The virtual wave flume - with the SPH method
Author
Hansen, Mads-Peter
Term
4. term
Education
Publication year
2008
Pages
97
Abstract
Dette projekt undersøger, om Smoothed Particle Hydrodynamics (SPH), implementeret via open source-koden SPHysics, kan bruges til at opbygge en todimensional virtuel bølgerende svarende til Aalborg Universitets laboratorierende og dermed reducere behovet for omfattende fysiske modelforsøg. Første del gennemgår SPH-teorien med enkle 1D-eksempler for at tydeliggøre centrale begreber som kernel-funktioner, smoothing length, partikelapproksimation og randbehandling samt sætte SPH i forhold til FEM/FDM. Anden del udvikler en 2D SPH-model af en bølgerende, hvor Navier–Stokes-ligningerne diskretiseres og løses eksplicit; modellen er opbygget i Fortran (SPHysics F95) med supplerende datahåndtering i MatLab, Excel og WaveLab. Valideringen sker ved at sammenligne tidsserier for bølgeegenskaber og bølgepåvirkning fra den virkelige rende med den virtuelle model, herunder bølgeopslaget under en platform. Resultaterne viser god overensstemmelse i de genererede bølgehøjder mellem virtuel og fysisk rende, hvilket indikerer at bølgebevægelsen og geometriens indflydelse fanges tilfredsstillende. Samtidig var det ikke muligt at opnå stabil beregning med en tilstrækkelig fin diskretisering til en præcis sammenligning af selve sammenstødsøjeblikket, hvilket peger på behov for yderligere arbejde med SPH-opsætning, viskositets-/turbulensmodeller og numerisk robusthed. Projektet demonstrerer potentialet i SPH-baserede virtuelle rendløsninger samt deres nuværende begrænsninger ved kraftig interaktion og brud af den frie overflade.
This thesis investigates whether Smoothed Particle Hydrodynamics (SPH), implemented via the open-source code SPHysics, can be used to build a two-dimensional virtual wave flume equivalent to Aalborg University’s laboratory flume, thereby reducing reliance on extensive physical model tests. The first part reviews SPH theory with simple 1D examples to clarify key concepts such as kernel functions, smoothing length, particle approximation, and boundary treatment, and contrasts SPH with FEM/FDM. The second part develops a 2D SPH model of a flume, in which the Navier–Stokes equations are discretized and solved explicitly; the model is coded in Fortran (SPHysics F95) with supporting data handling in MATLAB, Excel, and WaveLab. Validation compares time series of wave properties and impact from the real flume with the virtual model, including wave run-up beneath a platform. Results show good agreement in generated wave heights between the virtual and physical flumes, indicating that wave evolution and the influence of geometry are captured satisfactorily. At the same time, a stable computation with sufficiently fine discretization to allow precise comparison at the moment of impact was not achieved, highlighting the need for further work on SPH setup, viscosity/turbulence models, and numerical robustness. The project demonstrates the potential of SPH-based virtual flumes, as well as current limitations in modeling violent interactions and free-surface breakup.
[This summary has been generated with the help of AI directly from the project (PDF)]
Keywords