This thesis aims at developing novel methodologies to mitigate the limitations of the state-of-the art State-Space Substructuring (SSS) techniques and of the state-of-the-art approaches to estimate state-space models from experimentally acquired data. Moreover, the document also aims at showing the benefit of using state-space models and SSS methods to tackle time-domain Transfer Path Analysis (TPA) applications. In this thesis, the dual SSS formulation (also, denoted Lagrange Multiplier State-Space Substructuring method (LM-SSS)) is extended to perform dynamic substructuring (DS) operations with displacement and velocity models. A novel coupling form, tagged Unconstrained Coupling Form (UCF), tailored to perform DS operations with LM-SSS is also derived. Then, novel post-processing procedures to eliminate the redundant states originated from DS operations implemented with LM-SSS are proposed. Afterwards, the LM-SSS via compatibility relaxation method is developed to allow the inclusion of connecting elements (CEs) into SSS operations via compatibility relaxation. Post-processing procedures to eliminate the extra states originated from coupling operations implemented with this method are also derived. A novel method to impose Newton's second law on state-space models that does not rely on the use of undamped residual compensation modes (RCMs) is then introduced, while a novel strategy to impose stability on unstable coupled models is proposed. The methodologies developed are validated on numerical and experimental substructuring test cases. The identified state-space models representative of the substructures/assemblies under study showed to be highly accurate, while respecting Newton's second law. Reliable decoupling/coupling results are also obtained with LM-SSS enhanced by the developed post-processing strategies. Moreover, the LM-SSS via compatibility relaxation method showed to be accurate, provided that the CEs involved in the coupling operations verify the underlying assumptions of the Inverse Substructuring (IS) method. On top of this, the proposed approach to impose stability on unstable coupled models showed to be capable of determining an accurate stable coupled model from an unstable coupled model resultant from several DS operations. Finally, novel applications of LPV models in TPA are proposed and validated on an analytical assembly made of a source and a passive system. The dynamics of both systems is time-dependent. In a first instance, an interpolating LPV model describing the dynamics of the passive system is used with the matrix-inverse classical TPA method to estimate the time-domain connecting force acting on the interface between the source and the passive system. It is found that the estimated connecting force is reliable and can be used to accurately predict the responses at DOFs belonging to the passive system. Then, interpolated models computed from the LPV models representative of the source and passive systems are coupled at each time-sample by using LM-SSS to obtain coupled models characterizing the assembly. These coupled models are used with the in-situ component-based TPA method to estimate the time-domain equivalent force of the source. It is found that the estimated equivalent force is accurate and can be used to estimate the responses on the passive side of assemblies made of the same source linked to any passive system.
Questa tesi mira a sviluppare nuove metodologie per mitigare i limiti dello stato dell’arte nelle tecniche di State-Space-Substructuring (SSS) e negli approcci per stimare i modelli state-space dai dati acquisiti sperimentalmente. Inoltre, si vuole dimostrare il vantaggio dell'uso dei modelli state-space e dei metodi SSS per affrontare applicazioni di Transfer Path Analysis (TPA) nel dominio del tempo. In questa tesi, la formulazione duale SSS (indicata anche come metodo Lagrange Multiplier State-Space Substructuring (LM-SSS)) viene estesa per eseguire operazioni di Dynamic Substructuring (DS) con modelli di spostamento e velocità. Viene inoltre derivata una nuova forma di accoppiamento, denominata Unconstrained Coupling Form (UCF), adatta a eseguire operazioni DS con LM-SSS. Vengono poi proposte nuove procedure di post-processing per eliminare gli stati ridondanti originati dalle operazioni DS implementate con LM-SSS. In seguito, viene sviluppato il metodo “LM-SSS via compatibility relaxation” per consentire l'inclusione di elementi di connessione (CE) nelle operazioni SSS tramite l’approccio di “compatibility relaxation”. Vengono inoltre derivate procedure di post-processing per eliminare gli stati extra originati dalle operazioni di accoppiamento ottenute con questo metodo. Successivamente, viene introdotto un nuovo metodo per imporre la seconda legge di Newton ai modelli state-space che non si basa sull'uso di modi di compensazione residua (RCM) non smorzati, mentre viene proposta una nuova strategia per imporre la stabilità ai modelli accoppiati instabili. Le metodologie sviluppate sono state validate in applicazioni di substructuring numeriche e sperimentali. I modelli state-space identificati, rappresentativi delle sottostrutture/assiemi oggetto di studio, si sono dimostrati altamente accurati, rispettando la seconda legge di Newton. Risultati affidabili di disaccoppiamento/accoppiamento