Speaker: Amandine Fratani
Date: Thursday 25th of April 2024, 1:15pm.
Abstract:
Interpretation of faults is a requirement for a 3D geological modelling process. However, due to the incomplete observations caused by the gap between 2D seismic images or outcrops, the results of this stage can be ambiguous and uncertain. Recently, a proposition of solution based on a graph formalism has been expressed. In the graph, fault observations are represented as nodes, and edges carry the potential of pairwise associations computed from selected expert geological rules. Main limit of this work is that fault observations are not pairwise independent, therefore considering pair prevents the consideration of higher-order effects such as the distribution of the throw along several aligned nodes. We propose to consider a multiple-point likelihood computation to extend the graph where expert rules are replaced using machine learning on analog or partly observed data. The model is trained from a set of selected features such as the length of the fault trace or the throw value. Features are computed from fault traces extracted from 3D geological models. Association potential of k fault observations are then directly computed using the trained model. To prevent overfitting in our small geological models dataset, we propose to mimic a partly interpreted case: we split a 3D domain in two disjoint, contiguous sectors A and B, and use sector A as training and sector B for testing. This presentation will show first results on 2 and 3 points association using a Random Forest algorithm.
Speaker: Thomas Bodin
Date: Thursday 11th of April 2024, 1:15pm.
Abstract:
Earth sciences are by essence multidisciplinary as different fields study the same objects and processes at different scales. In seismology, the challenge is that tomographic models constructed from long period seismograms only “sees” the crust and the mantle as an effective medium, where the smallest scales are given by the resolution power of seismic waves. On the other hand, geodynamics and geochemistry “see” the Earth as a mixture of different chemical components, with the smallest scales given by chemical diffusion. In this presentation, I will show how modern tools from inverse theory, computational statistics, and data science can surpass qualitative interpretations of results obtained separately in different fields. Within a Bayesian framework, we will propose a fully probabilistic approach to explore the ensemble of small-scale earth models compatible with a set of observations and constraints.
Speaker: Matthieu Lusseyran
Date: Thursday 4th of April 2024, 1:15pm.
Abstract:
Understanding the damage processes in clay-bearing rocks is a decisive factor in geological engineering, and for instance considering nuclear waste deep geological repositories. But, more generally they may also contribute to localized deformation, and thus the rupture of fault gauges in seismic zones. However, owing to their complex mineralogy, multiscale microstructures and anisotropy, the mechanisms of clay-rich rock damage and their chronology are not yet well understood. Here we focus on the impact of micro-damage on ultrasonic wave propagation velocity, which is confronted with the corresponding full deformation fields calculated by digital image correlation (DIC). The aim is to associate the acoustic signature with the active deformation mechanisms identified by DIC. To this end, an integrated experimental approach is proposed to characterize localization and to identify the related deformation micro-mechanisms during uniaxial compression of natural clayey rock samples (Tournemire shales) with two simultaneous measurements: 1) the evolution of P-wave velocity within the sample by active acoustics, 2) the development of the 2D mechanical full field by digital image correlation. Both experimental techniques are well known, but the innovation of our approach is to combine simultaneously both measurements. Deformation localization is a multiscale problem, which obviously occurs at the sample scale, but also at the fines scales of the microstructure. Therefore, we developed two different experimental setups. On the one hand, during uniaxial compression with a standard MTS loading frame the macro-scale localization patterns are characterized by optical observations, which image resolution is well suited to the cm sample scale (sample diameter: 3.6 cm and double in length). On the other hand, in order to characterize the initiation of micro-damage at the microstructure scale of the composite type of rock, the same loading protocol is reproduced (while keeping the acoustic diagnosis) on smaller scale mm-sized specimens (sample diameter : 8 mm, double in length), using a home-designed miniature loading frame fit for an environmental scanning electron microscope (ESEM). The latter analysis is carried out under controlled relative humidity of RH = 80%, hence preventing the samples to dry out due to the high vacuum. A similar acoustic signature is identified at both scales of observation, in spite of the variations of experimental conditions imposed by the environmental SEM. We are therefore confident to be able to understand the fracturing process from micro-cracking initiation (microscale) to sample failure (macroscale), and to assess its impact on ultrasonic wave propagation.
Speaker: Janne T. Yliharyu
Date: Friday 22nd of March 2024, 1:15pm.
