4D seismic monitoring of subsurface fluids in the EI-330 Field of offshore Louisiana, Gulf of Mexico
3.1 Introduction
3D seismic images have improved to the point that they are being applied to reservoir production and engineering problems. 4D (time-lapse 3D) seismic imaging may soon be applied to oil field monitoring. However, this technology will require a better understanding of the link between stratigraphy, fluid content and changes in seismic response.
Time-dependence of individual reservoirs have long been observed in production wells. The use of 4D seismic datasets to monitor fluid movements dates back to more than a decade when thermal and CO2 injection enhanced oil recovery projects were carried out (Nur, 1982; Dunlop et al., 1988 and Breitenbach et al., 1989). Petrophysical studies indicate that seismic datasets may be used to monitor subsurface fluid movement in sandstone reservoirs saturated with medium-to-light weight hydrocarbons (Wang et al., 1992). The acoustic contrasts caused by lightweight, gas saturated hydrocarbons (oils with high gas:oil ratio, GOR) can produce even greater seismic amplitude changes.
Changes in seismic amplitude within a given reservoir can be caused by changes in gas:oil:water ratios, fluid pressures and/or fluid contact levels. Time-lapse 3D seismic surveys acquired can quantify these changes. These seismic datasets can be used to decipher both temporal and spatial distributions of oil, gas and water in reservoirs under production. We believe that such 4D seismic technologies, integrated with petrophysics and other related multi-disciplinary principles, will become the industry standard in hydrocarbon exploitation.
4D seismic technology is still in its infancy. New techniques are being developed yearly, as the great potential of using such datasets in reservoir surveillance and management is recognized (Nur, 1989; Lumbly, 1995; Anderson, et al., 1995a, 1995b and 1995c).
In order to extract 4D seismic changes, we examine successive 3D seismic surveys and examine their similarities and differences. The temporal changes allow us to predict the locations and compositions of bypassed hydrocarbons. We also use wireline logs (sonic, density, porosity, oil saturation variations over time) and production data (pressure, water cut, GOR variations over time) to calibrate our models of acoustic impedance changes. The model results will allow us to predict the quantitative changes in seismic response.
Because the time-lapse 3D seismic datasets used in this study were acquired for other purposes, we must pay special attention to their registration in space and time. The acquisition and processing parameters of different surveys were not the same, and thus we must correctly register the seismic datasets to preserve the maximum amount of common information of these surveys.
Despite these disadvantages, the results obtained from analyzing these "legacy" datasets are rather encouraging. Elimination of artifacts caused by differences in data processing enables us to distinguish the actual acoustic changes in hydrocarbon reservoirs. Many such legacy 4D seismic datasets are available, especially for the producing oil fields in the offshore Gulf of Mexico. We expect that the application of 4D seismic techniques to these seismic datasets will result in the recovery of a large volume of bypassed oil. Nevertheless, much research will be needed to advance 4D seismic techniques so that they can be used to view of the true dynamic of the drainage of oil and gas from reservoirs.
In this chapter, we present a seismic amplitude analysis technique, applied to the LF reservoir in the Eugene Island 330 Field. A 4D seismic dataset consisting of two 3D seismic surveys (acquired in 1985 and 1992, respectively) is used to locate regions where seismic amplitudes have changed over time. These changes are qualitatively analyzed in term of brightening, dim-outs, and unchanged areas that might contain bypassed hydrocarbons (pay). We then examine the spatial and temporal changes in acoustic impedance, as determined by the nonlinear 4D seismic inversion techniques described in Chapter 2. The changes in the estimated 4D acoustic impedances are better imaged than those changes observed in seismic amplitudes alone.
3.2 Seismic prediction of dynamic changes in hydrocarbon reservoirs
Laboratory experiments indicate that the acoustic reflection coefficients varies with oil, gas and water mix, effective pressure and temperature (Wyllie et al., 1958; King, 1965; Domenico, 1976; Gregory, 1976 and Wang et al., 1988). This behavior is also observed in the field in water and steam floods, which have produced noticeable acoustic differences over time (Nur, 1982 and 1989; Dunlop et al., 1988; Breitenbach et al., 1989 and Wang et al., 1992).
