Subsurface models are essential to build knowledge and support decisions about natural resources, energy transition and hazard management, but they can be affected by large uncertainties due to the lack and incompleteness of subsurface data. We perform Research for Integrative Numerical Geology to better account for geological concepts in interpretation of geoscientific data and to sample and reduce the associated uncertainties. Our ultimate objective is to effectively describe geological objects with evolving and adaptative numerical models, integrating geometry and physics at various space and time scales, with real-time updates. This translates into theoretical work and novel technologies to efficiently represent, build, update and ask questions to deterministic and stochastic subsurface models.

Our research primarily focuses on the geometry, topology and properties of geological objects. It is declined into four main research areas:

Stochastic structural and stratigraphic modeling

The goal is to capture the main uncertainties in the interpretation of tectonic and stratigraphic structures by combining data and ancillary conceptual and regional knowledge. We see this research as an essential component of geomodels because fractures, faults and folds exert a tremendous control on subsurface heterogeneities.

In spite of the steady quality increase of subsurface imaging methods, the identification of tectonic and stratigraphic structures often raises interpretation challenges due to uncertainties. The goal of this research is to capture these uncertainties by combining data and ancillary conceptual and regional knowledge. It also aims at putting more objectivity during structural interpretation by explicitly and quantitatively stating hypotheses.

We see this research as an essential component of geomodels because fractures, faults and folds exert a tremendous control on subsurface heterogeneities and often have a first order impact on flow, wave propagation and deformation in the underground.

Accounting for uncertainties in structural interpretation requires to increase the robustness and automation of structural modeling methods. It is also essential for the meaningfull integration of indirect subsurface data such as reservoir production curves.

Stochastic sedimentary and diagenetic objects description

The goal is to describe or simulate geological objects created by depositional and diagenetic processes consistently with observations and genetic or pseudo-genetic concepts. Such realistic descriptions are important to capture the spatial structures of hydrodynamic and mechanical heterogeneities.

 

Heterogeneities in the subsurface result from a succession of sedimentological, diagenetic and mechanical processes. Classical geostatistical techniques tend to ignore these concepts and are generally implemented on a fixed-geometry grid whose resolution has to be decided a priori. The goal of this research is to describe and simulate sedimentary and diagenetic objects consistently with observations and genetic or pseudo-genetic geological concepts. This translates into innovative ways of combining points, lines, surfaces and volumetric grids to allow for flexible representations of geological objects at multiple resolutions.

The resulting models aim at capturing spatial structures and the most relevant hydrodynamic and geomechanical heterogeneities. They can either be used directly or as training images for multiple-point geostatistical simulations.

We are also working at producing a workflow of modelling of channelized systems which would propose the integration of conditioning data and geologic processes. The aim is to propose a model reproducing the complexity and connectivity of the geological bodies for a better precision of flow simulation.

Adaptive gridding and scale management

The goal is to establish simple links between geological models and the physical processes involved in several indirect subsurface measurements (e.g., reservoir production, deformations, waveforms, gravity anomalies, etc). Controlling the level of detail and consistently simplifying the geomodel is essential to ensure efficient and accurate physical computations such as flow simulation.

The goal of this research line is to establish simple links between geological models and the physical processes involved in several indirect subsurface measurements (e.g., reservoir production, deformations, waveforms, gravity anomalies, etc). Controlling the level of detail and consistently simplifying the geomodel is essential to ensure efficient and accurate physical computations such as coupled thermo-hydro-mechanical and chemical simulations.

The accurate representation of geological features in the computational support is an essential aspect of this research. This motivates the use of unstructured meshes, which can adapt to complex geological features and technical objects such as boreholes while offering local adaptiveness. An important aspect of RING research concerns the generation of these meshes based on prior knowledge.

Because the needed level of detail and computational resources may vary from one application to another, we also do research on the scaling of petrophysical properties, both in unstructured and structured meshes.

Physical processes:

We investigate the relations between geological structures and the physical processes in the subsurface (mechanical evolution, wave propagation, flow and coupled physical problems). This research concerns the past and current geometry and physical state of the subsurface.

  

We are interested in using static geomodels to explore the impact of geological heterogeneities on various physical processes such as fluid flow, wave propagation and mechanical deformation. As all these processes are non-linear, a small geomodel perturbation may, in principle, have a relatively large impact on the physical response. Our goal is to understand which geological parameters have an impact for a given physical process. This paves the way for reducing geological uncertainty by solving inverse problems. In addressing these issues, we consider the complementarity between classical bottom-up approaches (which mainly defines model parameters based on prior geological knowledge) and the top-down approach (where spatial complexity emerges from the inverse process).

Restoration and Geomechanics

Geological structures originate from and exert a control on the mechanical evolution of the subsurface. This research line aims at defining new ways to assess the relations between the geometry and the stress state of the subsurface at different time scales.

At geological time scale, 3D structural restoration has implications for testing the structural consistency of 3D structural interpretations and for quantitative modeling of the paleo-geometry of basins and reservoirs. It can also provide guidelines to understand how strain localized and how fractures developed through time. 

At human and natural resource management time scale, this research aims at better using prior geological knowledge in stress predictions. This has implications for drilling operations and for modeling the generation or reactivation of fractures and faults due to variations of pore pressure and effective stress.

Waves

We are working on effective elastic medium computation for elastic wave propagation. Depending on the wave bandwidth, we have developed two codes to compute equivalent medium properties from fine-scale descriptions of the medium. We also look at new ways to appraise static geomodels from seismic data. this can be useful for veryfying the consistency of interpretations, and also to test the value of new seismic experiments in reducting geological uncertainty.

Flow

We generate different kinds of flow simulation grids (corner-point, unstructured control-volumes or unstructured CVFE grids) to model fluid flow in the subsurface using external flow simulators such as Schlumberger's Eclipse, Stanford's GPRS or Melbourne/ETH's CSMP++.  Ahead of the flow simulation, we also work on using connectivity information to characterize complex geologic media, e.g., using percolation theory.