Random-interval hybrid reliability analysis for fan performance in the presence of epistemic uncertainty

Abstract

With the aim of reducing fan performance failure, an effective method of fan reliability analysis based on a hybrid uncertainty model was investigated using a combination of orthogonal test design, an approximation model, and computational fluid dynamics (CFD) simulation and analysis technology. By introducing interval uncertainty, this method can effectively solve the problem of quantifying model uncertainty owing to the lack of experimental samples and can greatly expand the applicability of reliability analysis technology in fluid machinery research. Random and interval uncertainty parameters were introduced to describe the fan system and a mixed random-interval reliability analysis model of fan performance was established based on the traditional first-order reliability analysis method. Using CFD analysis of the renormalization group (RNG) model, the flow, pressure, shaft power, efficiency, and other fan performance parameters were calculated, and the performance function was obtained. Owing to the low efficiency of the CFD calculation, the effective iterative algorithm of response surface to solve hybrid reliability model was studied, and the failure probability interval of fan performance was calculated. The method was used to study the fan performance of a rail transit train, and the failure probability, reliability index, and performance parameter sensitivity and uncertainty were analyzed. The influence of parameter uncertainty on system reliability was determined, and the engineering measures necessary to improve the fan’s performance reliability were proposed.