Professor Dr. G. Keith Still FIMA 9 thDecember 2011
Contents Legal Outlines - page 3
The Brief - page 3 Overview - page 4
Expert Witness - Background and Experience - page 8
Crowd Dynamics and Event Planning - page 9 Crowd Dynamics - page 9
The Event Plan - page 10 Event planning - page 10 The Approval Process - page 10 Crowd Density and Crowd Safety - page 11 Modelling and Mathematics of Crowds - page 12 Crowd Behaviour (phases and influences) - page 12
Site Plans - page 13 CAD Plans - page 14 Modelling the Site - page1 5
Event Organizers Numerical Data - page 15 Interpretation - page 16 Conclusions - page 19 References - page 20
1. Legal Outlines 1.1. The legal outlines have been agreed by all parties on receipt of the Public Prosecutors definition of terms.
2. The Brief
2.1. The brief was to review the information provided (translated documents and video footage) with specific focus on the crowd flows and behaviour of the crowds into and around the area of the accident.
2.2. The brief included a review the crowd density (pressure and risk) around the area of the incident.
2.3. The brief also included a request to define the crowd behaviour, the reasons why the crowds head towards the only visible means of egress (the stairs).
2.4. The crowd dynamics for events are defined in three primary phases.
Ingress (inflow, Gesamt Zu, entering the main event space),
Circulation (moving around the site) and:
Egress (out flow, Gesamt Ab, leaving the eventspace),
2.5 Typically an event space requires consideration of the overall capacity of the site and the capacity of the ingress/egress system.
2.6. Ingress/Egress movement and flow rates (from the data provided) on the ramp and tunnels were examined and calculated and it can be demonstrated that these would have failed when the ingress and egress crowds merged.
2.7. Reasonable care needs to be taken to ensure both a free flowing system and sufficient monitoring and crowd flow management to prevent rapid high density build up and crowd crushing.
2.8. The Duisburg site used the ramp and tunnels as both the ingress and the egress system with considerable utilization expectedbeyond the upper limit of safe capacity on the ramp.
2.9. Within this report it will be demonstrated that the utilization was beyond the safe capacity on the ramp.
3. Overview
3.1. The event capacity is based on number of people per square metre.
3.2. This is typically between 2-3 people per square metre, based on Fruin's level of service for crowd movements and event capacity.
3.3. If crowd density reaches 6-7 people (touching) per square metre there is significant risk of shock waves, crushing and personal injury.
3.4. High density in areas of crowd movement; such as the entry and exit systems, are historically areas where accident occur (see references on international accidents and incidents - summary table below).
[image, table+content cannot be scanned]
3.5. The vast majority of international incidents are a consequence of too many people in too little space.
3.6. There are some simple calculations that can be applied to measurements from maps, plans and diagrams, assess both the crowd capacity (number of people per square metre) and flow rates (time to reach capacity).
3.7. For over 30 years Fruin's Levels of Service has been used as a benchmark upon which crowd calculations have been developed. This has been recognized as best/safest practice within the industry.
3.8. For crowd flow/movement the maximum crowd flow corresponds 82 people per metre per minute at 2 people per square metre.
3.9. To illustrate crowd density, and the relationship to crowd risk, we use the plan View of the average individual in a 4 square metre grid.
[image, persons per square meters, cannot be scanned]
3.10. Between 2-4 people per square metre there is space for crowd movement.
3.11. Above 5 people per square metre the crowd can become congested (crowd flow slows down considerably) and there is risk of slips/trips/falls and subsequent personal injury.
3.12. The density in the area of the incident exceeds the 6 people per square metre as the crowds can be seen to be touching (packed) on all sides. This level of packing is extremely dangerous and the crowd would be trying to escape.
3.13. However, at eye/ground level. the crowd can only see the stairs as a means of escape.
3.14. People heading towards the stairs will increase the pressure at the base of the stairs and along the approach to the stairs.
3.15. The crowd cannot see the problem a few metres in front of their position. The people are unaware of the extreme pressure ahead.
[image, photo of crowd, cannot be scanned]
3.16. Above -- image from the arch of the incident (Camera" 13). This is estimeted at 8-10 people per square metre where the crowd is trying to escape
[image, photo of crowd, cannot be scanned]
3.17. At ground level the crowds cannot see beyond a few metres ahead of their position.
3.18. The stairs and light towers are on a higher sightline and appear as means of escape.
3.19. The noise, confusion and congestion are all factors which lead the crowds to seek escape from their immediate area.
[image, photo of crowd, cannot be scanned]
3.20, This is due a perception of risk caused by a lack of adequate crowd management and control. The situation was allowed to reach dangerous levels of congestion and the crowd reacts to the threat of excessive crowd pressure and tries to escape.
[image, photo of crowd, cannot be scanned]
[credentials] Background and experience of Prof dr. G. Keith Still
7. The Event Plan
7.1. It is difficult to envisage a safe, robust crowd management plan without some form of crowd modelling, including site plans, operational plans, risk assessment for normal and emergency contingencies.
7.2. Crowd modelling does not require complex computer simulation, it's a process of defining how the crowds arrive, over time, where queuing may occur and how to manage the crowds under both normal and emergency situations.
7.3. Any areas where high density, moving crowds should be assessed for risks.
7.4. Management, communications (PA, signage, maps, plans and diagrams to assist the crowd in way finding) need specific attention.
7.5. This is generally defined as. the "Event Plan" and/or "Safety Concept" and normally includes a risk assessment and analysis and a risk register (who is responsible for each category of risk).
7.6. Equally it is important to employ some form of crowd monitoring (monitoring the arrival flow rates, how the queues build up, areas of high crowd density and different types of crowd behaviour, both normal and emergency).
7.7 In the Duisburg case there are references crowd monitoring but it is based on capacity on the site (percentage of full) and there are differences between observers.
7.8. Effective crowd modelling and crowd monitoring are both essential to the development of a safe, robust crowd management plan.
8. Event Planning
8.1. Event planning is an essential element for event safety.
8.2. Typically event plans need to include contingencies plans for emergency situations, risk registers and operational parameters - such as command and control (chains of command - who does what? Who takes key decisions, how are these transmitted/acknowledged?).
9. The Approval Process
9.1. Once an event plan is submitted to the local authority it is then checked for compliance with local regulations.
9.2. For the Duisburg Love Parade event independent experts were commissioned to check the event plan and the safety concept and a simulation was commissioned for means of escape.
9.3. The event plan and safety concept were approved by the relevant authorities.
9.4. We have no records of any official objections to the "event plan" and "safety concept" during any of the desktop exercises, during the overall approval process or from the experts reviewing the plans.
9.5. There are no specific details relating to the safety of the ramp, tunnels or the combined flows into/out of the event site.
9.6. A simple mathematical process calculating the area of the ramp and identifying both the ingress and egress flows would have highlighted that the ramp area could not cope with the combined flows at the peak period.
10. Crowd Density and Crowd Safety . 10.1. Although industry standards vary it is generally acknowledged that 2-3 people per square metre (illustrated below) can be safely accommodated.
[image, persons per square meters, cannot be scanned]
10.2. The event plan defines the crowded spaces as 2.5 and 3 people per square metre placing this in the correct region for crowd safety. However this should have applied to ALL parts of the system (specifically the ingress/egress ramp and tunnels).
10.3. If crowd density exceeds 4-5 people per square metre there is a risk of rapid crowd congestion in areas where high volume crowd flow is expected and a subsequent high risk of trips/slips and falls and subsequent personal injury.
10.4. In the evidence presented there were no specific references to monitoring and regulating the areas of high volume (throughput) in the system.
10.5. Failing to control crowd flow in areas of high throughput has been the primary cause of historical accidents.
11. Modelling and Mathematics of Crowds 11.1. Crowd flow and crowd density are related as the following graph shows.
[image, graph]
11.2. This.is a fundamental relationship between crowd flow and crowd safety.
11.3. High density crowds have a high risk due the potential of tripping, slipping or falling and care must be taken to mitigate this risk. In other words, avoid any trip hazards in areas of high crowd flow.
11.4. It is noted the gully cover in the area of the incident was broken and covered by a section of Heras fencing. This is a significant trip hazard.
11.5. In a high density crowd, if someone trips, slips or falls the pressure of crowd can easily cause serious injury to the masses due to rapid pressure build up.
11.6. Historically ingress and egress systems experience high density crowd flow and these are the areas that require specific focus for crowd management.
12. Crowd Behaviour (phases and influences)
12.1. Events have three primary phases of crowd movement (ingress, circulation and egress).
12.2. During ingress the crowd has only one primary objective - to enter the event space. This is dictated by many factors including opening times, entry accessibility and management systems.
12.3. Crowd safety is the responsibility of all participating authorities involved in managing the event. Individual roles and responsibilities should be defined in the event plan.
12.4. In any environment there are three primary influences on crowd behaviour; Design (barriers, gates, entry points), Information (posters, intemet, loudspeakers/PA) and management (operational functions such as search process and procedures).
12.5. There are also two primary modes of crowd behaviour - normal and emergency. Typically event plans have both normal and emergency contingency plans in place so that, in an emergency, the operational management teams are prepared.
12.6. There are many ways of defining the event plan - the table below is used at the UK Cabinet Office Emergency Planning College to teach the relationships between ingress-circulation- egress, design-information-management under normal and emergency situations.
[image, graph]
12.7. For the purposes of crowd analysis the event plan would typically be based on site maps, plans and diagrams.
13. Site Plans
13.1. To understand the crowd build up at the area of the fatalities we need to explore the planning information available before the event.
13.2. In order to support the findings we have examined CAD plans from the site and we also walked the site. It is clear from these observations that not all the information found on the site is present on the maps/plans and diagrams (such as site condition, trip hazards, sight lines).
14. CAD Plans 14.1. From the CAD plan - grRampe masse O2.dwg
[map, plan]
14.2. The minimum width in the CAD is the area circled (11.38m + 6.9m = 18.28m).
14,3. Using the calculation (Fruin) of 82 people per metre per minute we have an UPPER limit of 18.28m * 82 people per metre per minute * 60 minutes = 89,790 people per hour.
