Reduced order modelling in stochastically parametered acousto-elastic system using arbitrary PCE based SEREP. (April 2018)

• Record Type: Journal Article
• Authors: Lal, Hridya P.; Jith, Jithin; Gupta, Sayan; Sarkar, Sunetra
• Is Part Of: Probabilistic engineering mechanics. Volume 52(2018)
• Page Start: 1
• Page End: 14
• Publication Date: 2018-04
• Subjects: Stochastic reduced order modelling -- Acousto-elastic system -- Arbitrary polynomial chaos -- SEREP -- Fluid–structure interaction -- Proper orthogonality check; Engineering -- Statistical methods -- Periodicals; Mechanics, Applied -- Statistical methods -- Periodicals; Probabilities -- Periodicals; Ingénierie -- Méthodes statistiques -- Périodiques; Mécanique appliquée -- Méthodes statistiques -- Périodiques; Probabilités -- Périodiques; 620.100727
• DOI: 10.1016/j.probengmech.2018.02.002
• Languages: English
• ISSNs: 0266-8920

Abstract

Abstract: This study focusses on developing a novel reduced order model for dynamic analysis of an acousto-elastic system, such as an impeller housed in a pressurised incompressible fluid filled chamber. Resolving the coupled dynamics of such systems require high-fidelity numerical models for both the structure and the fluid and are typically modelled using finite elements (FE) leading to FE equations with large ordered asymmetric matrices. The parameters for the system are obtained from measurements which are susceptible to uncertain errors and biases. This study proposes the development of a stochastic reduced order FE model based on integrating arbitrary PCE with SEREP. Numerical results reveal that the proposed method results in significant reduction in computational costs for dynamic analysis without compromising on accuracy. The reduced order model is shown to be useful for proper orthogonality check for quantifying the degree of correlation between experimentally observed and model based modes. Highlights: Stochastic ROM for acousto-elastic system developed. SEREP approach applied for asymmetric FE matrices. PCE framework for arbitrary input distribution developed. Savings in Computational cost as high as 95%.