Mapping reservoir pressure and saturation changes using seismic methods - possibilities and limitations

In this paper, travel-time and amplitude information from time-lapse seismic data are combined to improve the mapping of an over-pressured reservoir segment. The increased pore pressure is caused by water injection into the segment. According to RFT measurements in the well, the pore pressure increase is of the order of 50-80 bars. The observed amplitude increase (between the base line and the monitor survey) associated with the top reservoir event is of the order of 100-200%. A rough estimate of the velocity changes needed (assuming that the density change is negligible) to explain this amplitude change is a 20% reduction in the P-wave velocity. The observed pull-down effect (time-lapse travel-time shift) measured over the whole reservoir thickness is of the order of only 4-6 ms, indicating a gradual change in velocity below the top reservoir event, decreasing with depth. Interpretation of time-lapse seismic data depends on knowledge of the relationship between the seismic parameters and the typical reservoir parameters we want to map, such as, for instance, pore pressure changes and fluid saturation changes. The established link for bridging the gap between the two types of parameter is rock physics. For saturation effects, the Gassmann equations form a reasonable working basis, enabling us to make quantitative estimates of how much each of the seismic parameters changes with, for instance, water saturation. However, for pore pressure changes, the theoretical basis is weaker, and we have to rely on ultrasonic core measurements in order to establish a link between pore pressure changes and the corresponding changes in seismic parameters. One major weakness of the core measurements is that the core sample is damaged during the coring process (Nes et al. 2000). Artificial cracks are probably formed during the anisotropic stress unloading process. Even if the core sample is reloaded to simulate in situ stress conditions for the ultrasonic core measurements, it is not very likely that the original crack state of the sample is re-established. Furthermore, we have to deal with the upscaling problem for our core measurements: Are the ultrasonic measurements made on the small core sample representative at a seismic scale?