COMBINING DATA HELPS PINPOINT INFILL DRILLING TARGETS IN TEXAS FIELD

John R. Frank Jr., Edward Van Reet, William D. Jackson
Chevron U.S.A. Production Co.
Midland, Tex.

Kingdom Abo field has a complex reservoir geometry, with significant vertical and areal variations in reservoir rock at interwell spacing.

Complexity of this reservoir suggests that an infill drilling program may recover hydrocarbons not recoverable at the present well spacing.

The initial success of the Kingdom Abo infill well project is directly attributed to the synergistic combination of geostatistics, 3-D seismic data, and well log data.

This paper discusses the geostatistical spatial characterization of this field, the limits this method has when used alone, and the additional spatial resolution realized with 3-D seismic.

THE FIELD

Kingdom Abo field is located in northwest Terry County, Tex., on the Northwest shelf of the Permian basin (Fig. 1). The field produces oil from the Abo formation at an average depth of 7,550 ft.

Kingdom Abo field was discovered by Gulf Oil in 1975 with the drilling of the Mallet Land & Cattle No. 46. In the next 11/2 years, 60 producing wells were drilled on 40 acre spacing.

Starting in 1978, 50 water injection wells were drilled on 20 acre, 5 spot patterns to provide support to the producing wells. The study area encompasses 2,240 acres (Fig. 2).

Kingdom Abo Field contains a weak water drive and is undersaturated (Table 1). The field had a rapid decline in production prior to installation of the waterflood in 1978 (Fig. 3).

In 1991 production increased from a workover program, which added perforations to provide increased reservoir sweep efficiency between the injection wells and the producing wells.

Most of the 60 producing wells had a dual induction-laterolog and a compensated neutron-formation density open hole log suite run. The majority of the 50 injection wells had a cased hole gamma ray-neutron log run during completion procedures.

These wells were logged between 1975 and 1978. Fourteen wells were cored through the productive interval of the Abo formation.

All porosity logs were digitized and loaded on a VAX VMS workstation. Cross plots of core porosity vs. log porosity were made, per logging company, and each porosity trace was adjusted to core porosity.

GEOLOGY

Kingdom Abo is elongate and sinuous, approximately'/2-1 mile wide and 10 miles in length. Abo formation massive dolostone is a time-transgressive lithofacies, with up to 400 feet of closure mapped (Fig. 2).

Dip is 3-6 on the flanks of the massive dolomite.

The Abo formation is adjacent to the Leonardian shelf margin. The Abo aggraded as the sea level slowly rose. At times, the sea inundated the shelf margin, depositing mudstone. When the carbonate production kept pace with rising sea level, grainstones and packstones were deposited in wave base.

The Abo has been completely dolomitized. It consists of dolomitized mudstones, wackestones, packstones, and grainstones. Multiple episodes of anhydrite dissolution and cementation have been documented.

Examination of the 14 cored wells reveals that the productive portion of the Abo formation can be split into four intervals. These intervals are described below in ascending order (Fig. 4).

• D marker to C marker. The lower Abo formation contains several separate mudstone through grainstone shallowing-upward parasequences. These formed in localized areas where carbonate production was able to keep pace with sea level rise and fall. The D marker is picked at the present-day oil-water transition zone and is not a stratigraphic marker.

• C marker to B marker. This interval contains a large volume of anhydrite, of probable sabkha origin, that has been mobilized numerous times throughout its diagenetic history, eradicating the original depositional texture. Preserved porosity tends to be minor and localized.

• B marker to Abo marker. This interval is composed of dolomitized grainstones, deposited in a high energy shallow marine environment. This interval is the most continuous throughout the field. Its reservoir rock contains the best reservoir quality.

• Abo marker to A marker. The uppermost interval has limited areal extent and the least amount of reservoir rock. The development of this facies above the Abo marker is characterized by mixed carbonates and siliciclastics. A series of grainstones, interbedded with siliciclastics, was deposited on the protected northwest flank of the structure.

GEOSTATISTICAL STUDY

Geostatistics is a tool for describing the spatial correlation of a variable, such as porosity or permeability.

It is assumed that the spatial distribution of reservoir properties can be described as a realization of a random process; therefore, it can be studied from a probabilistic viewpoint.

