Measuring Permeability Trends and Correlation Structures in Heterogeneous Carbonates, San Andres Formation, New Mexico

GOGGIN, D. J., C. W. GRANT, and P. M. HARRIS, Chevron Oil Field Research Company, La Habra, California

Detailed permeability measurements were obtained using a mechanical field permeameter (MFP) on a 210-ft core of the upper San Andres Formation. Analyses of these measurements were used to arrive at an optimal combination of sampling method, number of measurements, and measurement spacing.

Parasequences are considered to be a fundamental geologic unit in the shelf carbonate depositional setting. In this case, the parasequences typically are composed of a thin mudstone base, overlain by bar-flank, burrowed wackestones-packstones, and capped by thick barcrest planar- to cross-bedded oolitic packstones-grainstones. The presence and physical scale of vugs, burrows, cemented zones, and moldic porosity makes the task of obtaining representative petrophysical measurements, such as permeability, difficult at best in these facies.

To address this complexity, two basic sampling strategies were applied to the core for MFP-scale measurements. The first approach used a vertical transect of closely spaced (1.2 in) measurement points along the length of the core. The second approach used clusters of five points (pads) located at less frequent 6-in intervals.

Data show the basal mudstones to be tight, with a generally increasing permeability trend toward the top of each parasequence. Within each facies, permeability is highly variable. These trends were obtained by applying pad or vertical boxcar filters to the data using

both arithmetic average and median estimators. The median filter showed the least amount of bias, sensitivity to extremes in the data, over a wide range of permeability values and is considered to be the best estimator for MFP data in the most heterogeneous core segments.

Vertical correlation structures were estimated on the unfiltered data sets using variogram analysis. The loyal, small-scale permeability variations yield a nearly "pure noise" model with the exception of strong correlation features at 18- and 48-ft data spacings. These correlation features are attributed to facies and diagenetic pattern similarities within and between successive parasequences.

In a comparison of sampling schemes, closely spaced transects and pads yield similar local estimates of the mean; however, the closely spaced transect yields more correlation information because of its superior vertical coverage for a fixed overall number of measurements. Therefore, we recommend using closely spaced, vertical transects of measurements over pad sampling schemes.

AAPG Search and Discovery Article #91004 © 1991 AAPG Annual Convention Dallas, Texas, April 7-10, 1991 (2009)