The following study focuses on the dynamic modelling of fluid-structure interaction in thin-walled liquid propulsion stages. Indeed, for launchers using liquid propulsion, the mass of liquid propellant is a high percentage of the total mass, therefore its coupling to the launcher structure must not be neglected to accurately represent the dynamic behavior of the {launcher+spacecraft} assembly. Previously, a software called FABE (Fluid Analysis with Boundary Elements) was used within the European Space Agency to model liquid propellant tanks during coupled loads analyses for missions launched on Ariane-5. However, this software is no longer supported, an alternative is needed since Vega-E will feature a liquid propulsion third stage. Since 2019, the real coupled modes method is available within MSC Nastran allowing to model liquid-filled tanks with a finite element model and accurately represent the fluid-structure interaction. The main objective of this study is to perform a first validation and some tuning and testing of this method, considering various benchmarks: from a simple cubic tank filled with water to a model of Ariane-5 EPC. The first topic is the incompressibility of the fluid. A condition is given by MSC Nastran - and has been verified in this study - in order for the fluid to be considered as incompressible. For cases where this condition is not respected - typically for larger tanks and at higher excitation frequencies - a generic MATLAB code has been developed - and verified - in order to generate a Multi-Point Constraint equation that enforces the incompressibility of the fluid explicitly. A first validation of the real coupled modes method was performed, by running normal modes analyses on various benchmarks and comparing the mode sets to the ones obtained with FABE and the ones computed analytically. Consistent results were obtained. With the real coupled modes method, superelements can be created from finite element models of the liquid-filled tanks. These superelements, based on the Craig-Bampton reduction method, allow to reduce the size of models and so the computational costs. Different frequency and transient responses analyses were performed (using the direct and modal method) on various benchmarks to compare the response spectra (displacement, acceleration, element forces) obtained when considering the finite element models to the ones obtained when considering the superelement models. In the meantime, different damping models were defined, in particular a decoupled equivalent viscous damping that can be used for superelements. Overall, the different analyses results show consistency between the finite element and superelement models and the different damping methods considered. The response spectra, computed with decoupled equivalent viscous damping, could exhibit overly damped responses in free-free condition but the results are still considered correct. However, some issues arose with the modal method. Because of the missing high-frequency truncated modes, the input forcing function could not be recovered correctly. Moreover, for the finite element models, the steady-state acceleration recovered with the modal method was incorrect because of the treatment of the rigid-body modes and constant-pressure modes of the fluids by MSC Nastran during the analysis. The key contributions and sources that form the basis of this study are clearly cited within each chapter, with particular attention given to their roles in guiding the methodology and analysis.
Lo studio si concentra sulla modellazione dinamica dell’interazione fluido-struttura negli stadi a propulsione liquida a pareti sottili. Infatti, per questo tipo di lanciatori, la massa del propellente liquido rappresenta una percentuale importante quella totale, pertanto, il suo accoppiamento con la struttura del lanciatore non deve essere trascurato per simulare accuratamente il comportamento dinamico del sistema {lanciatore+veicolo spaziale}. In passato, l'Agenzia Spaziale Europea disponeva del software FABE (Fluid Analysis with Boundary Elements) per la modellizazione di serbatoi a propellente liquido. In particolare, veniva impiegato per le analisi dei carichi accoppiati per missioni lanciate con Ariane-5. Tuttavia, poiché questo software non è più supportato, è stato necessario trovare un’alternativa per le simulazioni del lanciatore Vega-E e il suo terzo stadio a propulsione liquida. Dal 2019, è disponibile in MSC Nastran il metodo dei modi accoppiati reali, che consente di modellare i serbatoi a liquido con un modello agli elementi finiti e rappresentarne accuratamente l’interazione fluido-struttura. L’obiettivo principale di questo studio è eseguire una prima validazione di questo metodo, condurre alcuni aggiustamenti e infine testarlo, considerando vari benchmark: a partire da un semplice serbatoio cubico riempito d’acqua fino ad un modello dell’EPC di Ariane-5. La prima parte del lavoro riguarda l’incomprimibilità del fluido. Una condizione è fornita da MSC Nastran - e verificata in questo studio - affinché il fluido sia considerato incomprimibile dal software. Per i casi in cui questa condizione non è rispettata - tipicamente per serbatoi più grandi e a frequenze di eccitazione più elevate - è stato sviluppato, e verificato, un codice MATLAB generico per creare un’equazione di vincolo multipunto che impone esplicitamente l’incomprimibilità del fluido. In secondo luogo, è stata eseguita una prima validazione del metodo dei modi accoppiati reali, attraverso un'analisi dei modi normali sui vari benchmark e confrontando i set di modi con quelli ottenuti con FABE e quelli calcolati analiticamente. I