The standard methodology to model turbulent flows in multidimensional internal combustion engine (ICE) simulations is still the Unsteady Reynolds Averaged Navier Stokes (URANS) approach. It allows both sufficient accuracy and limited computational cost modelling the whole turbulence spectrum. However, for several applications, RANS models cannot provide the degree of accuracy required during the design process (i.e. aero-acoustics simulations and massive separation phenomena). The rapid spread of cost-effective HPC resources has increased the use of Large Eddies Simulations (LES) turbulence treatment, especially for the analysis of cold flow, combustion, spray and cycle to cycle variability (CCV). LES approach offers accurate spatial and temporal descriptions of turbulence, resolving only the large scales (problem-dependent, anisotropic, and full of energy) and modelling the small ones (universal and isotropic). However, the application of LES models for High Reynolds (HR) number simulations, is still limited because of the high computational costs. For free shear flows, the number of grid points required for a LES simulation scales according to O(Re0.4). In wall-bounded flows, the turbulence length scale near the walls becomes very small compared to the boundary layer thickness and the high gradients impose a very small computational grid size in the wall-normal direction. Moreover, to resolve the small isotropic scales in the boundary layer, a high-resolution grid in the tangential direction is required as well, leading to more rigid scaling, i.e. O(Re1.8). This kind of LES approach is called Wall-Resolved LES (WRLES) and it is prohibitively expensive even for moderate HR number applications, such as ICEs simulations. A possible alternative to WRLES is represented by hybrid URANS/LES models, which are more and more diffused for HR number applications and include a great number of subclasses, such as Detached Eddy Simulation (DES) and its variants, Scale Resolving Simulation (SAS) and the more recent Stress-Blended Eddy Simulation (SBES) models. Focusing on DES, it is based on the principle that large eddies are resolved away from walls while the small ones, in the boundary layer, are modelled with RANS closures which have decades of validation in boundary layer description. It requires a fine grid at walls only in the wall-normal direction, where y+≈1 is required for the RANS Low Reynolds turbulence model, and a coarse grid in the tangential direction. It was originally proposed for the RANS Spalart-Allmaras (S-A) one equation model, but it can be applied to any others eddies viscosity models in which the switching criterion acts on either turbulent viscosity or turbulent kinetic energy. In general, all DES formulations performance to reduce the background RANS approach to a Smagorinsky-type SGS one-equation model (or similar such as the Yoshizawa one). In this framework, the present work aims at finding a stable and robust methodology to apply the Zonal-DES hybrid turbulence treatment in CFD-3D ICEs simulations. The proposed methodology acts on both the production and dissipation terms of the turbulent kinetic energy transport equation and the switching criteria applies to all grid nodes, both near-walls and bulk flows, without any kind of interface between the two approaches. A dedicated near-wall treatment is added to preserve the well-established HR approach in the first near-wall cell, based on the use of wall function. The methodology has been tested, at first, on a reference test case widely reported in literature and subsequently for multi-cycle simulations of two reference single-cylinder optical research engines. The engine simulations results demonstrate consistency and efficiency of the proposed methodology, which is a suitable candidate for affordable scale-resolving analyses aiming at evaluating turbulence-governed phenomena in direct-injection engine.

