Year 3 Progress

Technical Progress Report, Year 3: 4-D High-Resolution Seismic Reflection Monitoring of Miscible CO₂ Injected into a Carbonate Reservoir

by Richard D. Miller, Abdelmoneam E. Raef, Alan P. Byrnes, and William E. Harrison

KGS Open File Report 2007-11
Aug. 23, 2006

Abstract

The objective of this research project is to acquire, process, and interpret multiple high-resolution 3-D compressional wave and 2-D, 2-C shear wave seismic data in an attempt to observe changes in fluid characteristics in an oil field before, during, and after the miscible carbon dioxide (CO₂) flood that began around December 1, 2003, as part of the DOE-sponsored Class Revisit Project (DOE #DE-AC26-00BC15124). Unique and key to this imaging activity is the high-resolution nature of the seismic data, minimal deploy-ment design, and the temporal sampling throughout the flood. The 900-m-deep test reservoir is located in central Kansas oomoldic limestones of the Lansing-Kansas City Group, deposited on a shallow marine shelf in Pennsylvanian time. After 30 months of seismic monitoring, one baseline and eight monitor surveys clearly detected changes that appear consistent with movement of CO₂ as modeled with fluid simulators and observed in production data.

Executive Summary

Efficiency of enhanced oil recovery (EOR) programs relies heavily on accurate reservoir models. Movement of miscible carbon dioxide (CO₂) injected into a thin (~5 m), shallow-shelf, oomoldic carbonate reservoir around 900 m deep in Russell County, Kansas, was successfully monitored using high-resolution 4-D/time-lapse seismic techniques. High-resolution seismic methods show great potential for incorporation into CO₂-flood management, highlighting the necessity of frequently updated reservoir-simulation models, especially for carbonates. Use of an unconven-tional approach to acquisition and interpretation of the high-resolution time-lapse/4-D seismic data was key to the success of this monitoring project.

Interpretations of geologic features from seismic data have provided location-specific reservoir properties that appear to strongly influence fluid movement in this production interval. Linea-ments identified on seismic sections likely (based on time-lapse monitoring and production data) play a role in sealing and/or diverting flow through the reservoir. Distribution and geometries associated with similarity seismic facies and seismic lineament patterns are suggestive of a complex ooid shoal depositional environment. By incorporating these features, using properties consistent with core data, a more realistic reservoir simulator results, honoring the production and core properties. Flow models after simulator updating (sealing lineaments and preferential permeability mani-fested by faster progression of the CO₂ bank) show improvement in detail and provide correlation with the material balance.

Amplitude envelope attribute data possess changes in texture generally consistent with expecta-tions and CO₂ volumetrics. Arguably, a multitude of different boundaries could be drawn to define the shape of the CO₂ plume, but the shapes suggested match the physical restraints, based on engineering data and the estimated amplitude response. Focusing on the injection well area and continuity of the characteristics defin-ing the anomalous area, it is not difficult to identify a notable change in data character and texture likely associated with the displacement of reservoir fluids with CO₂.

Advancement of the CO₂ from the injector seems to honor both the lineaments identified on baseline data and changes in containment pressures. Overlaying the amplitude envelope attribute map with the lineament attribute map provides an enhanced view, and therefore perspective, of the overwhelming variability in the reservoir rocks and the asso-ciated consistency and control these features or irregularities have on fluid movement.

Increased northerly movement of the CO₂, as interpreted on seismic data and inferred from production data, after several months of CO₂ injection and oil production, stimu-lated an increase in injection rates at the water flood wells. After several months of increased water injection, the CO₂ advancement to the northwest was halted and some regression was observed on seismic data.

Production of CO₂ at a well significantly outside the designed flood to the northwest was cons-istent with the 4-D seismic, but a complete surprise to the petroleum engineers operating the flood. Seismic data early on suggested preferential northerly movement and once the CO₂ plume moved outside the high-fold footprint of the seismic volume it could no longer be tracked. Clearly the seismic monitoring could have dramatically improved the flood efficiency if the operators would have put more confidence in and heeded the images produced during the early stages of this flood.

Shortness of turnaround time of time-lapse seismic monitoring in the Hall-Gurney field provided timely support for reservoir-simulation adjustments and flood-management of the pilot study. Initial reservoir flow simulations utilized models based on pre-CO₂ oil production history, measured rock properties from core, water injectivity testing, and interwell testing. These data did not completely constrain the possible permeability architecture in the reservoir and CO₂-flood performance did not match pre-CO₂ injection predicted performance. 4-D seismic data, obtained and interpreted while the CO₂ flood was ongoing, was interpreted independent of simulations updated with the most current production data; therefore, to a limited extent, the interpretations of CO₂ movement based on seismic data were performed without field production input. Seismic predictions of CO₂ breakthrough at well 12 and the delay at well 13 were based on seismic data alone after the second monitor survey. Following initial seismic prediction, seismic and flood performance data were integrated to both validate the 4-D interpretation and confirm it was not inconsistent with flood performance, and to provide seismic input of flood progress to the flood management process. In general, seismically predicted changes in the CO₂ plume and measurements at production wells have been consistent throughout the flood.

Interpretations of time-lapse seismic data are consistent with and have assisted understanding of field response for the pilot. In a similar fashion, 4-D seismic have provided input to reservoir simulations investigating full-field EOR-CO₂ floods. Key observations from seismic data include

• accurate indication of solvent "CO₂" breakthrough in well 12,
• predicted delayed response in well 13,
• interpretation of a permeability barrier between wells 13 and CO2I#1, and
• consistency with reservoir simulation prediction of CO₂ movement and volume estimated to have moved north, outside the pattern.

Time-lapse seismic monitoring of EOR-CO₂ can reveal weak anomalies in thin carbon-ates below temporal resolution and can be successful with moderate cross-equalization and attention to consistency in acquisition and processing details. Most of all, methods applied here avoid the complications associated with inversion-based attributes and extensive cross-equalization techniques. Spatial textural, rather than spatially sustainable magnitude, time-lapse anomalies were observed and should be expected for thin, shallow carbonate reservoirs. Non-inversion, direct seismic attributes proved both accurate and robust for monitoring the development of this EOR-CO₂ flood.

Weak-anomaly enhancement of selected non-inversion, 4-D seismic attribute data repre-sented a significant interpretation development and proved key to seismic monitoring of CO₂ movement. Also noteworthy was the improved definition of heterogeneities affect-ing the expanding flood bank. Among other findings, this time-lapse seismic feasibility study demonstrated that miscible CO₂ injected into a shallow, thin carbonate reservoir could be monitored, even below the classic temporal seismic resolution limits.

Complete Report

• Acrobat PDF, 200 kB