Inclusion of Fatigue Effects in Human Reliability Analysis

Abstract

This research investigates a quantitative methodology for the inclusion of fatigue, induced by sleep deprivation, in human reliability analysis (HRA). A statistical approach is developed for the use of empirical data in deriving performance shaping factor (PSF) coefficients that could be used in quantitative HRA methods. PSFs are used in human reliability analysis to modify general human error probabilities to situation specific conditions. Current HRA methods mostly rely on expert opinion to select PSF coefficients; the use of empirical data will reduce this reliance on expert opinion and help develop an improved technical basis for HRA. A review of existing HRA methods revealed that fatigue or sleep deprivation was not explicitly included. Literature on sleep deprivation effects on performance (e.g., reaction time) was reviewed. Data were extracted from the identified studies according to the meta-analysis research synthesis method and used to analyze the effect of sleep deprivation on performance. The data were used to estimate effect sizes, performance trends, and error probabilities. The error probabilities under sleep deprived and control conditions were compared, and the resulting probability ratios are suggested for use in informing the selection of PSF coefficients in HRA methods.