| Popis: |
The paper proposes a more sophisticated approach to the understanding of risk by approaching risk as the interaction of distributions on orthogonal dimensions. This approach enables novel understanding of not only incident causation but also of the solution strategies employed to prevent incidents. It also proposes a different approach to the measurement of the safety of an operation that avoids simply equating safety with not having incidents. All organisations make money by taking risks and reducing them to a point where the outcomes are profitable for the organisation. In general the main risks an organisation takes are seen as financial risks, however this is an over-simplification. Organisations also take risks in all their operational decisions as well. For high-hazard organisations this is brought to the fore as failures in quality and safety are more pronounced and have much higher profile outcomes. Risk assessments combine scenario outcomes and expected frequencies. Outcomes are often simplified, such as loss of primary containment or fatality, reducing the wide range of possible variations and causes; the entire outcome distribution may be collapsed to 4, 5 or 6 categories of severity. Frequencies are often simplified to orders of magnitude as point probabilities; a more sophisticated approach uses Bayesian conditional probabilities as distributions. The Rule of Three, a simple but powerful risk management tool originally developed to understand helicopter accidents, uses a number of distinct dimensions that capture the multi-dimensionality of a variety of conditional causal factors. Risk space typically has about 7 or 8 major orthogonal dimensions (e.g. equipment, people, weather, plan, environment etc.). This complex interaction paradigm provides added insights and understanding not only about complex process incidents, but also the increasing difficulty of attaining "goal zero", as the tools managing the risks are built with less sophisticated methodologies. In short, trying harder no longer delivers improved performance. Risk space plots the system's sensitivity to upset as the result of a triggering event. One major advantage of the multi-dimensional risk space approach is that we can now define safety without reference to having or not having accidents. Safe operations can still have accidents, if the trigger is severe enough, unsafe operations can still survive if the triggers are small enough. By identifying distinct dimensions that can be combined, risk space enables us to plot smaller high-risk areas that are missed in more broad-brush approaches where most activity is safe but unexpected risks can present themselves. It is also possible to examine the trigger effects that may take normally safe systems into danger. The paper will briefly discuss how these operational risks are understood in modern day accident causation models. It will also address how a change of paradigm may enable organisations to improve their understanding of the risks. The paper introduces a multi-dimensional approach to understanding and managing risk. The approach provides greater explanatory power for incident causation and allows measurement of safety that is less sensitive to ‘good’ or ‘bad’ luck and more accurately reflects the true level of safety of an organisation. |
