human psychological behaviour and evidence taken from real footage (albeit of different circumstances) - shows two random sets of people (the dots have been coloured for clarity) setting off, each walking in the opposite direction.

At first, everyone is jumbled up, struggling to walk against their immediate neighbour; the dots are frenetic, and slow, as they try to fight to keep going forward against a tide of dots coming the other way. However, as the simulation runs, a pattern begins to form, as dots fall in behind other dots - shadowing the dot in front and minimising the likelihood of bumping into dots coming in the opposite direction.

Before long, the chaotic patterns of dots scrabbling to keep going forward regardless of anyone else turns into a pattern of self-disciplined follow-my-leader.Within a few minutes, the simulation shows at least two rows of dots in an orderly 'queue', forming a column of dots walking round in distinct patterns and in distinct locations - some on the outer circumference and some on the inner or in the centre of the circular corridor. Notably, hardly anyone bumps into each other and people move around much more quickly than before.

This is a simple, but abstract, example to test out the predictive power of the Legion system; it is not borne out in direct practical evidence, since the conditions of a circular corridor don't really exist.

However, Legion prides itself on 'understanding how people make choices'. The flow of pedestrians on real public concourses - such as LUL's Upton Park tube station (see box) - shows the merit of replicating and analysing real footage of how people move around so that predictive flows can be inserted to improve conditions.

Congestion charge Given that congestion and constriction of the pedestrian flow is understood to be one of the constraints on the efficient workings of the Underground system, this means that instead of having to design and build real renovations, alterations or additions at a particularly congested hotspot to see if they will work, architects can now mess around with plans to find out, before the event, how people are likely to move around it.

Theoretically, Legion's software could even be a small business tool, advising concession-stand owners on railway platforms or in shopping malls where best to place themselves to maximise the flow past their display of wares.

The program ensures that design time, construction effort and management costs can be kept to a minimum. In an ideal world, clients and architects need not opt for a design solution until the simulation identifies the optimum practical design. The information can be printed in a flow diagram showing flaring pressure points or as a bar chart.

Former deputy director of the Health & Safety Executive David Eves describes Legion's pedestrian simulation software as 'a better solution to managing crowd safety' when compared with the various standards and models of computing pedestrian flows in transport buildings and other large-scale public environments. Legion's chief executive, Martin Band, certainly believes it has come up with a winner. 'By applying the right simulation system before implementation, informed decisions can be taken to protect people and maximise returns on assets, ' he says.

My prediction is that computer flow programs such as this, based on thorough and developing R&D, will improve our understanding of the way buildings don't always work. By helping us learn from our mistakes, Legion's software is a very useful addition to an architect's toolbox.

Have your say

You must sign in to make a comment.

The searchable digital buildings archive with drawings from more than 1,500 projects

AJ newsletters