Abstract:
In many applications, it is essential to know what happens to the properties of a material if the water content is changed, i.e., if the material is wetted or dried [1]. One such example is bentonite clay, which is planned to be used as a buffer material in the geological final disposal of radioactive waste. Safety assessment of the multibarrier system, including the bentonite buffer, requires modelling and investigating the release barrier over long periods. These models benefit from information on the local water content of the bentonite layer. To facilitate modelling, we have developed X-ray tomographic methods to measure the time evolution of the local water content and applied the methods to bentonite clay samples in various conditions. This approach provides direct experimental data on the wetting and hydromechanical behavior of small-scale compacted bentonite samples, which can be used to validate material models. We present two approaches for measuring local water content. The first one is based on comparing successive tomographic images obtained during the wetting of the bentonite clay sample. This method involves measurement of the internal deformations of the sample, which is vital because, in general, materials will deform if the water content is changed. The second method is based on dual-energy imaging. Energy information is utilized to obtain quantitative information about the spatial distributions of bentonite and water. In both cases, we have applied a nonstandard beam hardening correction and careful calibrations of the mass attenuation coefficients. The deformation measurement method has been applied extensively to bentonite samples with different dry densities. Also, the influence of temperature on water transport has been studied. Based on the observations, the evolution of the water distribution is qualitatively reminiscent of diffusive transport. The first results from the dual-energy CT are promising, but further studies and validations are required before more profound conclusions can be drawn. These results suggest that deformation measurement enables more accurate water content estimates, but dual-energy CT can be applied in a broader range of material combinations.
Speaker: Claire Bossennec
Date: Thursday 21st of March 2024, 1:15pm.
Abstract:
La caractérisation et la modélisation précises des milieux géologiques hétérogènes sont essentielles pour optimiser l'exploration et l'exploitation des ressources énergétiques, telles que la géothermie, le stockage thermique en profondeur et les processus de co-production d'énergie (lithium, hydrocarbures). Cette présentation propose une approche intégrant des analyses multi-échelles et multi-méthodes, depuis l'observation de terrain jusqu'à la modélisation numérique, incluant également des expérimentations de laboratoire. La combinaison avec de l’imagerie géophysique et l’analyse des propriétés thermo-physiques et pétrophysiques permet de couvrir un spectre allant du microscopique au macroscopique, et d'englober des techniques avancées pour examiner en détail les structures et mécanismes régissant les transferts de chaleur et de matière dans des milieux géologiques complexes à différentes échelles temporelles. La caractérisation des propriétés de transfert des grès et de leur évolution par diagénèse et déformation constitue la base des travaux de thèse et sont une première étape des travaux présentés.
Les études de cas variées illustrant l’approche développée, issues de projets tels que INTERREG DGE-ROLLOUT, MEET, SKEWS, PotAMMO, PUSH-IT et AMPEDEK, complètent et explorent des contextes géologiques divers (sédimentaires, volcano-sédimentaires et cristallins), permettant une meilleure compréhension de la complexité et de l'hétérogénéité des formations géologiques visées. Ceci facilite une évaluation plus précise du potentiel géothermique et de stockage. Les résultats obtenus jouent également un rôle clé dans l'amélioration de la production d'énergie géothermique, le perfectionnement des stratégies de stockage thermique et l'évaluation des opportunités de co-production d'énergie, tout en respectant les contraintes environnementales et économiques.
L’ensemble des travaux souligne l'importance vitale d'une analyse détaillée des propriétés et processus régissant les milieux géologiques pour un développement responsable des ressources énergétiques souterraines. En ouvrant la voie vers des stratégies de gestion énergétique plus efficaces, ces travaux proposent plusieurs innovations concrètes pour répondre aux défis de la transition énergétique et à la nécessité d'une sécurité énergétique renforcée par une production plus locale et intégrée.
Speaker: Charlie Garayt
Date: Thursday 14th of March 2024, 1:15pm.
Abstract:
This work aims to simulate 2D structural geological models, or geomodels, that respect given knowledge and data. By definition, geomodeling is an ill-posed problem due to the limited quantity and quality of available data. Current geomodeling methods struggle to both characterize uncertainties and produce realistic geomodels.To achieve this goal, a deep generative adversarial network (GAN) has been implemented. GANs, which are usually used in image generation, need a large training dataset. For instance the ImageNet dataset contains more than 1.5 million images. In geomodeling, such a large real dataset does not exist. Fortunately, geological structures are a consequence of physical and chemical processes, so creating a synthetic dataset is feasible from the simulation of these processes. The training dataset is created from Noddy, which can be viewed as an object-based simulator. The use of advanced GANs like Least-Square GAN (LSGAN) and Wasserstein GAN allows training a deep neural network called the Generator. The Generator defines an implicit distribution of geological models. This is a function that transforms a random vector into a geomodel similar to the training dataset. However, the Generator produces unconstrained geomodels. A Bayesian approach is used in order to generate geomodels that fulfill constraints, or conditioned geomodels. Thanks to the versatility of the Bayesian approach, constraints can be of different types and quality, for instance rock type, rock orientation or geophysical data. A modified Metropolis adjusted Langevin algorithm (MALA), which is a Markov chain Monte Carlo method, enables the acquisition of conditional geomodels. Finally, the combination of MALA and the GAN allows the generation of conditioned geomodels. In addition, this approach enables handling uncertainties and to perform computation, since the final results is an implicit distribution of conditioned geomodels.