14.4. However on the day the narrowest point' on the system were the barriers (not shown on the plan) indicated in the photograph below.
14.5. A broken gully, covered by a section of Heras fence is highlighted below (bottom left) and in an area where it would pose a significant trip hazard.
[Photo shows: narrowest point messured 10.59 meters + broken gully with Heras fence]
14.6. The operating capacity of the narrowest point in the system is 82 people per metre per minute * 10.59m * 60 minutes = 52,103 people per hour.
15. Modelling the Site
[Photo, ramp from above ]
15.1. Crowds would converge from the West and East and combined ingress and egress flows would combine in the tunnels and ramps
15.2. These parts of the system would experience high volumes of pedestrian movements.
16. Event Organizers Numerical Data 1
16.1. Numerical (Source - Besucherprognose Loveparade 2010 - Duisburg -- folder FVorauswertung SpohrEx1021_1 l_DO66__2_ehd2, Pl GSP08_Produktionsleitungx - e Stephan from Public Prosecutor - slide 15)
16.2. From Published Data (source -- file from Public Prosecutor -- Besucherprognose Loveparade 2010 -- Duisburg - folder FVorauswertung SpohrExl02l_1l__DO66_2_ehd2, PlGSP08_Produktionsleitungx - Stephan from Public Prosecutor - slide 16. This shows site occupancy (sum of flow in minus flow out).
[graph]
17. Interpretation
17.1. To clarify this graph I have shown the data as ingress flow (Zu = Green) and egress flow (Ab = Red)
[graph]
17.2. The above data and graphs clearly shows the capacity of the event (blue line) does not exceed the defined capacity (less than 250,000 at any given time).
17.3. However it does not show the flow rates through the tunnels/ramps. Which is an important area to calculate. Ingress and egress numbers should be calculated for all areas, throughout the event, during the planning of events that have high crowd attendance.
17.4. Using the same data I have shown the ingress (Zu = Green) and egress (Ab = Red) in the graph below. Note the vertical scale is the same as the previous graph.
[graph]
17.5. The blue line shows the expected throughput in the tunnel/ramp combination
[graph]
[graph]
17.6. The demand (number of people flowing down the ramp and through the tunnels is the sum (addition) of the red and green values above. Flow in PLUS flow out.
[graph] [High Risk]
17.7. This shows an expected flow THROUGH the tunnel/ramp combination of 145,000 people during the period 17:00 - 18:00 (as the parade finishes and the evening crowd arrrives).
17.8. The scale has been adjusted in the graph below.
[graph]
17.9. The tunnels (from the Event Organizers own data) are expected to handle a combined in and out flow of 145,000 during the peak period.
17.10. The crowd flow is typically (HCM/Fruin) estimated (for design purposes) as 82 people per metre per minute.
17.11. This is typically an applied calculation to establish the minimum width (from plans).
17.12. Therefore the minimum required width is 145,000 / (82 * 60 minutes) = 29.5 metres.
17.13. From the CAD plans and site survey the dimensions are less than the minimum required width for the expected two way crowd flow.
17.14. Using the calculation (Fruin) of 82 people per metre per minute we have an UPPER limit of 18.28m * 82 people per metre per minute * 60 minutes = 89,790 people per hour. This is a one-way flow calculation.
17.15. On the day of the event the smallest part of the system was 10.59m
17.16. The operating capacity of the narrowest point in the system is 82 people per metre per minute *10.59m * 60 minutes = 52,103 people per hour.
17.17. The conclusion is that the minimum width section of the system (CAD plan - grRampe Masse 02.dwg) would fail a basic mathematical analysis for safe passage of crowds during the peak period.
17.18. Failure of the system was predictable.
17.19. The combined flow would have exceeded the minimum width and crowd congestion would have built up to dangerous (high risk) levels at a predictable time (end of Parade).
18. Conclusions
18.1. From this report it is clear from the evidence presented that there are questions regarding the following area:
18.1. Crowding: from the evidence presented clearly the tunnel and ramp system was not able to cope with the number of attendees arriving and leaving the event.
18.2. This has, in turn, a relationship to the management, validation and monitoring of the event and who is responsible and who has primacy for managing and controlling the event itself.
References
signed, signature
The information given in this report is an expert witness opinion based on the evidence presented by the Office of the Public Prosecutor in Duisburg. Some of the documentary evidence was translated into English by the Office of the Public Prosecutor. Other documents were translated by Sabine Funk on express wish of the Office of the Public Prosecutor. The University reviewed the report to ensure that firstly that it answered only the questions posed by the Office of the Public Prosecutor and also whether any comments in the report had potential implications for the reputation of the International centre for Crowd Management & Security Studies at the University and the University itself. I can confirm that no content, structure, theme or changes to the conclusions were made during my review. This is a true record of the report for the Duisburg Public Prosecutor compiled and presented by Professor Doctor G. Keith Still on the 14th December 2012.
(sic, er had gestaan moeten hebben: 14th December 2011)