Using 4C data for predicting lithology and fluid type

Three case studies show how advances in marine 4C-data analysis allow clearer imaging of reservoirs, and discrimination of fluids and rock properties

Eivind Fromyr and Santi Randazzo, PGS Reservoir Services, Inc., Houston

n recent years, development of marine four-component (4C) acquisition and processing technology has made it possible to apply the technique in most geographical areas and water depths. Due to higher cost when compared to conventional streamer surveys, future success depends on identifying appropriate use, and on demonstrating that results can be interpreted and used in ways that still make the technology cost efficient.

This article does not address processing and interpretation issues on a detailed technical level, but it does discuss results of three case histories, each showing a different application of marine 4C data. The applications cover both imaging and characterization, including improved imaging in low-impedance P reservoirs, how sand / shale content can be inferred from multicomponent attributes, and fluid-discrimination problems.

Introduction

A conventional marine source gives rise to the full complement of wave types through conversions, chiefly upon reflection at any impedance contrast. 4C acquisition makes it possible to record the entire vector wavefield. Imaging of structures beneath gas clouds is a well-established application of this technology. Additionally, acquisition with stationary receiver systems on the seabed enables better imaging of complex structures due to improved aperture and application of dual-sensor summation to remove waterborne multiples.

Generally speaking, in addition to a P-wave section, recording and processing an S-wave section gives a second set of independent information that can be combined with P-wave velocity (V_P) to yield Poisson’s ratio, or simply the velocity ratio V_P/V_S. This allows better determination of lithology and pore fluids.¹ This cannot usually be done with P-wave data alone, because many different rocks with different fluids have similar P-wave responses.

Multiple Removal / Subsalt Imaging

Three aspects of marine 4C data are the focus of the following case history. First, seabed sensors offer semistationary acquisition with free geometry that allows optional wide azimuth and a full range of offset, including true-zero offset. Second, it is well known from shallow-water OBC data that waterborne reverberations can be removed by combining hydrophone and vertical-geophone wavefields.² Third, in deepwater seabed acquisition, the effective aperture for a given nominal source-offset range increases relative to streamer geometry.

A 2D/4C line can prove additional reserves due to improved imaging beneath a salt structure. Such data was collected in the Gulf of Mexico. The field was covered by a 3D streamer survey, and a comparison of the 2D/4C line and an extracted 2D line from the 3D cube is shown in Fig. 1.

Fig. 1. Comparison of streamer data (left) and 2D dual-sensor data (right).

The illumination advantage, combined with the multiple suppression technique (no destructive FK or Radon), are responsible for the improvement that allows further mapping of the reservoir reflector beneath the salt structure. Imaging of both the base of the salt overhang and the salt diaper flank was also improved. Standard prestack time processing was applied, as well as a spectral-balancing technique prior to dual-sensor summation.

Projecting the additional pay into the 3D volume indicates the potential increase in estimated reserves, Fig. 2. Full 3D acquisition would be required to prove this in detail.

Fig. 2. Potential extent of additional reserves.

Low-Impedance Reservoir Characterization

Many Paleocene sandstone reservoirs in the North Sea represent combined stratigraphic and structural traps. This particular case has a low-relief anticline, and the reservoir is shaling out toward the east. On conventional seismic, the reservoir reflector is characterized by relatively good reflectivity in the west, which gradually weakens moving eastward, Fig. 3.

Fig. 3. Conventional P data. One observes decreasing reflection strength due to increasing shale content. It is difficult to see the shale-out line.

Typically, these Paleocene reservoirs are difficult to interpret, since the reservoir reflector becomes too weak to identify over the whole field. In Amplitude vs. Offset (AVO) terms, this is a Class-II situation. These reservoirs tend to be very heterogeneous, and shaling out of the sand makes delineation and exploitation difficult. They are typically associated with a high number of delineation wells to minimize uncertainty in reserve estimates.

Converted-wave information will help in the structural imaging of such reservoirs and better define shaling out of the sand. This example explains why this is so.

