The turbulent pipe flow of non-Newtonian power-law fluids plays a key role in many civil and industrial applications. An accurate fluid dynamic characterization improves the understanding of physical processes and supports technological progress. Due to the complexity of acquiring detailed experimental data, Computational Fluid Dynamics (CFD) becomes a valuable alternative. Direct Numerical Simulations (DNS), which resolve all turbulence scales, offer high-fidelity data but are too computationally expensive for practical engineering use. Reynolds-Averaged Navier–Stokes (RANS) simulations, which solve only for the mean flow, are widely used due to their efficiency and reasonable accuracy. However, modelling non-Newtonian fluids with RANS is more complex than Newtonian ones, as it involves fluctuating apparent viscosity correlations. This introduces additional uncertainty, which adds to that from the turbulence model used to compute the Reynolds stress tensor. This thesis investigates the impact of two elements of RANS modelling for shear-thinning, power-law fluids: the turbulence model and the model for the mean viscous stresses tensor. The study focuses on turbulent pipe flow cases at varying friction Reynolds numbers and fluid properties. DNS data from literature serve as a benchmark to evaluate the accuracy of different RANS configurations. The work is structured in three stages of increasing complexity. First, three low-Reynolds turbulence models—the Two-Layer k-ε, the Lam–Bremhorst k-ε, and the k-ω SST—are tested for Newtonian fluids, where only turbulence modelling affects the results. Next, the same models are applied to non-Newtonian shear-thinning fluids using the built-in viscous stress model from commercial CFD tools. Lastly, each turbulence model is combined with three literature-based mean viscous stress models. Various flow variables are analyzed to gain insights beyond typical experimental limits. The results highlight a strong interplay between turbulence and viscosity models, making it difficult to define a universally optimal combination. However, clear trends emerge, offering valuable guidance for future RANS model development.
I flussi turbolenti di fluidi non-Newtoniani a legge di potenza rivestono un ruolo importante in varie applicazioni civili e industriali. Una descrizione dettagliata delle caratteristiche fluidodinamiche consente una maggiore comprensione del processo fisico e supporta quello tecnologico. A causa della difficoltà di ottenere dati sperimentali, la fluidodinamica computazionale (CFD) rappresenta una valida alternativa per lo studio di tali fenomeni. Le simulazioni numeriche dirette (DNS), che risolvono tutte le scale della turbolenza, forniscono dati ad alta fedeltà ma con costi computazionali elevati. Le simulazioni che adoperano le Equazioni di Navier-Stokes mediate (RANS), risolvono solo il flusso medio, e sono usate per la loro efficienza e accuratezza. Tuttavia, nel caso non-Newtoniano, esse risultano più complesse a causa delle fluttuazioni della viscosità apparente, che si sommano all’incertezza introdotta dai modelli per il tensore degli sforzi di Reynolds. Questa tesi analizza l’impatto di due aspetti nella modellazione RANS di fluidi pseudoplastici: il modello di turbolenza e quello per il tensore degli sforzi medi viscosi. Lo studio si concentra su flussi turbolenti in condotti, variando il Numero di Reynolds di parete e le proprietà del fluido. Dati DNS di letteratura sono usati come riferimento per valutare l’accuratezza dei risultati. Il lavoro si articola in tre fasi a complessità crescente. Nella prima, tre modelli di turbolenza per bassi Re, il Two-Layer k-ε, il Lam–Bremhorst k-ε e il k-ω SST, sono testati su fluidi Newtoniani, dove solo il modello influisce sui risultati. In seguito, gli stessi modelli sono applicati a fluidi non-Newtoniani utilizzando un modello commerciale per gli sforzi viscosi. Infine, ogni modello di turbolenza è combinato con tre modelli di letteratura per lo sforzo viscoso medio. Sono state analizzate diverse variabili di flusso per superare i limiti sperimentali. I risultati evidenziano una forte interazione tra i modelli di turbolenza e quelli di viscosità, rendendo difficile definire una combinazione ottimale universale. Tuttavia, emergono delle tendenze, che offrono indicazioni per lo sviluppo futuro di modelli RANS.
Numerical investigation of turbulent transport of shear-thinning, power-law fluids in a circular pipe
Maraschi, Federico Vincenzo
2024/2025
Abstract
The turbulent pipe flow of non-Newtonian power-law fluids plays a key role in many civil and industrial applications. An accurate fluid dynamic characterization improves the understanding of physical processes and supports technological progress. Due to the complexity of acquiring detailed experimental data, Computational Fluid Dynamics (CFD) becomes a valuable alternative. Direct Numerical Simulations (DNS), which resolve all turbulence scales, offer high-fidelity data but are too computationally expensive for practical engineering use. Reynolds-Averaged Navier–Stokes (RANS) simulations, which solve only for the mean flow, are widely used due to their efficiency and reasonable accuracy. However, modelling non-Newtonian fluids with RANS is more complex than Newtonian ones, as it involves fluctuating apparent viscosity correlations. This introduces additional uncertainty, which adds to that from the turbulence model used to compute the Reynolds stress tensor. This thesis investigates the impact of two elements of RANS modelling for shear-thinning, power-law fluids: the turbulence model and the model for the mean viscous stresses tensor. The study focuses on turbulent pipe flow cases at varying friction Reynolds numbers and fluid properties. DNS data from literature serve as a benchmark to evaluate the accuracy of different RANS configurations. The work is structured in three stages of increasing complexity. First, three low-Reynolds turbulence models—the Two-Layer k-ε, the Lam–Bremhorst k-ε, and the k-ω SST—are tested for Newtonian fluids, where only turbulence modelling affects the results. Next, the same models are applied to non-Newtonian shear-thinning fluids using the built-in viscous stress model from commercial CFD tools. Lastly, each turbulence model is combined with three literature-based mean viscous stress models. Various flow variables are analyzed to gain insights beyond typical experimental limits. The results highlight a strong interplay between turbulence and viscosity models, making it difficult to define a universally optimal combination. However, clear trends emerge, offering valuable guidance for future RANS model development.| File | Dimensione | Formato | |
|---|---|---|---|
|
2025_07_Maraschi.pdf
accessibile in internet per tutti a partire dal 02/07/2026
Descrizione: testo tesi
Dimensione
4.96 MB
Formato
Adobe PDF
|
4.96 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/10589/240598