Résumé
Quantitatively assessing attenuation and dispersion of elastic‐wave velocities in partially saturated reservoir is difficult because of its sensitivity to fluid distribution. We conducted experiments on homogeneous Indiana limestone samples, partially saturated by two methods: drying and imbibition which lead to different fluid distribution for a given saturation. Forced oscillations (from 0.004 to 100 Hz) and ultrasonic (1 MHz) measurements were done under confining pressure to measure the change of elastic moduli with frequency and their attenuation. Our measurements show that compressional (P‐)velocities are strongly sensitive to the sample’s saturation method. For high saturations (above 80%), obtained by drainage, compressional velocities are frequency dependent, and clear peaks of attenuation can be observed. However, at the same saturations obtained by imbibition, no dispersion or attenuation is observed. In addition, shear velocities show little variation with frequency, saturations, and fluid distribution. The dispersion and attenuation of P‐velocities are shown to be influenced by the pore fluid distribution, which was investigated using micro‐computer‐assisted tomographic (CT) scans. Furthermore, a numerical model developed within the framework of poroelasticity’s theory predicts well the experimental results, using the fluid distribution obtained from CT as an input. Our results show that the velocity dispersion was related to wave‐induced fluid flow at mesoscopic scale controlled by the geometry and distribution of the gas patches. Plain Language Summary It is difficult to quantitatively assess frequency‐dependent elastic‐wave velocities in partially saturated reservoir, due to its sensitivity to fluid distribution. The topic of our study aims at investigating the effect of fluid distribution on frequency‐dependent elastic velocities using rock‐physics experiments. In the experiments, dispersion and attenuation curves observed for a water–air‐saturated porous limestone exhibited distinct variations for the same saturation. The variations were proved to be related to different saturation methods, thus providing physical evidence of the effect of fluid distribution on the frequency‐dependent elastic velocities. Furthermore, micro‐computer‐assisted tomographic scan was used to calculate the fluid distributions from different saturation methods. The scanned fluid distributions, as the inputs of poroelastic model, were able to interpret the dispersion and attenuation observed. The conclusion provides the possibility to quantify the relationship between fluid distribution and seismic attenuation and could be a useful tool for upscaling at reservoir scale. In addition, the study improves our understanding of seismic wave propagation in partially fluid‐saturated rock formations and thus has potential applications in seismic exploration, reservoir production, and for CO2 sequestration in aquifers. Key Points Saturation obtained by drainage and imbibition methods lead to different dispersion and attenuation of elastic‐wave velocities Experimental results on the attenuation and dispersion highlight the importance of the fluid distribution which was investigated using computer‐assisted tomographic (CT) A numerical model based on Biot’s theory using the fluid distribution—from the CT scan as an input—reproduces the experimental results