Speaker: Boris Kratz
Date: Thursday 16th of March 2023, 1:15pm.
Abstract:
Les métropoles sont confrontées à une saturation de leurs axes de mobilité, ce qui a entraîné la mise en place de nouveaux grands projets d'infrastructures tels que la construction de tunnels. Cependant, la construction de tunnels comporte des risques, notamment les déplacements surfaciques causés par les travaux d'excavation. Pour évaluer ces risques et protéger les bâtiments situés à la surface, il est nécessaire de prédire et surveiller l'ampleur des tassements. Pour cela, différentes méthodes de prédiction ont été développées, allant des méthodes empiriques et analytiques aux méthodes numériques. Les modèles 3D par éléments finis (EF), avec des lois de comportements mécaniques des sols réalistes, sont considérés comme étant les plus prédictifs, mais sont complexes et coûteux en temps de calcul. Un méta-modèle, ou modèle de substitution, peut fournir une approximation suffisamment précise d'un modèle (EF), avec l'avantage d'être beaucoup moins coûteux en temps de calcul. Cette présentation vise à expliquer les étapes nécessaires pour développer un tel méta-modèle pour le problème de l'excavation par un tunnelier. En particulier, nous discuterons de la précision des méta-modèles en fonction du nombre de simulations effectuées pour la construction du méta-modèle.
- Details
- Category: Seminar
Speaker: Christophe Reype
Date: Thursday 9th of February 2023, 1:15pm.
Abstract:
The analysis of hydrogeochemical data aims to improve the understanding of mass transfer in the sub-surface and the Earth's crust. This work focuses on the study of fluid-fluid interactions through fluid mixing systems, and more particularly on the detection of the compositions of the mixing sources. The detection is done by means of a point process: the proposed model is unsupervised and applicable to multidimensional data. Physical knowledge of the mixtures and geological knowledge of the data are directly integrated into the probability density of a Gibbs point process, which distributes point patterns in the data space, called the Hug model. The detected sources form the point pattern that maximises the probability density of the Hug model. This probability density is known up to the normalisation constant. The knowledge related to the parameters of the model, either acquired experimentally or by using inference methods, is integrated in the method under the form of prior distributions. The configuration of the sources is obtained by a simulated annealing algorithm and Markov Chain Monte Carlo (MCMC) methods. The parameters of the model are estimated by an approximate Bayesian computation method (ABC). First, the model is applied to synthetic data, and then to real data. The parameters of the model are then estimated for a synthetic data set with known sources. Finally, the sensitivity of the model to data uncertainties, to parameters choices and to algorithms set-up is studied.
- Details
- Category: Seminar
Speaker: Augustin Gouy
Date: Thursday 9th of March 2023, 1:15pm.
Abstract:
Karst systems, which are often present in carbonate rocks, are characterized by a complex ensemble of underground drains which strongly control the water flow in the massif. It is therefore necessary to take into account the karst network to model the hydrogeological functioning of karstified rocks. Information on the position and the geometry of those caves is however often scarce, indirect and imprecise. A discrete karst network simulation approach is proposed, capable of representing the diversity of possible models for a given dataset, while exploiting those data to reduce the ensemble distribution. This method solves a shortest path algorithm between sinks and springs –respectively the inlets and outlets of the network- with the use of an anisotropic cost function defined in the whole domain. The cost represents in a simplified way the physico-chemical processes of karstification, while facilitating control of the geometry and topology of the generated networks. Examples derived from a real case study show how the position, density, connectivity rules and hierarchy of the network’s inlets and outlets can be used to create realistic karst networks under various underlying hypotheses.
- Details
- Category: Seminar
Speaker: Behshad Koohbor
Date: Thursday 2nd of February 2023, 2:30pm.
Abstract:
The modelling of flow, mass/heat/phase transport in fractured porous media is of great interest to engineers and geoscience researchers. As a young researcher with a growing research career, I have been greatly interested in the general topic of environmental fluid mechanics (i.e. with a focus on mathematical modelling and regards to experimental studies) I would like to present my educational and research background, current activity and future research perspectives. The first part of the presentation will be on my research project during my Ph.D. studies at the University of Strasbourg on the topic of simulation of seawater intrusion and flow in variably saturated fractured porous media. Then, a brief description of my postdoctoral research project at BRGM (the French Geological Survey), Orleans, France will be presented. The topic mainly focuses on the numerical and experimental study of the evolution of electrical resistivity and multiphase flow in porous media for the monitoring of contamination and remediation of NAPLs. The second part of the presentation will be related to one of the projects that I am involved in HydroSciences Montpellier related to the numerical application of hydro-geophysics in karst media. And finally, a few insights related to my personal future research interests related to developing an ensemble of criteria for choosing the most relevant approach when simulating flow and transport processes in fractured porous media.
- Details
- Category: Seminar
Speaker: Paul Cupillard
Date: Thursday 2nd of March 2023, 1:15pm.
Abstract:
Tectonic processes and the industrial exploitation of the subsurface induce brittle deformations in the earth crust, leading to fractures at all scales. Geological observations have evidenced that a power law is appropriate to describe the density of a fracture set as a function of the fracture size. Nevertheless, for either theoretical or computational reasons, studies on seismic wave propagation in fractured media have been restricted to a short range of fracture sizes so far. In particular, effective medium theories all rely on an Elementary Representative Volume (ERV) consisting of a host matrix in which cracks are included. Assuming the ERV small with respect to the wavelength, then it is in a quasi-static regime of stress so that techniques from micromechanics can be used to compute an equivalent medium. In the frame of a 18-month post-doc, Anaïs Ibourichène started working on the application of the non-periodic homogenization method to fractured media to go beyond the ERV setting and explore the effect of a wide distribution of crack sizes on seismic wave propagation. I here present the preliminary results obtained by Anaïs in 2D. First, the effective properties computed using the homogenization in the case of Eshelby problems are compared to analytical solutions provided by various effective medium theories. Then, the homogenization is applied to a synthetic mass rock which contains different fracture sets, each set being characterized by a length l (up to λmin/4) and a density d ∼ l−3. Finally, some short-term and long-term perspectives of this work are discussed.
