Target individual involuntary risk in the UK is set one order of magnitude lower at 1 in 100,000 per year with the tolerability risk one order of magnitude lower again at 1 in 1,000,000.

Another approach is to take the average risk of death of an average 35-year-old male as 1 in 1,000 from all causes. A high proportion of his risk budget is from natural causes and voluntary risk and it is a reasonable postulation that no more than 10% (1 in 10,000) should come from involuntary risk. If a new hazardous installation is planned near to him (for which he receives no tangible benefit) only adds 1% to this involuntary risk, this would not seem too unreasonable. In other figures, if his additional risk from the planned installation is no more than 1 in 1,000,000, that is not unreasonable.

One of the main advantages of the concept of individual risk is that it is, more or less, comprehensible to the average person who is generally familiar with betting odds. Additionally, it is straightforward to produce a simple comparison (as Table 1 below) which shows common risks and most people can slot in. More precise data on risk presented in the table below can be found in Ref. 1.

Table 1: Examples of individual risks

For the high speed Channel Tunnel Rail Link in the UK, the following individual risk criteria have been adopted. It can be seen in Table 2 that these are broadly in line with Table 1 above.

Table 2: CTRL individual risk criteria

4 Societal Risk

Societal, or community, risk is much more opaque and troublesome. It seeks to measure the risks where there is the potential for multiple, simultaneous fatalities. One common way to express it is on an F/N curve which is a cumulative plot of frequencies (F) and consequences (N) on a log/log plots. Frequencies are usually presented on the Y-axis and consequences in the X-axis.

"One of the problems with the societal risk has been the term itself, which, as the word risk means different things to different people at different times; leading to some misunderstanding and confusion. For instance, from an engineering perspective, societal risk is often regarded as no more than the relationship between the frequency and number of people suffering a specified level of harm from a particular hazard. Alternatively, others see societal risk as a much broader concept incorporating many other dimensions of harm, in some cases even the socio-political response in the aftermath of major accidents, or even lesser accidents where these might give rise to a significant expression of public concern". Ref. 3.

In case anyone is not familiar with F/N curves, an example is shown overleaf. It is the outcome of a quantified risk analysis which will, typically, have identified the hazards, created a number of event trees to model the progression from the potential hazard to its manifestations. Each branch of the event trees has an associated frequency and consequence. The results of the all event trees will be entered into a database and sorted so as to group all events with the same consequences together. The F/N curve is then plotted by adding up all the frequencies of events which cause 'N' or more fatalities and this frequency is plotted against 'N' on the x-axis.

It is common to put criteria on an F/N curve. The one in the figure above is split into 3 zones by the two diagonal lines. The zone above the upper line is often known as the Intolerable Zone and risks which cause the F/N curve to stray into this zone must be mitigated to bring the curve back below the upper line.

The zone below the lower line is often called the Tolerable zone and risks contributing to the F/N curve in this zone probably need no further mitigation measures.

The As Low as Reasonably Practical (ALARP) zone is where further risk reduction measures should be considered and implemented if practicable. The way in which practicability is judged using cost benefit analysis is discussed later.

The position and the gradient of the zone boundary lines vary from country to country. In the UK they are 3 orders of magnitude apart and have a gradient of -1, commonly regarded as 'risk-neutral' approach. In effect, this means there is no aversion, so that killing 100 people in an accident every 1000 years is the same as killing 1 person every 10 years.

Several countries (Denmark and Holland) apply a 'risk-averse' approach towards multiple fatalities and the zone boundary gradients are minus 2.

However, as the Social Risks study (Ref. 3) indicates, that in view to the practical implication of a steep societal risk FN line, e.g. a difficulty in a justification of the robustness of the risk results, the selection of a gradient of minus 1 for FN based criteria appears reasonable.