Rigorous uncertainty quantification with correlated random variables from multiple sources. (March 2021)

• Record Type: Journal Article
• Authors: Liu, Xi; Wang, Rongqiao; Hu, Dianyin; Chen, Gaoxiang
• Is Part Of: Engineering failure analysis. Volume 121(2021)
• Publication Date: 2021-03
• Subjects: Multidimensional parallelepiped model -- Correlation analysis -- Quantification of margins and uncertainties -- Turbine disc -- Fatigue life; System failures (Engineering) -- Periodicals; Fracture mechanics -- Periodicals; Reliability (Engineering) -- Periodicals; Pannes -- Périodiques; Rupture, Mécanique de la -- Périodiques; Fiabilité -- Périodiques; Fracture mechanics; Reliability (Engineering); System failures (Engineering); Periodicals; Electronic journals; 620.112
• DOI: 10.1016/j.engfailanal.2020.105114
• Languages: English
• ISSNs: 1350-6307

Abstract

Highlights: A method of quantification of margins and uncertainties for structures with correlated random variables. The multidimensional parallelepiped model is applied to quantify the uncertainties of correlated random variables. Structure uncertainties is quantified through rigorous probability bounds by using the McDiarmid's inequality. The confidence factor is used as a reliability measure to certify the design specifications of structure of turbine disc. Abstract: A method of quantification of margins and uncertainties (QMU) is developed for structures with correlated random variables originated from multiple sources. The QMU analysis aims to certify the concerned structure with a prescribed confidence factor (CF), and further determine an adequate design margin. After partitioning correlated random variables into different groups based on their uncertain sources, the multidimensional parallelepiped model is established to describe the uncertainty domain of these correlated random variables. Based on the grouped independent random vectors, McDiarmid's inequality is applied to realize rigorous uncertainty quantification of the structural performance. The computation of structural uncertainty is formulated as an optimization problem of variables within the multidimensional parallelepiped uncertainty domain. The QMU framework is then established and applied to low cycle fatigue life assessment of aero-engine turbine disc made of GH720Li superalloy. Stochastic parameters involving loads, geometries, and material properties, are chosen as input random variables. It is demonstrated that the turbine disc can be rigorously certified under the given design specification. By contrast, it will provide a wrong certification decision when the correlation of variables is not considered. The assessment provides an effective basis for the design with enough confidence in engineering applications from the perspective of QMU analysis.