Skip to Main content Skip to Navigation

Origine, caractérisation et distribution prédictive des structures karstiques. De la karstologie aux modèles numériques 3D.

Abstract : The characterization of underground heterogeneities is one of the main concerns for qualifying and quantifying petroleum or water reservoir behaviors. In the case of karst reservoirs, the regional heterogeneity is mainly due to the presence of a hierarchical network of conduits, connecting a recharge zone to a restitution zone, and behaving as drains for underground fluid flows. However, the humanly observed karst conduits represent only a limited part of the complete karst conduit system. Therefore, many uncertainties exist concerning the location, the geometry and the dynamic behavior of a karst network at the massif scale. Geostatistical stochastic simulations represent an interesting tool to study the different three-dimensional (3D) probable configurations of karst networks and then, to determine the uncertainties on the reservoir behaviors. This problematic is a major challenge not only in the field of water, hydrocarbon and mineral resources, but also in civil engineering. This approach first requires understanding the successive stages of karst structuring of a reservoir and then to numerically reconstruct the 3D organization of karst structures, given the uncertainties that may exist. Thus, a multidisciplinary and methodological approach was proposed in this work to address these issues. It is based on (i) field geology and karst studies and (ii) on numerical geology, including 3D structural and geostatistical modeling. The hierarchical organization of the karst networks within the geological host represents a key particularity of these features. Their distribution and organization is closely related to static (geological) and dynamic (hydrological) parameters, and depends on the geomorphological evolution of the studied massif. The analysis of the karst structuring steps of a reservoir is here carried out with two main objectives: the determination of karst structuring phases to understand and predict the functioning of the system, and the definition of conditioning data for simulating realistic karst networks and reducing uncertainties. Karstological determinations allow identifying the morphogenesis of the exo- and endokarst forms. Then, the speleogenetic reconstitutions make it possible to identify the successive phases of the karst system structuration (epigenic or hypogenic speleogenesis, evolution of base level, etc.) allowing to define the type and the zonation of karst pattern occurrence (karst zonation : vadose, epiphreatic or phreatic zone). In parallel, a quantitative analysis of the geometries and the topologies performed on 3D cave surveys permits to compare the different organizations of the karst network patterns, related to speleogenetic processes. This morphometric analysis provides a quantitative database of morphological characteristics according to their speleogenetic processes. Finally, two geostatistical approaches were tested to generate karst networks. They correspond to two classical pixel-based geostatistical methods: the sequential indicator simulation (SIS) and the multipoint simulations (MPS). This work focused on the integration of different data sources (geology, karstology, hydrology) as well as on the ability to reproduce the observed hierarchical organization of karst systems. For instance, the aim was to integrate a priori knowledge as hard data (e.g., inlet and outlet points, known cavities) and as secondary data (trends, proportions, probability of occurrence). These approaches make it possible to simulate various karst patterns inherent to a karst reservoir zonation in relation to the speleogenetic history. They have been tested on synthetic cases for validation and one of them on a case study.
Complete list of metadata

Cited literature [393 references]  Display  Hide  Download
Contributor : Bruno Arfib <>
Submitted on : Wednesday, July 10, 2019 - 10:18:16 AM
Last modification on : Wednesday, October 14, 2020 - 3:49:29 AM


Files produced by the author(s)


  • HAL Id : tel-02178707, version 1


Johan Jouves. Origine, caractérisation et distribution prédictive des structures karstiques. De la karstologie aux modèles numériques 3D.. Géologie appliquée. AMU - Aix Marseille Université, 2018. Français. ⟨NNT : 2018AIXM0153⟩. ⟨tel-02178707⟩



Record views


Files downloads