Understanding Reservoir Architecture: Combining Continuous Fluid Facies Mapping, Pressure Measurements, Downhole Fluid Analysis, and Geochemical Analyses*
Daniel McKinney1, Hani Elshahawi1, Matthew Flannery1, Mohamed Hashem1, Lalitha Venkatramanan2, and Oliver Mullins2
Search and Discovery Article #40229 (2007)
Posted February 4, 2007
*Adapted from extended abstract prepared for presentation at AAPG 2006 International Conference and Exhibition, Perth, Australia, November 5-8, 2006
1Shell International, E&P, Houston, TX
2Schlumberger Oilfield Services, Houston, TX
Identifying compartmentalization and understanding reservoir structure are of critical importance to reservoir development. Traditional methods of identifying reservoir compartmentalization, such as drill stem tests and extended well tests, often become impractical in deepwater settings with costs approaching the costs of new wells and emissions becoming increasingly undesirable. Thus, compartments often have to be identified by some other means. Identification of reservoir compartmentalization by pressure gradient analyses, downhole fluid analysis (DFA), and geochemical fingerprinting are all means for identifying barriers, with DFA being a recently introduced novel approach. Independently, each technique has its limitations, but, together, they are a powerful tool for providing insights into reservoir architecture. This paper presents two case studies where the authors have used these techniques in a single well penetration (i.e., vertical barrier identification) and comparison of data in two wells in the same structure (i.e., lateral variability).
Case I: Assessment of Vertical Barriers in a Single Well Penetration
Figure 1 displays the gamma ray, resistivity and formation pressure data for Case I along with sampling stations F through I. The level of OBM filtrate contamination and GORs computed downhole in real-time from optical absorption spectra for the different fluids are shown in columns 3 and 4 in Table 1. The last column indicates the true GOR of the fluid, found at a later date from a surface PVT laboratory.
Several interesting features are observed in Table 1. First, F and G appear to have similar fluid density, whereas an apparent fluid density inversion between fluids G and J is observed. Second, a gentle fluid compositional grading is observed with the GOR varying gradually from fluid F to fluid I. Lastly, although the GORs computed from downhole spectra are about 15% lower than the true GORs computed in the lab (last column), the trend in the downhole GORs matches the trend observed in the measured GORs.
Application of Fluid Comparison Algorithm (FCA described by Venkatramanan et al., 2006, and H. Elshahawi et al., 2006) to analyze fluids G and J showed that the fluid density inversion is statistically significant with probability 0.99. Pressure gradients shown in Figure 1 confirm the fluid density inversion between fluids G and J. Densities of fluids J and H are found to be around 0.62 g/cm3 and less than density of 0.665 g/cm3, corresponding to fluids F and G. This fluid density inversion is also consistent with mud-gas logging data, which shows the presence of relatively more methane for fluid J. Mud-gas analysis determined a concentration of methane at J, double that at G and a discontinuity in the isotopes, supporting the conclusion of a density inversion and a sealing barrier between these fluids.
The gentle composition or GOR gradient seen between fluids J, H, and I is expected for an oil column in vertical communication. Open-hole logs suggest that this vertical span is fairly homogeneous. Analysis of gradient densities is in agreement with this assessment. It is suggested that this part of the well is in vertical communication. 13C/13C ratio determination on the mud-gas methane isotopes across this interval supports this interpretation; a clear increase in isotopic trend with depth suggests increasing contribution from thermogenic-sourced methane (Figure 2). The gradual decline in methane concentration in the mud-gas indicates wetness increasing with depth with no step-changes, supporting the gradual density gradient observed by the MDT.
The confirmation of vertical compartmentalization between G and J and a compositional gradient between J and I directly impacts reservoir modeling, reserves booking, and development planning. Drainage projections and reserve calculations cannot treat the whole interval as a continuous unit, and scope for recovery could be significantly reduced. Gas injection and water-flood scenarios would have to treat each zone independently, increasing design complexity and cost, while reducing sweep volume and time-to-water break-through. The compositional gradient in the lower reservoir likely extends downdip from the penetrated zone, and topsides facility design plans will have to anticipate production that will drop in GOR with time, even before depletion drive is exhausted. Medium- and long-term economic decisions can hence be revised at a much earlier stage of exploration.
