The present work introduces a new strategy for performing an exact tuning of the parameters governing the Von Karman - Pao isotropic energy spectrum for generating a stochastic turbulent velocity field. The tuning is carried out by numerically solving a single-variable equation defined through the confluent hypergeometric function of second kind. This procedure effectively captures the entire turbulent energy content for each point of the domain, yielding distinct and optimized outcomes based on the local turbulence Reynolds number. This stands in contrast to the approach commonly adopted in the same applications, where the spectrum parameters are calculated under the assumption of infinite Reynolds number at each point, assumption not always valid and therefore potentially leading to significant errors, in particular close to physical walls. Additionally, the strategy used to select the parameters defining the numerical integration of the spectrum is innovative as well, enabling an optimal balance between error committed and computational cost. The synthesis of the turbulent velocity field represents the most peculiar step of the Stochastic Noise Generation and Radiation (SNGR) method, in which, starting from the results provided by a RANS or URANS simulation, the aeroacoustic sources are reconstructed and then the radiated acoustic field is computed to estimate the farfield broadband noise. Stochastic Noise Generation (SNG) represents a lower-fidelity but much more computationally efficient strategy compared to directly obtaining the acoustic sources through a LES simulation or a hybrid RANS-LES simulation such as DES/DDES. The first part of the present work illustrates the significant benefits brought by the overall tuning procedure for the cases of a round jet or a 2-D NACA0012 airfoil at AOA = 8°. Subsequently, a direct comparison between the temporal history of the velocity field generated through SNG post-processing (applied both to a RANS and a URANS) and that obtained via a DDES is presented for a 3-D NACA0021 airfoil at AOA = 17◦. Despite having a good correspondence of the signals from a statistical point of view, the frequency behavior of those obtained with SNG remains the most limiting aspect of this method in its current version. The simulations have been performed with the open-source software SU2.
Il presente lavoro di tesi introduce una nuova strategia per effettuare il tuning esatto dei parametri che governano lo spettro energetico di Von Karman - Pao, utilizzato poi per la generazione di un campo di velocità turbolento stocastico. Il tuning viene operato risolvendo numericamente un’equazione ad una sola incognita definita tramite la funzione ipergeometrica confluente di secondo tipo. Tale procedura riesce a cogliere efficacemente l’intero contenuto energetico turbolento in ciascun punto del dominio, garantendo un esito ottimizzato e differente a seconda di un numero di Reynolds propriamente definito. In questo si discosta dall’approccio comunemente adoperato in letteratura, dove i parametri dello spettro vengono calcolati sotto l’ipotesi di numero di Reynolds infinito in ciascun punto, ipotesi però non sempre valida e dunque potenzialmente associata a notevoli errori, in particolare a ridosso di pareti fisiche. Inoltre, la strategia utilizzata per selezionare i parametri che definiscono l’integrazione numerica dello spettro risulta anch’essa innovativa, consentendo di ottenere il miglior compromesso fra errore commesso e costo computazionale. La sintesi di un campo di velocità turbolento rappresenta lo step più caratteristico del metodo Stochastic Noise Generation and Radiation (SNGR), nel quale a partire dai risultati forniti da una simulazione RANS/URANS le sorgenti aeroacustiche vengono ricostruite e successivamente viene calcolato il campo acustico radiato da esse. Il metodo Stochastic Noise Generation (SNG) rappresenta una strategia meno affidabile ma molto più efficiente dal punto di vista computazionale rispetto all’ottenimento diretto delle sorgenti tramite una simulazione LES o simulazioni ibride quali DES/DDES. La prima parte di questa tesi illustra i significativi benefici apportati dalla nuova procedura complessiva di tuning nei casi di un getto assialsimmetrico e di un NACA0012 bidimensionale a AOA = 8°. In seguito, un confronto diretto fra la storia temporale del campo di velocità generato tramite SNG (applicato sia ad una RANS che ad una URANS) e quello ottenuto tramite una DDES viene presentato per un NACA0021 tridimensionale a AOA = 17°. Nonostante una buona corrispondenza dei segnali dal punto di vista statistico, il comportamento in frequenza di quelli ottenuti con SNG rimane l’aspetto più critico di questo metodo nella sua versione attuale. Le simulazioni sono state effettuate tramite il software open-source SU2.
Exact tuning of Von Karman - Pao energy spectrum for stochastic noise generation in turbulent flows
Guglielmi, Riccardo
2022/2023
Abstract
The present work introduces a new strategy for performing an exact tuning of the parameters governing the Von Karman - Pao isotropic energy spectrum for generating a stochastic turbulent velocity field. The tuning is carried out by numerically solving a single-variable equation defined through the confluent hypergeometric function of second kind. This procedure effectively captures the entire turbulent energy content for each point of the domain, yielding distinct and optimized outcomes based on the local turbulence Reynolds number. This stands in contrast to the approach commonly adopted in the same applications, where the spectrum parameters are calculated under the assumption of infinite Reynolds number at each point, assumption not always valid and therefore potentially leading to significant errors, in particular close to physical walls. Additionally, the strategy used to select the parameters defining the numerical integration of the spectrum is innovative as well, enabling an optimal balance between error committed and computational cost. The synthesis of the turbulent velocity field represents the most peculiar step of the Stochastic Noise Generation and Radiation (SNGR) method, in which, starting from the results provided by a RANS or URANS simulation, the aeroacoustic sources are reconstructed and then the radiated acoustic field is computed to estimate the farfield broadband noise. Stochastic Noise Generation (SNG) represents a lower-fidelity but much more computationally efficient strategy compared to directly obtaining the acoustic sources through a LES simulation or a hybrid RANS-LES simulation such as DES/DDES. The first part of the present work illustrates the significant benefits brought by the overall tuning procedure for the cases of a round jet or a 2-D NACA0012 airfoil at AOA = 8°. Subsequently, a direct comparison between the temporal history of the velocity field generated through SNG post-processing (applied both to a RANS and a URANS) and that obtained via a DDES is presented for a 3-D NACA0021 airfoil at AOA = 17◦. Despite having a good correspondence of the signals from a statistical point of view, the frequency behavior of those obtained with SNG remains the most limiting aspect of this method in its current version. The simulations have been performed with the open-source software SU2.File | Dimensione | Formato | |
---|---|---|---|
2024_04_Guglielmi_Tesi_01.pdf
solo utenti autorizzati a partire dal 13/03/2027
Descrizione: Testo della tesi
Dimensione
23.03 MB
Formato
Adobe PDF
|
23.03 MB | Adobe PDF | Visualizza/Apri |
2024_04_Guglielmi_ExecutiveSummary_02.pdf
solo utenti autorizzati a partire dal 13/03/2027
Descrizione: Executive Summary
Dimensione
2.44 MB
Formato
Adobe PDF
|
2.44 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/10589/219224