The computer program used in this study is Geolith, a commercially available geostatistical program developed by Chevron Petroleum Technology Co. The steps in developing the cross sections in this paper are briefly reviewed below:

• Assumption of stationarity. The assumption of stationarity must be made before a geostatistical study can begin. Since the interval under study is a dolostone with common properties, this assumption is deemed to be appropriate. 2

• Data preparation. All normalized logs had marker beds picked on each log trace. These stratigraphic markers form the boundaries of a system of cells for interpolating porosity.

• Experimental semi-variogram analysis and theoretical modeling. Semi-variograms are statistical tools that measure the spatial correlation of a variable.

  The term "experimental" implies that the points on the semi-variogram are computed from the actual data. After an experimental semi-variogram is calculated, theoretical semi-variogram points are fit to the data, which mimics the character of the shape that the experimental points form. This step is required, since the kriging estimation procedure requires a data point at every possible lag.'

  Two experimental semi-variograms were modeled in this study. One captured the structural component on the Abo marker, and another was used to capture the cross sectional porosity variation.

• Kriging. Kriging is a linear point estimation method, which uses the spatial correlation structure of a variable (from the theoretical sen-d-variogram) to calculate values at each grid node. 2

  Geolith 3-D cross sections were constructed between each well in both dip and strike directions. The 3-D cross section option utilizes data from other wells out of the plane of the cross section, so a fuller representation of the geology is depicted.

  The distance between cross sections is approximately 1,000 ft, or about the standard well spacing between wells on a regular 20 acre, 5 spot waterflood pattern. A high degree of reservoir discontinuity throughout the length of the study area is apparent (Fig. 5).

  Visual examinations of this and other cross sections reveal that porosity pinchouts are predicted between wells in many portions of the field.

LIMITATIONS

There are limitations in using well logs to construct geostatistical cross sections between wells. For example: 0 The dominant limitation is the spatial data sampling. The spatial variance cannot be reconstructed at a scale smaller than the sample interval.

There is certainly interwell variability between the current well spacing ( 1,000 ft) that the semi-variogram does not capture.

• The geology at Kingdom Abo is the end result of a multitude of complex deterministic geologic processes. Since these processes are numerous and complex, a probabilistic approach is used to model the spatial relationships.

  The underlying reality may not be accurately represented, since geostatistics assumes geology is a result of random processes.

• The random variables are calculated from various types of porosity logs.

  Although these logs are of recent vintage and were normalized, it is possible that they do not accurately depict actual distribution of reservoir porosity.

3-D SEISMIC PROGRAM

To improve estimates of interwell reservoir geometry in the Abo reservoir, a 3-D seismic survey was acquired in late 1991 and early 1992 over Kingdom Abo field.

Recent geologic and engineering work indicates that the reservoir is far more complex than originally thought and that efficient depletion of the reservoir calls for a detailed understanding of its architecture and continuity.

The recent seismic acquisition will aid in the evaluation of reservoir volumetric calculations, porosity zone continuity, infill well locations, and flood pattern realignments.

Geophysical modeling completed prior to acquisition indicated that the 3-D survey would be able to image details in the Abo reservoir. The vertical resolution of the 3-D data is not, nor was it thought to be, near that of the geologic models.

For purposes of geophysical interpretation, the Abo reservoir is divided into three intervals. The upper porosity interval correlates to the previously described A to B zones. The lower porosity interval includes the C to D zone, and separating the two is the tight zone B to C. These intervals are illustrated on a theogram constructed from well logs run on the Mallet Land & Cattle No. 51 (Fig. 6).

Two key points related to successful interpretation of seismic data in the Abo reservoir are:

• The Abo reservoir is a clean dolostone. The absence of shales, and thus the related ineffective porosity zones, simplifies the interpretation. In simple terms, the changes observed in the seismic record reflect changes in porosity within the reservoir. These changes are related to the amount of porosity and the distribution of the porosity zones.

• The abundance and distribution of "modern" open hole logs provide excellent control points to be used in interpolating the interwell data.

INTEGRATION

Work still remains in order to fully integrate the 3-D seismic, geologic, and engineering data sets.

Currently, analysis of the seismic attributes is under way to determine which combination of attributes best characterizes the reservoir.

Even as this work is in progress it is apparent that the amplitude information correlates to porosity information. A well to well model using sonic logs was constructed. A geostatistical cross section was also constructed through the same wells.

Close correlation of the amplitude data and the geostatistical model is apparent (Figs. 7 and 8). An actual user tract from the seismic data, which runs through Mallet Land & Cattle Nos. 108 and 109 locations is shown in Fig. 9.

The corresponding geostatistical cross section (constructed without the well logs from 108 and 109) is shown in Fig. 10. In this example, the geostatistical model and the 3-D data do not agree on the interwell reservoir picture.

At the time this work was under way, the Mallet Land & Cattle No. 108 location was selected as an infill producer location. The proposed location was equidistant from four offset wells.

The selection of the infill location followed traditional geologic and engineering principles based on reservoir volume, continuity, drainage, and flood front predictions.

The importance of an accurate reservoir model to these types of evaluations is obvious. The 3-D data raised a warning that the reservoir model is not correct. The seismic data indicate a decrease in reservoir quality and continuity at the Mallet 108 location.

An amplitude map for the upper and lower porosity intervals reveals the areal size of the amplitude anomaly (Figure 11). Re-evaluation of the location confirmed the need for tighter well spacing; however, to improve the odds of the upper interval porosity being present, the Mallet 108 location was moved out of the low porosity anomaly.

RESULTS

A Geolith cube was built in the study area, and volumetrics from each producer-pattern were compared to the ultimate recovery as calculated from decline curve analysis.

In those areas where greater remaining volumetric reserves existed relative to remaining decline reserves, infill locations were picked. Two of these have been drilled to date:

• The Mallet Land & Cattle No. 108. This well was drilled during July-August 1992. The Abo section exposed in the Mallet 108 well is essentially as expected based on the 3-D data.

  A revised geostatistical cross section using the Mallet 108 log data is shown in Fig. 12. Clearly in this case, the 3-D seismic data improved the interwell reservoir picture and impacted the 108 location. This well penetrated 286 ft of net pay and had an initial potential test of 86 b/d of oil without fracture stimulation.

• Mallet Land & Cattle No. 109. Another producer infill location was apparent to the east of Mallet 108, where porosity pinches out between the location and the offsetting water injector (Fig. 8).

Further review of the volumetric reserves in place vs. the ultimate recovery, as determined from decline curve analysis, indicates that the current well spacing of one producer well/40 acres cannot effectively drain all of the recoverable oil. The well was drilled equidistant between Mallet Land & Cattle Nos. 51 and 53, since the 3-D survey indicated that this would be a good location. Mallet Land & Cattle No. 109 was drilled during August 1992. This well penetrated 218 ft of net pay and had an initial potential test of 90 b/d of oil without fracture stimulation.

FUTURE WORK

Geostatistics is a powerful tool for reservoir characterization, since it utilizes all well data in a manner that adheres to a model based on statistical and user-defined spatial correlations.

When coupled with 3-D seismic, the end product is an interpretation of the reservoir that can be used to pinpoint additional development drilling locations.

The ultimate goal is to statistically merge the well data (hard data) with the seismic data (soft data). In such a process (called cokriging), the well data would be held inviolate over the seismic data at the well locations, and the seismic data would be used to influence the interwell spatial interpolations.

ACKNOWLEDGMENTS

The authors thank Chevron U.S.A. Production Co. for permission to publish this paper. S. L. Russell provided the initial reservoir characterization work. R. L. Genter, R. F. Lindsay, L. M. Roe, C. M. Keefer, J. D. Gillespie, and D. J. Goggin of Chevron provided constructive input.

REFERENCES

1. Araktingi, U. G. , Hewett, T. A., and Tran, T. T., "GEOLITH: An Interactive Geostatistical Modeling Program," SPE 24432, presented at the SPE Computer Conference, Houston, July 19-22, 1992.

2. Isaaks, E. H., and Srivastava, R. M., "An Introduction to Applied Geostatistics," Oxford University Press, Chapts. 12 and 16, page 349.

3. Ward, R. F., Kendall, C. G. St. C., and Harris, P.M., "Late Permian (Guadalupian) facies and their association with hydrocarbons-the Permian Basin, West Texas, and New Mexico," American Association of Petroleum Geologists Bull., Vol. 70, 1986, pp. 239-262.

Copyright 1993 Oil & Gas Journal. All Rights Reserved.