La metodologia standard utilizzata per modellare flussi turbolenti multidimensionali in simulazioni di motori a combustione interna (ICE) si basa sull’approccio “Unsteady Reynolds Averaged Navier Stokes” (URANS). Esso garantisce sufficiente accuratezza con un costo computazionale limitato modellando l’intero spettro turbolento. Tuttavia, per diverse applicazioni tale approccio non fornisce il grado di dettaglio richiesto in fase di progettazione. La rapida diffusione di efficienti risorse di calcolo ha consentito l’uso di approcci “Large Eddies Simulations” (LES), soprattutto per analisi di flussi, combustione, spray e variabilità ciclica (CCV). L’approccio LES offre una descrizione spaziale e temporale dettagliata della turbolenza risolvendo le scale grandi (problema-dipendenti, anisotrope e ricche di energia) e modellando le più piccole (isotrope e universali). Tuttavia, il suo uso, per simulazioni ad alto numero di Reynolds (HR), è ancora limitato a causa dell’alto costo computazionale. In flussi lontani da pareti il numero di nodi di griglia stimato per simulazioni LES è legato al numero di Reynolds secondo la legge N=Re0.4. Per flussi intorno le pareti, la scala integrale della turbolenza è molto piccola, confrontabile con lo spessore dello strato limite e i forti gradienti in direzione normale alla parete vincolano l’altezza di cella in tale direzione. Inoltre, la risoluzione delle strutture più piccole richiede il raffinamento della griglia anche in direzione tangenziale ottenendo così la legge N=Re1.8. Tale approccio è noto come “Wall-Resolved LES” (WRLES) e esige uno sforzo computazionale eccessivo anche per applicazioni con numeri di Reynolds moderati, quali sono le simulazioni motoristiche. Una possibile alternativa all’approccio WRLES è fornita dai trattamenti ibridi URANS/LES, sempre più diffusi per simulazioni HR ed includono un gran numero di sottoclassi come Detached Eddy Simulation (DES) e sue varianti, Scale Resolving Simulation (SAS) e la più recente Stress-Blended Eddy Simulation (SBES). Soffermandosi sul modello DES, esso prevede la risoluzione delle scale grandi, lontane da parete, mentre quelle piccole, nello strato limite, sono modellata in ambiente RANS, che vanta decenni di studi nella descrizione dello strato limite. Tale approccio richiede griglie raffinate in direzione normale la parete, dove y+≈1 è limitato dall’uso del trattamento RANS Low Reynolds, ma meno raffinata parallelamente ad essa. Il trattamento DES, originariamente definito sul modello RANS ad una equazione di Spalart-Allmaras (S-A), può essere applicato a qualsiasi modello “eddy viscosity” in cui il criterio di switch agisce o sulla viscosità turbolenta o sull’energia cinetica turbolenta. In generale, in tutte le formulazioni DES il modello di background RANS emula il modello LES di Smagorinsky (o altri come Yoshizawa). In tale contesto, il presente lavoro di tesi mira a sviluppare una metodologia robusta e stabile per l’uso del trattamento turbolento Zonal-DES in simulazioni CFD-3D di motori a combustione interna. La metodologia proposta agisce su entrambi i termini di produzione e dissipazione dell’energia cinetica turbolenta e il criterio di switch è attivo su tutti nodi di calcolo, senza alcuna interfaccia tra i due approcci. Inoltre, è stato usato uno trattamento di parete dedicato, al fine di preservare l’approccio HR, basato sull’uso di wall functions. La metodologia è stata testata prima su un caso test di riferimento ampiamente riportato in letteratura e in seguito adottata in simulazioni multi-ciclo di due motori da ricerca ad accesso ottico. I risultati delle simulazioni motoristiche dimostrano l’affidabilità e l’efficienza della metodologia proposta, adatta ad l’analisi “Scale-Resolving” e volta a indagare fenomeni dominati dalla turbolenza nei motori ad iniezione diretta.

SVILUPPO DI UN MODELLO IBRIDO AVANZATO URANS/LES E TRATTAMENTO DI PARETE DEDICATO PER FLUSSI TURBOLENTI COMPLESSI E CONFINATI / Clara Iacovano , 2022 May 16. 34. ciclo, Anno Accademico 2020/2021.

