[FR] Ceci est la version de TEST de DIAL.mem. Veuillez utiliser l'URL suivante pour déposer votre mémoire: 'https://thesis.dial.uclouvain.be'. [ENG] This is the TEST version of DIAL.mem. Please use the following URL to submit your master thesis: 'https://thesis.dial.uclouvain.be'.
 
Loading...
Thumbnail Image

Assessing the morphological acceleration method using a GPU code-based model

Colinet, Alexandre
(2024)

Files

Colinet_79831600_2024.pdf
  • UCLouvain restricted access
  • Adobe PDF
  • 3.85 MB

Details

Supervisors
Legat, Vincent ; Lambrechts, Jonathan
Faculty
Ecole polytechnique de Louvain
Degree label
Master [120] : ingénieur civil mécanicien, à finalité spécialisée
Abstract
The subject of this master thesis was to study the influence of the morphological acceleration factor in the simulation of bed modification due to sediment transport. This factor is used to linearly increase the modification of the bed for each iteration, allowing for accelerated simulations by scaling up bed modifications. For instance, if one hour of simulation can represent 5 years, a morphological acceleration factor of 20 allows for simulating 100 years. Such simulations are time-consuming as they need to cover decades to observe the impact of sediment on the bed and reach a state of equilibrium. To conduct this study, we utilized the Second-generation Louvain-la-Neuve Ice-ocean Model (SLIM), employing a discontinuous Galerkin Finite Element Method on GPU. The main challenge was to familiarize ourselves with SLIM, implement the bed modification and sediment equations, and finally run the simulations on LUCIA to analyze the results. However, a new challenge emerged: to compare our results with the literature, we needed to implement wetting-drying in the simulations. Wetting-drying is a phenomenon where parts of the domain switch from wet to dry during the simulation, which can lead to ill-posed equations requiring special treatment. While this had already been implemented for the shallow water equations in SLIM, it needed to be extended to the tracer responsible for sediment transport and bed erosion simulation. Due to issues and the long time required to run the simulations, even with LUCIA, the results of this master thesis only highlighted the difference between simulations with and without the morphological acceleration factor. Ultimately, this master thesis provides an initial exploration into the problem and attempts to identify the causes of these differences, but it remains a broad and interesting subject that warrants more in-depth research.