Smoothed Particle Hydrodynamics (SPH) is a purely Lagrangian method developed at the end of seventies to cope with the limitations of finite difference methods in modeling some physical phenomena. Initially designed for astrophysical purposes, in the recent years SPH has been successfully applied to coastal engineering, fracture mechanics and elasticity, with also interesting results in gaming graphical simulation and film special effects. Compared to traditional finite difference methods, SPH is able to deal with free surface, deformable boundary and solid mechanics at nearly zero expense. Furthermore, several particle software such as DualSPHysics and GPUSPH exploit parallel computing and graphical processing unit calculation reducing considerably the simulation running time. Despite the actual researches are mainly focused on the analysis of coastal engineering problems, these features makes SPH suitable for the study of fluid machines. Indeed, the Lagrangian nature of the method allows a simplified pre-processing and an easy handling of the rigid motion. The study of these characteristics has been the subject of this work. In particular, this thesis focuses on the evaluation of the performance of SPH in dealing with confined flows and internal fluid dynamics. Initially, it is performed a detailed analysis of the Hagen-Poiseuille flow. Then, it is investigated the simulation of more complicated cases as the description of gear or piston pump operations. Despite some software limitations, this work has found that weakly-compressible SPH formulation and GPU computing are able to return quick and accurate results in almost all the tested phenomena. Furthermore, huge stability issues have been observed only with the simulation of laminar viscous flow at high Reynolds number where the retention of the initial homogeneous particle distribution is still critical.
Il metodo Smoothed Particle Hydrodynamics (SPH) è un metodo Lagrangiano puro che è stato sviluppato alla fine degli anni settanta per far fronte alle limitazioni dei metodi alle differenze finite nella modellazione di alcuni fenomeni fisici. Inizialmente pensato per problemi astrofisici, negli ultimi anni l'SPH è stato impiegato con successo nell'ingegneria costiera, nella meccanica della frattura e nell'elastica, con anche interessanti risultati nelle simulazioni grafiche dei giochi e negli effetti speciali dei film. Rispetto ai tradizionali metodi alle differenze finite, l'SPH è capace di gestire il pelo libero, i bordi deformabili e la meccanica dei solidi in maniera naturale. Per di più, diversi software a particelle come DualSPHysics e GPUSPH sfruttano il calcolo parallelo e su scheda video riducendo considerevolmente il tempo di simulazione. Nonostante le attuali ricerche siano maggiormente improntate all'ingegneria costiera, queste peculiarità rendono l'SPH adatto allo studio delle macchine a fluido. Infatti, la natura Lagrangiana del metodo permette un pre-processing semplificato e una facile gestione del moto rigido. Lo studio di queste caratteristiche sono l'oggetto di questo lavoro. In particolare, questa tesi si focalizza sulla valutazione delle performance del codice SPH nella gestione di flussi confinati e problemi di idrodinamica interna. Inizialmente, è stata svolta un'analisi dettagliata del flusso di Hagen-Poiseuille. Poi, è stata indagata la simulazione di casi più complessi come la descrizione delle operazioni di una pompa a ingranaggi o di una pompa a pistoni. Nonostante alcune limitazioni del software, questa tesi ha evidenziato come la formulazione SPH debolmente comprimibile e il calcolo su scheda video sono in grado di restituire risultati accurati e in tempi brevi in quasi tutti i fenomeni testati. Per di più, problemi seri di instabilità sono stati osservati solo nella simulazione di flussi laminari viscosi ad alti numeri di Reynold dove il mantenimento della distribuzione omogenea iniziale delle particelle è ancora critica.
Towards the application of SPH technique to the simulation of fluid machines
OLIVA, ANDREA
2017/2018
Abstract
Smoothed Particle Hydrodynamics (SPH) is a purely Lagrangian method developed at the end of seventies to cope with the limitations of finite difference methods in modeling some physical phenomena. Initially designed for astrophysical purposes, in the recent years SPH has been successfully applied to coastal engineering, fracture mechanics and elasticity, with also interesting results in gaming graphical simulation and film special effects. Compared to traditional finite difference methods, SPH is able to deal with free surface, deformable boundary and solid mechanics at nearly zero expense. Furthermore, several particle software such as DualSPHysics and GPUSPH exploit parallel computing and graphical processing unit calculation reducing considerably the simulation running time. Despite the actual researches are mainly focused on the analysis of coastal engineering problems, these features makes SPH suitable for the study of fluid machines. Indeed, the Lagrangian nature of the method allows a simplified pre-processing and an easy handling of the rigid motion. The study of these characteristics has been the subject of this work. In particular, this thesis focuses on the evaluation of the performance of SPH in dealing with confined flows and internal fluid dynamics. Initially, it is performed a detailed analysis of the Hagen-Poiseuille flow. Then, it is investigated the simulation of more complicated cases as the description of gear or piston pump operations. Despite some software limitations, this work has found that weakly-compressible SPH formulation and GPU computing are able to return quick and accurate results in almost all the tested phenomena. Furthermore, huge stability issues have been observed only with the simulation of laminar viscous flow at high Reynolds number where the retention of the initial homogeneous particle distribution is still critical.| File | Dimensione | Formato | |
|---|---|---|---|
|
Thesis_AndreaOliva_836263.pdf
Open Access dal 22/09/2019
Descrizione: Thesis text
Dimensione
16.2 MB
Formato
Adobe PDF
|
16.2 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/10589/142941