Algorithms for EOR imaging using crosshole seismic data: an experiment with scale model data

Abstract

Seismic velocities are influenced by many enhanced oil recovery (EOR) techniques. The crosshole seismic survey is well suited to the mapping of injection fluids or fracture patterns following reservoir stimulation. Both traveltime tomography and wave-equation techniques can be used to monitor and map changes in the seismic velocities, given pre-stimulation and post-stimulation crosshole seismic data.

In order to evaluate algorithms for EOR mapping using crosshole seismic surveys, data were obtained from a scale model of a crosshole seismic survey. The epoxy resin model contained simulated geological structures with strong velocity contrasts. Two versions of the same model were constructed, both with and without a simulated “flood” zone of a known geometry. Traveltime tomography and tow wave-equation algorithms, the inverse generalized Radon transform (inverse GRT) and frequency-domain wave-equation imaging, were used to attempt to locate the extent and velocities of the perturbation.

The results of this experiment show that traveltime tomography suffices to locate the flood zone and to determine the magnitude of the velocity perturbations. However, images that resolve the geometry of the flood zone were only obtained when the full waveform was utilized, using either the inverse GRT or frequency-domain wave-equation imaging. In this experiment the best images of the flood zone were obtained using frequency-domain wave-equation imaging. This result is due to the (realistic) complexity of the model, which supports wave types not accounted for by the acoustic ray approximation used in tomography and in the inverse GRT.

Image quality depends on how the input data to the full waveform schemes are generated. For the inverse GRT the change in interface reflectivities due to the flood zone could be detected using preprocessed, “scattered” wave fields as input data. However, better images of the geometry of the flood zone were produced when the input data consisted of “difference” wave fields (the subtraction of preflood data from post-flood data). Although the inverse GRT contains an “obliquity factor” that will normally ensure a high image quality, a further result of the experiment is that the obliquity factor in the inverse GRT needed to be suppressed to image the flood zone directly from the difference data.