Quantitative 4D seismic interpretation can be successfully achieved by exploiting the causal link between the temporal variation in well activity and the 4D seismic signatures they induce. This is achieved by capturing in mathematical form the common interpretational practice of identifying the origin of dynamic signals in the 4D seismic volumes or maps on the basis of their association with a particular injector or producer. Thus, for example, a region of reservoir hardening (impedance increase) around a producer may be interpreted as a signal of pressure decrease. Similarly, an area of softening (impedance decrease) around an injector is interpreted as a signal of pressure increase when pressures are above bubble point. In the literature, a hardwired integration between the seismic and engineering domain has been obtained to some extent using methods such as seismic history matching, where the observed seismic and well production history data are simultaneously fit by predictions from a common simulation model. However, this approach is computationally expensive and suffers from nonuniquenesses, inaccuracies in the petroelastic model, and is ultimately only as accurate as the model itself. As an alternate approach, we reconcile here the well production history and time-lapse seismic data in the data domain without the need for a model. This approach involves the use of many frequently repeated seismic surveys shot over the same field, and mathematically correlates changes in the mapped seismic attributes directly to the fluid volumes injected and produced from the wells. Thus, well data normally used exclusively for history matching in the reservoir engineering domain can now also be directly integrated with the time-lapse seismic data.

You do not currently have access to this article.