Speaker: Baptiste Rousset
Date: Wednesday 21st of February 2024, 1:15pm.
Abstract:
Like earthquakes, transient slow slip events participate in the release of accumulated strain at plate boundaries. They occur in various tectonic contexts including shallow creeping sections of strike slip faults, the brittle ductile transition of subduction zones in concurrence with tectonic tremors and the seismogenic part of subduction zones, sometimes associated with seismicity. In this talk, I will present the advantages and limits of geodesy techniques including GNSS, InSAR and tiltmeters, to characterize slow slip events in these various contexts through three case studies. The first one will be on the Izmit section of the north Anatolian Faults in Turkey, the second on the Mexico subduction zone and the third on the Sagami trough in Japan. I will finally present an ongoing project to measure the transient aseismic deformation associated with deep geothermal wells in Alsace.
Speaker: Augustin Gouy
Date: Thursday 15th of February 2024, 1:15pm.
Abstract:
In karst aquifers, groundwater flow is highly influenced by the interconnected underground cavities and conduits that form the karst network. Modeling karst flows requires the use of spatially distributed approaches accounting for these networks. Their exploration is, however, often complex, and mapping them using indirect methods such as geophysical ones has proven challenging. To overcome these limitations, stochastically simulating discrete karst networks should account for the uncertainties on conduit position and geometry. Only a few existing methods can reproduce realistic and diverse karst morphologies. We present a public C++ code, KarstNSim, for simulating discrete karst networks, that incorporates field data to generate a range of possible karst network geometries. It relies on the computation of the shortest path between the inlets and outlets of the network with the use of an anisotropic cost function defined on an n-nearest neighbor graph conformal to geological and structural heterogeneities. This cost function represents the physico-chemical processes that govern speleogenesis – such as erosion and chemical weathering – providing simplified control over the geometry of the generated networks. Our approach reproduces the vadose-phreatic partition visible in the karst networks, by generating sub-vertical conduits in the unsaturated zone and sub-horizontal ones in the saturated part. It encompasses geological parameters such as inception surfaces, fractures, permeability, and solubility of layers, along with considering the hydrological context of recharge by assigning relative weights to the inlets. A demonstration of how to use the code will be provided, covering the process from defining inputs to displaying outputs on a geomodeling software.
Speaker: Yu Zhang
Date: Friday 26th of January 2024, 1:15pm.
Abstract:
Natural gas hydrate is regarded as a potential unconventional alternative clean energy with development prospects because of its huge reserves, high calorific value, and environmental protection. The Outline of China's Medium-and Long-Term Science and Technology Development Plan (2006-2020) clearly puts forward "breaking through the safe exploitation technology of natural gas hydrate"; China's 14th Five-Year Plan and the long-term goal outline for 2035 also put forward "trial production of natural gas hydrate in South China Sea and other areas". In the process of gas hydrate exploitation, the reservoir pores are in the state of gas hydrate solid-decomposed liquid-decomposed gas coexisting and filling, showing the characteristics of "high porosity, gas-water permeability and slightly low strength"; Macroscopic performance is reservoir compaction and wellbore collapse, and microscopic performance is pore structure and gas-water permeability evolution, which directly affects wellbore stability; The above problems are bottlenecks faced by efficient exploitation of natural gas hydrate
Speaker: Guillaume Pirot
Date: Thursday 18th of January 2024, 1:15pm.
Abstract:
The mining industry faces various challenges to satisfy the increasing demand of minerals for clean energy transitions. One of them is to improve geological characterization at the stage of undercover exploration. This usually requires costly drilling operations. Can we reduce our exploration footprint and costs while reducing geological uncertainty by optimising our drilling location design? Here I propose two ways to look at this problem, based on data from the Hamersley basin, WA, Australia. One examines legacy drillholes in the area and investigate the impact of removing drillholes on the quality of geological models. The other one explores an iterative drilling optimisation approach to reduce modelled geological uncertainty, based on a synthetic case.
Speaker: Franck Sfiligoi-Taillandier
Date: Thursday 11th of January 2024, 1:15pm.
