Sediment transport in alpine rivers represents a significant issue for hydropower plants, as suspended solid particles can cause severe erosion phenomena on hydraulic machin- ery. This work is presented as a case study on one of the two Francis turbines of the Mompantero hydropower plant: under current operating conditions, the turbine is not significantly affected by suspended sediment-related issues. However, future sediment management strategies are expected to adopt a sediment continuity approach, allowing the controlled passage of sediment-laden flows through the turbine as part of a continuous desedimentation process. In general, erosion caused by sediments leads to a non-negligible material loss over time, a reduction in performance, and consequently an increase in maintenance costs and a decrease in the service life of components, as well as a drop in electrical energy production due to plant shutdowns required for maintenance. This Master’s thesis aims to predict sediment-induced erosion in a Francis turbine by means of Computational Fluid Dynamics (CFD) simulations using the solver provided by Ansys, through the CFX module, which is highly optimized for turbomachinery applications. After an initial phase of geometry reconstruction, several multiphase ap- proaches are implemented and compared: the homogeneous Eulerian–Eulerian model, the Eulerian–Lagrangian model, and the non-homogeneous Eulerian–Eulerian model. The simulations are limited to the turbine runner and the guide vanes, in order to investigate particle transport, impact locations, and erosion intensity. The results show that all approaches are capable of capturing the main erosion pat- terns observed in Francis turbines, including the characteristic “shark bite” damage on the runner blades, as well as erosion on guide vanes and near hub and shroud regions. However, significant differences show up in terms of particle distribution, predicted erosion levels, and computational cost. In particular, the Eulerian–Lagrangian approach provides the most detailed and physically accurate description of particle trajectories and impact phenomena, at the expense of a significantly higher computational cost. The homogeneous Eulerian–Eulerian model shows good numerical robustness and reduced computational time, but tends to smooth particle concentration fields, limiting the accuracy of erosion prediction. The non-homogeneous Eulerian–Eulerian approach, although more physically representative, presents numerical stability issues and convergence difficulties, which affect its reliability in complex turbo- machinery flows. Despite the limited number of studies applying Eulerian–Eulerian approaches to hy- draulic turbines, the results obtained in this work are promising and demonstrate their potential for erosion prediction with lower computational effort compared to Lagrangian methods. Overall, this study confirms that CFD-based multiphase simulations represent an effective tool for predicting sediment-induced erosion trends in hydraulic turbines and can support both design optimization and maintenance planning under high sediment concentration operating conditions.
Il trasporto di sedimenti nei fiumi alpini rappresenta una problematica rilevante per gli impianti idroelettrici, in quanto le particelle solide sospese possono causare fenomeni di erosione significativi sulle macchine idrauliche. Questo lavoro si presta come caso studio nell’analisi di una delle due turbine Francis dell’impianto idroelettrico presso Mompantero: allo stato attuale, la turbina non presenta criticità legate alla presenza di sedimenti sospesi nel flusso. Tuttavia, in prospettiva futura, è prevista l’implementazione di strategie di gestione dei sedimenti basate sulla loro continuità lungo l’impianto, che comporteranno il transito controllato di flussi sedimentati attraverso la turbina. In generale l’erosione causata dai sedimenti comporta un consumo di materiale non trascurabile nel tempo, una riduzione delle prestazioni e quindi porta ad un incremento dei costi di manutenzione e a una diminuzione della vita utile dei componenti, oltre che ad un calo nella produzione di energia elettrica dovuta a fermi impianto necessari alla manutenzione stessa. Questa tesi magistrale ha l’obiettivo di prevedere l’erosione da sedimenti in una turbina Francis mediante simulazioni CFD (Computational Fluid Dynamics) sfruttando il codice di calcolo fornito dal software di Ansys, tramite il modulo CFX, altamente ottimizzato per lo studio di turbomacchine. Dopo una prima fase di ricostruzione della geometria, vengono implementati e confrontati diversi approcci multifase: il modello Euleriano–Euleriano omo- geneo, il modello Euleriano–Lagrangiano e il modello Euleriano–Euleriano non omogeneo. Le simulazioni sono condotte limitatamente alla girante della turbina e al distributore mobile, al fine di studiare il trasporto delle particelle, le zone di impatto e l’intensità dell’erosione. I risultati mostrano che tutti gli approcci sono in grado di catturare le principali config- urazioni di erosione osservate nelle turbine Francis, inclusi i tipici danni a “shark bite” sulle