Changes in seismic data expected from the production of hydrocarbons in the LF hydrocarbon reservoir are summarized using a simple seismic modeling approach (Figure 3.1). Seismic amplitudes were modeled by varying the acoustic properties across a hypothetical, constant thickness sandstone reservoir undergoing pressure depletion and GOR changes. The 1985-1992 petrophysical changes match the production data for the LF reservoir. The acoustic property changes are computed using the experimental data from Ottawa sand (Domenico, 1976). The changing gas/water contact (G/W), gas/oil contact (G/O), and oil/water contact (O/W) also causes changes in seismic amplitudes across the reservoir boundaries. Effective pressure (overburden pressure minus pore fluid pressure) increases as hydrocarbons are drained from a reservoir, reducing seismic amplitudes in both oil and gas sands. If, however, a GOR increase occurs during production, as with the formation of a secondary gas cap, seismic amplitudes are predicted to increase (brighten) over time, because the acoustic affects of the fluid change dominate over the pressure depletion affects. Similarly, seismic amplitudes can dramatically decrease (dim-out), if, in addition to pressure depletion of a high GOR oil, the O/W contact migrates across a reservoir. The amplitude decrease caused by pressure depletion is further enhanced by the drop in the impedance contrast caused by the replacement of low velocity oil and/or gas by relatively high velocity water.
Our seismic modeling investigation suggests that bypassed hydrocarbons are associated with near-zero changes in regions of high seismic amplitude over time, if there has been little change in effective pressure (as within a water-drive reservoir). Areas of sustained high seismic amplitudes are either the highest permeability drainage pathways through which the hydrocarbons move, or isolated pressure compartments that are not connected to the wellbore drainage system. Bypassed hydrocarbon can be located on the basis of changes in seismic amplitudes only after the cause for impedance changes is understood.
3.3 4D seismic monitoring technology
Our 4D seismic analysis technology consists of imaging, feature extraction, and pattern recognition functions, coupled iteratively with high resolution, seismic forward and inverse modeling. Due to the practical limitations, systematic seismic forward modeling methods are only discussed briefly because they are still being developed. Pattern recognition, rather than interpolation and recomputation, is used to examine similarities and differences among the seismic amplitudes and acoustic impedances of the datasets in the hope that we can use large amount of legacy 3D seismic datasets to monitor hydrocarbon drainage and to locate bypassed pay in the mean time.
3.4 The Eugene Island Block 330 Field
The Eugene Island Block 330 (EI-330) Field is located approximately 270 Km south-west of New Orleans (Figure 3.2). This field has produced more than 7.9x107 m3 (equivalent to 5x108 barrels of oil) since 1972. The oil and gas are produced from more than twenty-five stacked sand reservoirs in roll-over anticlines abutting against the large growth fault system (the Red fault) and associated smaller antithetic faults (e.g., the F fault in the study area).
The ongoing development of this field and the remarkable amount of seismic, production, geological and wireline data available from the field make it an ideal location to research the effects of production and fluid migration on changing acoustic (seismic) signals over time. Multiple vintages of 3D seismic data, large amounts of digital wireline logging data, and production histories from multiple wells enable us to study the reservoir fluid dynamics of the LF in great detail within the EI-330 Field.
The study area in this chapter is centered along the Block 330/331 boundary where two vintages of 3D seismic surveys overlap. The two 3D seismic surveys are the 1985 survey centered on Block 330 and the 1992 survey centered on Block 331 (Figure 3.2) (b). Well logs and production data from nineteen wells are available in the overlap area (Figure 3.3). Combined stratigraphic, structural and seismic attribute analyses of the legacy 3D seismic datasets have shown how depositional features and faults compartmentalize reservoirs in this field. Geopressures in the study area are also predicted and characterized (He et al., 1995a and 1995b) as part of the larger study contained in Chapter 5. It is believed that hydrodynamic disequillibrium that has produced dramatic variations along the top-of-geopressure surface over short distances in the minibasin, coupled with periodic fluid release up the large regional growth fault system, is the primary cause for the active hydrodynamic fluid flow in this region. The well-connected aquifer supporting the water drive mechanism within the LF is one result from this dynamic system.
3.5 4D seismic data registration
Amplitude measurements from two 3D seismic surveys cannot be directly used because the data acquisition and processing parameters applied to them originally were very different. Seismic attributes, such as amplitude, phase, and frequency bandwidth, differ considerably. We use spectral matching, amplitude normalization, and phase corrections to correct the data.
The 1985 and the 1992 3D seismic surveys were first re-binned into the same common grid described in Chapter 2. When comparing the two 3D seismic volumes acquired seven years apart, navigation errors may have been introduced because the resolution of navigational systems at sea has been constantly improved. In our case, we found that there is a 62 m displacement in the east-west direction between the two seismic volumes. The offset was determined from a global cross-correlation between the two seismic volumes in both the north-south and the east-west directions. The global cross-correlations are computed from sequential time slices through the entire two volumes.
