The 2015 version of the COMAH Regulations, which came into force in June of this year, have brought about a number of changes, including adjusting the way domino effects are planned for. Alan Field discusses the implications of these wider duties and takes a look at domino effects and some related concepts.

Introduction

The Control of Major Accident Hazards Regulations 2015 (COMAH) — which arose out of the Seveso III Directive (2012/18/EU) — came into force on 1 June 2015. The 1999 version of these Regulations have now been revoked.

COMAH is often thought of in relation to petrochemical sites, but it can often have a wider application relating to the storage, creation and processing of hazardous substances.

As is outlined in more detail below, one big change in COMAH 2015 is the way that the accumulations of substances under the Regulations are defined, and the reference point is now based on European Regulation No. 1272/2008 on Classification, Labelling and Packaging of Substances and Mixtures (The CLP Regulation).

Mixing it up

One of the main drivers behind COMAH 2015 was the change in European chemicals classification, which arose out of the UN’s Globally Harmonised System of Classification and Labelling of Chemicals (GHS). This then led to the CLP Regulation.

Most usually, the Competent Authority (CA) that enforces the Regulation in England and Wales will be the Health and Safety Executive (HSE). In Scotland, the Environment Agency, the HSE and the Scottish Environmental Protection Agency are most prominent, while in Northern Ireland the Health and Safety Executive for Northern Ireland and the Northern Ireland Environment Agency are frequently the CAs.

The COMAH Regulation continues to divide sites that come under the Regulations into upper and lower tier sites. The CA will discuss with COMAH operators how the changes in chemical classifications may change their upper or lower tier status and, indeed, a small number of lower tier sites may no longer under come under COMAH due to CLP. This could arise because CLP classifies chemicals differently to the previous COMAH 1999 Regulations and, in particular, the way of accumulation of dangerous chemicals may be viewed under COMAH.

In the short term, this means that all sites that fall under COMAH need to re-assess their substances inventories. This is especially true where there is lack of information about the toxicity levels of certain chemicals or the environmental impacts of combinations of chemicals. It should always be remembered that while COMAH is often associated with safety risks, it also aims to prevent pollution and other major environmental accidents. To be aware of the safety aspects of chemicals — but not then to assess their ecological impacts in the event of an accident — is something to be avoided.

One example would be the Buncefield incident. While the immediate consequences of the explosion and fire is always remembered, one of the most serious impacts was having to deal with many thousands of gallons of contaminated fire water runoff mixed with various hydrocarbons and fire fighting foam, some of which were protein compounds, acutely toxic to aquatic life.

Playing dominos?

One new area of the regulations is the increased obligations concerning “domino effects”. These are described in more detail in Regulation 24 of COMAH.

In broad terms, “domino effects” refers to the increase in the potential risk or consequences that emerges where a major accident at one COMAH site could impact on another. This could be, but would not necessarily be, an adjoining site, and the increased risk or consequences could be safety and/or environmental related.

The CA will be responsible for defining domino groups, as they already do in some instances. However, this duty is now more specific; under COMAH 2015 the CA now has the legal right to obtain information from COMAH site operators. More importantly, the CA can also require sites designated as being in a domino group to exchange information with one another. This is to ensure that the overall hazard is assessed and managed across the domino group, and to plan for the potential of combined consequences of a major accident.

This will not be an issue for some site operators as they already co-operate with one another. One such example would be the Essex Petrochemical Mutual Aid Group; the Essex County Fire and Rescue Service have regular meetings with all the COMAH site operators in their area, which has led to co-operation and sharing of information and emergency response resources amongst the different sites and at all levels of their management.

However, where there isn’t currently such a level of co-operation between COMAH sites, one interesting aspect of the “exchanging information” requirement of COMAH 2015 is that different sites may have slightly different – and potentially ill-fitting – approaches to assessing risk and managing their controls. This is something that needs to be considered by all members of a domino group.

While a number of risk modelling techniques exist, the wider the scope of co-operation, information exchange and potential domino effect being defined, the greater the need for a broader understanding of risk assessment.

Models for risk assessment

At high level, the way that risk assessment is approached by top management may need to be reviewed. For example, there are various models to consider risks that are perceived as unknown or completely unexpected; one model is known as “black swan”, as coined by Nassim Nicholas Taleb.

While it was devised for financial markets, it has been used for other major risks and could be helpful when starting to look critically at the technical and engineering basis that many high-risk processes are considered from.

One particularly interesting element of the black swan model is the idea that all risk analysis is limited by the psychological bias that can, individually and collectively, blind those who are assessing risk to potential major events. This can be especially true when having to consider the risks of an adjoining site where, perhaps, completely different dangerous substances could react with those being assessed, or where different emergency planning assumptions have been made.

Another strategic approach to risk is sometimes referred to as the “Swiss cheese model” or the “cumulative act effect.” The original model, proposed in 1990 by James T. Reason of Manchester University, has gone through a number of adaptations and applications and is used widely in a number of sectors, including healthcare and aviation safety. It is used as a way of looking at how simultaneous (or near simultaneous) events could circumvent individual levels of controls in place, leading to a major failure or accident.

The Swiss cheese model moves away from the idea of a “single point of failure” – the idea that failure in an isolated part of a system can then compromise the entire system – that had previously been at the centre of risk analysis. The Swiss cheese model anticipates that it may be varying combinations of events and/or failures, not necessarily serious in their own right, that could lead to a major accident. In other words, different holes within the layers of risk anticipation (imagined as Swiss cheese) may line up in different combinations. This is where psychological bias may impact on the potential scenarios defined or controlled.

To use an example with hazardous chemicals; where it is unknown or uncertain what the toxicity levels of certain chemicals or combinations of chemicals across a domino group would be in the event of an incident, this would lead to more controls being considered at each “hole” defined in the cheese, where these “unknowns” are identified. In the black swan model, the approach might be more to consider any major risks not normally perceived as arising from the interaction of substances, or assumptions made about the effectiveness of controls.

When looking at domino groups, the cumulative act effect model could be beneficial in that the Swiss cheese refers then to more than the potential risks at just one site – the layers of protection across all sites (and the potential holes in these layers, and the manner in which they might align) can be considered together. It could also be used as a tool to evaluate risk information shared by different domino group members. The model might also impact on offsite emergency planning — another familiar concept within COMAH — where cumulative act effect would consider what events might compromise the efficacy of these plans across the whole domino group, It could also be used as a tool to critically test these plans through desktop scenarios.

Conclusion

While domino groups defined by the Competent Authority and co-operation between COMAH sites are not new concepts, there are now wider duties concerning them within COMAH 2015. This might encourage wider and more strategic concepts of risk analysis — over and above the usual technical and engineering one.

Yet, co-operation can lead to confusion, and, equally, it can be very powerful in terms of critically testing previous assumptions and even shibboleths that have emerged. Tools like cumulative effect analysis can be part of a catalyst to aid such discussions across management teams.

Last reviewed 12 August 2015