Propagation of Certainty Levels

Abstract

Uncertainty in basic structural parameters is characterized by objective and subjective information. A method was developed to quantify this uncertainty and propagate its effect through a structural reliability analysis. Certainty levels are established on the resulting failure probabilities. The concept of confidence intervals from classical statistics is extended to include both objective uncertainty, e.g., that due to limited sample size, and subjective uncertainty, e.g., data interpretation and extrapolation. The uncertainty is propagated by separating the objective and subjective information in the basic input random parameters and, also, in the calculated random probabilities. An example truss problem illustrates an application of the method.