The frequency bandwidth of the 1985 volume differs from that of the 1992 seismic volume because seismic sources have been steadily improved over the years. To match the frequency bandwidth, we first computed the global frequency spectrum of each volume, determined the bandwidth that common to both, and used it to bandpass filter both datasets. A zero-phase filter is used (Figure 3.4).
Different static shifts were applied to seismic datasets in the original processing. We correct for this shift by cross-correlating corresponding traces to determine the relative offset. The offset is then averaged over all the seismic traces, and the average is applied to the 1985 seismic survey to correct it to that the 1992 survey (Figure 3.5).
Phase shifts between legacy seismic surveys are determined using a cross-correlation technique to determine the phase angle difference between the two volumes. We held the seismic traces from the 1985 survey fixed, while phase-shifting the traces of the 1992 survey with all possible angles (0-360 degree). The phase angle that maximizes the correlation between the two traces is applied to the 1992 survey to correct it (Figure 3.6). Numerical experiments carried out in our volume matching techniques indicate that the matching of conventional post-stack seismic volumes can be best performed by applying time-variant filters to different time windows because the seismic waveform characteristics vary with traveltime (Figure 3.7) and (Chapter 2).
Amplitudes of the two seismic volumes are normalized by requiring that the cumulative amplitude histograms of the two seismic volumes be similar. The two surveys have rather different amplitude distributions (compare (Figure3.8) (a) with (Figure 3.8) (b)). We choose the 1992 seismic volume as the amplitude reference, performed a global histogram renormalization of the 1985 survey (Figure 3.8) (c) and (Figure 3.8) (d). The amplitudes are rescaled by a factor appropriate for real acoustic variations in the earth by comparison with synthetic traces computed from sonic and density logs.
The corrected 4D dataset are shown in (Figure 3.9) and (Figure 3.10), and two cross-sections are shown in (Figure 3.11). Amplitude of the two surveys are generally similar, but significant differences occur within the producing reservoirs. We will next focus on the time window from 1.7 to 2.2 seconds, which includes the LF reservoir, to perform our 4D seismic analysis.
3.6 The LF reservoir in the study area
The LF reservoir is centered along the boundary of blocks EI-330/331 (Figure 3.12). It dips gently to the west from the crest of the roll-over anticline in the center of EI-330, and is bounded on the north by the south-dipping B fault and on the south by the north-dipping F antithetic fault. It is one in a stack of reservoirs (JD-KE-LF) that are main oil and gas producers in the field. Structural dips are of the order of 10-20 degrees, and the sand top deepens from about 2,012 m (6600 ft) to over 2,316 m (7,600 ft) in this fault block. (Figure 3.16) also shows the locations of production wells, and the positions of estimated fluid contacts at the beginning of production in 1972 and those estimated at the beginning of our 4D study in 1992, indicating their movement with time as interpreted from the production data. The regions in red are the original gas cap. Regions in pink are areas that have had oil replaced by gas due to expansion of a secondary gas cap down-dip. Regions in green are predicted to be oil-filled and are thought to have undergone no change in fluid composition. Regions in dark blue are water. Regions in light blue in the west are water sweep occurred since 1972. This figure shows a large region to the west that has been water swept between 1972 and 1992, whereas between the wells 330_PZ_C-18 and 330_PZ_A-16ST to the east of the 330/331 block boundary, a secondary gas cap has grown between the time production started in 1972 and 1992. Low permeability gas is also detected from logs near the crest of the structure (regions in dark pink). The original oil-water contact in the LF reservoir is interpreted to be at the 2,307 m (7,568 ft) level from the production data. Uniform, gravity-driven, up-dip movement of the oil-water contact is predicted from production data. The oil-water contact as deduced from the sparse well data is not horizontal in 1992, but appears to cut across structural contours in the south of the F fault. The dip could indicate that the thick LF sand located in these zones is more permeable, and fluids are being preferentially drawn from these high quality sands.
