On the June 23^th , 2015, Charline Julio presents her work on conditioning in stochastic structural modeling.

Summary

Faults are discontinuities in rock volumes that affect mechanical properties and flow paths of hydrocarbon reservoirs. Three-dimensional modelling of faults is often critical to the understanding and the simulation of physical phenomena such as fluid flows. However, subsurface modelling remains limited by the incompleteness and resolution of available data, so that uncertainties remain on the geometry and the connectivity of fault networks. To assess fault network uncertainties, several stochastic approaches have been introduced in the literature. These methods generate a set of possible fault models conditioned by reservoir data. Nonetheless, the conditioning of fault networks by both hard data (e.g., seismic data, well data) and geological concepts remains a delicate point of stochastic simulations.

In this thesis, we investigate two main conditioning strategies of stochastic fault modelling methods. The first one takes into account the observations of the fault absence, for instance, as indicated by seismic reflector continuity. To do this, the reservoir volume is divided into two sub-volumes delimited by a 3D envelope surface : (1) a volume where no faults occur, and (2) a potentially-faulted volume. Then, faults are simulated and optimized in such a way as to be entirely confined to the potentially-faulted volume. The second presented strategy deals with the uncertainties related to the seismic interpretation of fault segmentation. It generates a set of fine-scale segmented faults from a larger-scale and continuous interpretation of the fault. The method uses the orientation variations of the continuous fault to subdivide it into several possible fault segments. The effects of the different segmentation configurations on flow simulations are studied. The results show relationships between fault connectivity and reservoir dynamic behaviour.

The different studies presented in this thesis highlight the importance and the difficulty in generating a set of models coherent with both hard data and geological concepts. The introduced strategies ll a part of the gap related to stochastic fault simulation conditioning and open new avenues for the development of new reservoir work flows that take structural uncertainties into account.

PhD committe members:

• Pierre Thore (Total)
• Catherine Homberg (UPMC, Paris)
• Roger Soliva, Université de Montpellier
• Jean-Paul Chilès, Mines Paris Tech
• Paul Cupillard, Université de Lorraine

Advisors:

• Guillaume Caumon, Université de Lorraine
• Mary Ford, Université de Lorraine

What is next?

Charline Julio is now working for PDS in The Hague.