Prioritising incident learning by formally estimating potential severity
08 July 2015
Use of potential severity to classify incidents is not yet embedded beyond very large multi-national companies. In this article, Michael Schön of Syngenta Crop Protection AG explains the company’s approach to using potential severity to classify both process safety and occupational incidents, and also benefits and learning.
Starting at the very beginning, what do we mean by potential severity? Consider two incidents that have resulted in a broken wrist:
1. An employee stumbles and falls while walking across a paved walkway, fracturing a wrist in bracing for the impact.
2. An employee reaches into a screw conveyor. The machine moves unexpectedly, catches the employee’s hand and fractures his wrist.
These two incidents have resulted in similar injuries so their actual severity is identical, but they have completely different potential severity.
* If Incident 1) were to happen 100 more times it is very unlikely to result in an outcome much worse than a fractured wrist.
* If Incident 2) were to happen 100 more times there is a very good chance of a lost limb or even fatality.
In a world of limited resources it makes much more sense to learn as much as possible from incident 2) to prevent reoccurrence. In comparison, incident 1) would have fewer resources dedicated to investigation and analysis.
Within Syngenta we apply the potential severity assessment both for process safety and occupational safety incidents to prioritise reporting, investigation and learning.
Serious Injury and Fatality Prevention
In 2012 Syngenta joined the European Serious Injury and Fatality (SiF) Prevention study group chaired by BST. The results were sufficiently promising to add potential SIF assessment into Corporate reporting requirements at the end of 2013.
Based on both external advice and analysis of our own incident data, SIF potential is defined using a rule-based decision tree. Investigation of an incident meeting any of the SIF criteria is prioritised because it had the potential to result in a fatality or serious Injury (life-changing or life-threatening accident). These criteria include:
* work in confined spaces
* any fire, flash fire or explosion
Fig 1. Actual serious injuries, recordable accident rate and SiF potential rate
* fall from height >2m (even if stopped by fall protection).
These are deliberately generic and broad criteria to maximize reporting. Based on professional judgment on regional or global level incidents might be downgraded afterwards.
The graph left (Fig. 1) is for the region Europe, Africa and Middle East (EAME)
* The actual serious injuries and fatalities (one fatality during a confined space entry and one finger amputation related to use of powered hand tools, both in 2011)
* The recordable injury rate (per 200.000 working hours)
* The rate of actual serious injuries and fatalities and recordable accidents with serious injury and fatality potential per 200.000 working hours
(N.B. Driver safety and process safety data are excluded)
Actual serious injuries and fatalities are rare events but it appears that 10 – 12 % of all recordable accidents have serious injury and fatality potential. Carefully analysing and learning from these cases should ultimately result in fewer actual serious injuries and fatalities.
Key learnings one year after implementation of the reporting requirements
* The rate of recordable accidents with SIF potential was much more stable than the overall incident rate including first aid cases and near misses.
* The number of incidents with serious injury and fatality potential reported globally has increased.
* Setting a target for reporting incidents with serious injury and fatality potential would probably be counterproductive. Using the concept maximizes usefulness of incident data beyond recordable injuries.
* There is an overlap to process safety incidents (e.g. release of an acute toxic gas) or driver safety incidents (e.g. vehicle roll overs).
Process Safety Incidents
Fig 2. Classification of Process Safety Incidents
Since 2011 our largest chemical manufacturing sites have classified process safety incidents based on both actual and potential severity. Syngenta defines process safety incidents as: ‘hazardous events caused due to loss of containment or control of a chemical process.’
Examples are: fire or explosion, loss of primary containment/ energy release, internal/ external damage to equipment by process or loss of process control like hazardous deviation of temperature, level, pressure etc. This definition is deliberately broad to maximize reporting! We then prioritize based on actual and potential severity.
Given the higher HSE maturity of these sites and their sound risk assessment knowledge, the classification is based on the related scenarios in the site’s process risk assessments. The process used to classify process safety incidents is a two-step approach and shown in the flowchart 1. For both actual and potential severity we use three levels: high, medium and low which are based on the severity ratings used in Syngenta’s process risk assessments. Severity 1 means catastrophic consequences for people, environment and / or supply disruption; Severity 4 means relatively low consequences.
1. Assess the actual severity of the process safety incident based on the matrix used for the process risk assessment PRA
2. Assess the potential severity of the process safety incident based on the failure of one additional barrier / control
From November 2013 to October 2014 approximately 3,000 process safety incidents (all low actual severity) were reported. Around 1% were classified as medium severity potential and only 2 incidents were classified as high severity potential.
These numbers clearly show one benefit of the concept; severity potential can be used to filter the huge number of incidents global companies have in their databases. High potential incidents can then be escalated to ensure that the response is proportionate.
* Incidents with low severity potential are handled locally, incident investigation is often done by first line managers using simple problem solving tools, and learnings are shared and actions tracked on site.
* Incidents with high severity potential are investigated at least by a member of the site management team often supported by a regional or global specialist. Root cause analysis is more thorough and learning is shared and actions tracked globally
Why have two different approaches?
From our experience, the concept of classifying incidents based on their potential severity needs to consider the HSE maturity and risk management competency of the reporting unit.
* Mature organisations with well-developed risk management competency (like chemical manufacturing units) will be able to classify incidents based on a risk matrix consistently after a period of global calibration.
* For less mature organisations with lower risk management competency a more rule-based decision tree proofed to be more effective.
Benefits of classifying incidents based on their potential severity
The concept of classifying incidents based on their potential severity can be applied to a range of different incidents and offers clear benefits for the organisation.
* Incidents with actual high severity are rare events and therefore the typical reaction “how could this happen on our site?” is not surprising. Visualising and using the still significant number of incidents which could have resulted in high severity events in discussion with line managers helps to maintain a sense of vulnerability.
* By escalating incidents with high severity potential to upper line management we escalate the incidents to a level in the organisation which is able to deploy learning and implement the necessary changes.
* By accepting less formal investigations for low potential severity incidents we make effective use of our resources and empower employees and front line managers. Nevertheless it is the role of upper line managers to ensure that these processes are effective and actions implemented.
* Intuitively employees and managers feel the difference between low and high potential severity. By responding to incidents proportionately to the potential severity (and not only the actual severity) we in HSE will gain more credibility within the organisation.
By improved tracking of actions from high potential severity incidents we ensure that they are implemented on time and are effective. Several of our sites have recognised that relying on “percentage of overdue actions” can result in a blind spot for actions with potentially higher importance.
Contact Details and Archive...