3.7 4D Seismic amplitude analysis technique
3.7.1 The region-growing algorithms
In practice, the amplitude normalization, frequency and phase matching techniques cannot totally correct for differences in original seismic surveys, so the match is never perfect. The inter-comparisons are accomplished in attribute-derivative space since we have found that only the lowest frequency in amplitude spectra preserve the best commonality among the differently-processed seismic datasets. Comparison based on wavefield envelope are more robust than those based on the seismic trace itself. Therefore, we use the reflection strength, or instantaneous amplitude (Taner et al., 1979; also see Appendix A) of the seismic amplitude in our analysis. These reflection strength volumes computed and divided into volumes of similar high amplitude regions (HARs) through the use of region-growing algorithms that we have developed. The conventional isosurfacing technique bounds regions by connecting data points with the same amplitude values on the surface (Figure 3.13) (a). The usefulness of this technique applied to as noisy seismic data is very limited. Our region-growing technique is more robust. The technique employs nonlinear, 3D operators to isolate HARs within the seismic datasets. Beginning from a set of initial "seed" points (with large amplitude values), we track the magnitude of the change-in-amplitude in 3D. A threshold value of amplitude gradient is selected to bound the regions. The HARs are bounded by surfaces of high amplitudes that exceed the threshold in the gradient of amplitudes away from the seed points. Data points outside HARs are excluded from analysis, downsizing the overall amount of data (Figure 3.13) (b). Rough-cut connectivity between HARs within each dataset are obtained by properly choosing threshold operator so that lowpass spatial filtering, dilation and erosion "grows" the connections between segmented HAR seed points. An example of applying our region-growing and differencing techniques to two traces (1D) extracted from the 1985 and 1992 seismic volumes is shown in (Figure 3.14) (Anderson et al., 1995b).
We then difference the two 3D seismic datasets within each of the HARs. Corresponding HARs defined from the two separate surveys are merged into a single set of HARs by a volume union operation (Anderson et al., 1995b). Each HARs is characterized as having undergone near-zero change, brightening (red in the figure), or dim-out (color coded as green/yellow, red, and blue in (Figure 3.15) (a)). Fine-scaled structure within a single HAR may be indicative of bypassed pay (green in (Figure 3.15) (a)). Intervals of sustained high amplitudes that are within 10% of each other can be used to reveal inter-connectivity (Figure 3.15) (b). The 3D pattern of bypassed hydrocarbons has a surprising inter-connectivity. The amplitude pattern is aligned 30 degrees to the major bounding faults in this portion of the LF reservoir. The apparent lineation may be due to small faults caused by westward shear in the vicinity of the LF reservoir. This pattern is not evident in either of the primary 3D seismic surveys.
Misregistration errors can cause edge effects on the differentiated seismic data, as can be created by velocity differences caused by the changing fluid composition. For example, the formation of a gas cap significantly lowers the velocities within the top of the reservoir, resulting in a mismatch between the two times. These edge effects must be removed from the final image of the reservoir (Figure 3.16), (Figure 3.17), and (Figure 3.18).
3.7.2 Volumetric analysis of drainage of the LF reservoir
Regions of near-zero seismic amplitude differences may possibly contain bypassed hydrocarbons (green in (Figure 3.18)). Dim-outs could be caused by water encroachment or pressure depletion (blue in (Figure 3.18)) and brightened regions could be from GOR increases and secondary gas dissolution (red in (Figure 3.18)).
This detailed hydrocarbon distribution image can be combined with production data to further analyze the drainage patterns caused by production. More than 7.5x104 (470,000 barrels) of oil with low GOR has been produced from this part of the LF reservoir from 1985 to 1992, mostly from wells A-12A, A-6, A-8, and B-4. The other thirteen wells (Figure 3.34) in both Blocks 330 and 331 stopped producing due to water intrusion or mechanical failure early in the production history (before the 1985 3D survey was acquired).
The estimated initial (1972) and current (1992) oil/water and oil/gas contacts from the production data are shown in (Figure 3.12). The oil/water contact has moved about 107 m (350 ft) in vertical distance up-dip and to the east, and the gas expansion should have deepened the gas/oil contact by about 15 m (50 ft) down-dip and to the west. From these production history interpretations, new wells would have to be placed in a very narrow (< 610 m or 2,000 ft wide) band along the structure contours in the fault block B.
However, the 4D image of the LF reservoir predicts a significant different drainage pattern from the interpretation derived based on the gravitational assumptions of the production history (Figure 3.19). The drainage is much more heterogeneous than indicated by the production data. The drainage patterns of the LF reservoir do not follow structure contours. Instead, the bypassed hydrocarbons appear to be related to fingered sands trending in the north-west to south-east direction. Water intrusions appear to have fingered into the reservoir in the same direction. The isolated regions with water invasion appear to be caused by water-conning.
3.7.3 Horizon amplitude extraction of the LF reservoir
We sample the 3D amplitude along the 2D top horizon of the LF reservoir, which corresponds to a strong seismic reflector (Figure 3.20) (a) and (b). The 1985 2D image clearly shows dim-outs that are approximately parallel to structural contours, especially in the west and near the bounding faults B and F in the north and south. These amplitude boundaries correspond to the 1985 oil/water contacts. The amplitude pattern is more continuous in the east. However, by 1992, the amplitudes have become discontinuous. The total area of high reflectivity has decreased with time, as expected for an actively-produced reservoir with decreasing fluid pressures and oil saturations.