- Details
- Category: Seminar
Speaker: Mustapha Zakari
Date: Thursday 26th of January 2023, 1:15pm.
Abstract:
Numerical models are powerful tools to forecast the behavior of aquifers and subsurface reservoirs. In these numerical models, geological grids approximate the subsurface geometry with structured or unstructured meshes. Because structured grids lead to relatively simple and efficient flow simulation algorithms, they have become the industrial standard to approximate geometries in reservoir modeling. Unstructured grids can produce more accurate geometries of complex geological features and are more flexible to mesh the simulation domain adaptively than Cartesian or Corner Point Grids (CPG). However, explicitly accounting for fine-scale heterogeneity leads to a number of grid cells too large to be directly manageable by flow solvers. Upscaling methods were developed to coarsen geological grids in order to produce manageable grids for flow simulations. Most upscaling methods were developed for structured grids. Here we propose a new upscaling strategy to upscale unstructured grids to non matching structured grids. It combines a control volume finite element (CVFE) method to compute fine scale pressure values and an incremental mesh intersection algorithm to upscale flow information to a non matching coarser structured grid. The intersection algorithm is inspired from computational geometry algorithms. We introduce upscaling methods and present our CVFE solver and mesh intersection algorithm. We apply the upscaling strategy to first geological grids. The obtained results are discussed and first conclusions are then presented.
- Details
- Category: Seminar
Speaker: Amandine Fratani
Date: Thursday 23th of February 2023, 1:15pm.
Abstract:
The CO2 storage interests several industries to reduce their net carbone missions. Therefore, reservoir simulation methods need to be adapted to the injection of CO2, in particular on unstructured grids to allow a geomechanical coupling. For three years, TotalEnergies has been developing a workflow for fluid flow simulation on unstructured grids. This presentation focuses on the addition of analytical aquifer in the workflow. Analytical aquifers are one method used to consider the behaviour of water outside the reservoir. Several analytical aquifers can be implemented, we only focus on the implementation of radial analytical aquifers in a ParaView plugin. Faults were already implemented in the workflow, but some modifications on this part were performed in order to bring consistency between the different parts. Finally, two applications are proposed. The first is a synthetic oil production case, which showed that the workflow is working as expected. In this case, aquifer behaviour in presence of fault is checked and a sensitivity analysis of the Carter-Tracy parameters is made. Some additional work remains to obtain the same results between the various grid types. The second application is a CO2 injection in a ‘’real’’ model, showing interesting results not fully understood yet.
- Details
- Category: Seminar
Speaker: Jeremie Giraud
Date: Thursday 12th of January 2023, 1:15pm.
Abstract:
We present a case study using geological and geophysical data from the Western Pyrenees. We use an inversion algorithm integrating an automated geological modelling component to perform geometrical gravity inversion. Starting from a pre-existing geological model of the study area, we deform it using our inversion algorithm to fit the gravity data while accounting for geological data to maintain geological realism. We then investigate different geological scenarios through the removal or insertion of rock units and consider uncertainties in the recovered geometries using a series of geophysically equivalent models generated with the so-called Hamiltonian null-space shuttles. The area we consider is located in the subduction zone forming the western Pyrenees across France and Spain in a region centred on the Mauleon basin. The subduction was generated during the Iberian and Eurasian collision, with a probable partial uplift of the upper mantle. Using a geophysical inversion driven approach, we focus on the potential presence of shallow mantle material. We assess the geophysical consistency of different geological and geophysical scenarios using the null-space shuttles to transition between several geophysically valid yet geologically different models. Results corroborate previous hypotheses about the presence of a probable mantle wedge, but suggest that its interpreted geometry may need to be revisited.
- Details
- Category: Seminar
Speaker: Danielle-Rita Pajiep-Ngongang
Date: Thursday 16th of February 2023, 1:15pm.
Abstract:
L’évaluation des risques liés à la présence d’ouvrages souterrains (carrières, mines, cavités naturelles) en cours d’exploitation et/ou abandonnés pose légitimement la question de la stabilité dans le temps de ces ouvrages. En janvier 1910 est survenu l’effondrement de la carrière souterraine dite « Beaulieu », située dans la commune de Château Landon (département de la Seine-et-Marne), après une série de précipitations et la forte crue du cours d’eau en pied de carrière. Avec le changement climatique et devant le risque de crues exceptionnelles et les conséquences possibles d'instabilités de ces ouvrages souterrains, l’Ineris a décidé d’approfondir les connaissances actuelles afin de pouvoir prédire l’ampleur et l’impact en surface de ces instabilités. Pour ce faire la carrière de craie « Royer » à Château Landon (77) a été sélectionnée car elle présente de fortes similitudes géologiques et géotechniques avec la carrière voisine Beaulieu qui s’est effondrée. Le comportement de cette craie a fait l’objet de plusieurs études expérimentales au sein de l’Ineris notamment dans la thèse de Lafrance (2016). Toutefois très peu de données ont pu être collectées en raison de sa forte sensibilité à l’eau. Les principaux objectifs de mes travaux visent donc à l’amélioration des connaissances du comportement hydromécanique de cette craie lorsqu’elle est soumise à des sollicitations hydromécaniques et au développement d’une approche de modélisation numérique pertinente pour l’analyse de la stabilité à long terme des cavités exposées à des conditions environnementales complexes.
- Details
- Category: Seminar