The cost of human error intervention

DOE has directed that cost-benefit analyses be conducted as part of the review process for all new DOE orders. This new policy will have the effect of ensuring that DOE analysts can justify the implementation costs of the orders that they develop. We would like to argue that a cost-benefit analysis is merely one phase of a complete risk management program -- one that would more than likely start with a probabilistic risk assessment. The safety community defines risk as the probability of failure times the severity of consequence. An engineering definition of failure can be considered in terms of physical performance, as in mean-time-between-failure; or, it can be thought of in terms of human performance, as in probability of human error. The severity of consequence of a failure can be measured along any one of a number of dimensions -- economic, political, or social. Clearly, an analysis along one dimension cannot be directly compared to another but, a set of cost-benefit analyses, based on a series of cost-dimensions, can be extremely useful to managers who must prioritize their resources. Over the last two years, DOE has been developing a series of human factors orders, directed a lowering the probabilitymore » of human error -- or at least changing the distribution of those errors. The following discussion presents a series of cost-benefit analyses using historical events in the nuclear industry. However, we would first like to discuss some of the analytic cautions that must be considered when we deal with human error.« less