Some illustrations of my results are displayed below, either as movies or pictures. Most of these results have been produced during my PhD thesis at the University of Geneva (see references below). The lattice Boltzmann method was used for all the CFD examples below, mostly through the Palabos library : www.palabos.org.
Articles published as main author
Thorimbert (2019). Lattice Boltzmann simulations of complex flows. University of Geneva. thesis. (link)
Thorimbert et al. (2019). Implementation of lattice Boltzmann free-surface and shallow water models and their two-way coupling.
Journal of Computational Science.
Thorimbert et al. (2019). Coupling of lattice Boltzmann shallow water model with lattice Boltzmann free-surface model.
Journal of Computational Science. (link)
Thorimbert et al. (2018). Lattice Boltzmann simulation of dense rigid spherical particle suspensions using immersed boundary method.
Computers & Fluids. (link)
Thorimbert and Chopard. (2016). Polynomial method for fast procedural terrain generation.
arXiv.org. (link)
Thorimbert et al. (2016). Virtual wave flume and Oscillating Water Column modeled by lattice Boltzmann method and comparison with experimental data.
International Journal of Marine Energy. (link)
Articles published as secondary author
Latt et al. (2020). Palabos: Parallel Lattice Boltzmann solver.
Computers & Mathematics with Applications. (link)
Marson et al. (2020). Enhanced single-node boundary condition for the Lattice Boltzmann Method
arXiv. (link)
Chliamovitch and Thorimbert (2018). Turbulence through the Spyglass of Bilocal Kinetics.
Entropy. (link)
Three-phase suspensions
Non-Newtonian fluids as blood or magma exhibit complex rheological properties, often incorporated as phenomenological models inside CFD codes. In this project, this behaviour emerges as a result of the interaction between the three different fluid constituents : a pure fluid phase, a solid phase (the particles) and a gaseous phase (the bubbles).
The immersed boundary method with a multi-direct forcing approach has been used to model fluid-particle interactions. For fluid-bubbles interactions, a free-surface volume-of-fluid method has been used.
Two-phase suspensions
The concentration of particles in a suspensions strongly determines its behaviour, leading to either dilute, semi-dilute or concentrated regimes. The latter constitutes a highly challenging problem in CFD, as chains of particles tend to form and to interact with the fluid in a still misunderstood fashion. In the above animation, the concentration of particle is 50%, close to the critical value at which a granular media approach should be employed.
This model has been validated by means of a comparison with well-known phenomenological laws in all three regimes. The graph on the right shows how the viscosity evolves as the concentration of particles grows. See: Thorimbert et al. (2018). Lattice Boltzmann simulation of dense rigid spherical particle suspensions using immersed boundary method. Computers & Fluids (link). The model is straightforwardly adaptable to assess the complex behaviour of multimodal suspensions, as depicted below.
.
Free-surface simulation of a wave energy converter
A free-surface, volume-of-fluid method is used as a tool to investigate the efficiency of an oscillating water column wave energy converter.
This model has beeen validated via direct comparison with experiments in a flume. See: Thorimbert et al. (2016). Virtual wave flume and Oscillating Water Column modeled by lattice Boltzmann method and comparison with experimental data. International Journal of Marine Energy (link).
Free-surface simulation of a dam flush scenario
Using bathymetry data from the Verbois dam near Geneva, simulations of miscellaneous scenarii can be run using a free-surface, volume-of-fluid method.
The shear stress exerted to the river bed can be assessed through the simulation, allowing for the prediction of sedimention and erosion occuring in such a crucial event for the region.
Procedural noise generation
A method allowing for the generation of coherent noise with arbitrary values and gradient smoothness is presented in https://arxiv.org/abs/1610.03525. Procedural terrain and textures can be generated faster than with prevailing fundamental methods as Perlin and Simplex noise.
Texture of wood, marbles and clouds generated using the same base algorithm with a different post-treatment:
This method has been exploited inside a 2D exploration game coded in one week for the Python Pyweek challenge, that can be found here: http://thorpy.org/applications/torus.html.