A 2D/4C seismic line was acquired passing through the two wells in Fig. 3. Well 1, on the left, has good reservoir-quality sand and high net / gross. Well 2, on the right, shows a reservoir section with very low net / gross and only a few sand lenses of good quality. Analysis of available logs from the two wells explains why the P-wave seismic is poor and why converted waves will give a better image.

Fig. 4 shows cross plots of Well 1 in terms of V_P/V_S vs. both P impedance and S impedance. These indicate that there is stronger impedance separation between the sand and the surrounding shale in the S domain than in the P domain. Fig. 5 shows P and S impedance for the reservoir interval in both wells; as shale content increases, impedance contrast decreases for both P and S.

Fig. 4. Cross plot of P and S impedance vs. VP/VS for the reservoir zone and surrounding shales.

Fig. 5. Cross plot of P vs. S acoustic impedance for the reservoir sand, both hydrocarbon and wet, for both wells. Color scale on right is according to well location. Color scale on left is based on VP/VS ratio

Due to higher initial separation in the S domain, a sufficient impedance contrast is still present there. P-impedance contrast has decreased so much that reflectivity between the reservoir formation and the surrounding shale disappears and, hence, the reservoir becomes transparent.

The seismic fully supports log observations, Fig. 6. In the PP section, the reservoir reflector weakens and becomes difficult to follow moving from left to right. In the PS section, however, the reservoir reflector is strong all the way through. In addition, one can identify the reservoir base and thereby confirm thinning of the reservoir formation observed in Well 2.

Fig. 6. PP and PS data around the reservoir. Reservoir top is visible all the way across only on PS; base is also more defined, which is absent on PP.

The logs were further analyzed to investigate the second objective – which is related to lithology – and especially to predict shale content. Fig. 5 shows the same cross plot between P impedance and S impedance for both wells, but color-coded according to well location and V_P/V_S ratio. One can observe good separation between sand and shale, about 20%, which should be observable on the seismic data.

Since this is a low P-impedance case, AVO on P data does not appear attractive. With a direct PS observation, a V_P/V_S attribute must be derived from the seismic data with sufficient resolution. This may be achieved by using PP and PS amplitudes in the following way.

For an angle-limited stack, it can be shown that:³

By combining this with the fact that V_P/V_S is proportional to the ratio of P and S impedance, the following procedure should work.

Correlate the main geologic markers and compress PS to PP time and then invert both the PP and PS data to give acoustic-impedance estimates. By dividing the two impedance sections, obtain the instantaneous V_P/V_S attribute. Since this is stacked, converted data, an amplitude correction must be applied to the PS data before inversion. The instantaneous V_P/V_S section is shown in Fig. 7; the reservoir interval is characterized by a low V_P/V_S value on the left side which increases gradually as one moves to the right. This dataset shows degradation of reservoir quality relatively close to Well 1.

Fig. 7. Instantaneous VP/VS attribute derived from inverted amplitude for PP and PS. Orange represents low values; green / blue, high values. No well calibrations have been done. One can observe increasing values along the reservoir sand to the right, indicating increasing shale content.

The above procedure works well in 2D with simple structures. A more general method that can also be used in 3D must be horizon based. It is clear that there are a number of challenges involved; among them is the attempt to correlate events from PP and PS in a common domain, e.g., depth. With well calibration, it should be possible to achieve quantitative estimates – in a geostatistical sense – that can be used in the final volumetric calculation. This would require 3D data.

Fluid-Discrimination Aspects

One of the ultimate 4C acquisition goals is to be able to discriminate lithology and fluids sufficiently to assist in determining the volume of gas, oil and water. This is a challenging task, and the industry needs to learn how to do this from theory and case studies. There is a complicated interplay of lithology factors and fluid factors. In general, one should be able to benefit from P and S imaging combined with AVO analysis of both wave types.

In this case study from a Gulf of Mexico field, it is shown that, despite inherent ambiguities, it may be possible to associate certain seismic attributes with the physical subsurface. Geostatistical methods combined with well control may offer a workable technique. Calibration based on field-specific knowledge may prove to be absolutely necessary. Techniques developed in this branch of geophysics have the potential to become fundamental elements in applying the same technology to reservoir monitoring. Since this is an immature area, the results that follow are indicative in nature and should be interpreted with caution.

