ECCOMAS 2024

Development of a multi-GPU Accelerated Code to Model the Complex Dynamics of Volcanic Plumes using the Lattice Boltzmann Method

  • LEMUS, Jonathan (University of Geneva)
  • Simionato, Riccardo (University of Geneva)
  • Coreixas, Christophe (University of Geneva)
  • Lätt, Jonas (University of Geneva)
  • Bonadonna, Costanza (University of Geneva)

Please login to view abstract download link

Among the numerous hazards related to volcanic events, explosive eruptions are as fas cinating as dangerous. They inject large quantities of volcanic ash particles (tephra) in the atmosphere. The dispersion and sedimentation of tephra represent a serious threat to communities, as it can notably affect human health, damage infrastructure, pollute ecosystems, and paralyse economic and transport sectors. Even though numerical dispersion models have now reached a significant level of sophistication, there is still a need for a better understanding of some processes such as particle aggregation or convective instabilities that enhance the premature sedimentation. Then, since the transport and sedimentation of tephra is strongly affected by the conditions at the source (i.e. at the volcano vent), a comprehensive description of the complex processes is crucial. Indeed, an explosive volcanic eruption involves compressible flows with possible transonic aspects, buoyancy effects, multiphase interactions and strong turbulence during the plume ascent. We then developed a numerical model in order to investigate the volcanic plume dynamics from the early stage to a larger scale dispersion by the wind [1]. The code makes use of the accuracy of the Lattice Boltzmann Method (LBM) to simulate complex flows. We also take advantage of the easiness to couple it with finite difference methods such as the low diffusive Weighted Essentially Non Oscillatory (WENO) scheme to simulate the transport of species along the fluid. Finally, another strong aspect is the capability of running the code with a multi-GPU computing acceleration in order to significantly improve the performances for large-scale 3D simulations. The results are validated by comparison with experimental data such as turbulent and thermal jets, showing also a strong stability for various configurations. The code is in-tended to be ”used” as a source term for a larger scale dispersion code thanks to a strong coupling using mesh superposition techniques. Eventually, using this numerical model, the study of complex processes in volcanic plumes and cloud such as particle aggregation and settling-driven gravitational instabilities will improve our understanding and help their parametrization in operational models.