risultati che ne derivano risultano coerenti. È possibile inoltre simulare con modelli di superelementi i modelli agli elementi finiti dei serbatoi pieni di liquido. Questi superelementi, basati sul metodo di riduzione Craig-Bampton, consentono di ridurre le dimensioni dei modelli e, di conseguenza il tempo computazionale total dell'analisi. Sono state eseguite diverse simulazioni su vari benchmark: sia risposte in frequenza, sia analisi transitorie (con metodo diretto e modale) al fine di confrontare gli output (spostamento, accelerazione e forze) ottenuti dai modelli agli elementi finiti, con quelli ottenuti considerando i modelli di superelementi. In parallelo, sono stati definiti diversi tipi di smorzamento, in particolare uno smorzamento viscoso equivalente disaccoppiato utilizzato per i superelementi. Nel complesso, i risultati delle diverse analisi mostrano coerenza tra i modelli agli elementi finiti, i modelli di superelementi e i vari smorzamenti considerati. Gli spettri di smorzamento viscoso equivalente disaccoppiato potrebbero mostrare risposte eccessivamente smorzate in condizioni non vincolate, ma i risultati vengono comunque considerati corretti. Tuttavia, con il metodo modale sono emersi alcuni problemi. A causa dei modi troncati ad alta frequenza, la funzione forzante in ingresso non viene recuperata correttamente. Inoltre, per i modelli agli elementi finiti, l’accelerazione in stato stazionario risulta errata, a causa di come MSC Nastran gestisce, durante l’analisi, i modi di corpo rigido e i modi di pressione costante dei fluidi. I contributi chiave e le fonti utilizzate durante questo studio sono citati in ogni capitolo, con particolare attenzione al loro ruolo nella guida della metodologia e dell’analisi.
Dynamic modelling of fluid-structure interaction in thin-walled liquid propulsion stages with the real coupled modes method
MAUGIS, NELLY MONIQUE PAULETTE
2023/2024
Abstract
The following study focuses on the dynamic modelling of fluid-structure interaction in thin-walled liquid propulsion stages. Indeed, for launchers using liquid propulsion, the mass of liquid propellant is a high percentage of the total mass, therefore its coupling to the launcher structure must not be neglected to accurately represent the dynamic behavior of the {launcher+spacecraft} assembly. Previously, a software called FABE (Fluid Analysis with Boundary Elements) was used within the European Space Agency to model liquid propellant tanks during coupled loads analyses for missions launched on Ariane-5. However, this software is no longer supported, an alternative is needed since Vega-E will feature a liquid propulsion third stage. Since 2019, the real coupled modes method is available within MSC Nastran allowing to model liquid-filled tanks with a finite element model and accurately represent the fluid-structure interaction. The main objective of this study is to perform a first validation and some tuning and testing of this method, considering various benchmarks: from a simple cubic tank filled with water to a model of Ariane-5 EPC. The first topic is the incompressibility of the fluid. A condition is given by MSC Nastran - and has been verified in this study - in order for the fluid to be considered as incompressible. For cases where this condition is not respected - typically for larger tanks and at higher excitation frequencies - a generic MATLAB code has been developed - and verified - in order to generate a Multi-Point Constraint equation that enforces the incompressibility of the fluid explicitly. A first validation of the real coupled modes method was performed, by running normal modes analyses on various benchmarks and comparing the mode sets to the ones obtained with FABE and the ones computed analytically. Consistent results were obtained. With the real coupled modes method, superelements can be created from finite element models of the liquid-filled tanks. These superelements, based on the Craig-Bampton reduction method, allow to reduce the size of models and so the computational costs. Different frequency and transient responses analyses were performed (using the direct and modal method) on various benchmarks to compare the response spectra (displacement, acceleration, element forces) obtained when considering the finite element models to the ones obtained when considering the superelement models. In the meantime, different damping models were defined, in particular a decoupled equivalent viscous damping that can be used for superelements. Overall, the different analyses results show consistency between the finite element and superelement models and the different damping methods considered. The response spectra, computed with decoupled equivalent viscous damping, could exhibit overly damped responses in free-free condition but the results are still considered correct. However, some issues arose with the modal method. Because of the missing high-frequency truncated modes, the input forcing function could not be recovered correctly. Moreover, for the finite element models, the steady-state acceleration recovered with the modal method was incorrect because of the treatment of the rigid-body modes and constant-pressure modes of the fluids by MSC Nastran during the analysis. The key contributions and sources that form the basis of this study are clearly cited within each chapter, with particular attention given to their roles in guiding the methodology and analysis.File | Dimensione | Formato | |
---|---|---|---|
2024_12_Maugis_Thesis.pdf
non accessibile
Dimensione
10.64 MB
Formato
Adobe PDF
|
10.64 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/10589/230450