74 All such techniques need to consider three aspects in making an assessment. The first is to identify events – fires, explosions and toxic releases – that could have major hazard potential. Next the consequences of each event need to be estimated in terms of human harm. Finally frequencies need to be assigned to such events using published sources of failure rates, for instance, and presented as the chance per year of such an event occurring. Generally speaking, the more severe the event the less its likelihood but the greater its consequences in terms of human harm.
75 From this it can be seen that risk assessment requires difficult judgements based on formulating assumptions, estimating probabilities of infrequent events and incorporating the uncertainty surrounding these. Underestimating the likelihood of major events may lead to unacceptable levels of risk to those in the vicinity of a major hazard site, while overestimating it may lead to unnecessary blighting of otherwise economically viable land. Risk assessments have to be conducted to a high standard to be credible to planning authorities and we therefore say more about the current techniques and judgements involved before introducing our conclusions and recommendations.
Simplified risk approach
76 In the years following the Flixborough incident HSE adopted a hazard-based approach for setting zone boundaries: a site-specific risk-based approach was not generally feasible at that time. This approach is nevertheless risk based, albeit in a simplified and semi-qualitative form. For clarity, we will use the term ‘simplified risk assessment’. The aim of this approach is to achieve a separation between
developments and the site which provides a very high degree of protection against the more likely smaller events, while also giving very worthwhile (sometimes almost total) protection against unlikely but foreseeable larger scale events. This approach was endorsed by the Advisory Committee on Major Hazards at the time, and became one of its fundamental principles for controlling major industrial risks in built-up areas.
77 To apply the approach in practice involves the identification and selection of a single major event – fire or explosion. Its effectiveness depends to a considerable degree on the identification of the most appropriate worst-case event that dominates the risk scenario. Then a separation distance is determined based on a ‘dangerous dose’, a term defined by HSE (see Annex 8) which spans fire, explosive and toxic events. The risk assessed in this way is that of an individual at a particular place being exposed to a dangerous dose or worse.
Quantified risk approach
78 Planning enquiries have, over the years, concluded that HSE’s advice should take specific account of the likelihood of death or injury to the public. HSE has
responded by developing techniques to quantify the risks associated with hazardous installations, based on a range of foreseeable failure scenarios. The method is widely used for toxic releases, such as the unintentional loss of containment of chlorine gas being stored on a chemical plant. The risk to an individual in a specific location is the summation of the risks arising from these different scenarios. This process is known as quantified risk analysis (QRA) and we refer to it as the QRA approach. The 1989 risk criteria document sets out HSE policy, and the reasons for criteria adopted in this approach, and the Cullen Report(ref 10)in 1990 on the Piper Alpha
disaster strongly advocated the use of QRA for operational safety and risk reduction purposes (see Annex 6). As a result numerous techniques have been developed for applying QRA to major hazard risk control.
Estimation of consequences
79 HSE, in doing both simplified and quantified risk assessments for land use planning purposes, uses ‘dangerous dose or worse’ as its measure of consequence in terms of human harm (see Annex 8). Most other organisations and countries use a measure based on the risk of fatality and as does HSE it in its tolerability of risk guidance.(ref 11)The use of ‘dangerous dose or worse’ has implications which we will
return to in paragraphs 97–99 in terms of being able to compare risks and therefore achieve broadly similar advice in response to broadly similar risks.
Comparison of the two approaches
80 In 1998 HSE initiated a fundamental review of its involvement in land use planning (Annex 9). A subsequent project to implement the findings of the fundamental review examined HSE’s use of simplified and quantified risk analysis. The resulting report identified key advantages and disadvantages of each approach and some of these are summarised below. It made a range of recommendations to improve the consistency of simplified risk analysis, where this is adopted, but also urged HSE to continue to perform research into QRA methods so that some of the reasons for having to resort to the simplified approach could be resolved.
81 In summary, simplified risk analysis is usually appropriate where:
M there is insufficient information for QRA or some of the available information contains a high degree of uncertainty;
M the surrounding population density and demand on land use are low; or
M it would generate similar results (in terms of the sizes of land use planning zones and the advice given) to those from QRA, as might be presumed where there is a single risk source. After Buncefield, this would be unlikely at COMAH top-tier sites.
82 There are also objective criticisms to the simplified risk approach, a number of them put in the 1989 risk criteria document,(ref 9)such as:
M vagueness in terminology, for example ‘a very high degree of protection’, ‘worthwhile (sometimes almost total) protection’, ‘unlikely but foreseeable’;
M the resulting protection can be excessively restrictive in terms of land use;
M some arbitrariness and lack of transparency in selection of the worst-case event, and through this, potential inconsistency in treatment between installations; and
M the difficulty of comparing the degree of protection achieved with that for other everyday risks.
83 A further review following the 1998 fundamental review noted that if a site has both hazards that are analysed by the simplified risk approach and hazards that are analysed by QRA it is very difficult to add the two together to get an overall risk from the site.
84 The advantage of QRA, coupled with use of appropriate risk criteria, is that it deals with the objections to the simplified approach set out in the two previous paragraphs. However, use of QRA is not without its own difficulties. For example, assigning frequencies to rare events, such as major equipment failures, can introduce a high degree of uncertainty. Also the resources required for performing full QRA will be greater than for the simplified approach (though the time and costs are much less than they once were).