5 Conclusions and recommendations regarding use of risk acceptance criteria in Denmark

Acceptance criteria in Denmark and the EU

This final chapter draws conclusions about the situation in Denmark and the EU in relation to the use of risk acceptance criteria for major hazard establishments, and comments on how the situation might be improved. Section 5.1 describes the current situation. Section 5.2 lists the requirements that should apply to risk acceptance criteria and the risk analysis methods used to generate the information compared against these criteria. Sections 5.3 to 5.5 contain recommendations for how risk criteria can be formulated, including the magnitude of acceptable risk levels. Finally, section 5.6 contains comments on the need for further initiatives in this area.

5.1 Status in Denmark and the EU

Most of the risk analyses performed since 1995 for major hazard establishments in Denmark have used qualitative methods, with widespread use of safety-barrier diagrams. These diagrams have been found to be useful for assessing the safety measures establishments have implemented. Unfortunately, these qualitative methods have not been suitable for defining uniform and generally applicable regulations governing appropriate protection of surrounding areas against the residual risk of accidents with consequences outside the establishment’s boundaries. This has meant that decisions about safety distances, for example, have been made on a case-by-case basis, without always giving general consideration to how to deal with accident scenarios with low or very low expected frequencies.

There has been no qualified discussion to date on acceptance criteria in Denmark for environmental damage. This situation is not very different from other EU Member States. With respect to personal injury, some Member States have developed good systems based on quantitative methods (e.g. the United Kingdom, the Netherlands, and Flanders), hybrid methods (France and Italy), or deterministic methods involving no probability considerations (only in Germany). No equivalent well-developed methods exist for dealing with environmental damage.

5.2 Risk acceptance criteria requirements

The first principle of risk acceptance is that all unnecessary risk should be removed. This means that the ALARA or ALARP principle are always followed, even where the level of risk already complies with the general risk acceptance criteria. The ALARA principle involves an assessment of whether the costs of a given safety measure are disproportionately large compared to the safety gain. This assessment will be different depending on whether the level of risk is high or low in relation to the risk acceptance criteria. It is therefore not necessary to highlight risk levels where ALARA is especially applicable.

Other risk acceptance criteria and methods to generate the data to be compared against these criteria must comply with the following requirements:

Consistency, proportionality, and transparency, as explained in the latest Commission guidelines;
It should be possible to evaluate the level of risk exposure for surrounding residents considering all activities at the plant, i.e. all relevant accident scenarios must be included in the assessment;
They should enable assessment of the risk of environmental damage into risk acceptance;
They should enable selection of one or more safety distances (for different types of land use), and a maximum consequence distance;
Whenever residential areas are permitted within the maximum consequence distance, assessment of some form of societal risk should be possible, i.e. risk levels for the areas with a concentration of population outside the safety distance, but within the maximum consequence distance, should be accounted for and handled appropriately.
Risk assessment should reflect the effects of any specific safety measures implemented by the establishment in question, i.e. risk assessment should not be based on generic accident frequencies alone.

5.3 Incorporating frequency criteria

The definition of risk involves a clear element of probability. It is necessary to incorporate probability or frequency into any analysis, either as numeric values or using qualitative descriptions or classes. Selection of reference accident scenarios to determine safety distances should be based on explicit quantitative or qualitative frequency criteria, and distinction should be made between safety distances and the maximum consequence distance, in order to achieve acceptable limitations on land use while also staying aware of possible consequences in very rare situations.

5.4 Protection of vulnerable objects

Safety distances, i.e. boundaries for areas with land-use restrictions, can be set for different objects and/or groups of people. The basic division should be as follows:

Employees at the major risk establishment will be protected on the basis of normal occupational safety requirements;
Workplaces at other establishments must not be exposed to a location-based risk of fatality (or equivalent qualitative criteria), greater than approx. 10^-5 per year. Employees at these establishments must be informed of the risk conditions and management of accident situations
General residential areas and other areas frequented by the general public, including schools, homes for the elderly, etc., must not be exposed to a location-based (individual) risk of death (or equivalent qualitative criteria) exceeding approx. 10^-6 per year.
Objects playing a role in public emergency services, such as hospitals, and fire and police stations, should be placed outside the maximum consequence distance.

These criteria should be supplemented with criteria for societal risk and environmental damage, in order to limit the cumulative frequency for major accidents to:

Approx. 10^-3 per year for major accidents (involving up to one fatality, or equivalent damage or injury as defined in Table 14);
Approx. 10^-5 per year for catastrophes (involving up to 10 fatalities, or equivalent damage or injury);
Approx. 10^-7 for disasters.

When calculating societal risk, people’s presence or absence and the protective effect of buildings, etc. should be taken into account.

The same criteria should be used for both existing and new situations. For existing situations that do not fulfil the requirements, a timeframe should be set within which the criteria must be fulfilled. This can be done by implementing preventative measures to reduce frequencies, and/or mitigating initiatives to reduce the consequences.

It is expected that where the ALARA principle is followed, most risks can be reduced to at least a factor of 10 less than the above acceptance criteria.

5.5 Risk analysis methods

Environment Project 112 proposes that quantitative and qualitative risk analysis methods should lead to comparable results. This review has found that the methods, as they have been applied to date, are quite difficult to compare.

It is possible to use quantitative risk analysis methods to meet the above risk acceptance criteria. However, it is recommended that the safety-barrier diagram method also be used with quantitative methods, as this is easily comprehensible and relatively simple to use. It also makes it possible to take into account site-specific circumstances.

Qualitative risk analysis methods should continue to be available. However, in order for the results to be applicable in relation to the above risk acceptance criteria, frequency classes and quantitative frequency intervals must be linked, as shown in Table 3 or Table 5. Guidelines should also be developed for the combined assessment of various accident scenarios (for example, the way to sum frequencies). These considerations suggest that the French method would be a good candidate to use as the basis for a Danish hybrid method.

5.6 Needs for further work

Qualitative or hybrid methods need to be developed that can be used to determine safety distances and societal risk. Section 5.5 above mentions that the French hybrid method could provide a basis for a Danish hybrid method. If such a hybrid method is applied, the French frequency and seriousness classes should be critically reviewed and possibly adjusted. It may also be appropriate to include meteorological factors that reduce exposure frequency in terms of emissions frequency, by re-evaluating and generalising Figure 2.2 in Annex A of Environment Project 112.

Acceptance criteria should be developed for natural and environmental damage. A starting point might be to compare the seriousness of personal injury and environmental damage, as listed in Table 14, adapting this to the Danish requirement of being able to assess environmental damage in salt-water areas, such as fjords, sounds and coastal regions, rather than in rivers and canals. This work should include a review of the methods proposed by some EU Member States (see section 3.1), and analysis methods previously used in Denmark, especially with the aim of accommodating Denmark’s particular interest in protecting groundwater aquifers.

There are currently no Danish guidelines for using the ALARA or ALARP principle. Inspiration for such guidelines can be drawn from experience in countries such as the United Kingdom.