Abstract:
Les modèles Agent (ou modèles multi-agents), issus du domaine de l’Intelligence Artificielle, sont particulièrement intéressants pour la modélisation et la simulation des systèmes sociaux et sociaux-techniques. Contrairement aux approches par apprentissage (Deep learning, Machine learning…), ils passent par une modélisation et une simulation explicite et naturelle des entités constitutives des systèmes et de leurs interactions, ce qui permet de faciliter la compréhension du système et de l’explorer. Ce côté intuitif et ouvert (par opposition aux boites noires), rend ces modèles particulièrement adaptés aux approches de modélisation et de simulation participatives. La modélisation participative est une approche dans laquelle les parties prenantes sont directement impliquées dans la construction du modèle. La simulation participative renvoie à une simulation interactive dans laquelle l’utilisateur peut modifier le cours de la simulation ; cette approche est beaucoup utilisée dans le cadre des jeux sérieux. Dans cette présentation, je reviendrai sur ces différents éléments en me basant sur des applications liées à la gestion des territoires, et en essayant de vous convaincre de l’intérêt de ces approches dans un dispositif d’aide à la décision ainsi que de leur valeur scientifique.
Speaker: Simon Daout
Date: Thursday 21st of December 2023, 1:15pm.
Abstract:
In this presentation, I will present you the results obtained in a recent publication entitled: « Along-strike variations of strain partitioning within the Apennines determined from large-scale multi-temporal InSAR analysis, https://doi.org/10.1016/j.tecto.2023.230076 ». In this paper, we produced and analyzed a large set of GPS and InSAR measurements over the Apennines (Italy), where the origin of the long-wavelength topography and the driving mechanisms of the extension are debated. Our continuous mapping of the surface displacements across the range allows to understand how the different deformation scales (fault tectonic, landslides, surface processes, hydrological loading, mantle processes) are imbricated and how faults behave in relation to driving mechanisms at the boundaries of the system. Although, I hope this presentation can be relevant from the scientific point of view of various fields (seismology, geomorphology, geodynamic, remote sensing), here, I would like to particularly focus on numerical models and the development of inversion tools to derive the spatio-temporal evolution of seismic and aseismic slip on faults.
Speaker: Thibault Faney
Date: Wednesday 13th of December 2023, 1:30pm.
Speaker: Ahmad Marvi Mashhadi
Date: Thursday 30th of November 2023, 1:15pm.
Abstract:
At the present time, however, the geological storage of H2 remains very little studied despite of the specific behaviour of this gas. A key point in the development of such technology is to characterize and constrain the biological processes that could alter qualitatively and quantitatively the resource within the storage framework in porous reservoir rocks. As first electron donor for life and crucial energy source for subsurface microbial processes, indeed, H2 allows the autotrophic growth of microorganisms under oligotrophic conditions (i.e. limited supply of carbon) in deep environments. In the presence of an available terminal electron acceptor such as nitrate (NO3-), ferric iron (Fe3+), sulfate (SO42-) or carbon dioxide (CO2), H2 is susceptible to be consumed by microorganisms to gain energy. To date, unravelling the contribution of H2-consuming microbes in the biogeochemical cycle of hydrogen is of high importance in a number of subsurface industries including H2 gas storage in the energy transition context. Particularly, bacterial activity is susceptible to produce methane (CH4) or hydrogen sulfide (H2S) to the detriment of H2. The main objective of this work will be to evaluate the kinetics of H2 consumption by bacterial model strains and multi-bacterial consortia under geological storage conditions in terms of temperature, pressure, salt concentration and electron acceptor availability. Batch and flow-through experiment will be designed to reproduce these storage conditions.
Speaker: Amandine Fratani
Date: Thursday 23rd of November 2023, 1:15pm.
Abstract:
During geological exploration, interpretation of faults can be ambiguous and uncertain because of disparate and often sparse observations such as fault traces on 2D seismic images or outcrops. The problem of associating partial fault observations was considered by Godefroy et al (2019), who decided to define a graph where each possible association of two fault observations (the graph nodes) are represented by an edge. The likelihood of this association was computed by using expert geological rules. However, fault observations are not pairwise independent, which limits the consideration of higher-order effects. For instance, the multiple-point association can be used to infer the evolution of the throw along the fault. In addition, the definition of rules in a multiple-point problem is also difficult because of the very large number of cases to consider. Here, we propose a machine learning approach to compute the likelihood of three-point fault data association. First, a computation of fault features (i.e. the length of the fault trace) from sections extracted from known 3D geological models is realized to create a data set of fault observations. The supervised machine learning problem is formulated as a classification problem to determine the probability that 3 fault observations belong to the same fault objects based on the feature vector. To prevent overfitting, we propose to mimic a partly interpreted case: we split the 3D domain in two disjoint sectors A and B, and use only data from sector A as training and data from sector B to test the method. However, the results are not conclusive, so an analysis of the features are proposed to choose the correct ones. At the same time, methods to deal with imbalanced dataset are explored.