sono stati ottenuti anche con LM-SSS migliorato dalle procedure di post-processing sviluppate. Inoltre, il metodo “LM-SSS via compatibility relaxation” si è dimostrato accurato, a condizione che gli CE coinvolti nelle operazioni di accoppiamento verifichino le ipotesi di base del metodo Inverse Substructuring (IS). Infine, sono state proposte nuove applicazioni dei modelli LPV nella TPA e sono state convalidate su un modello analitico composto da una sorgente e da un sistema passivo. La dinamica di entrambi i sistemi è tempovariante. Un modello LPV interpolante, che descrive la dinamica del sistema passivo, viene utilizzato con il metodo “matrix-inverse classical TPA” per stimare la forza di connessione nel dominio del tempo che agisce sull'interfaccia tra la sorgente e il sistema passivo. Si scopre che la forza di connessione stimata è affidabile e può essere utilizzata per prevedere con precisione le risposte alle DOF appartenenti al sistema passivo. Quindi, i modelli interpolati calcolati dai modelli LPV rappresentativi della sorgente e dei sistemi passivi vengono accoppiati a ogni campione temporale utilizzando LM-SSS per ottenere modelli accoppiati che caratterizzano l'insieme. Questi modelli accoppiati vengono utilizzati con il metodo “in-situ component-based TPA” per stimare la forza equivalente della sorgente nel dominio del tempo. Si è constatato che la forza equivalente stimata è accurata e può essere utilizzata per stimare le risposte sul lato passivo di assiemi costituiti dalla stessa sorgente collegata a qualsiasi sistema passivo.
Exploring State-Space Substructuring for time-domain dynamic substructuring applications
DA SILVA OLIVEIRA DIAS, Rafael
2024
Abstract
This thesis aims at developing novel methodologies to mitigate the limitations of the state-of-the art State-Space Substructuring (SSS) techniques and of the state-of-the-art approaches to estimate state-space models from experimentally acquired data. Moreover, the document also aims at showing the benefit of using state-space models and SSS methods to tackle time-domain Transfer Path Analysis (TPA) applications. In this thesis, the dual SSS formulation (also, denoted Lagrange Multiplier State-Space Substructuring method (LM-SSS)) is extended to perform dynamic substructuring (DS) operations with displacement and velocity models. A novel coupling form, tagged Unconstrained Coupling Form (UCF), tailored to perform DS operations with LM-SSS is also derived. Then, novel post-processing procedures to eliminate the redundant states originated from DS operations implemented with LM-SSS are proposed. Afterwards, the LM-SSS via compatibility relaxation method is developed to allow the inclusion of connecting elements (CEs) into SSS operations via compatibility relaxation. Post-processing procedures to eliminate the extra states originated from coupling operations implemented with this method are also derived. A novel method to impose Newton's second law on state-space models that does not rely on the use of undamped residual compensation modes (RCMs) is then introduced, while a novel strategy to impose stability on unstable coupled models is proposed. The methodologies developed are validated on numerical and experimental substructuring test cases. The identified state-space models representative of the substructures/assemblies under study showed to be highly accurate, while respecting Newton's second law. Reliable decoupling/coupling results are also obtained with LM-SSS enhanced by the developed post-processing strategies. Moreover, the LM-SSS via compatibility relaxation method showed to be accurate, provided that the CEs involved in the coupling operations verify the underlying assumptions of the Inverse Substructuring (IS) method. On top of this, the proposed approach to impose stability on unstable coupled models showed to be capable of determining an accurate stable coupled model from an unstable coupled model resultant from several DS operations. Finally, novel applications of LPV models in TPA are proposed and validated on an analytical assembly made of a source and a passive system. The dynamics of both systems is time-dependent. In a first instance, an interpolating LPV model describing the dynamics of the passive system is used with the matrix-inverse classical TPA method to estimate the time-domain connecting force acting on the interface between the source and the passive system. It is found that the estimated connecting force is reliable and can be used to accurately predict the responses at DOFs belonging to the passive system. Then, interpolated models computed from the LPV models representative of the source and passive systems are coupled at each time-sample by using LM-SSS to obtain coupled models characterizing the assembly. These coupled models are used with the in-situ component-based TPA method to estimate the time-domain equivalent force of the source. It is found that the estimated equivalent force is accurate and can be used to estimate the responses on the passive side of assemblies made of the same source linked to any passive system.File | Dimensione | Formato | |
---|---|---|---|
Tesi_Da_Silva_Oliveira_Dias.pdf
accesso aperto
Dimensione
17.34 MB
Formato
Adobe PDF
|
17.34 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/165887
URN:NBN:IT:UNIVPM-165887