Discrimination des fluides à partir de la sismique en réservoir gréseux à huile lourde peu consolidé par des techniques statistiques : applications à la caractérisation de réservoir et au monitoring sismique

摘要

This thesis studies lithology and fluid discrimination from well and seismic data in the unconsolidated heavy oil sand Morichal reservoir (Orinoco Oil Belt in Venezuela). In this kind of reservoirs, lithology and fluid discrimination is very complex. Lithologies are very hard to differentiate because of their similar petrophysical and elastic properties. As well, wet sand and oil sand have very similar elastic properties. In the first part of this thesis, we study lithology and fluid discrimination using well data by pattern recognition techniques. Then we analyze petrophysical and elastic relationships on well data : small variations in grain size or cement percentage between grains produce velocity changes that could be used for lithology and fluid discrimination. Furthermore a well-to-seismic tie is tried without success, due to strong variations in seismic phase and amplitude. To understand the reasons of this problem, the second part of the thesis is dedicated to the building of an integrated 3D model of Morichal reservoir, populated with facies, petrophysical and elastic properties. This model is then used for 1D and 2D seismic modeling, in order to study anisotropy and attenuation effects on the seismic signal. Anisotropy may induce polarity inversions for large incidence angles, thus affect Normal Move Out corrections. Attenuation due to heavy oil may produce phase delay and amplitude weakening. Therefore anisotropy and attenuation can lead to problems in angle stacks and well-to-seismic tie. From seismic modeling results, some hints are finally given to improve angle stacks and well-to-seismic tie in reservoirs from Orinoco Oil Belt.