ABSTRACT

Time-lapse seismic modeling is an essential step for any seismic reservoir monitoring prospect. Different algorithms with various degrees of accuracy can be utilized to image fluid flow. Highly accurate seismic modeling techniques, e.g., finite difference (FD), are capable of simulating different events including primaries, internal multiples, converted waves, diffractions, etc. However, the computational speed of such methods is considerably less than ray based-methods. Now, the question is do we require highly accurate modeling to seismically detect fluid flow or are approximate, ray-based and hybrid, modeling techniques adequate. To investigate this problem, we conducted a systematic modeling study over a recently developed petro-elastic model. A base and two monitor surveys were simulated and the corresponding time-lapse signatures were analyzed. Our analyses demonstrated that internal multiples behind the waterfront, flooded zones, partially subtract out in time-lapse differencing. We also found that for time-lapse seismic modeling, acoustic modeling of an elastic medium is a good approximation from near to middle offsets. From middle to far offsets, differences between elastic and acoustic wave propagation is the dominant effect rather than internal multiples and converted waves. We also found that time-lapse modeling of the reservoir using split-step Fourier plane-wave (SFPW) approach is computationally fast compared to FD. It is capable of handling higher frequencies than FD and provides an accurate image of the waterflooding process comparable to FD.

This content is only available via PDF.
You can access this article if you purchase or spend a download.