The problem was determining the oil / water contact. This turned out to be difficult based on any attributes derived from P-wave data. Fig. 8 shows the P-wave data through three wells with wet sand encountered in the down-flank well, indicated in blue.

Fig. 8. PP section with well trajectories.

Available log data was used to generate a simple stratigraphic model for PP and SS. The derived V_P/V_S attribute, shown in the middle of Fig. 9, illustrates P and S impedance in color on top and bottom, respectively. Note that the reservoir has been flattened.

Fig. 9. Stratigraphic model for PP and SS based on well-log data. Top view is P impedance; bottom, S impedance. Middle is VP/VS from the inverted amplitudes.

Fig. 10 shows a similar, higher V_P/V_S pattern for the wet sand. Two interesting aspects relevant to the above observations were investigated. First, the relative amplitude of PS with offset was modeled for each well based on well-log data, Fig. 11. The wells are annotated as gas / oil / oil / water, and a different response was observed than would be expected from a single-interface model with average properties and simple fluid substitution. This infers intricate response and tuning effects.

Fig. 10. Top shows seismic PP (left) and PS section. Bottom is instantaneous VP/VS attribute derived from inverted PP and PS data. Reservoir sand is outlined and possible oil / water contact shown. Orange represents low values; green / blue, higher values.

Fig. 11. PS AVO response based on well-log data. Plotted are AVO curves from top sand for the four wells.

A simple three-layer model was constructed that contained the following fluids / solids: gas / oil / water, gas / water / water and gas / shale / water. The model was coarse enough to avoid tuning effects. AVO response was calculated both for PP and PS, Fig. 12. Among the possible observations is one simple fact: Lithology changes, if present, will most likely be a significant factor in the total response.

Fig. 12. PP and PS AVO response for a three-layer model with (a) gas / oil / water (b) gas / water / water and (c) gas / shale / water. Average well-log data used as input.

In some cases, lithology prediction and fluid discrimination may be linked in ways that may simplify analysis. It will always be a challenge to discriminate between fluids and lithology simultaneously. Well-log information will, most likely, always be necessary. Ultimately, even if it can only be done in a geostatistical sense on a case-by-case basis, the economic benefits will be significant.

Conclusion

Use of dual-sensor data for P-wave imaging can be beneficial in deep and shallow water and lead to significant improvements, including improved imaging in low-impedance P reservoirs. Associated with this, it has been shown how sand / shale content can be inferred from multicomponent attributes. Fluid-discrimination problems will benefit from the additional converted-wave information, since lithology changes can now be analyzed.

Acknowledgment

This article was originally presented at the 2000 Offshore Technology Conference, Houston, Texas, May 1 – 4, 2000, as OTC 11907 and is used with permission.

Literature Cited

1. Tataham, R. H. and M. D. McCormack, "Multicomponent seismology in petroleum exploration," SEG Investigations in Geophysics No. 6, 1999.
2. Barr, F. J. and J. I. Sanders, "Attenuation of water-column reverberations using pressure and velocity detectors in a water-bottom cable," 29th SEG Annual Meeting, Dallas, Texas, 1989.
3. Stuart, R. R. and H. Bland, "An approximate relationship between Rps and Rss," University of Calgary, CREWES Annual Report, 1997.

The authors

Eivind Fromyr earned an MS in Physics from the University of Oslo in 1982. Thereafter, he worked for Geco Prakla until 1986. From 1987 to 1995, he was employed by Read Well Services. Then, in 1995, he joined PGS Reservoir, where he currently serves as vice president of Geophysical Technology, Reservoir Services, Houston.

Santi Randazzo earned his MS in Geophysics in 1986 from the University of Houston. From 1982 until 1989, he worked at INTERVEP. He subsequently joined Technica, where he remained until 1992. Thereafter, he worked at Simon Petroleum Technology until arriving at PGS Reservoir in 1994, where he is manager of Multicomponent Processing and Interpretation, Reservoir Consultants, Houston.