Reservoir simulation is an important tool for field planning and reservoir management, but it depends on input data and assumptions. If these conditions are not known accurately, which is the common case, the predictions from the simulation would be also inaccurate, or even misleading, in application. It is therefore essential to have an independent observation to guide the modelling process.
In this thesis, we use time-lapse seismic data as a constraint on reservoir simulation with the aim of improving the accuracy of reservoir description and simulation results. It appears based on the time-lapse seismic that two groups of shear fractures across the horizontal well bores were generated during the stage of injection and they may have prevailed in the entire recovery process. Through the fluid flow network of both fractures and well bores, steam in a few locations was driven to the bottom of the reservoir and further leaked to the underlying formations, even deep into Devonian carbonates. The break-through into the overlying formations may also have happened, as evidenced by reflection event disturbances. The interpretations suggest a new pattern of fluid flow within (and outside) the reservoir, and they are the crucial information for input data modification and assumption adjustment. Reservoir simulation coupled with geomechanical modeling was run with the interpretations applied, and the results are considerably different compared to those without the time-lapse seismic investigation.
Time-lapse seismic modelling based on the parameters acquired with the updated reservoir simulation indicates that the seismic observables of anomalies (newly created events, amplitude boost and event time delay) are consistent with those identified in the real time-lapse seismic. This is additional evidence to validate the reservoir simulation.
View full article as PDF (29.04 Mb)