One baseline and two time-lapse VSP surveys were acquired at the Aneth oil field in Utah for monitoring CO2 injection. Even thought the data were collected using a cemented geophone string, there are quite a bit of variations in the signal quality. We develop a processing methodology to best equalize the variations due to acquisition. The resulting images from the three acquisitions are compared with one another. The differences in images in the target zone are consistent with the CO2 injection program.
As a part of the effort of the Southwest Regional Partnership on Carbon Sequestration supported by U.S. Department of Energy, one baseline and two repeat VSP surveys were conducted from 2007 to 2009 at the Aneth oil field in Utah for monitoring CO2 injection (Figure 1). The aim of the time-lapse VSP surveys is to study the combined EOR and CO2 sequestration in collaboration with Resolute Natural Resources, Inc. VSP data were acquired using a cemented geophone string with 60 levels at depth from 805 m to 1704 m, and CO2 is injected into a horizontal well nearby within the reservoir at depth approximately from 1730 m to 1780 m. For each VSP survey, the data were acquired for one zero-offset source location and seven offset source locations (Figure 2). The baseline VSP survey was conducted before the CO2 injection. More than ten thousand tons of CO2 was injected between time-lapse VSP surveys (Huang et al., 2008; Zhou et al., 2008; Rutledge etal., 2008).
The initial baseline acquisition was done in October, 2007, subsequent time-lapse acquisitions were done in July 2008, and January 2009. The acquisition geometry is shown in Figure 2. Shot point 1 is the zero-offset source location, Shot points 2 to 8 are the seven off-set VSPs, arranged in a quarter circle on the Northwest side of the monitoring well. The horizontal injection well is shown in green. The black lines in Figure 2 shows the approximate azimuthal directions of the shot points from the monitoring well. VSP source location 5 is in a direct line with the injection.
The 96 geophones were cemented into the monitoring well just before the baseline VSP acquisition. The 96 phones consisted of 18 three-component geophones (at the bottom of the string) and 42 single vertical component phones. For this initial study, only the vertical geophone data were used.
Standard VSP processing is used to process the baseline VSP data. These include removing the reference signal, muting, sorting, vertical summing, reversing traces with reversed polarity, notch and
bandpass filtering, first break picking, T-gain, wavefield separation, up-wave enhancement, deconvolution, velocity model building, and VSP-CDP transform (Hardage, 2000). Because the dataset is to be used as a baseline, care is taken to get the best data out from a rather noisy raw dataset. The raw and final processed data are shown in Figures 3 and 4.
The pre-processed field data are then processed to separate the upgoing and downgoing wavefields using median filtering. The upgoing wavefield is then deconvolved using a source signature obtained from the downgoing wavefield.
