One of the most important limiting factors in the study of the dynamic response of a structure is represented by the computational effort. A comprehensive dynamic analysis requires demanding computer performances and a lot of time to be carried out, especially if the Finite Element (FE) model of the structure presents a large number of Degrees of Freedom. This issue can be so serious that the adoption of the full FE model may be computationally unfeasible. A solution to this problem is offered by Model Order Reduction (MOR) techniques. However, replicating the original system dynamics accurately enough with a reduced model, whose size can be orders of magnitude lower than the original size, is very challenging, as it is proved by the large amount of research on this topic. This work focuses on the analysis and comparison of a set of MOR methods belonging to two different families. The first family encompasses the Component Mode Synthesis or dynamic substructuring methods, developed within the framework of structural dynamics, while the second family includes techniques derived from systems and control theory. The MOR methods are applied to the FE model of the ESA Micro Vibrations Measurement System (MVMS) with the aim of determining which one provides the most accurate reduced-order system. The performance of each technique is assessed from the outcome of modal and dynamic response analysis. A general comparison is also performed in this work in order to characterize the different techniques, highlighting their strengths and limits in an investigation of a wider scope. The aim is to compare the methods performances with the ones of the Craig-Bampton (CB) reduction, which represents the standard approach employed in industry. The focus is on evaluating the possibility of employing MOR techniques of higher level of accuracy than CB in the space engineering industrial practice. The main contributions provided by this work regard this last topic and include the development of solutions aimed at overcoming possible practical problems that would arise from the replacement of the classical Craig-Bampton method with alternative techniques.
Uno dei più rilevanti fattori limitanti nello studio della risposta dinamica di una struttura è rappresentato dal costo computazionale. Un'esaustiva analisi dinamica richiede performance di calcolo elevate e molto tempo per essere eseguita, specialmente se il modello agli elementi finiti (FE) della struttura presenta un elevato numero di gradi di libertà. In molti casi, questa problematica è così seria che l'utilizzo del modello originale diventa infattibile o estremamente sconveniente. Le tecniche di riduzione d'ordine del modello (MOR) rappresentano una soluzione a questo problema. Tuttavia, riuscire a replicare la dinamica del sistema originale in modo sufficientemente accurato con un modello ridotto, le cui dimensioni possono essere ordini di grandezza inferiori di quelle dell'originale, è molto complicato, come dimostra la vastità della ricerca dedicata a questo argomento. Questo lavoro si concentra sull'analisi e comparazione di un insieme di metodi appartenenti a due diverse famiglie. La prima è costituita dai metodi Component Mode Synthesis o di sottostrutturazione dinamica, sviluppati nel contesto della dinamica strutturale, mentre la seconda famiglia include tecniche derivanti dalla teoria del controllo. I metodi MOR sono applicati al modello FE dell'ESA Micro Vibrations Measurement System (MVMS) per determinare quali generino i modelli ridotti più accurati. La performance di ciascuna tecnica è valutata basandosi sull'analisi modale e della risposta dinamica. Per caratterizzare le varie tecniche, si è eseguita anche una comparazione generale, evidenziandone i punti di forza e di debolezza in un'analisi di più ampio respiro. L'obiettivo è quello di comparare le performance dei vari metodi con quelle della riduzione di Craig-Bampton (CB), che rappresenta lo standard utilizzato nella pratica industriale. Il focus è posto sulla valutazione della possibilità di utilizzare i metodi MOR capaci di restituire un maggior livello di accuratezza di CB nella pratica industriale dell'ingegneria spaziale. I maggiori contributi derivanti da questo lavoro riguardano quest'ultima tematica e includono lo sviluppo di soluzioni mirate a superare possibili problematiche che nascerebbero dalla sostituzione del classico metodo di Craig-Bampton con tecniche alternative.
A comparison of model order reduction techniques from structural dynamics and control theory: application to the ESA Micro Vibrations Measurement System
Peri, Leonardo
2021/2022
Abstract
One of the most important limiting factors in the study of the dynamic response of a structure is represented by the computational effort. A comprehensive dynamic analysis requires demanding computer performances and a lot of time to be carried out, especially if the Finite Element (FE) model of the structure presents a large number of Degrees of Freedom. This issue can be so serious that the adoption of the full FE model may be computationally unfeasible. A solution to this problem is offered by Model Order Reduction (MOR) techniques. However, replicating the original system dynamics accurately enough with a reduced model, whose size can be orders of magnitude lower than the original size, is very challenging, as it is proved by the large amount of research on this topic. This work focuses on the analysis and comparison of a set of MOR methods belonging to two different families. The first family encompasses the Component Mode Synthesis or dynamic substructuring methods, developed within the framework of structural dynamics, while the second family includes techniques derived from systems and control theory. The MOR methods are applied to the FE model of the ESA Micro Vibrations Measurement System (MVMS) with the aim of determining which one provides the most accurate reduced-order system. The performance of each technique is assessed from the outcome of modal and dynamic response analysis. A general comparison is also performed in this work in order to characterize the different techniques, highlighting their strengths and limits in an investigation of a wider scope. The aim is to compare the methods performances with the ones of the Craig-Bampton (CB) reduction, which represents the standard approach employed in industry. The focus is on evaluating the possibility of employing MOR techniques of higher level of accuracy than CB in the space engineering industrial practice. The main contributions provided by this work regard this last topic and include the development of solutions aimed at overcoming possible practical problems that would arise from the replacement of the classical Craig-Bampton method with alternative techniques.File | Dimensione | Formato | |
---|---|---|---|
Master_Thesis_Leonardo_Peri.pdf
accessibile in internet per tutti
Descrizione: Master Thesis
Dimensione
14.49 MB
Formato
Adobe PDF
|
14.49 MB | Adobe PDF | Visualizza/Apri |
Executive_Summary_Leonardo_Peri.pdf
accessibile in internet per tutti
Descrizione: Executive Summary
Dimensione
1.45 MB
Formato
Adobe PDF
|
1.45 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/198762