Amplitude changes in the top of the LF reservoir also indicate brightened areas marking gas formation (red in (Figure 3.21)) and dim-out areas marking water invasion (blue in (Figure 3.21)). However, the boundaries are more complex and less clearly tied to structure than in the gravitational model derived from production data. Depletion of the gas cap from 1985 to 1992 is imaged as an amplitude dim-out in the eastern portion of the reservoir (up-dip in (Figure 3.21)), with brightening indicated within the newly depleted oil zone (down-dip to the west). Areas of near-zero amplitude change (green and yellow in (Figure 3.21)) are primarily located in regions where no active production wells were operated (e.g., Block 331). We predict that there are significant bypassed oil located in the LF reservoir of this fault block.
Significant edge effects are caused by traveltime misalignment of the LF reservoir in the southwest and south of the fault block (just north of the F fault). This misalignment is probably caused by velocity changes from the significant drainage in the shallower JD reservoir, which is a large gas producer.
3.7.4 Volumetric variations of seismic amplitude differences
The HAR analysis measures the acoustic thickness of the LF reservoir, and allows variations within the volume to be examined. We difference HARs to create volumetric representations of the seismic amplitude changes between the 1985 and 1992 seismic surveys (Figure 3.22). Because 3D variations are difficult to convey on paper, we slice the difference volume through its top, middle and bottom along planes parallel to the LF structure surface (Figure 3.23). We present both a fluid contact interpretation of the differences (Figure 3.22), and the absolute difference images (Figure 3.23).
We then compared these HAR slices with a slice made along the top of the reservoir, as defined by the seismic reflector (Figure 3.21). Both show the gas formation, but the HAR slices demonstrate that the high amplitude anomalies have migrated much deeper into the reservoir and to the west by 1992 (Figure 3.22). However, in the eastern portion of the reservoir (up-dip), water sweep has changed the original gas distribution dramatically due to the active production in Block 330. For example, the A-6 well (in Block 330) produced oil since 1981. The A-6 well was shut-in in 1987 because of a sudden water intrusion, and it was producing hydrocarbon at a GOR of 3,000 before the shut-in (Figure 3.34). The case of amplitude increase (brightening) in Block 331 is unknown-it contains no active production wells. Perhaps pressure decreases within the overall reservoir caused gas to come out of solution. However, the brightened amplitudes moved more than 305 m (1,000 feet) to the northwest from 1985 to 1992, so we believe the amplitude increases to be real.
3.8 4D acoustic impedance analysis technique
Volumetric reservoir descriptions based upon seismic amplitude data are less physical than those based upon acoustic impedance. Because acoustic impedance is more closely associated with the petrophysical and fluid properties of reservoirs than amplitudes. We apply the same region-growing technique to the estimated acoustic impedance volumes. We therefore inverted the amplitude-normalized, phase-matched seismic datasets to produce the estimated impedance volumes. The constrained, full-scale nonlinear seismic inversion technique discussed in Chapter 2 is applied. The estimated 4D acoustic impedance volumes from the 1985 and 1992 seismic volumes within the time window from 1.7 to 2.2 seconds are shown in (Figure 3.24) (a) and (b). The hydrocarbon reservoirs correspond to low impedance regions within these two impedance volumes.
3.8.1 Differences in acoustic impedance between 1985 and 1992 in the LF reservoir
Using our region-growing algorithms, the 1985 and 1992 acoustic impedance volumes are then segmented into similar Low Impedance Regions (LIRs) and data from outside these LIRs are excluded from future analysis (Figure 3.25).
The shape of the LIRs (derived from acoustic impedance volumes) differs from that of the HARs (derived from seismic amplitude volumes). Because seismic amplitudes are sensitive to reservoir boundaries, to fluid contacts, and to impedance contrasts, while acoustic impedances are more sensitive to internal continuity and heterogeneity within the reservoir itself. Reservoir thicknesses derived from the estimated acoustic impedance volumes are much more accurate than those resolved from seismic amplitudes. As we will see later in this section, our region-growing techniques are more robust when applied to the estimated acoustic impedance data.
