Abstract
The discovery of giant hydrocarbon reservoirs, easily exploitable and profitable is now rarely relevant. Today, reservoirs are smaller, more heterogeneous, so it is necessary to optimize their exploitation by complex and expensive production methods, such as the injection of water, gas, or chemicals. Despite the considerable advances in the fields of seismic acquisition and processing there is a lack of seismic resolution at depths compatible with hydrocarbon traps. Also, well data provide precise information, but they are available only locally. These limitations induce an uncertain knowledge of the petroleum reservoir. To overcome the lack of knowledge, geoscientists rely on multiple realizations brought by geostatistical methods. However, they are not always representatives of the reservoir geological reality.The PhD work seeks to exploit the 3D images provided by a shallow broadband seismic in similar sedimentary systems, to populate equivalent geomodels with realistic heterogeneities more representative of the reservoir.Our approach includes 3 stages:1. The first step is to rely on the high resolution of broadband seismic, especially in the shallow part, where it is possible to finely observe the geometry of the bodies and the connectivities in three dimensions to create 3D seismic analogs. 2, Once the 3D seismic analogs have been created, we propose to store them in a database as a continuous, non-discrete property. Thus, the seismic analog, under certain conditions can be edited (distorted or simulated) to adapt to the context of the target reservoir. 3, The third step is to transfer the geological information finely described in the 3D seismic analog to a geomodel based on a morphing algorithm developed during the PhD