pale della girante, nonché fenomeni erosivi sul distributore e nelle regioni prossime a mozzo e corona. Tuttavia, emergono differenze significative nella distribuzione delle particelle, nei livelli di erosione previsti e nel costo computazionale. In particolare, l’approccio Euleriano–Lagrangiano fornisce la descrizione più accurata e fisicamente dettagliata delle traiettorie delle particelle e dei fenomeni di impatto, a fronte però di un costo computazionale significativamente più elevato. Il modello Eu- leriano–Euleriano omogeneo mostra una buona robustezza numerica e tempi di calcolo ridotti, ma tende a smussare la distribuzione delle particelle, limitando l’accuratezza nella previsione dell’erosione. Il modello Euleriano–Euleriano non omogeneo, pur essendo più rappresentativo dal punto di vista fisico, presenta problemi di stabilità numerica e difficoltà di convergenza, che ne compromettono l’affidabilità in flussi complessi tipici delle turbomacchine. Nonostante il numero limitato di studi che applicano approcci Euleriani–Euleriani alle tur- bine idrauliche, i risultati ottenuti in questo lavoro risultano promettenti e ne dimostrano il potenziale per la previsione dell’erosione con un costo computazionale inferiore rispetto ai metodi Lagrangiani. Nel complesso, questo studio conferma che le simulazioni multifase basate su CFD rappresentano uno strumento efficace per la previsione delle tendenze di erosione nelle turbine idrauliche e possono supportare sia la progettazione sia la pianificazione della manutenzione in condizioni operative caratterizzate da elevate concentrazioni di sedimenti.
CFD Analysis of sediment transport in a Francis turbine through the application of an innovative Eulerian–Eulerian approach
TELLO ORTIZ, MARCO ANDREA
2025/2026
Abstract
Sediment transport in alpine rivers represents a significant issue for hydropower plants, as suspended solid particles can cause severe erosion phenomena on hydraulic machin- ery. This work is presented as a case study on one of the two Francis turbines of the Mompantero hydropower plant: under current operating conditions, the turbine is not significantly affected by suspended sediment-related issues. However, future sediment management strategies are expected to adopt a sediment continuity approach, allowing the controlled passage of sediment-laden flows through the turbine as part of a continuous desedimentation process. In general, erosion caused by sediments leads to a non-negligible material loss over time, a reduction in performance, and consequently an increase in maintenance costs and a decrease in the service life of components, as well as a drop in electrical energy production due to plant shutdowns required for maintenance. This Master’s thesis aims to predict sediment-induced erosion in a Francis turbine by means of Computational Fluid Dynamics (CFD) simulations using the solver provided by Ansys, through the CFX module, which is highly optimized for turbomachinery applications. After an initial phase of geometry reconstruction, several multiphase ap- proaches are implemented and compared: the homogeneous Eulerian–Eulerian model, the Eulerian–Lagrangian model, and the non-homogeneous Eulerian–Eulerian model. The simulations are limited to the turbine runner and the guide vanes, in order to investigate particle transport, impact locations, and erosion intensity. The results show that all approaches are capable of capturing the main erosion pat- terns observed in Francis turbines, including the characteristic “shark bite” damage on the runner blades, as well as erosion on guide vanes and near hub and shroud regions. However, significant differences show up in terms of particle distribution, predicted erosion levels, and computational cost. In particular, the Eulerian–Lagrangian approach provides the most detailed and physically accurate description of particle trajectories and impact phenomena, at the expense of a significantly higher computational cost. The homogeneous Eulerian–Eulerian model shows good numerical robustness and reduced computational time, but tends to smooth particle concentration fields, limiting the accuracy of erosion prediction. The non-homogeneous Eulerian–Eulerian approach, although more physically representative, presents numerical stability issues and convergence difficulties, which affect its reliability in complex turbo- machinery flows. Despite the limited number of studies applying Eulerian–Eulerian approaches to hy- draulic turbines, the results obtained in this work are promising and demonstrate their potential for erosion prediction with lower computational effort compared to Lagrangian methods. Overall, this study confirms that CFD-based multiphase simulations represent an effective tool for predicting sediment-induced erosion trends in hydraulic turbines and can support both design optimization and maintenance planning under high sediment concentration operating conditions.| File | Dimensione | Formato | |
|---|---|---|---|
|
TelloOrtiz_MarcoAndrea.pdf
embargo fino al 17/10/2027
Dimensione
29.8 MB
Formato
Adobe PDF
|
29.8 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/107545