Reservoir drainage from the shallower JD reservoir causes vertical traveltime delays in the LF reflections. Up to 20 ms of delay in the 1992 survey versus the 1985 survey can be seen in (Figure 3.26). We compensate the traveltime delays by applying a horizon-balancing technique. We interpreted the top of the LF reservoir sand on both estimated acoustic impedance volumes. The traveltime differences in the study area caused by production can be seen in (Figure 3.27). We then corrected the traveltime differences between the 1985 and 1992 LF reservoir tops by shifting the 1985 data. The edge effects shown in (Figure 3.18) are thus eliminated.
The differences within the LIRs of the LF reservoir between 1985 and 1992 were then computed (Figure 3.28). Changes in acoustic impedance over time are quantitatively related to the drainage of hydrocarbons. Near-zero impedance differences indicate locations where there was minimal change within the LIRs of the surveys, which we interpret to be possible bypassed hydrocarbons (green in (Figure 3.28)). Increases in impedance are likely caused by water encroachment or pressure depletion between the times of the two surveys (blue in (Figure 3.28)), and decreases from GOR increases and secondary gas dissolution (red in (Figure 3.28)). A visualization of the volumetric changes within the LF reservoir is created by slicing through the upper-, mid-, and lower-intervals of the LF LIR (Figure 3.29). Comparing this result to the similar figure made from seismic amplitude differences (Figure 3.19), more coherences are observed within the LIR than in the seismic amplitude HAR. In the upper-interval slice, green regions can be from either oil or gas, since the original reservoir in 1985 contained a gas cap in this fault block. In the mid-interval slice, red and green regions are the primary features, which indicate that both gas and oil are still present in 1992. In the lower-interval slice, blue regions are predominant, indicating the water sweep started from the base of the reservoir, as expected. However, water appears to have encroached in what might be fingers through high permeability sand from several directions, not just from the deeper portion of the reservoir in the west.
3.8.2 Horizon acoustic impedance extraction of the LF reservoir
The acoustic impedances extracted from along the LF reservoir top structure from each impedance volume are shown in (Figure 3.30). The extracted impedances can be directly used to delineate distributions of hydrocarbons remaining within the reservoir.
The reservoirs are outlined by the boundaries between high (blue) and low (red) impedances. The reservoir dim-outs and brightenings have higher coherence than in the corresponding analysis that used seismic amplitudes. The hydrocarbon contacts are clearly seen on both images. The similarities and differences between the 1985 and 1992 extractions (Figure 3.30) (a) and (b) show the quantitative hydrocarbon changes that have occurred within the LF reservoir. Such changes indicate reservoir dim-outs and brightening inside the reservoir from 1985 to 1992.
As with the qualitative results obtained from the amplitude analysis, the dim-outs (increased impedance) are approximately parallel to structural contours, particularly in the western part of reservoir (Figure 3.31). These impedance boundaries correspond to the oil/water contact. No obvious water intrusion is detected perpendicular to structural contours in the northwest. However, long-distance water encroachment perpendicular to the structural contours has occurred along the north of the F fault. In 1985, the region of low impedances show continuity to the eastern part of the reservoir. But by 1992, the continuity has markedly decreased. In general, the area of low impedances decreases with time, as might be expected for an actively-produced reservoir.
3.8.3 Volumetric variations of acoustic impedance differences in the LF reservoir
Impedance differences within the LIRs are related to hydrocarbon movement (Figure 3.32). Brightened volumes mark gas cap formation (red in (Figure 3.32)) and volumes with dim-outs mark water sweep (blue in (Figure 3.32)). But the boundaries are more complex and less tied to the reservoir structure than described by production data analysis. Low impedance regions above the initial gas/oil contact (red and pink regions in (Figure 3.12)) in fault block B are considered to be gas. Depletion of the gas cap from 1985 to 1992 is not obvious in these regions as shown in (Figure 3.28) and (Figure 3.29). However, areas with decreasing impedances occur further down-dip to the west within the oil zone (Figure 3.12) and (Figure 3.32), suggesting down-dip migration of the gas cap and gas dissolution from oil. The areas with near-zero impedance change (green regions in (Figure 3.32) (b)) in Block 331 are greater than those in Block 330, indicating that there may be more bypassed hydrocarbon reserves in Block 331 of the fault block. However, as what we have noted previously, the cause of these regions with decreased impedance is unknown because there was no well producing from 1985 to 1992.
We present a fluid contact interpretation of the differences (Figure 3.33) based on the analysis of images shown in (Figure 3.32). Both volumetric impedance extractions and planar extractions show the low impedance anomalies migrated much deeper down-dip. This information is not present in the 2D surface extraction analysis (Figure 3.31).
The total volume of decreased impedance (shown as red) and near-zero impedance change (green) is larger than those observed in the HAR analysis, and these volumes are more continuous as well. This improvement occurs because traveltime delays caused by production of shallower reservoirs have been eliminated.
The regions with near-zero impedance changes (green and yellow in (Figure 3.33)) within the original gas cap to the east should still be gas. However, significant water encroachment is detected, and the original gas distribution (gas/oil contact) has been changed dramatically by the production in Block 330.
Production data from the nineteen wells used in this study are shown in (Figure 3.33). The well data in Block 330 are consistent with the predicted hydrocarbon drainage patterns from the 4D amplitude and impedance analyses. For example, in 1985, A-8 well produced oil, gas, and water (Figure 3.34) as the water front moves up-dip toward east and south, the well was shut-in in mid 1987 because of water encroachment. This well is located in the area of water sweep predicted by our 4D analysis (Figure 3.22) and (Figure 3.33). The nearby A-6 well (in the up-dip direction to the south of A-8) has a similar history, but it saw water about six months earlier (instead of later) than A-8, indicating that either the water intrusion in the LF reservoir is preferentially oriented in certain directions or it is caused by water-conning. Another example is the B-4 well (Figure 3.34), which came into production in 1984, it has produced oil steadily with low GOR into 1991. It is on the edge of the area of bypassed oil predicted by our 4D analysis (Figure 3.33).
The water-conning is observed much more clearly in the Block 331 (isolated blue areas in (Figure 3.22) and (Figure 3.33). Although production from wells in Block 331 has stopped because of water encroachment or sand-flow induced mechanical failures before 1985, there are water encroachment "halos" detected from our 4D analysis existing around the watered-out wells (e.g., wells A-4, A-7 and A-5). Nevertheless, the predicted water encroachment are not radially isotropic, instead, the drainage appears to have been significantly affected by sand quality variations and other permeability heterogeneities.
3.9 Quantitative 4D seismic monitoring technique
In an effort to further quantify our 4D seismic monitoring techniques (He et al., 1994 and 1995b), we have used a multi-dimensional seismic forward model based on the finite element method (FEM) and fluid substitution technique (Sun, 1994) to verify whether the hydrocarbon drainage changes can produce the observable seismic amplitude changes. We constructed a velocity model of the LF reservoir as it appeared in 1985. The 1985 seismic response is shown in (Figure 3.35) (a). To change the reservoir petrophysical properties, we use Gassman's equation combined with the dynamical theory of porous media (Sun, 1994) for estimating the acoustic velocity of rock containing a multi-phase pore fluid. We incorporate changes in gas cap development, water intrusion, GOR increase and unchanged oil saturation into the 1992 model. We then compare its FEM seismic response to the 1985 response.
The modeled seismic amplitudes of the LF reservoir in 1992 are shown in (Figure 3.35) (b), and the differenced seismic model results in (Figure 3.35) (c). Dim-outs (blue in (Figure 3.35) (c)), brightening (red in (Figure 3.35) (c)), and unchanged differences (<10% change, green in (Figure 3.35) (c)) are all observed. The FEM model results indicate that surface seismic data can be used to monitor the hydrocarbon drainage process.
3.10 Conclusion
When applied to data from the LF reservoir in the EI-330 Field of offshore Louisiana, our 4D seismic monitoring techniques are effective in determining the locations and volumes of bypassed hydrocarbons. Corrections for spatial data registration and traveltime from delays caused by drainage above the reservoir are very important. Without them, the resulting edge effects may cause the 4D analysis to fail.
The application of our 4D seismic monitoring techniques to seismic amplitude and estimated acoustic impedance data suggested that: a) the amplitude analysis is the better prediction of reservoir drainage patterns; and b) the impedance analysis is better prediction of reservoir petrophysical property changes. The integration of the two techniques produces a quantitative understandings of the drainage of hydrocarbon reservoirs over time.
Our analysis suggests that drainage patterns are very sensitive to heterogeneities in lithology, porosity, and permeability of the reservoir. Water intrusion and gas cap expansion are found to be almost independent of the LF reservoir structure. Instead, we found that they are determined by potential fluid pathways provided by faults and fine-scale sediment fairways with high permeability.
Production wells within the LF reservoir in fault block B were found to have produced very limited, heterogeneous and contorted drainage radii. The drainage pathways in the LF reservoir are elongated along directions of high permeability (northwest-southeast) and shortened along directions with lower permeability (northeast-southwest), which indicates that the production wells have greater drainage efficiencies in selected directions. The wells that were shut-in because of water intrusions were seen to have "suffered" from much more water-conning from below than from lateral, up-dip gravity driven water invasion. The use of 4D seismic data has significantly improved the understanding of reservoir dynamics of the LF reservoir, compared to what is possible with the traditional interpretations based on well logs and production history.
References
Alexander, L., and P.B. Flemings, 1995, Geologic evolution of a Plio-Pleistocene salt-withdrawal minibasin: Eugene Island Block 330, Offshore Louisiana: American Association of Petroleum Geologists Bulletin, v. 79, p. 1737-1756.
Anderson, R.N., E. Bagdonas, A. Boulanger, L. Xu, and W. He, 1995a, 4D seismic interpretation techniques for the detection of drainage and migration in oil and gas reservoirs: US. Letters Patent Pending.
Anderson, R.N., A. Boulanger, W. He, Y.F. Sun, L. Xu, 1995b, 4D seismic imaging of drainage and pressure compartmentalization in the Gulf of Mexico, Oil and Gas Journal, part 1, March 28, part 2, April 4.
Anderson, R.N., P.B. Flemings, and S. Losh, 1995c, The GBRN/DOE Eugene Island Dynamic Enhanced Recovery Project, CD-ROM, Lamont Press.
Breitenbach, E.A., G.A. King, and K.N.B. Dunlop, 1989, The Range of Applications of Reservoir Monitoring, Society of Petroleum Engineers Paper 19853.
Domenico, S.N., 1976, Effect of brine-gas mixture on velocity in an unconsolidated sand reservoir, Geophysics, v. 41, p. 882-894.
Dunlop, K.N.B., G.A. King, and E.A. Breitenbach, 1988, Monitoring of oil/water fronts by direct measurement, Society of Petroleum Engineers Paper 18271.
Gregory, A.R., 1976, Fluid saturation effects on dynamic elastic properties of sedimentary rocks: Geophysics, v. 41, p. 895-921.
He, W., R.N. Anderson, and X. Wang, 1994, Constructing 3D porosity and fluid pressure models in the Eugene Island Block 330 Area offshore Louisiana: American Association of Petroleum Geologists Annual Convention Official Program, v. 3, p. 167.
He, W., and R.N. Anderson, 1995a, Seismic Attribute mapping of the geopressure surface in the Gulf of Mexico, Journal of Geophysical Research, submitted.
He, W., R.N. Anderson, X. Wang, and Y.-C. Teng, 1995b, Application of finite element seismic modeling in the Eugene Island Block 330 Field offshore Louisiana, Gulf of Mexico: American Association of petroleum geologists Annual Convention Official Program, v. 4, p. 41A.
Holland, D.S., J.B. Leedy, and D.R. Lammlein, 1990, Eugene Island Block 330 Field-U.S.A. Offshore Louisiana: in Structural traps III: Tectonic fold and fault traps: Treatise of Petroleum Geology, Atlas of Oil and Gas Fields, E.A. Beaumont and N.H. Foster, compilers, American Association of Petroleum Geologists, Tulsa, p. 103-143.
Lumbly, D., 1995, Seismic monitoring of oil production: a feasibility study: presented at the 1994 Society of Exploration Geophysicist Annual Conference, Los Angeles.
King, M.S., 1965, Wave velocities in rocks as a function of changes in overburden pressure and pore fluid saturants: Geophysics, v. 31, p. 50-73.
Nur, A.M., 1982, Seismic Imaging in Enhanced Recovery, Society of Petroleum Engineers Paper 10680.
Nur, A.M., 1989, Four dimensional seismology and (true) direct detection of hydrocarbons: the petrophysical basis: The Leading Edge of Geophysics, v. 9, p. 30-36.
Sun, Y.F., 1994, On the foundations of the dynamical theory of fractured porous media and the gravity variations caused by dilatancy, Ph.D. Dissertation, Columbia University.
Taner, M.T., F. Koehler, and R.E. Sheriff, 1979, Complex seismic trace analysis: Geophysics, v. 44, p. 1041-1063.
Wang, Z., and A.M. Nur, 1988, Effect of temperature on wave velocities in sands and sandstones with heavy hydrocarbons: Reservoir Engineering, v. 3, p. 158-164.
Wang, Z., and A.M. Nur, 1992, Aspects of rock physics in seismic reservoir surveillance: in Reservoir Geophysics, R.E. Sheriff, Editor, Society of Exploration Geophysics, p. 285-300.
Wyllie, M.R.J., A.R. Gregory, and G.H.F. Gardner, 1958, An experimental investigation of factors affecting elastic wave velocities in porous media: Geophysics, v. 23, p. 459-493.
