DIM-ICE and RAMP: Still’s risk toolkit

Introduction

After a crowd disaster, the first story is usually about the crowd: they surged, they panicked, they behaved badly. Keith Still has spent a career analysing such incidents, and his conclusion is blunt. “After major incidents when facts are analysed the crowd is rarely the cause.”

If the crowd is rarely the cause, what is? Still’s answer is the places people are put in, what they are told, and how the event is run. This lesson teaches his two tools for asking those questions before an event rather than after it. Neither needs software. Both need you to think in specifics: this gate, this street, this time of night.

DIM-ICE: three influences, three phases, two modes

Still describes the model in his own words:

“The DIM-ICE risk model was developed from the analysis of accidents and incidents from around the world. Understanding the design (D) limitations (capacity, throughput), the information (social media, signage, PA announcements, news) elements and the management (M) systems (processes and procedures) are key to understanding how to influence crowd behaviour in places of public assembly.”

DIM-ICE is Still’s crowd-risk matrix: three influences (Design, Information, Management) crossed with three phases (Ingress, Circulation, Egress), assessed separately for normal and emergency modes. In his formulation: “Three primary phases of crowd behaviour – Ingress, Circulation and Egress / Three primary influences on crowd behaviour – Design, Information and Management / Two primary modes of crowd behaviour – Normal and Emergency.” The matrix is populated for each event and each cell is colour-coded high (red), medium (amber) or low (green).

That gives 3 × 3 × 2 = 18 cells, each a specific question. “Design, egress, emergency” asks: if we must empty this place in a hurry, is there physically enough exit capacity, even with one route lost? “Information, ingress, normal” asks: do people know when to arrive and which gate to use?

Why these three influences

The influences come from Still’s categorisation of disasters: “Design problems (too many people, not enough space), Information issues (sudden changes in the environment, poor or no information, confusing signage etc) or management problems (good design, good information, badly managed).” His 2020 paper with Papalexi, Fan and Bamford adds, citing his earlier work, that “the design element was the fundamental causality in over half of the incidents”. Too many people for the space is the commonest root of all.

Still also classifies crowds into five types: casual, cohesive, expressive, anti-social and incident. A cohesive crowd of one band’s fans, all wanting the front, behaves differently from a casual crowd of shoppers, and the matrix should be filled in with the actual crowd in mind. Module 3’s psychological crowds lesson gives the research behind that intuition.

RAMP: four questions

RAMP is Still’s four-question analysis of a site and its crowd: Routes, Areas, Movement over time, and People (or Profile). Still uses both “Profile” and “People” for the last letter. The questions, in his words:

1. Routes. “What direction will crowds approach and depart the area?”
2. Areas. “How much Area do you have? How will the area be used?”
3. Movement. “What crowd Movements do we expect? Over what period of time?”
4. People. “What type of People/Profile, are you expecting – what do you know of the crowd demographics?”

Still teaches RAMP and DIM-ICE as separate tools, in separate course modules and book chapters, and this course has found no published method for combining them formally. A practical order, and the one this lesson uses, is RAMP first, because it produces the facts (routes, areas, numbers, timings) that DIM-ICE then asks you to judge.

Four heuristics

Still pairs the tools with a few rules of thumb. Each is short enough to remember and specific enough to use.

• Plan for three crowds. “First assess the maximum capacity then prepare a plan for 50% of that number, 100% of that number and 500% of that number… Have a SEPARATE plan for each option… break down the problems into phases (ingress-circulation-egress).” The 500% plan is the one people skip: what happens if five times as many turn up as the site can hold?
• Reject what fails the rough cut. “If a system fails a first pass analysis (sometimes called a rough cut capacity analysis) then it should be rejected as being ‘not fit for purpose’.” Do the arithmetic before the detailed design, and do not let detailed design talk you out of a failed rough cut.
• Risk is dynamic. “risk is dynamic. Risks can change with time.” Still recommends red-amber-green risk maps for each time slot.
• Watch corners and changes of speed. “In any high density crowded environment the areas of greatest risk are where the crowd change direction (corners) and/or change speed.”

Behind all four sits Still’s diagnosis of why things go wrong: “Typically, the underlying cause is a lack of awareness of the problems related to crowd density and crowd flow.”

Worked example: a city-square concert

The event is fictional and the numbers are illustrative, chosen to exercise the tools. A free concert is planned in a city square, with 20,000 people expected. The square is surrounded by buildings, with four streets leading out of it; the north street leads to a tube station.

RAMP

• Routes. About half the audience is expected by tube, arriving down the north street; the rest on foot from the other three streets. After the show the same proportions reverse, so the north street and the station will carry the heaviest outflow.
• Areas. After the stage, sound desk, bars and toilets, about 9,000 m² of the square is usable for standing.
• Movement. Arrivals spread over three hours, peaking in the last hour before the headliner. Departures are concentrated into the quarter of an hour after the encore.
• People. Mostly young adults, many of them fans of the headliner (a cohesive crowd, wanting the front), with alcohol on sale and some families.

The rough cut

At the 2 p/m² that is typical for UK event planning, 9,000 m² holds 18,000 people, not 20,000. The plan fails its first pass. By Still’s rule, it is not fit for purpose as it stands: the organisers must either find more space (an adjoining street closed to traffic and fenced in as an extension) or cap the audience (ticketing a free event to 18,000). Only then does detailed design make sense.

The 50/100/500% rule adds three more plans: 9,000 people (a wet evening), 18,000 (the design case), and 90,000 (word spreads that the headliner is playing free). The last cannot be accommodated; its plan is about stopping people reaching the square, with closed tube exits, outer cordons and messaging.

DIM-ICE

The matrix below is pre-filled with one judgement of the design case. Change the ratings and notes as you think; they are saved in this browser, and “Export as text” copies the whole matrix. Use the Normal and Emergency tabs to switch mode.

Notice where the reds cluster: design and management at egress, and circulation at the stage front. That is a typical pattern for a concert, and it tells the planner where to spend effort. Notice too that the emergency column for design–egress needs a number: with the north street lost, can the remaining width clear 18,000 people in the time the council expects? The next lesson provides the method.

Risk is dynamic

A single matrix hides how risk moves around the site through the evening. Still’s red-amber-green map by time slot makes it visible. Figure 2 is an illustrative map for the concert: the gates peak during the build-up, the stage front during the headliner, the exits and station at egress. Each red cell is a place and a time to put stewards, cameras and decisions.