Nicolas Cherpeau is working on structural uncertainties. During the process of reserves estimation, structural uncertainties may be as important as the estimation itself. Indeed they play a key role in the reservoir volume since a small variation in the localization of a horizon or fault may change the reservoir volume in an important manner. While most of the current techniques only deal with the geometry, the aim of Nicolas Cherpeau’s research is to generate models that modify not only the geometry of the horizons or faults but also the topology of the 3D models. Indeed, structural uncertainties relative to faults are also important in term of fluid flow simulations as faults usually play a role of sealing barriers or drains. Consequently, a change in the connection between two faults may bias fluid flow predictions. To handle this issue, a change in the fault geometry is not sufficient, it is necessary to modify the topology of the fault network. An implicit approach is used for its advantages for topological changes. Fault relationships can be described by a binary tree, each fault dividing the model in two blocks. The generation of several trees with specific constraints enables to obtain fault networks with different topologies. Once the topology has been modified, it is possible to perturb the geometry using geostatistical methods. Then, the generated models may be validated using restoration techniques or history matching.

Supervisors: Bruno Lévy (LORIA), Guillaume Caumon (CRPG-ENSG) and Jean-Jacques Royer (CNRS)