High performance computing for hydrothermal systems or geothermal reservoirs using MULTI-GPU technology: Application to the Lusi geyser systems in Java, Indonesia
Reza Sohrabi · 2018
Hydrothermal systems and geothermal reservoirs are of major interest for natural and engineering sciences. Their fascinating complexity spurs multidisciplinary efforts for supplying society with heat and electricity production. Their applications require the correct simulation of fluid dynamics and thermodynamics processes in fractured media, and many of these media host a certain degree of complexity. This thesis aims to understand the heat transfer and fluid dynamics within a newborn geyser system, named Lusi, which began erupting in 2006 in East Java, Indonesia. Geyser systems are ubiquitous, with a wide-range of processes, making the development of a general model difficult. My goal is to establish 3D conceptual and numerical models capable of simulating fluid dynamics and thermodynamics within large-scale geyser reservoirs such as the Lusi region, considering porosity and permeability evolution through time. These models can also address a number of other renewable applications, including Enhanced Geothermal Systems (EGS) and CO2 sequestration (Carbon Capture and Storage CCS). I developed a new mesh algorithm to create hexahedral octree meshes to transfer the structural geological information using binary space partitioning (BSP) of the input geometry and octree refinement on the grid. The algorithm provides a new method for hexahedral mesh generation for any 3D numerical simulations to quantify Thermal-Hydraulic-Mechanical-Chemical (THMC) physical processes. Further, I present a new high performance 3D tool using Graphics Processing Unit (GPU) workstations or cluster technology. The physical processes implemented into the code are those associated with deep hydrogeological complexes where high fluid pressures generated by dehydration reactions can be sufficient to induce hydro-fractures that significantly influence porosity and permeability structures within geological formation. Finally, I use the developed numerical models to investigate Lusi using 3D geological context and complexity with multiphysics processes considering High Performance Computing (HPC) on parallel computing. I present the first 3D numerical model of the Lusi geyser system using multi-GPU technology.