Case II: Cross Well Application.
Figure 3 displays a well schematic cartoon of Case II. In this example, there are two suspected flow-barriers that intersect the main borehole and side-track, as illustrated in Figure 4. To test the presence of these barriers, the formation testing and sampling tool was run with two probes positioned to straddle the suspected barrier. The top probe was used to pump fluid above the suspected barrier and the bottom probe was used to assay fluid below the suspected barrier.
Application of FCA to analyze fluids in the main borehole led to a probability matrix shown in Table 2. As a rule of thumb, when the output probability of FCA is less than 0.5, we classify the two fluids being compared as being “statistically similar,” referring to downhole optical fluid properties being within the error-bar of the measurement. When the probability is between 0.5 and 0.95, the fluids are classified as “statistically indeterminate”, referring to the lack of clear optical distinction between the two fluids. From the probability matrix in Table 2, we infer that fluids #1A and #1B (across the top suspected barrier, see Figure 4) are statistically similar to each other. Similarly, fluids #2A and #2B and fluids #3A and #3B are statistically similar to each other. The differences between fluids #1B and #2B, which are above the second barrier, and fluids #3A and #3B, which are below the second barrier, are statistically indeterminate. Fluid differences (if any) can only be distinguished from other measurements.
Geochemical fingerprinting, which has similarities to the FCA methodology by comparing variations in fluid compositions, confirms these observations. Figure 5A shows that the two fluid samples collected in the original hole are nearly identical to one another on the spider plot. In addition, pressure-gradient analysis does not indicate any obvious discontinuities or possible barriers in the original hole. Two pieces of data that do indicate the significance of these barriers are the strontium residual salt analysis (SrRSA) and vertical interference testing (VIT). SrRSA data (Figure 5B) show clear changes in the 87Sr/87Sr ratio exactly where the calcite streaks are evident on the logs; indicative of changes in the paleo oil-water contact at the time of hydrocarbon filling. Detailed analysis of the VIT results indicates that there is a significant reduction in kv/kh and that the lateral extent of the calcite layer in one of the wells is at least 150 ft.
Application of FCA to analyze fluids in the side-track yields the probability matrix shown in Table 3. Again, the fluids assayed in the side-track are statistically similar to each other in their optical properties.
The fluids in the main-hole (first column) are compared to fluids in the side-track (first row) in Table 4. In this case, because different spectrometers were used to assay the fluids downhole, a larger uncertainty in the measurement ( e s = 0.02) was used in data analysis with FCA. Again, though, geochemical fingerprinting between fluids collected in the original hole and side-track are nearly identical (Figure 5B). Also, statistical analysis of the pressure gradients between the two wellbores does not indicate any significant difference that cannot be described by expected compositional grading. Thus, the sand unit is still expected to be in communication between the two wellbores. This may indicate some of the limitations of the FCA methodology in that variability observed in a given wellbore on the same MDT run is more easily interpreted compared to multiple runs and differing wellbores within a given field.
A practical problem that most oil companies face is determining the number of sampling stations because of the associated cost. They need to know if fluid A is different from fluid B before committing themselves to sampling and further detailed analysis in the laboratory. Careful pressure gradient analysis and FCA addresses this by providing a framework to quantify uncertainties and identify if fluids are statistically different, thus resulting in optimized sampling and a continuous downhole fluid log. In general, the integration of pressure data, DFA measurements including FCA analysis, and discrete fluid sampling provide valuable insights to reservoir architecture and compartmentalization issues.
Elshahawi, H., L. Venkataramanan, D. McKinney, M. Flannery, O.C. Mullins, and M. Hashem, 2006, Combining continuous fluid typing, wireline formation tester, and geochemical measurements for an improved understanding of reservoir architecture, during SPE Annual Technical Conference and Exhibition, 24-27 September, San Antonio, Texas, USA: SPE Paper No. 100740-MS.
Venkataramanan, L., H. Elshahawi, D. McKinney, M. Flannery, and M. Hashem, and O.C. Mullins, 2006, Downhole fluid analysis and fluid comparison algorithm as an aid to reservoir characterization, during SPE Asia Pacific Oil & Gas Conference and Exhibition, 11-13 September, Adelaide, Australia: SPE Paper No. 100937-MS.