Dynamic simulations on generic unstructured meshes generated with GeoModeller
Simon Lopez1, Gabriel Courrioux1, Feng Xing1,2, Roland Masson2, Bernard Bourgine1, Philippe Calcagno1, Cécile Allanic1, Sunseare Gabalda1, Christelle Loiselet1
1 BRGM, 2 INRIA/Laboratoire J.A. Dieudonné – Université de Nice
The geological risk which is related to the limited knowledge of the subsurface is well known to all deep resource exploitations. Subsurface complexity has to be considered both in terms of the spatial distribution of petrophysical properties and of the structural aspects of the underground geology. When it comes to understand and quantify deep mass and energy transfers (e.g. geothermal energy), the traditional workflow is to bring all available information into a consistent conceptual model of the subsurface with the help of static geomodeling techniques. Then, dynamic modeling is performed to reproduce the initial state of the system and quantitatively compare different development and exploitation strategies. Our work deals with trying to tighten the integration of static and dynamic modeling and make their interaction possible to reduce exploration risks through dynamic validation of conceptual models. More generally, evolutionary conceptual quantitative dynamic models of the subsurface are expected to represent an integrative and shared vision of the subsurface that greatly improves both exploration and development/optimization of subsurface resources.
Over the years, implicit surface mapping, as it is implemented in GeoModeller has proven a very efficient framework to quickly build complex structural geological models involving geological bodies of any shape and with the occurrence of discontinuities such as faults (Lajaunie et al, 1997). These techniques have been developed for nearly 20 years. Yet, when it comes to produce quality meshes of such models the implicit nature of surfaces make 3D meshing a non-trivial task. We recently used the Computational Geometry Algorithms Library (CGAL) to build conformal simplicial (tetrahedral) meshes that exactly match any geological 3D object and its boundaries or internal 2D features such as fault surfaces (which are triangulated with triangles that are faces of the tetrahedrons from the 3D mesh). As the mesh generation algorithm is based on progressive Delaunay mesh refinement its completion is guaranteed but it may produce a few bad quality cells (slivers) which are removed through an optimization phase.
Fig. 1 3D Geological model of a faulted aquifer and subsequent mesh for flow simulation.
Then, the general meshes that are produced require specific numerical schemes to correctly reproduce multicomponent multiphase subsurface heat and mass transfers. Over the last few years, several research projects resulted in numerical schemes with the desired mathematical properties. These are able to deal with unstructured meshes that take into account the geological complexity (subsurface geo-bodies, heterogeneities, discontinuities, wedges) or to discretize facilities (wells, galleries) or more recently fracture networks. We will present recent developments concerning these different aspects (generic tetrahedral meshes generation and flow simulation) which have been tested on “toy models”. We will focus on the main achievements, the remaining milestones, and what already is and what could be the integration the future of such techniques within GeoModeller.
Lajaunie, C., Courrioux, G. and Manuel, L. (1997) Foliation fields and 3D cartography in geology; principles of a method based on potential interpolation. Mathematical Geology, 29, 571-584.
4th meeting on 3D Geological Modelling, Orléans 21-23 February 2018