SVILUPPO DI UN MODELLO IBRIDO AVANZATO URANS/LES E TRATTAMENTO DI PARETE DEDICATO PER FLUSSI TURBOLENTI COMPLESSI E CONFINATI

IACOVANO, CLARA
2022

Abstract

The standard methodology to model turbulent flows in multidimensional internal combustion engine (ICE) simulations is still the Unsteady Reynolds Averaged Navier Stokes (URANS) approach. It allows both sufficient accuracy and limited computational cost modelling the whole turbulence spectrum. However, for several applications, RANS models cannot provide the degree of accuracy required during the design process (i.e. aero-acoustics simulations and massive separation phenomena). The rapid spread of cost-effective HPC resources has increased the use of Large Eddies Simulations (LES) turbulence treatment, especially for the analysis of cold flow, combustion, spray and cycle to cycle variability (CCV). LES approach offers accurate spatial and temporal descriptions of turbulence, resolving only the large scales (problem-dependent, anisotropic, and full of energy) and modelling the small ones (universal and isotropic). However, the application of LES models for High Reynolds (HR) number simulations, is still limited because of the high computational costs. For free shear flows, the number of grid points required for a LES simulation scales according to O(Re0.4). In wall-bounded flows, the turbulence length scale near the walls becomes very small compared to the boundary layer thickness and the high gradients impose a very small computational grid size in the wall-normal direction. Moreover, to resolve the small isotropic scales in the boundary layer, a high-resolution grid in the tangential direction is required as well, leading to more rigid scaling, i.e. O(Re1.8). This kind of LES approach is called Wall-Resolved LES (WRLES) and it is prohibitively expensive even for moderate HR number applications, such as ICEs simulations. A possible alternative to WRLES is represented by hybrid URANS/LES models, which are more and more diffused for HR number applications and include a great number of subclasses, such as Detached Eddy Simulation (DES) and its variants, Scale Resolving Simulation (SAS) and the more recent Stress-Blended Eddy Simulation (SBES) models. Focusing on DES, it is based on the principle that large eddies are resolved away from walls while the small ones, in the boundary layer, are modelled with RANS closures which have decades of validation in boundary layer description. It requires a fine grid at walls only in the wall-normal direction, where y+≈1 is required for the RANS Low Reynolds turbulence model, and a coarse grid in the tangential direction. It was originally proposed for the RANS Spalart-Allmaras (S-A) one equation model, but it can be applied to any others eddies viscosity models in which the switching criterion acts on either turbulent viscosity or turbulent kinetic energy. In general, all DES formulations performance to reduce the background RANS approach to a Smagorinsky-type SGS one-equation model (or similar such as the Yoshizawa one). In this framework, the present work aims at finding a stable and robust methodology to apply the Zonal-DES hybrid turbulence treatment in CFD-3D ICEs simulations. The proposed methodology acts on both the production and dissipation terms of the turbulent kinetic energy transport equation and the switching criteria applies to all grid nodes, both near-walls and bulk flows, without any kind of interface between the two approaches. A dedicated near-wall treatment is added to preserve the well-established HR approach in the first near-wall cell, based on the use of wall function. The methodology has been tested, at first, on a reference test case widely reported in literature and subsequently for multi-cycle simulations of two reference single-cylinder optical research engines. The engine simulations results demonstrate consistency and efficiency of the proposed methodology, which is a suitable candidate for affordable scale-resolving analyses aiming at evaluating turbulence-governed phenomena in direct-injection engine.
DEVELOPMENT OF AN ADVANCED HYBRID URANS/LES TURBULENCE MODEL AND NEAR-WALL TREATMENT FOR COMPLEX WALL-BOUNDED TURBULENCE FLOWS
16-mag-2022
D'ADAMO, Alessandro
File in questo prodotto:
File Dimensione Formato  
TESI DEFINITIVA - IACOVANO CLARA.pdf

embargo fino al 15/05/2025

Descrizione: TESI DEFINITIVA - IACOVANO CLARA
Tipologia: Tesi di dottorato
Dimensione 14.33 MB
Formato Adobe PDF
14.33 MB Adobe PDF   Visualizza/Apri   Richiedi una copia
Pubblicazioni consigliate

Licenza Creative Commons
I metadati presenti in IRIS UNIMORE sono rilasciati con licenza Creative Commons CC0 1.0 Universal, mentre i file delle pubblicazioni sono rilasciati con licenza Attribuzione 4.0 Internazionale (CC BY 4.0), salvo diversa indicazione.
In caso di violazione di copyright, contattare Supporto Iris

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11380/1277161
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact