Dynamic application of seismic migration in field development

Seismic anisotropy can result in improperly defined reservoirs and mis-positioned wells. Post-SDM incorporating initial well velocities can quickly improve reservoir definition

Exploration in the Gulf of Mexico has been greatly influenced by use of seismic-amplitude anomalies. However, the spatial position of these anomalies can be changed significantly by varying the migration velocity field. Furthermore, recent developments in ray-trace modeling have shown that shooting direction during seismic acquisition can alter the apparent spatial distribution of these anomalies. This was evidenced in the development of Garden Banks Block 260 Baldpate field. Drilling results showed that in the initial migration, the amplitude anomaly associated with the pay sand was swept 1,500 to 2,000 ft beyond the updip reservoir truncation.

A comparison of well-checkshot velocities from a walk-away VSP against seismically derived velocities indicated severe anisotropy in the area, resulting in overmigration. A series of target-oriented, post-stack depth migrations were run and evaluated against ongoing development drilling until an optimum migration was identified. During the course of this evaluation, it was determined that acquisition direction could affect the amplitude image, creating an apparent change in shape and position of the anomaly.

This article describes how the anisotropy problem was discovered, and how its effects were remedied to allow accurate migration of the reservoir and an associated salt body, thus allowing proper field development.

BACKGROUND

Garden Banks Block 260 lies within the Plio-Pleistocene Flex Trend, 110 mi offshore, in 1,650 ft of water. In 1991, Amerada Hess and Oryx Energy drilled GB 260-1 to evaluate an Upper Pliocene, amplitude-associated event on a salt-flank structure, mapped from a 1989 time-migrated 3-D volume, Figs. 1 and 2. The well encountered 180 ft of net vertical Pliocene pay. Two major zones had 103 ft and 64 ft of net pay, Fig. 3. The upper pay sand was tied to the primary amplitude anomaly using a VSP and synthetics.

[Figures 1-3 ILLUSTRATION OMITTED]

Sidetrack 1ST 1 of the discovery well, drilled to test the apparent extreme downdip limit of the anomaly, encountered 243 ft of net pay in the same two major zones, Figs. 1 and 2. A subsequent sidetrack (1ST 2) was designed to test the reservoir's amplitude in an updip position. However, the well unexpectedly drilled into salt, which truncated the reservoirs, Figs. 4 and 5. Subsequent evaluation suggested that the seismic velocities used were inadequate for proper structural positioning, due to abnormally high anisotropy in this area.

[Figures 4-5 ILLUSTRATION OMITTED]

SEISMIC EVALUATION

Based on early drilling results, it was apparent that the reservoir was not positioned properly in the original 1989, 3-D time migration. Evaluation of velocity surveys acquired in GB 260-1 and 1ST 1 wells showed that the original migration velocities were much too high, resulting in mis-positioning of seismic events.

A walkaway VSP in the 1ST 1 well was studied in detail to better understand the velocity field. Evaluation showed that, over this field, longer offset, higher-reflection-angle raypaths had much higher velocity than the more vertical travelpaths, implying an abnormally high degree of anisotropy. Due to the anisotropy, velocities derived from seismic data are typically faster than those recorded from checkshot/well data. Using typical seismic velocities in this area resulted in overmigration of reservoir anomalies, sweeping the events into salt.

Having determined the problem, the partners worked jointly to generate a more accurately migrated volume. This was accomplished by using regional well velocities--combining checkshot data with seismic velocities--to design a velocity field, which was then applied to generate a new, time-migrated, 3-D volume. As expected, the amplitude anomaly shifted downdip and tied the existing well control much more accurately. The anomaly's shape now conformed more clearly with the known salt truncation on the reservoir's eastern flank.

Unfortunately, this new migration was adversely impacted by steeply dipping, undermigrated salt reflections that passed through the reservoir anomaly, creating bands of constructive and destructive interference in the amplitude extraction, Fig. 6. Further, the amplitude anomaly decreased in total area by about 15%. However, it was agreed that the spatial position of the reservoir in this migration was much more reliable.

[Figure 6 ILLUSTRATION OMITTED]

FIELD DELINEATION

Due to the high cost of a major deepwater development, further reservoir delineation was deemed necessary before project sanctioning. To test the concept of reserves that might not be imaged due to salt/raypath complexities, an exploratory well (260-3) was drilled into a non-amplitude-supported, sub-salt location southeast of the discovery. As before, the well emerged from salt into a section older than the reservoir sands.

However, due to the risk associated with testing a non-amplitude-associated, sub-salt target, the well had been pre-designed to facilitate sidetracking into the extreme eastern amplitude area. This sidetrack encountered the full pay section, supporting the new migration's accuracy of the amplitude position, Fig. 7.

[Figure 7 ILLUSTRATION OMITTED]

With three penetrations into the revised amplitude anomaly--all confirming the presence of the pay sands--confidence in the new migration was increasing. A final delineation well was necessary to determine the proper facility sizing and the optimum development-well count.

A weak band of amplitude was present in the primary amplitude near the anomaly's downdip limit. Termination of this event closely coincided with the mapped, lowest-known oil seen in 1ST 1. Isochron mapping indicated that the upper event was thinning downdip, which could explain the amplitude dimming as a function of the sand thinning below seismic resolution. In addition, the deeper reservoir appeared to strengthen in amplitude as it emerged from under the upper anomaly, suggesting it was masked under the upper reservoir.

A well was designed to test the absolute downdip edge of the upper amplitude, structurally very near the mapped lowest-known oil, for an oil/water contact or thinning/missing sand. It would also test amplitude that appeared to be associated with the deeper reservoir.

It became apparent that more velocity work was necessary when target sands came in more than 300 ft deeper than predicted, and wet. A checkshot in the well verified that velocities were much higher than in the updip wells, revealing a velocity gradient across the field. Fortunately, the sands were full thickness, disproving the thinning model and supporting a basin-floor, fan type, sheet-sand model. Because this appraisal well was designed to test the extreme downdip limits of the reservoir in an unfavorable drainage position, a sidetrack to an updip development location had been anticipated. The velocity gradient was interactively mapped and incorporated into the structure maps, and the sidetrack found reservoir depth and thickness as predicted. Critical sand control provided by this well and sidetrack provided confidence that the estimated reserves were sufficient to proceed with project development.

POST-STACK DEPTH MIGRATION

In 1995, a new 3-D volume was licensed over the area. As before, the contractor time data was overmigrated due to anisotropy. At the same time, the field development was entering the most difficult phase--drilling the attic wells. The optimum location for the final well would be at the crest of the structure, under the salt overhang. Ideally a pre-stack depth migration would be used to properly position the salt face and reservoir distribution. Unfortunately, with a fig in the field there was not adequate time to complete a pre-stack depth migration. However, Amerada Hess had by this time developed an efficient, 3-D post-stack depth migration that uses parallel-distributed processing. This allows generation of multiple-migration volumes quickly and inexpensively, without impacting other ongoing projects. Seismic velocities were picked and the data was post-stack depth migrated, resulting in overmigration similar to the results of the contractor migrations.

A second migration was performed, reducing velocities to 90% of the seismic-velocity field. This resulted in a volume that was undermigrated but fit much closer with the well data. Additional migrations were run in increasing 1% increments, from 90% up to 93% of the seismic-velocity values. These results were then compared with existing well penetrations, which provided updip and eastern salt limits, lowest-known oil and highest-known water. Using these control points, the best migration velocity was determined to be 92% of the seismic-derived velocity field, Figs. 8 and 9. Use of this migration to select drilling targets resulted in the remaining development wells encountering the reservoirs as predicted, including a difficult attic well under the salt.

[Figures 8-9 ILLUSTRATION OMITTED]

After calibrating these results with regional checkshot data, a relationship of velocity reduction vs. time/depth was derived. Consequently, target-specific, post-stack depth migrations were generated for other developing fields in the area.

SEISMIC IMAGING

On all depth migrations of the 1995 data, it was difficult to image the reservoir's extreme eastern flank. This area was best imaged on the time migration of the 1989 data, with revised velocities. Comparison of the original-contractor time migration shows that amplitude from the N-S oriented survey extends slightly farther east than amplitude from the E-W oriented survey, Fig. 10.

[Figure 10 ILLUSTRATION OMITTED]

Ray--trace models, or "illumination" models, are now being used in the industry to determine the relative number of raypaths that can actually image a horizon. Input to the illumination study is the acquisition geometry (shooting direction and cable length), velocity model and the horizon to be evaluated. The velocity model must accurately define subsurface velocities, including a detailed salt model. In this study, the velocity model was derived from several iterations of 3-D pre-stack depth migration. The output is the relative number of "hits" on the horizon from various offset ranges. Results are loaded to a workstation as a horizon for comparison to seismic-amplitude maps.

Models were run over the reservoir with both N-S and E-W shooting. Ray-trace modeling showed results consistent with the observed amplitude distribution. Both models showed a poorly illuminated area on the east-southeast side of the reservoir, but the N-S data imaged slightly farther east, Fig. 11. Therefore, loss of amplitude in the 1995 data was a function of imaging with E-W acquisition, not improper migration positioning. Similar results were observed in adjacent field developments, with the preferred shooting orientation changing from block to block, based on salt geometry. It can, therefore, become necessary to acquire additional data of a different orientation to better evaluate a given prospect.

[Figure 11 ILLUSTRATION OMITTED]

CONCLUSION

While seismic-amplitude anomalies can greatly decrease prospect risk, improper migration can result in significantly mis-positioned anomalies, resulting in dry holes or inefficient development locations. Use of post-stack depth migration can be a fast, inexpensive and yet reliable means of incorporating well control and checkshot data: Incorporating such data may be critical to generate improved migrations in a dynamic, real-time development scenario.

Certain combinations of salt geometry and seismic-acquisition direction can adversely impact reservoir imaging. Due to imaging complexities, it may be necessary, value permitting, to obtain additional 3-D data over a given field to fully evaluate reservoir potential. Raypath modeling can be used to better understand reservoir amplitude variations and to determine optimum shooting direction for a given prospect.

ACKNOWLEDGMENT

The authors thank partner, Oryx Energy, for allowing us to publish the field data; Diamond Geophysical and Western Geophysical for granting permission to present the seismic data; and Schlumberger for assisting in evaluation of the anisotropy in the VSP data. This paper, OTC 10843, was presented at the 1999 Offshore Technology Conference held in Houston, Texas, May 3-6, 1999.

D. Bret Walker earned his BS in physics from Southwestern Oklahoma State University in 1981. He has worked for Gulf Oil and Amoco Production Co., exploring in diverse environments, including the Mid-continent, Gulf of Mexico, Africa and Middle East regions. In 1990, he joined Amerada Hess, working in the Offshore Gulf of Mexico Exploitation and Development Group. He is currently the geophysicist assigned to the Amerada Hess Deepwater Production Unit.

Ron R. Pressler earned a BS in geology from the University of Texas at Austin in 1976, and conducted post-graduate work at the University of Houston. During the past 21 years, he has worked in various E&P capacities in a number of domestic petroleum basins. He has worked onshore/offshore Gulf of Mexico and East Texas for both Ashland Oil (1974-1979) and Union Texas Petroleum (1979-1985). He joined Amerada Hess in 1985, working in various exploration and development capacities. He is division geoscientist for the newly created Deepwater Production Unit.

Steve Checkles earned his BS in geology from the University of Texas in 1982, and received an MS in geophysics from the University of Houston in 1988. Steve spent seven years at Western Geophysical before joining Amerada Hess in 1991. He is currently manager of Specialized Seismic Processing, which includes subsalt, pre-stack depth imaging for E&P projects.

Fuhao Qin earned a BS in geophysics from the University of Science and Technology of China in 1983. After receiving an MS, in 1986, from the Institute of Geophysics, Chinese Academy of Science, he remained there as research assistant from 1986 to 1989. In 1994, he obtained a PhD from the University of Utah in applied geophysics and worked there for a year as a post-doc fellow. He joined the Geophysical Development Group at Amerada Hess in December 1994, where his work includes seismic imaging, seismic modeling and inversion. Qin developed ray-trace modeling "illumination" software, which has become routinely used to evaluate imaging problems typically associated with salt-flank and sub-salt plays.

COPYRIGHT 1999 Gulf Publishing Co.
COPYRIGHT 2000 Gale Group