10Single and multiphase flows in fractured porous media at the scale of natural 11 reservoirs are often handled by resorting to homogenized models that avoid the heavy 12 computations associated with a complete discretization of both fractures and matrix blocks.

Delay To cite this version:André Fourno, Christophe Grenier, Abdelhakim Benabderrahmane, Frédérick Delay. A continuum voxel approach to model flow in 3D fault networks: a new way to obtain up-scaled hydraulic conductivity tensors of grid cells. Journal of Hydrology, Elsevier, 2013, 493, pp.68-80. <10.1016/j.jhydrol.2013 Modelling transfers in fractured media remains a challenging task due to the complexity of the 15 system geometry, high contrasts and large uncertainties on flow and transport properties. In the 16 literature, fractures are classically modelled by equivalent properties or are explicitly represented. 17The new Fracture Continuum Voxel Approach (FCVA), is a continuum approach partly able to 18 represent fracture as discrete objects; the geometry of each fracture is represented on a regular 19 meshing associated with a heterogeneous field of equivalent flow properties. The mesh-20 identification approach is presented for a regular grid. The derivation of equivalent voxel 21 parameters is developed for flow simulated with a Mixed Hybrid Finite Element (MHFE) 22 scheme. The FCVA is finally validated and qualified against some reference cases. The resulting 23 method investigates multi-scaled fracture networks: a small scale homogenized by classical 24 methods and large discrete objects as that handled in the present work. 25 26

Most industrial and field studies of transport processes in Discrete Fracture Networks (DFNs) involve strong simplifying assumptions, especially at the meshing stage. High-accuracy simulations are therefore required for validating these simplified models and their domain of validity. The present paper proposes an efficient workflow based on open-source software to obtain transport simulations. High-quality computational meshes for DFNs are first generated using the conforming meshing approach FraC. Then, a tracer transport model implemented in the open-source code DuMux is used for simulating tracer transport driven by the advection-dispersion equation. We adopt the box method, a vertex-centered finite volume scheme for spatial discretization, which ensures concentration continuity and mass conservation at intersections between fractures. Numerical results on simple networks for validation purposes and on complex realistic DFNs are presented. An a-posteriori convergence study of the discretization method shows an order of convergence O(h) for tracer concentration with h the mesh size.

Résumé -Inversion conjointe des propriétés d'un modèle de fractures pour le monitoring d'un stockage de CO 2 ou le calage d'un historique de production -Que ce soit pour la production de pétrole ou le stockage de CO 2 , le terme réservoir est utilisé régulièrement. Un réservoir désigne une structure géologique hétérogène en types de roches ou en termes de propriétés de perméabilité et de porosité. Dans le domaine pétrolier, de nombreux réservoirs présentent des fractures ou des failles. Les réservoirs fracturés occupent une part importante de la production de pétrole dans le monde (Moyen Orient, Golfe du Mexique, etc.). Malgré la présence de fractures, ces réservoirs n'en restent pas moins de bons réservoirs. Des études de réservoirs dédiés au stockage de CO 2 ont montré de plus la présence de fractures diffuses ou de failles ainsi qu'un fort impact de celles-ci dans le transfert du CO 2 . Un facteur clé de la connaissance des réservoirs fracturés est la compréhension de la géométrie et de la conductivité hydraulique du réseau formé par les fractures. Cette compréhension nécessite la construction d'un modèle de réservoir intégrant l'ensemble des connaissances conceptuelles et des données disponibles sur le terrain. Le présent article expose une méthodologie permettant d'effectuer un calage d'historique d'un modèle réservoir par la modification des propriétés et de la géométrie d'un réseau de failles sub-sismiques. Un modèle original de réseau de failles sub-sismiques est présenté. Il est fondé sur une caractérisation fractale de la géométrie du réseau des failles à partir des données sismiques ou des affleurements. Ce modèle peut être simulé en étendant l'organisation du réseau des failles sismiques à l'échelle inférieure des failles sub-sismiques. Le réseau de failles obtenu peut alors être déformé de façon à refléter le comportement hydrodynamique du réservoir. Cette propriété permet d'effectuer le calage à l'historique de production : les positions et les propriétés hydrauliques des failles incertaines sont modifiées par un algorithme d'optimisation, permettant de réduire l'écart avec les données dynamiques observées. La cohérence géologique du modèle de failles est préservée. La mise en oeuvre des différentes étapes de l'approche proposée est illustrée par une application sur un réservoir synthétique. Abstract

The Fracture Cut (FraC) approach to mesh three-dimensional (3D) Discrete Fracture Networks (DFN) is presented. The considered DFNs consist of a network of planar twodimensional (2D) fractures sharing intersections that can in turn intersect themselves, resulting in highly complex meshing issues. The key idea of FraC is to decompose each fracture into a set of connected closed contours, with the original intersection traces located at the boundaries of the contours. Thus, intersection segments can be more easily accounted for when building a conforming mesh. Three distinct strategies for intersection points management are also proposed to enhance the quality of resulting meshes. Steady-state singlephase flow simulations are performed to validate the conform meshes obtained using FraC.The results from flow simulations as well as from a mesh quality analysis on a benchmark case show that a flexible AoM strategy (Adding or Moving intersection points) appears to be the best choice to generate ready-to-run meshes for complex DFN. This approach also allows accounting for tiny features within the fracture networks while keeping a good mesh quality and respecting DFN connectivity. Finally, a scalability of the mesh generator is conducted to assess the performance of the approach.

Summary Among various ways to extend the lifetime of mature fields, chemical enhanced-oil-recovery (EOR) processes have been subject of renewed interest in the recent years. Oil-wet fractured reservoirs represent a real challenge for chemical EOR because the matrix medium does not spontaneously imbibe the aqueous solvent of chemical additives. The present paper deals with chemical EOR by use of wettability modifiers (WMs). The kinetics of spontaneous imbibition of chemical solutions in oil-wet limestone plugs and mini-plugs was quantified thanks to X-ray computed-tomography (CT) scanning and nuclear-magnetic-resonance (NMR) measurements. Despite the small size of samples and the slowness of experiments, accurate recovery curves were inferred from in-situ fluid-saturation measurements. Scale effects were found quite consistent between mini-plugs and plugs. During a second experimental step, viscous drive conditions were imposed between the end faces of a plug, to account for the possibly significant contribution of fracture viscous drive to matrix oil recovery. The recovery kinetics and behavior, especially the occurrence of countercurrent and cocurrent flow, are interpreted through the analysis of modified forces in the presence of a diffusing or convected WM that alters rock wettability and reduces water/oil interfacial tension (IFT) to a lesser extent. This work calls for an extensive modeling study to specify the conditions on chemical additives and recovery-process implementation that optimize the recovery kinetics.

In the oil industry, complex workflow is used to match or predict fluid production. The large uncertainty of the data can lead to large variability of the simulation results, notably because of the strongly heterogeneous nature of fluid flows. The particular case of naturally fractured reservoirs is well known to be especially difficult to match.This paper presents a method to improve an initial geological analysis, carried out in 2008, through the integration of hydrodynamic data in the fractured reservoir model. Dynamic data such as production information, production logging tools and well tests are used to determine fractures properties by calibrating the fluid flow or reservoir pressure measured at wells. The approach, applied to a real field case, respects both the statistical geological analysis and the dynamic analysis of the production history. Using this methodology the geological structural model based on static characterization are preserved and available.In the study we analysed uncertainties and, in respecting the initial geological model, we prove the presence of compartmentalization in the reservoir by matching 1 year of production and three well tests. Both analytical and numerical flow simulations were used at different scales for time and space: near the well and on the whole reservoir.

No abstract