RISK ASSESSMENT

In an informal sense risk analysis has been used throughout the history of mankind. Risk is always associated with decision. Something has to be done; an action has to be taken. It might be trivial, as with crossing the street, or of great importance, as with siting and designing a major dam. In either case a choice has to be made as to what to do. The outcome is in the future, and is uncertain. Some outcomes will be better than others. Some might be good, some downright disastrous.

A choice of possible actions must be made, and the choice means taking a risk. We assess the risks associated with each action and make the decision. The process is mostly informal, using whatever information, experience and intuition is available. Much of the time this works well enough. However, there are also times when the risks are not well assessed and the resulting decisions are neither understandable nor rational. Particularly where there may be loss of life, this is not acceptable to a society demanding high levels of safety.

Large commercial losses or damage to the environment must also be avoided. In such circumstances risks need to be better controlled. There are two main strategies for controlling risk. One is to be more conservative in design to allow for uncertainties, which is, in other words, to assign more resources of money and materials to the engineering work. Indeed, engineers have many methods of tacitly controlling risk, such as the use of safety factors, permissible stresses or quality assurance.

The second overall strategy is to put more effort into careful risk assessment, to maintain safety levels and reduce risk while refining designs and reducing costs. Both strategies are legitimate. The second has led to the development of methods for formal risk assessment. Explicit techniques for risk assessment had their origins in the 1930s and 1940s. There were two main trends. One began in structural engineering and the other related to developments in operations research at the time. In the structural field, the main early work was due to Pugsley,' based on a background in aircraft structures, and Fre~denthal.~,~ Significant contributions were also made by Wierzbicki, Baker, Torroja, Johnson, Shinozuka and others: Turkstra4 gives a useful historical summary.

The structural engineering thrust had a particular and idiosyncratic flavour for two reasons. Firstly, structural engineering deals with risk problems in which the capacity of the system being analysed and the demand on it could be separated. Structurally, capacity and demand are represented by resistance and load. Secondly, apart from major exceptions such as offshore platforms or aerospace structures, most structural engineering projects are relatively inexpensive and do not warrant the cost of individual risk assessment.

As a result the greatest thrust in the development of risk analysis for structural engineering has been towards dealing with many structures at the same time by using risk analysis techniques to form better codes of practice. The requirements for code development are specialized and have given their own flavour to structural engineering risk assessment. The issues are interesting and important so will be explored in more detail later in the chapter.

Military and space requirements in the 1960s saw the development of a different approach deriving from earlier operations research work (for example, by Von Neumann5 ). The aerospace and electronics industries faced system reliability problems requiring assessment of the reliability of systems made up of many components, each with a known failure rate or reliability. The two groups followed somewhat different paths in that fault tree methods (developed in 1961 by H. A. Watson at the Bell Telephone Laboratories) were used at an early stage by aerospace engineer^,^ while the electronic industry used equivalent alternative approaches for another decade before using fault trees7 At this stage risk analysis techniques became particularly important in assessing the risks associated with nuclear power plant^.^ The chemical and process industry also began to use formal risk assessment methods at this time as they were ideal for dealing with the safety problems of large and complex chemical plants.