|
 DvC2-09
In July, 2002, as part of a management-assigned test of the D3DSP’s ability to “find oil and gas for the company”, EPL commenced its fourth D3D-reprocessing effort on a multi-client, Fairfield Industries "speculative" 3D seismic data set, recorded in 1995, over 1.5 OCS blocks, using an unusually fine 3 ms sampling interval. A desired attempt to increase resolution, made re-sampling to 2 ms the first of many unusual D3D-reprocessing steps. The reprocessing took three months to complete, and by the end October, 2002, many displays had been generated showing the high potential of the prospective reservoir, barely connected to the up-time-dip, 8.2 bcfge Odeco #4 watered-out well. The KB of this well was being carried as a "probable" +40 feet, and a speculative west-to-east velocity gradient was invoked to put the newly chosen (Bright Spot) proposed location, up-dip to the Odeco #4 well. But the Minerals Management Service office was found to have conflicting records of the KB elevation, with 73 feet above sea level being the highest one. The depth-to-time conversions of the well paths and formation tops shown in this Figure are based on the +73 feet KB. This value allows the tops (measured in depth below KB) to fit the time-dip nicely, using a Time-to-Depth relation from the EI-26 velocity survey ... and without the use of any increasing-to-the-west velocity gradient, unsupported by D3D-reprocessing stacking velocity analyses.
The left-hand panel shows a conventionally processed NW-SE arbitrary line connecting the Odeco #4 watered-out, abandoned well, the D3DSP-proposed drill-site, and the slightly down-dip, conventionally selected, EPL #1 CIB CARST gas sand discovery well. The timing lines are spaced every 10 ms, or 40-50 feet, depending on the velocities of the rocks at this range of depths. Note that the polarity is SEG "normal", and the high amplitude gas sand (under the EPL #1) is expressed as a black/blue-peak-over-pink/red-trough. By zooming in, it can be seen that the TWT (two-way-time) to the base of the CIB CARST sand "marker", tracks nicely just below the (peak-over-trough) zero-crossing using the EI-26 velocity function. This is especially true for the drilled Odeco #1 well (left) and the EPL #1 well (right). No velocity gradient is necessary if +73 feet is the Odeco #1 KB elevation. The Fourier frequency Amplitude Spectrum, shown beneath the seismic section, shows a typical shape for (4-ms-resampled) Gulf Coast seismic, with a "dominant frequency" of 30 Hz (one over the dominant period of 33 ms), a peak frequency of about 12 cycles per second (Hertz, or Hz), and sudden drops at 20 Hz and again at 50 Hz. This is the spectrum of the conventionally desired, layered-earth-model seismic wavelet, and it does a good job of presenting an "uncluttered" layered picture. But 8.2 bcfge produced from the left-hand well, and then watering out (presumably by the down-dip aquifer water), makes the right-hand, proposed well location look likely to find the depleted-gas (fizz-water) aquifer.
The middle panel is shown for a direct comparison of conventional-reflectivity vs. D3D-reflectivity. The D3DSP teaches the use of D3D-impedance volumes, and the D3D-reflectivity volume is merely an intermediate step in reaching this more useful Pseudo-impedance product, so its "wavelet/tuning/change-in-impedance" use is not too highly recommended by the D3DSP. But this Figure DvC2-09 shows a direct comparison between both the "twin" vertical-slice arbitrary lines and their Amplitude Spectra. The 4 ms conventional data wavelet has a highest amplitude ("peak") frequency of about 12 Hz, and the 2 ms D3D version has essentially no "peak frequency, because it was processed to be flat, or "white", out to 120 Hz. The D3D dominant frequency is close to 100 Hz, because (as shown by the small black arrows) the dominant, blue-peak-to-blue-peak period is 10 ms. Note, too, that the polarity convention shown here is preferred by the D3DSP, and is "SEG reverse". This polarity is referred to as Positive Physical Polarity (PPP) because, when traveling from lower- to higher-impedance rocks, a down-going compressional pulse (assumed to be a positive-value) gets reflected back as another compressional pulse, and not flipped over to a "rarefaction" (negative-value).
The right panel shows the D3D-impedance twin line, with its Amplitude Spectrum below. The low-impedance CIB CARST gas sand is clearly visible above 2.7 seconds, as is the very flat base of the low-impedance (yellow) CIO anomaly. Every fifth wiggle-traces (225 feet) is displayed, and the flat base can be seen to extend over nearly 650 feet to the left side of the NW-SE arbitrary line. No such flat "base" is evident on the conventional data on the left. In fact, because the sand thickness is everywhere thinner than the "tuning thickness" for this particular conventional wavelet,
{the tuning thickness equals the velocity (use 8000 f/s) multiplied by ¼ the dominant period (use .033 seconds), or 66 feet}
the TWT to the base of the CIB CARST sand would have to be calculated from the change in peak-to-trough amplitude along the anomaly, assuming some known wavelet shape and gas sand velocity. All of this is very difficult and prone to assumption errors in the conventional process, but becomes almost a simple geometry exercise when the D3D-processor has done a correct job of creating a D3D-impedance volume. And the D3D-QC geophysicist was right there to ensure that the D3D-processing rules were followed, and (in this case) the base of the gas sand was clearly "interpretable".
This well is still producing (11/2004), and seems to be on its way to producing something near the D3D-predicted EUR, but engineering data has shown little if any significant aquifer pressure support. So, either the visible gas-water contact has become cemented, and immovable, over the millions of years it has existed, or the exquisite D3D predictions are all coincidences. Because no well has yet been regulatory-permitted, to directly test the presence of a gas-water contact where the D3DSP predicts it exists as a flat base on the D3D-CIO, this question will remain unanswered for (perhaps) quite some time.
Comparing the Amplitude Spectra in the center panel (D3D-reflectivity) with the right panel (D3D-impedance) also shows the effect of the running-sum trace sample "integration" (or "inversion") on the Amplitude Spectrum of the traces. All the high frequency character preserved by the D3D-processor is there, but it is riding on the low dominant-frequencies contained in the local geology. The wavelet has been fairly well converted to a "spike", making it
(A) More difficult to make time-structure and seismic amplitude maps using a D3D-reflectivity volume (which is a primary reason it is taking so long to gain acceptance), but
(B) Much easier to see the valuable, buried 3-D objects and their diagnostic shapes and impedance distribution (and connectivity to existing wells and other possible targets), using the D3DSP.
Volume visualization and analyses are keys to obtaining maximum economic value from D3D data sets.
|
