The ever-increasing computational demands in structural dynamics models for the aerospace industry together with the growing complexity of modern engineering systems increase the necessity of developing more and more efficient model reduction techniques. As complexity grows, Model Order Reduction techniques are becoming indispensable tools, enabling the simplification of systems without sacrificing critical dynamic behavior. This thesis investigates a wide array of model order reduction methods, from those coming from the structural dynamics heritage that is those of the family of Component Modes Synthesis, including the Craig-Bampton and Enhanced Craig-Bampton approaches; to methods belonging to different fields, like the balanced truncation and the Krylov subspace-based methods. Aiming to integrate them into the unified mathematical framework of the subspace projection, each method is analysed in terms of its computational efficiency, accuracy in capturing dynamic behavior, and suitability for specific engineering scenarios. Central to this study is the development of hybrid reduction techniques that combine the strengths of individual methods already introduced to achieve improved performances across a range of simulation conditions. Hybrid methods will be incorporated into the same mathematical framework as the other methods presented, allowing for an easier and more rigorous comparison of the potential and limitations of conventional and hybrid approaches. A detailed case study, drawn from aerospace applications, is presented to illustrate how different techniques can be effectively applied to models. It will serve as a standard on which the techniques presented throughout the thesis are tested and numerical results gathered and compared.
Le crescenti richieste computazionali dei modelli di dinamica strutturale specifici dell'industria aerospaziale e la crescente complessità dei sistemi ingegneristici moderni rendono necessario lo sviluppo di tecniche di riduzione dei modelli sempre più efficienti. Con l'aumento della complessità, le tecniche di riduzione dell'ordine del modello stanno diventando strumenti indispensabili, consentendo la semplificazione dei sistemi senza sacrificare parti critiche del comportamento dinamico. Questa tesi indaga un'ampia gamma di metodi di riduzione dell'ordine, da quelli provenienti dal patrimonio della dinamica strutturale, ovvero quelli appartenenti alla famiglia Component Modes Synthesis, tra cui gli approcci Craig-Bampton ed Enhanced Craig-Bampton; a metodi appartenenti a campi diversi, come la balanced truncation ed i metodi basati sui sottospazi di Krylov. Con l'obiettivo di integrarli nel quadro matematico unificato della proiezione su sottospazi, ogni metodo viene analizzato in termini di efficienza computazionale, accuratezza nel catturare il comportamento dinamico e idoneità a specifici scenari ingegneristici. Al centro di questo studio c'è lo sviluppo di tecniche di riduzione ibride che combinano i punti di forza dei singoli metodi già introdotti per ottenere prestazioni migliori in una serie di condizioni di simulazione. I metodi ibridi saranno rappresentati con lo stesso formalismo matematico degli altri metodi presentati, consentendo un confronto più semplice e rigoroso delle potenzialità e dei limiti degli approcci convenzionali e ibridi. Un caso di studio dettagliato, tratto da applicazioni aerospaziali, viene presentato per illustrare come le diverse tecniche possano essere efficacemente applicate ai modelli. Esso servirà da standard per testare le tecniche presentate nel corso della tesi e per raccogliere e confrontare i risultati numerici.
Study on promising reductions techniques for the development of hybrid reduction methods
Andrenacci, Leonardo
2023/2024
Abstract
The ever-increasing computational demands in structural dynamics models for the aerospace industry together with the growing complexity of modern engineering systems increase the necessity of developing more and more efficient model reduction techniques. As complexity grows, Model Order Reduction techniques are becoming indispensable tools, enabling the simplification of systems without sacrificing critical dynamic behavior. This thesis investigates a wide array of model order reduction methods, from those coming from the structural dynamics heritage that is those of the family of Component Modes Synthesis, including the Craig-Bampton and Enhanced Craig-Bampton approaches; to methods belonging to different fields, like the balanced truncation and the Krylov subspace-based methods. Aiming to integrate them into the unified mathematical framework of the subspace projection, each method is analysed in terms of its computational efficiency, accuracy in capturing dynamic behavior, and suitability for specific engineering scenarios. Central to this study is the development of hybrid reduction techniques that combine the strengths of individual methods already introduced to achieve improved performances across a range of simulation conditions. Hybrid methods will be incorporated into the same mathematical framework as the other methods presented, allowing for an easier and more rigorous comparison of the potential and limitations of conventional and hybrid approaches. A detailed case study, drawn from aerospace applications, is presented to illustrate how different techniques can be effectively applied to models. It will serve as a standard on which the techniques presented throughout the thesis are tested and numerical results gathered and compared.File | Dimensione | Formato | |
---|---|---|---|
Master_thesis_Leonardo_Andrenacci.pdf
accessibile in internet per tutti
Dimensione
29.47 MB
Formato
Adobe PDF
|
29.47 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/10589/234847