Two of our students will defend their PhD on the same day at the Nancy School of Geology - the defenses will also be broadcasted online on Teams (links upon request).
Title: Interface insertion in 3D geomodels: from meshes to numerical simulations
Abstract:
Numerical subsurface models allow to understand the organization of structures and are adapted to simulate physical processes in order to study and predict the physical behavior of the subsurface.
The equations describing the physical phenomena are solved using a discretization of space: the mesh.
In the course of subsurface modeling projects, one often needs to revise an existing interpretation, integrate new spatial data, and perturb a geomodel to reflect subsurface uncertainty and to ultimately reduce this uncertainty using inversion methods.
In this thesis, I developed a mesh-based approach for local updating of meshed geomodels in 2D and 3D.
Local modifications are performed in a particular region of the model by changing the unstructured meshes of geomodels.
In particular, we focused on the insertion of lines and interfaces representing horizons or fractures.
During the modifications of the meshes, a particular attention is given to the mesh quality, especially near the intersections between interfaces, to maintain a conformal and valid mesh for numerical simulations.
In order to compare the impact of local modifications on physical simulations, I present three examples of application of this method: (1) wave propagation for the detection of a fluid contact on a 2D section, (2) the impact evaluation of a 2D fracture network structure on fluid flows in a porous medium, and (3) the impact of structural uncertainties in a 3D reservoir for the injection and CO2 storage.
I show that my method of local modifications (1) allows to insert interfaces in 2D and 3D models while keeping the conformity and validity of the meshes for numerical simulations, (2) shows a great flexibility to adapt to simulators of various physical phenomena, and (3) can be used in an inversion process to reduce uncertainties.
******************************************************
Title: Towards the characterization of earthquakes using time reversal: interpretation of focal spot and impact the mirror sampling
Abstract:
The localization and the characterization of seismic events are essential to understand the origin of the earthquake. Once it has been determined, it is possible to generate maps of different areas’ seismic risks and understand the impact of human activities. Several ways exist to locate seismic events, they use seismic recordings (i.e., seismograms). In those recordings, some methods use the time arrivals of particular waves and other parts or the entire signal. In this thesis, we are interested in one method: time reversal.
This method used the entire signal. The principle is to backpropagate the recordings reversed in time. The generated waves focus on the source location and create a focal spot. The time reversal has four terms of application i) the receivers form a closed surface, called a time reversal mirror, ii) the time reversal mirror does not perturb the wave propagation, iii) the medium is well known. Errors on the velocity model or interface positions would generate modifications in the wave path as compared to the path in the forward process, iv) the anelasticity is negligible. Anelastic attenuation would make a first-order time derivative appear in the wave equation so that the time reversibility would not be verified. In this thesis, we are first interested in the theoretical interpretation of the focal spot, that is to say in its physical meaning. To do so, we have used an existing software of wave propagation (SPECFEM2D) and we lean on the theory of point source homogenization. We have shown mathematically and numerically in several examples that the focal spot is a sum of the displacement generated by two homogenized point sources. This result is the main contribution of the thesis and opens a significant perspective which is to determine the source parameters by doing an inverse problem with the focal spot as the input data.
However, the conditions of time reversal make its application difficult in practical cases. It is almost impossible to have a close mirror of receivers. We have tried to quantify the impact of an incomplete mirror on time reversal simulation. To do so, we use a realistic geological case: the Groningen gas field. The results show that the incomplete mirror has a huge impact on the time reversal wavefield but in every case, we tested, a focal spot is obtained. The focal spots are in these cases deformed. Therefore, we illustrate the positive impact of the heterogeneities on the wavefield reconstruction. These results open multiple perspectives such as integrating surface data to improve the wavefield reconstruction when the mirror is open or completing the mirror.