Parametric Reduced-Order Models of Structural Dynamics for Design, Damage Detection and Structural Optimization.

Parametric Reduced-order Models;Design and Uncertainty;Damage Detection;Structural Optimization;Mechanical Engineering;Engineering;Mechanical Engineering

Hong, Sung-KwonOldham, Kenn Richard ;

University of Michigan

关键词: Parametric Reduced-order Models;    Design and Uncertainty;    Damage Detection;    Structural Optimization;    Mechanical Engineering;    Engineering;    Mechanical Engineering;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/91454/sungkwon_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

The main goal of this work is to develop new modeling and fast reanalysis techniques for predicting the dynamic response of complex structures with parameter variability at component level. The novel models allow for accurate reanalyses and are useful in many applications where the model of the pristine structure may not capture the changes in thesystem-level response due to component level parameter variations. Herein, such modelsare obtained by using a novel approach based on a modified concept of component modesynthesis.Two new types of models are developed. The first type, referred to as multiple component parametric reduced-order models (MC-PROMs), are designed for multiple substructuresmodeled with shell-type elements with parameter variabilities. Three types of parametricvariabilities are considered: (a) geometric variability, (b) structural deformations,and (c) cracks. For validating the MC-PROMs, dynamic responses predicted MC-PROMsare shown to agree very well with results obtained using full-order models.The second type of models are developed to address two important accuracy and performancechallenges of MC-PROMs, namely: (a) the transformation matrix is not alwaysmathematically stable, (b) the Taylor series parameterization techniques do not capturethickness variations of the structure modeled with solid-type elements due to the highlynonlinear dependence on thickness changes. Thus, herein, a new transformation matrixand novel parameterization techniques are proposed.Usual reduced-order models have difficulty handling the interface degrees of freedom.Thus, a novel way of reducing the interface degrees of freedom is proposed also. Thenovel models are referred to as the next-generation parametric reduced-order models (NXPROMs).The vibration responses predicted by NX-PROMs are in excellent agreementwith the responses predicted by full-order models.Both types of PROMs are used in three applications: (a) to establish a robust signal processingapproach for damaged vehicles (crack) with structural variability, (b) to developa damage detection algorithm to identify the size of a crack in complex structures, and(c) to provide a methodology for choosing joining locations for attaching structural reinforcementsonto complex structures. Numerical results for each application are presentedfocusing on a high mobility multipurpose wheeled vehicle with structural variabilities.