A novel framework for recording and evaluating incidents at large outdoor music events


White Paper


Accurate data on incident type, frequency and staff response at large scale outdoor music events are lacking. This is partly due to unreliable data and partly due to the confidentiality of such data not made available to the general public.  This study’s purpose is to present a novel framework for recording and evaluating the potential damage incidents are likely to cause at large scale outdoor music events.


Accidents during popular events cause physical and psychological suffering to the victims involved. In addition, bad publicity for the event’s organizers can lead to loss of revenue, compensation payments, high insurance costs and possible prosecution. All of the above can have negative long-term effects on individuals and businesses.  The risk of individual and professional damage highlights the ever so need for a comprehensive event risk assessment and emergency management plan.


Risk Assessments and Emergency Management Plans

Risk assessments and emergency management plans describe response procedures in the unlikely scenario of an incident occurring.  Prior to staging an outdoor music event a risk assessment is a pre-requisite and even a legal requirement in countries such as the United Kingdom [Management of Health and Safety at Work Regulations, 1999] and Australia [Environmental Health Service, 1992].

The individual or team conducting the risk assessment must:

  1. Identify all potential incidents however minor or major including personal injury, illegal activity, natural disasters etc.
  2. Analyse the impact of such incidents i.e. what are the consequences?
  3. Evaluate the riskiness of identified incidents e. determine to what extent an incident is likely to occur
  4. Develop plans, precautions and actions for minimizing or preventing potential damage that can be caused by the identified incidents.

An emergency is an unplanned event that can cause significant injuries or deaths; disrupt operations; cause environmental damage; threaten the reputation of the event; decrease the revenue from the event [Government of Western Australia, 2009].  An emergency management plan outlines the process to reduce the negative impact of emergencies. It describes the coordination of responses with partner agencies and local authorities to be executed in an emergency.

 An emergency management plan involves:

  1. Emergency and disaster identification
  2. Emergency and disaster planning and preparedness
  3. Emergency and disaster response
  4. Evacuation plan

Both risk assessments and emergency plans are confidential, not to mention costly to produce. Specialised firms are employed by event managers to conduct the risk assessment and the emergency plan for an event. These documents are very difficult to acquire, even for research purposes.   Even more difficult to acquire, due to insurance purposes, are post concert incident reports.


A Proposed Framework for Recording and Evaluating incidents.

Incident related data are not widely available, however the need for a standardised industry-wide framework for recording and evaluating incidents is necessary, since such information will enhance our understanding of potential safety threats. Furthermore, a standardised framework provides accurate incident related data which will serve as a useful reference point for safety and security systems that aim to avoid damages and at worst casualties at large scale music events.

The authors propose a three-level framework. The first level would allow a user to design an event profile and determine all applicable incidents from an existing standardised database. As no database can be exhaustive because of the wide variety of events (i.e. type of crowd, participants age/gender, music type etc) users will also have the possibility to add their own incident options. The second level of the proposed framework would filter the data collected from level 1 based on various formulation and normalization processes. Measures such as incident health impact, event reputation impact, individual staff involved, incident response, likelihood of occurrence would be calculated. Finally, the third level would utilize the measured values generated in level 2, compute metrics such as incident type occurrence, frequency of all incidents, incident impact metric and categorise incidents in a quantitave manner.

The proposed framework’s architecture is depicted in the following Figure.


  1. Event Profile generation and Incident’s logging process: Qualified users start by creating an event profile. First they select a music event type of their choice (i.e. rock, country and folk, jazz, rap, classical, children and family, other). Next they decide the event’s general features such as country, number of spectators, average age of audience, venue type. Finally they specify other attributes such as weather and the staff hired for the event safety and security. Once the event profile is created users select from an existing incident database those applicable to their event. The user’s selected incidents are then recorded. User data are recorded in an online open access database. The data recording communicates with the database with a use of an XML based schema. This is possible either in real time or in post processing mode.
  1. Filtering of user data: All data selected by users is filtered and processed based on predefined measures. Measures are quantified based on a nominal quantitave scale i.e. likelihood of occurrence (0, 1, 2…n).
  1. Clustering: The clustering process includes metrics computation based on the selected incidents and their respective measures and incidents categorization and ranking. Metrics are non nominal values that quantify physical features and attributes i.e. frequency. Metrics are also defined as a combination of specific measures in the non nominal space. The purpose of clustering user data is firstly to determine incident categories with quantitave accuracy, secondly to include relevant incidents in the right categories (incidents - categories correlation), thirdly to rank incident event impact using quantitave metrics.


 Proposed Measures

Several measures are proposed to describe possible risks of an incident which reflect every event’s profile and the authorities/teams that are present at a particular event. Incident measures are expressed with the use of either nominal or non-nominal discrete values ranging in a bounded scale [k...n]. Measurement values cannot be expressed accurately with real continuous values, unless a discretization process follows afterwards. Based on the literature review the authors conducted on Public Safety Guidelines the following incident measures are proposed:

  • Incident Health Impact (IHI): Authors define as IHI, the impact of an incident on a person (i.e. staff, spectator or artist). IHI is expressed as follows:

Value=1(Insignificant): No injury.

Value=2(Minor): First aid treatment on site required.

Value=3(Moderate): Medical treatment on site required.

Value=4(Major): Accidental death, extensive injuries or person disability involved in the incident.

Value=5(Catastrophic): Multiple deaths involved or permanent disability.


