Stuart G. Reid

Perception and communication of risk, and the importance of dependability

Abstract Studies of risk perception and risk communication often focus on differences between ‘expert’ and ‘non-expert’ points of view, and these differences are often attributed to flawed perceptions of risk and flawed cognitive processes. However, an alternative view is presented in this paper, centred on a dependability hypothesis that most observers view risk rationally with regard to the dependability of relevant safety measures. The paper reviews the historical development of work on risk perception and risk communication, and discusses the limitations of quantitative approaches, with particular reference to applications in the field of structural engineering. The fundamental importance of dependability is discussed, and it is concluded that risk analysis and risk communicators need to address the critical issues of dependability and trust.

Acceptable risk criteria

This article reviews work relevant to the identification and use of acceptable risk criteria for risk-based decision-making in engineering, particularly in relation to engineering safety standards. The article includes a brief outline of some relevant work in the fields of psychology, sociology and statistical decision theory, and it traces the development of quantitative risk acceptance criteria for engineering risk management and regulation. Work on quantitative methods to evaluate the acceptability of engineering risks is briefly reviewed, and limitations of the quantitative methods are noted. The use of quantitative risk acceptance criteria in engineering regulations and in engineering practice is discussed with particular reference to applications in the field of structural engineering. Also, the main risk acceptance issues that are currently attracting interest in the field of structural engineering are noted.

Specification of design criteria based on probabilistic measures of design performance

To date, reliability-based structural design standards have been developed using deterministic design procedures that are formulated to achieve target levels of structural reliability (target safety indices) without requiring explicit reliability calculations. Target safety indices have been selected to match typical safety levels in previous design standards. However, the current levels of structural safety are not necessarily the most appropriate, and it has been suggested that explicit risk-based measures of structural performance should be used to optimise the target levels of design performance. Explicit risk-based measures could be used to underpin deterministic design requirements or they could be used directly for explicit risk-based design. The paper discusses the use of risk-based measures in the development and application of design Standards. The paper also discusses broader issues related to the development of design Standards, including the influence of international standardisation.

Safety factors for bridge falsework by risk management

Determination of safety factors logically depends upon the cost of providing safety, the consequences of failure, the failure probability or rate, and the time of exposure to load. In the case of conventional structures these are indirectly accounted with safety factors prescribed by codes, and based on probabilistic studies and calibration. Bridge falsework may consist of structural systems that support relatively costly construction (with high consequences of failure) for very short time periods. Structural design codes generally do not prescribe safety factors for such temporary works. In the absence of prescriptive codes, contractors are left with decisions about safety factors. Some common policies are: to use factors or stresses and loads the same as for permanent construction; to use increased allowable stresses; or to use reduced return periods for environmental loads. This investigation applies basic principles of decision analysis and risk management to determine safety factors for several common bridge falsework or erection situations. The results indicate that substantial differences in the appropriate safety factors can be found, especially when low cost falsework supports costly permanent structure. In particular, the common practice of using increased allowable stresses or reduced load return periods can be seriously unconservative.

Random response to stochastic floor loading

Abstract In order to model random structural responses to floor loads, floor load-effects are modelled with regard to the randomness of the loads and of the influence functions relating loads to load-effects. The floor loading is represented by a stochastic process model, and influence functions are represented as random functions of distributed structural stiffnesses. Approximate expressions are obtained for the means and variances of floor-load-effects and related equivalent uniformly distributed loads. These expressions involve load parameters, stiffness parameters and influence factors derived from influence functions. Influence factors for deflections of several floor systems are tabulated and the practical application of the results is illustrated with calculations for the means and variances of the long-term deflections of two reinforced-concrete floor slabs. The proposed model is distinguished from a well-known load-effect model by its ability to account for the response randomness due to local fluctuations of stiffness. This ability is useful in modelling deflections and other load effects which might be significantly affected by random stiffness fluctuations.

Probability-based patterned live loads for design

Abstract Probability-based patterned live loads for design are discussed with regard to probabilistic models of load-effects due to sustained live loads on floors. Statistics are presented for equivalent uniformly distributed loads (EUDLs) and equivalent patterned loads (EPLs) for a continuous beam and for continuous floor slab systems. It is shown that EPLs (rather than EUDLs) are required to realistically model structural load-effects that are sensitive to random load patterns. Furthermore, it is shown that an appropriate set of EPLs (for load-effects that are compounded by loading in the vicinity of the load-effect) can be approximated using a standardised EUDL (dependent on the area over which it acts) on the critical structural bay(s) in the vicinity of the load-effect, together with a constant (mean) uniformly distributed load acting on the other structural bays. Accordingly, ‘chequerboard’ design loads are not required. Indeed, it is shown that chequerboard EPLs are not appropriate for design, because their statistics are sensitive to the structural type and they are significantly different from the statistics of the standard EUDLs that are the basis of the standard design loads specified in design codes.

Influence on structural reliability of uncertainty in estimated extreme values of load-effects

Extreme values of time-varying loads are often estimated to serve as design loads for the purposes of structural design. Uncertainty in the estimation of the design loads inevitably leads to uncertainty in the resultant levels of structural reliability. Uncertainty is assessed for estimates of extreme wind loads calculated using statistical methods based on the average conditional exceedance rate (ACER), fitting of a Gumbel distribution and Peaks-over-Threshold (POT). The ACER method gave the best results, but all the methods gave results which would normally be considered to be sufficiently accurate for engineering applications. However, for structures designed on the basis of the estimated values of V100, the uncertainty in the estimated design loads produced very uncertain probabilities of failure with a significant increase in their expected value. It is concluded that the uncertain distribution of the probabilities of failure must be taken into account when evaluating structural safety and a ‘fiducial confidence function’ is proposed for this purpose.

Acceptance criteria for risks of disasters with widespread effects

The paper discusses special issues that arise in engineering decision-making for engineering projects that involve a risk of potentially disastrous outcomes with widespread effects. For such projects, the costs, risks and benefits may spread beyond the jurisdiction of the relevant regulatory authorities, and the dispersion of the costs, risks and benefits may limit the accountability and liability of the decision-makers. The paper briefly reviews the conventional methods for determination of risk acceptance criteria, and it includes a critical review of ‘societal risk’ criteria based on FN curves for risks involving potentially disastrous outcomes. The paper discusses the limitations of conventional engineering risk acceptance criteria and identifies additional factors that must be considered for rational risk-informed decision-making whenever there is a risk of a disastrous outcome.