Bureau of Economic Geology, The University of Texas at Austin (www.beg.utexas.edu).
2004 West Texas Geological Society Fall Symposium, Midland, Texas, October 2729.
Permeability Estimation Using Porosity logs and Rock Fabric Stratigraphy: An Example from the Sacroc (Pennsylvanian)Field, Scurry County, Texas
F. Jerry Lucia and Charles Kerans
This equation requires values for interparticle porosity and rock fabric number. Rock fabric number is an expansion of petrophysical classes 1, 2, and 3 described by Lucia (1995). This information can be obtained from wireline logs if the proper logs are available and if the reservoir has not been water flooded. Many West Texas oil fields have been water flooded and most of the well logs are simple gamma ray – neutron logs. Therefore, the most common wireline log measurement available is total porosity and interparticle porosity and rock fabric information can not be obtained directly.
This is the case in the Sacroc field were few logs other than porosity logs are available. An approach has been developed that integrates stratigraphic information, thin section descriptions, core analysis data, and porosity logs to estimate permeability. In this approach total porosity is used instead of interparticle porosity and an apparent rock fabric number (ARFN) is used instead of rock fabric number of petrophysical class. The ARFN is calculated from core porosity and permeability values and averaged over a specific reservoir volume defined vertically by stratigraphy and laterally by facies mapping. Within the reservoir volume permeability is estimated from the global transform using porosity from wireline logs and stratigraphically defined ARFNs.
Permeability = 38520 × phi (5.0923) (3)
The mud-dominated fabrics are mostly dense. Mud-dominated fabrics are petrophysical class 3 and, if porous, would require a second transform to characterize their petrophysical properties. However, in this reservoir they are basically impermeable because they typically have very low porosity, and permeability values calculated from the above transform are below reservoir quality. Therefore, a class 3 transform is not required.
A comparison between core and log calculated porosity and permeability values from well Sacroc 27-7 shows a good match to permeability where core porosity and log porosity match. This emphasizes the need for good porosity logs. In this study we used the porosity logs given to us by the operator and did not attempt to improve them.
There are 14 core analyses from the Canyon, nine of which have core available for inspection. Log calculated permeability was compared to core permeability in all these wells. This comparison showed an interval in the southern portion of the Northern Platform in which the calculated permeability was consistently too high. To correct this error a second transform, presented below, was developed using an ARFN of 2.7, which is within the class 3 field. This transform was used in the southern-most portion of the Northern Platform. The most likely explanation for this shift in ARFN is the presence of a porous mud-dominated fabric in this volume. Unfortunately, no core was available to test this interpretation.
Permeability = 38520 × phi (5.0923) (4)
Canyon sequence 3 is characterized by moldic ooid grainstone, grain-dominated packstone, and mud-dominated fabrics. The average ARFN calculated from core analysis data is 2.5, which is the class 2-3 boundary. The resulting transform is
Permeability = 97628 × phi (5.3696) (5)
The reason these varied rock fabrics group along the class 2-3 boundary is that petrophysical classes are based on interparticle porosity and the ARFN uses total porosity. Whereas grainstone is a class 1 fabric, the presence of pore space within the grains reduces the permeability from what would be expected if all the porosity were interparticle. As a result, grainstones with separate-vug porosity always plot to the right of the class 1 field, in this case near the class 2-3 boundary. Most of the mud-dominated fabrics are dense and the few that are porous appear to have their permeability improved by microfractures. Therefore, they tend to move to the left of the class 3 field and plot near the class 2-3 boundary.
A comparison between core and log calculated porosity and permeability values from well Sacroc 27-7 shows a good match to permeability in some intervals and a significant difference in others. In one 20 ft interval the calculated permeability is a factor of 10 higher than the measured permeability. This is an ooid grainstone interval and the low neasured permeability may result from very high moldic porosity values or micritization. In a 10 ft interval the calculated permeability is lower than measured permeability suggesting a non-moldic grainstone. These examples illustrate the uncertainty inherent in estimating permeability in a moldic grainstone unit were the proper logs are not available to estimate separate-vug porosity.
The Cisco is divided into two units, early and late Cisco. The early Cisco is characterized by fusulinid/crinoidal/peloid grain-dominated packstones, grainstones, and wackestones. Most of the measured data plots in the class 2 field and the average ARFN is 1.9. The transform used for this unit is
Permeability = (2.690 ×106) × phi (6.3584) (6)
A comparison between core and log calculated porosity and permeability values from well Sacroc 27-7 shows a good match to permeability. The late Cisco has a wide variety of fabrics making permeability estimation from porosity logs very difficult. The wackestones are generally low porosity with little permeability. However, there are low porosity microbial boundstones, Phylloid algal grain-dominated packstones, and debris flows that are permeable. Also, there are intervals of touching-vug porosity which complicate permeability estimation.
The porous intervals tend to be composed of grain-dominated fabrics (grainstones and grain-dominated packstones). A cross plot of core data shows that they plot in the class 1 and 2 fields as expected. However, the proper logs and reservoir conditions are not present to separate grainstones from grain-dominated packstones. Therefore, all the data was averaged resulting in an ARFN of 1.7. The resulting transform for the late Cisco porous intervals is
Permeability = (1.031 ×107) × phi (6.7592) (7)
A 100 ft interval of leached crinoidal grain-dominated packstone and wackestone is present in well Sacroc 17-5. This leaching is interpreted to be the result of karsting. The cross plot of core data shows low porosity and scattered low to high permeability values. A reduced major axis RMA transform was fitted to this data and used to estimate permeability in only the highest portion of the Cisco.
Permeability = (2.1625 × 106) ×phi (3.8844) (8)
A comparison between
core and log calculated porosity and permeability values from well Sacroc
17-5 shows that the calculated values match measured values reasonably
well in the porous intervals but tend to underestimate permeability in
the lower porosity intervals. This most likely due to the fact that low
porosity-high permeability facies can not be identified using wireline
There is a wide variety of rock fabrics ranging from grainstone to wackestone with a wide variety of diagenetic effects including grain dissolution, compaction fracturing, micritization, and recrystallization. Despite the high degree of variability, the ARFN’s vary from 2.7 to 1.7 with most values lying within the petrophysical class 2 field. It appears that grain dissolution, micritization, and recrystallization have caused the ooid grainstones to move from the class 1 field into the class 2 field and sometimes the class 3 field. Most of the class 3 mud-dominated fabrics are dense and tight an do not require a specific transform for quantification. The few porous class 3 samples contain microfractures that enhance permeability and cause the samples to move from the class 3 field into the class 2 and sometimes class 1 field.
This approach results in a number of intervals in which the permeability is either underestimated or overestimated and a better fit could be made if the logs were available to make a more accurate determination of rock fabric. However, this approach provides better permeability values than would be obtained by using a single class 2 transform. One reason for underestimation of permeability in the late Cisco is the lack of a good geologic model for this complicated depositional system and our inability to identify and characterize touching-vug pore systems.
Jennings, J.W., Jr., and Lucia,
F.J., 2001, “Predicting Permeability from Well Logs in Carbonates
with a Link to Geology for Interwell Permeability Mapping,” paper
SPE 71336 presented at the 2001 SPE Annual Technical Conference and Exhibition,
New Orleans, Sep. 30-Oct. 3.