  • Reputation Impact Factor (RIF): Authors define as RIF the measure that describes the non financial damage caused by an incident to the event’s reputation and its organizers. RIF is expressed as follows:

Value=1(Insignificant): Non substantial, no impact, no media coverage.

Value=2(Minor): Low impact with no media coverage.

Value=3(Moderate): Moderate public embarrassment with low media coverage.

Value=4(Major): Substantial public embarrassment with media coverage, event delays and possible third party involvement (police, fire brigade, incident report required).

Value=5(Catastrophic): Significant public embarrassment, event cancellation, widespread media coverage, mandatory third party involvement (police, fire brigade, incident report required).


  • Individual Staff Involved (ISI): Authors define as ISI the number of different staff roles responsible for a particular incident’s response. Assuming that all roles are equally responsible for an incident’s response ISI is expressed as the number of individual staff responsible.


  • Incident Response (IR): Authors define IR as the time required to resolve an incident. Since time is a continuous value, appropriate discretization is required. IR is expressed as follows:

 Value=1(Little impact):

Value=2(Inconvenient delays): 10-30min resolve time.

Value=3(Significant delays to event deliverables): The time here is not the main issue but it is confirmed that the event will be delayed.

Value=4(Major): Non achievement of event’s deliverables – delay of the event more than 1hour.

Value=5(Catastrophic): Non achievement of event’s deliverables- multiple hour event delay or cancellation.


  • Likelihood of incident’s occurrence (LIO): This measure is expressed by the sum of number of users multiplied by the rating value they provided divided by the total number of users. Likelihood values are expressed as follows:

Value=1(Incident Very Unlikely to occur).

Value=2(Unlikely to occur incident).

Value=3(Incident that may happen in event but usually is not the case).

Value=4(Incident likely to happen).

Value=5(Incident very likely to happen).

Value=6(Certain or imminent incident).


Proposed Metrics

The authors propose the following metrics that will determine new incident categories or confirm the six incident categories presented in this study. In addition metrics will examine the correlation strength between incidents and the categories these will be assigned to.

  • Incident Type Occurrence (ITO): This metric is expressed as an absolute number i.e. the number of times an incident occurred during a specific event. It can take values from [0... k], k≤M, where k is the number of times a specific incident occurred and M is the total number of all incident occurrences.
  • Frequency of all Incidents Occurrence (FIO): This metric is expressed by the ratio:, where k is the ITO metric value, M the total number of incidents occurred for a specific event and n is the total number of events (i=1...n events of the same type for example rock festivals, or the same event over the years).
  • Incident Impact Metric (IIM): This metric is a combination of three proposed measures IHI, RIF and RIR. It is expressed by the following equation:


Where i is the number of events of the same type and C = [c0, c1, c2] for 0≤c0,c1,c2 ≤1 is the weight vector.   The values of the weight vector C are set by the event organizer based on the following two assumptions:

Case 1: If one of the incident weights (c0, c1, c2) is equal to 1 or if the norm ||C|| of the vector matrix converges to or is equal to 1, then the incident is considered of major impact. If one of the incident weights (c0, c1, c2) is equal to 0 then the incident is considered of minor impact.

Case 2:  If all incident weights (c0, c1, c2) are equal to 1 then the incident is considered of catastrophic impact.



Incidents occurring at large scale outdoor music events may have a significant impact on an event’s reputation let alone spectators’ well being. Presently in depth research into incident causes and an evaluation of their impact is not adequately performed. Professionally done risk assessments are confidential and not available to the public. Incident recording and evaluation of potential damages need a systematic approach, based on defined measures and metrics, that could combine all or some of the above pre-defined measures according to an event’s profile. This framework of measures and metrics presented in this paper is a novel approach in evaluating incidents for large scale outdoor music events. This approach could enhance our understanding on the nature of incidents and lead to the design of more efficient reponses. Future work by the authors involves the implementation and testing of this framework.




  1. Crowd Management Strategies. (2001) Concert Safety Survey Report.
  2. The Management of Health and Safety at Work Regulations 1999.
  3. Environmental Health Service. (1992) Health (Public Buildings) Regulations.
  4. Government of Western Australia, Department of Health. [2009] Concert and Mass Gathering Guidelines.
  5. Witt Associates. (2012) An Independent Assessment of the August 13th, 2011 Indiana State Fare Collapse.
  6. SafeWork SA. (2006). Event Safety Risk Assessment (small to medium sized community events).
  7. Ministry of Civil Defence & Emergency Management. (2003). Safety Planning Guidelines for Events.
  8. Health and Safety Executive Department. (1999). The event safety guide.
  9. Health and Safety Executive Department. (2000). Managing Crowds Safely.
  10. Government of Ireland, Department for Education. (1996). A Code of Practice for Safety at outdoor pop concerts and other outdoor musical events.
  11. S. Department of Justice Office of Community Oriented Policing Services. (2007) Planning And Managing Security For Major Special Events: Guidelines for Law Enforcement.
  12. Hanna J. (1994). Emergency preparedness Guidelines for Mass, Crowd-intensive Events. Office of Critical Infrastructure Protection and Emergency Preparedness, Government of Canada.
  13. Darlington Public Event Safety Advisory Group. (2012). A4 Form - Event Risk Assessment.
  14. Dublin City Council. (2012). Guidelines for Event Organizers.

Laird P. N. (1983). Mental Models: Towards a Cognitive Science of Language, Inference and Consciousness. Cambridge University Press