Uncertainty Analysis on a Pump Assembly Using Component Mode Synthesis

Abstract eng:
The combined use of the component mode synthesis and the probabilistic uncertainty analysis is studied on an industrial structure. Component mode synthesis (CMS) methods are useful for the analysis of structures that are built-up of several components. The components are modeled individually and their dynamic models, represented by modal bases, are assembled to produce a much smaller model of the whole structure. In the component mode synthesis framework, uncertainties in properties can be naturally and straightforwardly introduced at the component level; properties concerned can be either component or joint model parameters, such as material characteristics, or component modal characteristics. Variations on component modal characteristics both reflect the model and parametric component uncertainties. The structure considered is a free-free two pump component assembly. The whole pump was designed by SULZER Pumps and is used in EDF thermal units. The two frame and bearing support components are assembled using two bolted joints. The variables of interest are the eigenfrequencies of the built-up structure. The uncertain properties are the eigenfrequencies of the two sub-structures: they are modelled in a probabilistic frame as independent random variables defined on intervals determined as a result of 2008-2010 SICODYN international benchmark. A Monte Carlo approach with 1000 runs is applied in order to estimate eigenfrequency statistics on the built-up structure. The following tendencies can be shown: - the coefficient of variation of the assembly eigenfrequency values is much lower than the coefficient of variation introduced in the sub-structure eigenfrequency values; - the first assembly eigenfrequencies depend on the frame eigenfrequencies and not on the bearing support eigenfrequencies, which is a stiffer component. The present research work has been carried out within the FUI 2012-2015 SICODYN project.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket