Anhydrite distribution within a shelf-margin carbonate reservoir: San Andres Formation, Vacuum Field, New Mexico, USA

Transcription

1 Anhydrite distribution within a shelf-margin carbonate reservoir: San Andres Formation, Vacuum Field, New Mexico, USA Matthew J. Pranter,, Neil F. Hurley and Thomas L. Davis Colorado School of Mines, Department of Geology and Geological Engineering, Golden, Colorado, USA ( matthew.pranter@colorado.edu; nfhurley@mines.edu; tdavis@mines.edu) Present address: University of Colorado at Boulder, Department of Geological Sciences, Campus Box 399, 0 Colorado Avenue, Boulder, Colorado, USA ABSTRACT: Anhydrite cement causes significant heterogeneity within the San Andres reservoir of the Vacuum Field where it is associated with faults, fractures, karst zones and highly cemented dolomite intervals. The primary reservoir rocks within the Central Vacuum Unit of the Vacuum Field are dolomitized peloidal packstones, skeletal and ooid grainstones and fusulinid packstones. These rocks alternate with lower reservoir-quality dolomite intervals with variable amounts of anhydrite cement. Nodular and pore-filling fabric-selective anhydrite cements are common within the reservoir interval. Cross-plots of apparent matrix grain-density versus apparent matrix volumetric cross-section (ρ maa U maa cross-plots), combined with V p /V s seismic attributes and core data, provide insight into the vertical and lateral distribution of anhydrite within the San Andres reservoir. Anhydrite is generally concentrated in thin depositional cycles or intervals that are separated by relatively anhydrite-free cycles that exhibit relatively higher porosity and permeability. Lower V p /V s values correspond to higher percentages of anhydrite and are useful for mapping anhydrite distribution. This evaluation of anhydrite distribution provides an estimate of the significant cementation-related heterogeneities within the reservoir that is useful for development-well planning and to target areas of bypassed pay. KEYWORDS: San Andres, carbonate, anhydrite, cement INTRODUCTION The Vacuum Field was discovered in 99 and is one of the larger oil fields in the Permian basin that produces from the Permian San Andres and Grayburg formations. It is located in southeast New Mexico and is part of a major productive trend along the Northwest Shelf of the Permian basin (Fig. ). Stratigraphic, structural and diagenetic variability within the shelf-margin carbonates of the San Andres and Grayburg formations form a heterogeneous reservoir. Detailed characterization and modelling of heterogeneities within the reservoir are necessary so that supplemental recovery techniques are more effective and areas of bypassed pay can be targeted. In addition, such characterization provides a better means of locating infill development wells, including horizontal wells. The study area (Figs and 3), located within the Central Vacuum Unit (), is under waterflood operations and was converted to a partial-field CO flood and monitored to evaluate the effect on reservoir performance and recovery. During the CO programme, time-lapse, multi-component (4D, 9-component) seismic data were acquired to aid in the reservoir characterization effort. The main goal of this study was to evaluate the distribution of anhydrite within the San Andres reservoir. In doing so, it was determined that V p /V s measurements from multi-component surface seismic data can be used to detect the lateral distribution of anhydrite within the reservoir. When combined with Petroleum Geoscience, Vol. 0 04, pp United States Guadalupe Mountains N Diablo Platform 0 Miles 50 0 Kilometres Vacuum Field Northwest Shelf New Mexico Texas Delaware Basin Northern Shelf Eastern Shelf anhydrite estimates at well locations from ρ maa U maa crossplots, 3D models of anhydrite cement distribution can be Central Basin Platform Midland Basin Fig.. Location map showing the Vacuum Field on the Northwest Shelf of the Permian basin. Major San Andres and Grayburg fields are shown as black areas. Modified from Hills (984) /04/$ EAGE/Geological Society of London

2 44 M. J. Pranter et al Central Vacuum Unit W 9 5W W W W North West Vacuum Unit / Mile (6 ft) Kilometre (000 m) W 56W 7W W 5W W 345 WI W WI-94 WI WI WI WI Vacuum Grayburg San Andres Unit Fig.. Base map that shows the western portion of the Vacuum Field and the location of the Central Vacuum Unit. The large circle shows the extent of time-lapse, multi-component, 3D seismic surveys. Small solid circles indicate production wells. Small open circles with arrows indicate injection wells. WS WS66 WS67 67 SO SO6 66 WS74 58 WS68 SO WS WS5 63 SA6 6 SA SA8 Central Vacuum Unit WS WVG8 74 WS4 46 SB7 WS WS88 SO SO5 WS94 83 WVG SO SD WVG0 WVG SE North WVG4 VGS57 VGS57 SL6 VGS49 VGS50 SL8 VGS4 VGS34 VGS7 VGS7 VGS9 VGS SL VGS58 VGS58 SL7 VGS50 VGS35 VGS SR7 VGS43 VGS43 VGS8 VGS SL9 ST VGS8 VGS WS WS 5 97 WS9 WS WS0 4 WS WVG WVG WS4 WS WS Producer Injector Cored well SR0 96 SR 95 SR8 95 WVG NW LATERAL 0 7 / Mile (6 ft) Kilometre (000 m) SR9 SR7 0 NE LATERAL SAB5 SR 30 0 SR SAB SAB SVAA Prod. / Inj. from other reservoir CO Injector 7 Well used for!maa -U analysis maa Fig. 3. Base map of available data in the reservoir characterization area. The dashed square indicates the area where the 3D geological model was constructed. The large circle shows the extent of the time-lapse, multicomponent, 3D seismic surveys. Open circles indicate cored wells and CO injection wells are indicated by triangles. Vertical seismic profiles (VSPs) were acquired in wells -0 and WS-5 (central part of study area). A Formation MicroImager log (FMI) was acquired in well WS-6 (southwest part of geological model area). The location of the dual-lateral horizontal well (-0) is shown in the southeast part of the time-lapse area. Wells with names underlined were used for ρ maa U maa analyses. generated. These models can be combined with carbonate facies models or other petrophysical models to produce a more representative reservoir model. Available data Within the study area (Fig. 3), 3D surface seismic volumes consist of four compressional-wave (p-wave) volumes, four

3 Anhydrite distribution in a carbonate reservoir, USA N S T 00 T GB 00 Permeability (md) Time (ms) 700 USA LSA KZ 0. Oolite Grainstone Peloidal Packstone Wackestone Mudstone Siltstone Porosity (fraction) Fig. 4. Porosity permeability cross-plot based on core data from well m 500 ft Fig. 6. North south compressional-wave seismic profile showing small-scale faults (black lines). Displayed horizons include the Grayburg (GB), Upper San Andres (USA), Lower San Andres (LSA) and the base of a laterally extensive karst zone within the Lower San Andres (KZ). Red and blue colours correspond to changes in peak and trough amplitudes, respectively (SEG reverse polarity). Well paths are shown as yellow vertical lines. The location of the seismic line is shown in Figure 5. 0 (Formation MicroImager) in well WS-6, vertical seismic profiles (VSPs) in wells -0 and WS-5, and log data from a medium-radius, dual-lateral horizontal well (-0). Fifty of the wells are present within the smaller 3D geological model area (Fig. 3). Six cores were available (Fig. 3) and neural-network-estimated permeability curves, and injection and production data were provided by Texaco, the unit operator, for the majority of the wells within the study area. Fig. 5. Top San Andres structure contour map. Contour interval is ft (6 m). The location of the seismic line of Figure 6 is shown. The San Andres shelf margin is clearly depicted by the closely spaced contours. fast shear-wave (S-wave) volumes, and four slow shear-wave (S-wave) volumes. These data were acquired as part of a time-lapse seismic monitoring project associated with one phase of the CO injection programme within the. The data were also used to evaluate the static reservoir characteristics within the area. A regional p-wave seismic volume that covers the study area was also available. Other data include conventional logs (commonly gamma ray, neutron, density, resistivity) from wells, one FMI Geological setting Within this portion of the Vacuum Field, the Permian San Andres Formation (Guadalupian) consists of approximately 500 ft (457 m) of dolomites interbedded with a few thin dolomitic siltstones at a depth of approximately 4500 ft (37 m). Only the upper 0 0 ft (83 44 m) of the San Andres comprise the main hydrocarbon-bearing interval. The overlying Grayburg Formation (Guadalupian) consists of approximately ft (76 9 m) of interbedded dolomite, sandstone, anhydrite and shale. In general, the Grayburg Formation exhibits much lower reservoir quality than the San Andres Formation in this area. Permeability increases with porosity within the reservoir interval; however, similar to many San Andres reservoirs within the Permian basin, there is considerable scatter about this trend (Fig. 4). The Vacuum Field is associated with an anticlinal feature that developed due to a combination of sediment drape, differential compaction and fault development (Purves 990). Shelf-margin depositional relief is dominant to the south (Fig. 5), and small-scale faults (throw=0 5 ft; 3 8 m) compartmentalize the reservoir laterally to varying degrees (Fig. 6). This anticlinal feature, combined with the high-frequency cycles that are characteristic of the San Andres Formation, created a stratigraphic-structural trap for hydrocarbons at Vacuum Field. Evaporites, supratidal carbonates, and low-permeability siltstones provide the seal for the reservoir.

4 46 M. J. Pranter et al. Fenestral algal laminated MS Laminated MS Skeletal Ooid Peloidal PS/GS Middle Shelf --- Shelf Crest --- Outer Shelf --- Shelf Margin Supratidal Restricted Intertidal Intertidal Subtidal Mean Sea Level Peloidal PS Fusulinid WS/PS Crinoidal- Fusulinid WS/PS Fig. 7. Facies model of the San Andres Formation in the Central Vacuum Unit. Four main facies tracts include middle-shelf to shelf-crest supratidal, shelf-crest restricted intertidal and intertidal, outer-shelf moderate-to-high energy subtidal, and outer-shelf to shelf-margin low-to-moderate energy subtidal. Core photographs illustrate key facies showing a general decrease in interpreted depositional energy basinward. Black scale bar represents in. (.5 cm) in all photographs. MS, mudstone; WS, wackestone; PS, packstone; GS, grainstone. Four main facies tracts are defined for the San Andres Formation within this portion of the Vacuum Field (Fig. 7). These facies tracts include: () middle-shelf to shelf-crest supratidal; () shelf-crest restricted intertidal and intertidal; (3) outer-shelf moderate-to-high energy subtidal; and (4) outershelf to shelf-margin low-to-moderate energy subtidal. The primary reservoir rocks consist of dolomitized peloidal packstones, skeletal grainstones and fusulinid packstones (Adams 997; Scuta 997; Pranter 999). These rocks alternate with lower reservoir-quality dolomite intervals that exhibit variable quantities of anhydrite cement. The Lovington siltstone separates the Upper and Lower San Andres in the northwest portion of the study area and is characterized by very low matrix permeability. The Lovington represents aeolian silts and sands that were deposited on the platform. The San Andres in this area represents an overall shallowing-upward interval composed of numerous high-frequency depositional cycles that subdivide the reservoir into alternating zones of high and low reservoir quality. Significant faults, fractures and features resulting from pervasive diagenesis overprint the primary depositional fabric (Adams 997; Scuta 997). The prominent diagenetic processes that were active include dolomitization, karstification and cementation. Diagenetic effects increase reservoir complexity. The San Andres Formation along the Northwest Shelf was subdivided using core, log and seismic data into two third-order composite sequences, herein referred to as the Upper and Lower San Andres (Fig. 8). Four San Andres high-frequency sequences comprise the most productive reservoir interval within the study area. These sequences include the upper two high-frequency sequences of the Lower San Andres and the upper two high-frequency sequences of the Upper San Andres. Individual high-frequency sequences are further divided into numerous higher-order (fifth-order) depositional cycles that are generally represented by distinct vertical lithofacies successions. Analysis of the sedimentology and sequence stratigraphy from core and 3D seismic data suggests that the San Andres first developed in an overall aggradational (vertically stacked) pattern, followed by a decrease in accommodation and basinward progradation (seaward-stepping facies tracts). This interpretation is in agreement with several other interpretations of the San Andres Formation from outcrops within the Guadalupe Mountains (Sarg & Lehmann 986; Sonnenfeld & Cross 993; Kerans 995; Kerans & Kempter 0). MULTICOMPONENT SEISMIC DATA Compressional and shear waves respond differently to variations in rock and fluid properties. V p /V s measurements, for example, can be affected by several petrophysical properties but are influenced significantly by porosity, pore geometry and mineralogy (Tatham & McCormack 99). At Joffre Field, Arestad (995) showed that an increase in anhydrite content within the reservoir corresponded to reduced porosity and to a local decrease in V p /V s. Within the study area, seismic amplitude and average reflection strength data correlate well with average porosity data for the reservoir interval (Pranter 999). Based on the work of Arestad (995), the question arose whether V p /V s measurements could be used to predict porosity. To compute V p /V s values for the reservoir, interval time measurements were used for the top of Grayburg to the middle of the Lower San Andres

5 Anhydrite distribution in a carbonate reservoir, USA 47 Fig. 8. North south profile through wells -, -345 and -00 showing gamma ray, neutron porosity, general lithofacies description from core analysis and interpreted sequence-stratigraphic framework. Profile datum is the top of the Lower San Andres. From core analysis, a simplified description of an ideal vertical lithofacies succession is illustrated. Depth is in feet. Lithofacies for -345 and -00 were modified from Capello de Passalacqua (995), Adams (997) and Scuta (997). interval from both compressional and shear-wave volumes. This interval was used because it coincides with the portion of the reservoir that is commonly logged in this part of the field. Also, the interval coincides with two significant horizons that exhibit relatively high-amplitude, semi-continuous reflections that are relatively simple to interpret on 3D seismic data. The interval covers most of the San Andres reservoir and it is thick enough so that the sensitivity to small differences in seismic interpretation is minimized. Because it is possible that mineralogy affects V p /V s values, it was necessary to estimate mineralogical distribution within the reservoir. ρ maa U maa cross-plots (discussed later in the section entitled petrophysical analysis ) were used to evaluate mineralogical composition. These estimates were then used to determine the effect of lateral changes in mineralogy on multi-component seismic response. This evaluation was necessary before V p /V s information could be used in reservoir modelling to estimate porosity. Anhydrite is an abundant mineral, second to dolomite, within the San Andres reservoir at the Vacuum Field. Therefore, the focus of the mineralogical estimation was on anhydrite. Zones of massive anhydrite that range from several centimetres to more than 30 cm in thickness were observed in cores, and intervals greater than 5 cm in thickness were identified through well-log analysis. Based on log correlations and the abundance of anhydrite within certain depositional cycles, it appears that these zones extend laterally across the study area. Anhydrite is abundant as a pore-filling cement and as a replacement mineral for dolomite. Anhydrite cement can be nodular (Fig. 9) and, in some cases, it completely fills pores of fusulinid packstones and peloidal grainstones. It is also common for fractures and karst breccia porosity to be completely filled with anhydrite. PETROPHYSICAL ANALYSIS OF ANHYDRITE Anhydrite (CaSO 4 ) is a mineral with a high bulk density (average.98 g cm 3 ). As such, anhydrite-rich intervals commonly appear as zones of high density, low neutron porosity and high photoelectric index (PEF=5) on the appropriate well logs. Because most anhydrite-bearing intervals in the Vacuum Field are mixed lithologies with dolomite, a range of porosity values and less than 00% anhydrite, this study chose to use a combination of logs (photoelectric, density and neutron) to make quantitative estimates of anhydrite composition. This is considered to be a superior technique to simple density-log or density-neutron cross-plot interpretation because more logs can help resolve more variables. Identification of common minerals and lithologies can be achieved by visual inspection of a photoelectric index curve combined with neutron-density curves. To evaluate mixed

6 48 M. J. Pranter et al. A B ρ maa gcm Kaolinite Quartz Dolomite K-Feldspar Illite Anhydrite Calcite cm inch (a) U maa Barns cm-3 cm inch cm inch C D ρ maa gcm Kaolinite Quartz Dolomite K-Feldspar Illite Anhydrite Calcite lithologies and to semi-quantitatively estimate mineral volumetric proportions, a cross-plot of the log response for an interval is recommended (Doveton 994). A cross-plot technique with well log data was used to identify lithology (or mineralogy) within the reservoir interval. The method involved plotting apparent matrix grain density, ρ maa (in g cm 3 ), versus apparent matrix volumetric cross-section, U maa (in barns cm 3 ). To compute the apparent matrix properties, ρ maa and U maa, the contribution of the pore fluid was removed. The ρ maa U maa cross-plot utilizes log measurements of the photoelectric index, neutron-density cross-plot porosity and bulk density for matrix mineral evaluation. The apparent matrix grain density is computed as: ρ maa = ρ b φ ta ρ f φ ta () where ρ b (g cm 3 ) is the bulk density from the density log, φ ta (fraction) is the apparent total porosity from neutron and density logs and ρ f (g cm 3 ) is the flushed zone pore fluid density (>.0gcm 3 for fresh-water mud filtrate). In practice, φ ta is the average of the density and neutron porosity logs. The cm inch Fig. 9. Diagenetic features that affect reservoir quality: (A) karst breccia filled with anhydrite cement; (B) nodular anhydrite cement; (C) anhydrite replacement of fibrous gypsum crystals; (D) open fractures. (b) U maa Barns cm-3 Fig. 0. ρ maa U maa cross-plots for top-of-grayburg to middle of Lower San Andres interval in wells (a) -96 and (b) -3. Apparent matrix grain density (ρ maa ) is plotted against apparent matrix volumetric cross-section (U maa ). Each point represents a 0.5 ft (5 cm) data sample from the well logs in this well. apparent matrix volumetric cross-section (Doveton 994) is computed from the photoelectric index and bulk density log measurements as: U maa = P e ρ e φ ta U f φ ta () where P e (barns/electron) is the photoelectric index, ρ e (g cm 3 ) is the electron density, and U f (barns cm 3 )isthe volumetric photoelectric absorption or volumetric cross-section of the flushed zone pore fluid (>0.398 barns cm 3 for freshwater mud filtrate). The photoelectric index, P e, must be converted to a volumetric measure by multiplying it by the electron density (Doveton 994), which is defined as: ρ e = ρ b (3).0704 The ρ maa and U maa values are generally presented on a cross-plot with a percentage-scaled template or grid. ρ maa and U maa cross-plots were generated for 7 wells in the study area that had the required log data (Fig. 3). Fifteen of the wells were located in the geological model area (Fig. 3) and they were used for more detailed analyses. The areal coverage was

7 Anhydrite distribution in a carbonate reservoir, USA 49 WVG SD SE WS WS3 SR SR 96 6 WVG 37 WVG WVG WS4 WS5 SR9 Fig.. Interpreted ρ maa U maa cross plot for cored well -345 (top of Grayburg middle of Lower San Andres). A % cutoff was used to highlight points that correspond to anhydrite zones in the core. Each point represents a 0.5 ft (5 cm) data sample from the well logs in this well. 345 U maa Barns cm-3 Gamma Ray m p Porosity >% Anhydrite Indicator 0 0 s w g 30 % 0.5 gm cm Umaa Fig.. Log curves and lithology for well The anhydrite indicator log (derived from the ρ maa U maa cross-plot for well -345) is shaded to show intervals in the well that plot beyond the % cut-off on the ρ maa U maa cross-plot. These intervals correlate with anhydrite zones observed in the -345 core. The cross-plot data are also depicted as ρ maa and U maa curves on the figure. Depth is in feet. judged to be broad enough to be representative of the study area. The interval from the top of Grayburg to the middle of the Lower San Andres was evaluated for each of the 5 wells. This interval was used for the analysis, because several of the wells did not penetrate deeper than that zone. For direct ρ maa ρmaa 36 North 05 WS Contour Interval: 5 percent / Mile (6 ft) 3 Kilometre (000 m) 07 Fig. 3. Map of average anhydrite percent for the top-of-grayburg to the middle of the Lower San Andres interval. For each well, average anhydrite percent was calculated based on the presence of anhydrite from ρ maa U maa cross-plots. Average anhydrite percent values are shown in white at each well. comparison, V p /V s maps were generated for the same interval using the multi-component seismic data. Cross-plots for wells -96 and -3 illustrate the dominance of dolomite within the reservoir and the variability in anhydrite content between the two wells (Fig. 0). On the ρ maa and U maa cross-plot, standard reference points are shown for common end-point minerals (e.g. dolomite and quartz). Data points are plotted on joined percentage-scaled triangular grids to show the amount of each mineral within a zone. To estimate the proportion of anhydrite within the analysed interval, the ρ maa U maa cross-plots were first calibrated to core data. To calibrate the cross-plot information, linked cross-plots and density, neutron and photoelectric log curves were evaluated within Petroworks log analysis software (Landmark Graphics Corporation). Data points on the ρ maa U maa crossplot were selected using various percentage cut-offs (Fig. ; e.g. all points that plot >% on the anhydrite, dolomite, calcite triangle) and the corresponding intervals were evaluated on gamma ray, porosity, ρ maa and U maa log curves. The log curves were then visually compared to lithologies from core. This process was repeated until a percentage cut-off on the ρ maa U maa cross-plot that represented the major anhydrite intervals in the core was determined. Using core and log data from well -345, anhydrite content defined by data points that were greater than the % cut-off line on the ρ maa U maa cross-plot (Fig. ) corresponded to the anhydrite zones observed in core. Likewise, data points that plotted less than the % cut-off corresponded to intervals in the core that contained negligible anhydrite. Using this method, the % cut-off on the ρ maa U maa cross-plot was found to correspond approximately to insignificant anhydrite within the reservoir. The % cut-off was applied to the other cross-plots to estimate the occurrences SR7

8 50 M. J. Pranter et al. North.4 V p/v s.4 / Mile (6 ft) Kilometre (000 m) Fig. 4. V p /V s map for the Grayburg to middle of the Lower San Andres interval (Blaylock 999). This map covers approximately the same area as Figure 3. Black dots and open circles represent well locations. ρ Anhydrite Percent (from maa -U maa Analysis) r = r = 0.66 n = V p /V s Fig. 5. Cross-plot of anhydrite percent versus V p /V s (derived from the multi-component 3D seismic data) for the top-of-grayburg to the middle of the Lower San Andres interval. of anhydrite within each well. Figures and illustrate the calibration methodology and the results of using a % anhydrite cut-off in well Using the % cut-off, the most significant anhydrite zones are represented (Fig. ). The percentage of anhydrite within the reservoir interval was determined for each of the 7 wells and mapped to show an estimate of the anhydrite distribution within the study area (Fig. 3). Even though the number of wells with the appropriate logs was limited, the wells were scattered throughout the study area and clear trends could be discerned. The anhydrite distribution computed from well data corresponds to general trends of V p /V s computed from surface seismic data across the same interval (Blaylock 999; Fig. 4). This is especially true in the southern portion of the mapped area. In addition, a cross-plot of V p /V s (based on multi-component 3D seismic data) versus anhydrite percent from well data shows a decrease in V p /V s with increasing anhydrite content (Fig. 5). To estimate the spatial distribution of anhydrite within reservoir zones, anhydrite indicator log curves were computed for each well with the appropriate log suite. To generate the anhydrite indicator curve, any point on the ρ maa U maa plot that fell on or above the % cut-off was flagged as an anhydritebearing interval. If the point fell below the % cut-off, the interval was considered to be relatively free of anhydrite (based on calibration to core). Using the indicator log curves, numerical models of anhydrite cement distribution were generated. Finely layered (30 cm or ft thick layers) models were built to preserve the vertical detail from the log data. Figure 6, a north south cross-section through an anhydrite cement model, illustrates one possible view of the estimated vertical and lateral distribution of anhydrite within the reservoir. The anhydrite model was built under the assumption that anhydrite-dominant intervals are somewhat laterally continuous for several hundred metres within high-frequency depositional cycles. DISCUSSION Within the southern portion of the study area, the lower V p /V s trend corresponds to higher percentages of anhydrite, similar to the observations of Arestad (995) at the Joffre Field. At the Vacuum Field, this also corresponds to a trend of higher average porosity. Through the analysis of anhydrite distribution, it was determined that anhydrite is generally concentrated in thin depositional cycles or intervals that are separated by relatively anhydrite-free cycles that exhibit high porosity and permeability. In general, it has been shown that V p /V s increases with increasing porosity (Vernik & Nur 99). Within the study area, average porosity increases to the south and southeast. Several of the reservoir flow units that are present only in the south exhibit high porosity and cause the average porosity in the south to be relatively high. In addition, the low porosity Lovington siltstone is present only in the northern portion of the study area and contributes to the relatively lower average porosity. The decrease in V p /V s to the south is most likely due to the increase in anhydrite cement to the south (Fig. 6), which overshadows the lateral change in porosity. Anhydrite commonly occurs as a pore-filling cement, and when present, increases rigidity relative to porous dolomite, increases shear wave velocity, and, therefore, decreases V p /V s values. V p /V s maps can be used to condition reservoir models to improve estimates of the lateral distribution of anhydrite cement. The model of anhydrite cement can ultimately be combined with carbonate lithofacies models and used to improve porosity and permeability predictions for the San Andres reservoir. The origin of anhydrite is most likely related to faults and fractures that acted as conduits for the movement of sulphaterich fluids during exposure of the shelf margin and after burial. Leary & Vogt (990) proposed an alternative or additional hypothesis for anhydrite origin. They suggested that the precipitation of calcium sulphate within the San Andres was a by-product of dolomitization. The release of calcium (Ca + ) during dolomitization could supersaturate pore waters with respect to calcium sulphate (CaSO 4 ) and result in precipitation

9 Anhydrite distribution in a carbonate reservoir, USA 5 N S 4300 Grayburg Dolomite Grayburg Sandstone Upper San Andres Lovington Lower San Andres 5000 V.E. ~ 3.5 > % Anhydrite < % Anhydrite 000 ft 50 m Fig. 6. North south cross-section from a 3D model that shows the distribution of anhydrite within reservoir zones. The cross-section shows the increase in anhydrite to the south. This model is conditioned to the anhydrite indicator log curves for 7 wells based on ρ maa U maa analysis. An inverse-distance weighted-average interpolation algorithm was used. of anhydrite and gypsum (Leary & Vogt 990). Although evaporites are common deposits within the supratidal environment, normal marine fossils, such as crinoids and fusulinids, in the San Andres subtidal lithofacies preclude a hypersaline environment during deposition (Leary & Vogt 990). Therefore, within subtidal and possibly intertidal lithofacies, anhydrite precipitation was most likely post-depositional. CONCLUSIONS Anhydrite cement is a significant cause of heterogeneity within the San Andres reservoir at the Vacuum Field. Zones of massive anhydrite that are several centimetres to greater than 30 cm in thickness are common. Anhydrite cement is abundant along key horizons that exhibit karst and other exposure-related features. ρ maa U maa cross-plots from well log data and V p /V s measurements from surface seismic data are useful for semiquantitative estimates of anhydrite percent. Although V p /V s values generally increase with an increase in porosity, the presence of nonporous anhydrite within reservoir zones results in a relative decrease in V p /V s values. When combined with anhydrite estimates at well locations from ρ maa U maa crossplots, these data can be used to generate 3D models of anhydrite cement. These can be combined with carbonate facies models or other petrophysical models to produce more representative reservoir models. We thank the industry sponsors of the Colorado School of Mines Reservoir Characterization Project for funding and input to this study. The consortium members include AGIP, Amoco Production Company (now bp), Anadarko Petroleum Corporation, ARCO (now bp), Chevron Petroleum Technology Company (now Chevron- Texaco), China National Petroleum Corporation, Compagnie Generale de Geophysique, Conoco Inc. (now ConocoPhillips), Dawson Geophysical Company, Exxon Production Research Company (now ExxonMobil Upstream Research Company), Gas Research Institute (now Gas Technology Institute), GeoQuest/Schlumberger/Geco, Golden Geophysical/Fairfield Industries, Grant Geophysical, Inc., Input/Output, Inc., INTEVEP, S.A., Japan National Oil Corporation, Landmark Graphics Corporation, Marathon Oil Company, Maxus Energy Corporation, Nambe Geophysical, Inc., Occidental Oil and Gas Corporation, Oyo Geospace Corporation, PanCanadian Petroleum Limited (now EnCana), Phillips Petroleum Company (now ConocoPhillips), Paradigm Geophysical (formerly CogniSeis), Shell E& P Technology Company, Discovery Bay Company (now Rock Solid Images), Silicon Graphics Corporation, Solid State Geophysical, Texaco Group, Inc. (now ChevronTexaco), Union Pacific Resources Company (now Anadarko Petroleum Corporation), UNOCAL/Sprint Energy, Western Geophysical and Veritas DGC, Inc. The study was also supported through research grants and funding from the American Association of Petroleum Geologists, Geological Society of America, Society of Professional Well Log Analysts and the Department of Geology and Geological Engineering at the Colorado School of Mines. REFERENCES Adams, S.D Sedimentology and diagenesis of the San Andres Formation, Vacuum Field, New Mexico. MS thesis. Colorado School of Mines, Golden, Colorado. Arestad, J.M An integrated multicomponent three-dimensional seismic characterization of Joffre Field, Alberta, Canada. PhD thesis. Colorado School of Mines, Golden, Colorado. Blaylock, J.J Interpretation of a baseline 4-D multicomponent seismic survey at Vacuum Field, Lea County, New Mexico. MS thesis. Colorado School of Mines, Golden, Colorado. Capello de Passalacqua, M.A Geology and rock physics of the San Andres Formation in Vacuum Field, New Mexico. MS thesis. Colorado School of Mines, Golden. Doveton, J.H Geologic log interpretation. SEPM Short Course No, 9. Society for Sedimentary Geology, 9. Hills, J.M Sedimentation, tectonism and hydrocarbon generation in Delaware Basin, west Texas and southeastern New Mexico. AAPG Bulletin, 68, Kerans, C Stratigraphic framework of the San Andres Formation, Algerita Escarpment, Guadalupe Mountains, New Mexico. In: Moussa, M.T. (ed.) The San Andres in Outcrop and Subsurface, Guidebook to the Permian Basin. SEPM Annual Field Conference, Kerans, C. & Kempter, K. 0. Hierarchical stratigraphic analysis of a carbonate platform, Permian of the Guadalupe Mountains. AAPG/Datapages Discovery Series No, 5. Leary, D.A. & Vogt, J.N Diagenesis of the San Andres Formation (Guadalupian), Central Basin Platform, Permian Basin. In: Bebout, D.G. & Harris, P.M. (eds) Geologic and Engineering Approaches in Evaluation of San Andres / Grayburg Hydrocarbon Reservoirs Permian Basin. Texas Bureau of Economic Geology, 48. Pranter, M.J Use of a petrophysical-based reservoir zonation and multicomponent seismic attributes for improved geologic modeling, Vacuum Field, New Mexico. PhD thesis. Colorado School of Mines, Golden, Colorado. Purves, W.J Reservoir description of the Mobil Oil Bridges State Leases (Upper San Andres reservoir), Vacuum Field, Lea County, New Mexico. In: Bebout, D.G. & Harris, P.M. (eds) Geologic and Engineering Approaches in Evaluation of San Andres / Grayburg Hydrocarbon Reservoirs Permian Basin. Texas Bureau of Economic Geology, 87.

10 5 M. J. Pranter et al. Sarg, J.F. & Lehmann, P.J Lower-middle Guadalupian facies and stratigraphy, San Andres/Grayburg Formations, Permian Basin, Guadalupe Mountains, New Mexico. In: Moore, G.E. & Wilde, G.L. (eds) San Andres/Grayburg Formations, Guadalupe Mountains, New Mexico and Texas. SEPM, Permian Basin Section, publication, 86-5, 8. Scuta, M.S D reservoir characterization of the Central Vacuum Unit, Lea County, New Mexico. PhD thesis. Colorado School of Mines, Golden, Colorado. Sonnenfeld, M.D. & Cross, T.A Volumetric partitioning and facies differentiation within the Permian Upper San Andres Formation of Last Chance Canyon, Guadalupe Mountains, New Mexico. In: Loucks, R.G. & Sarg, J.F. (eds) Carbonate Sequence Stratigraphy: Recent Developments and Applications. AAPG Memoir, 57, Tatham, R.H. & McCormack, M.D. 99. Multicomponent seismology in petroleum exploration. Investigations in Geophysics No, 6, Vernik, L. & Nur, A. 99. Petrophysical classification of siliciclastics for lithology and porosity prediction from seismic velocities. AAPG Bulletin, 76, Received July 0; revised typescript accepted August 03.