Sterling James Crabtree

Petrographic Image Analysis, I. Analysis of Reservoir Pore Complexes

ABSTRACT There exists a need to relate the petrology of reservoirs (pore geometry, surface areas of mineral phases and pores) to geophysical and petrophysical data. The end result is improved assessment of reservoir quality as well as better interpretation of well logs and seismic data. Petrographic Image Analysis (PIA) was developed from the beginning to interface with petrophysical/geophysical data. PIA relies upon computer-based image analysis using pattern recognition/classification programs, and so information can be obtained very rapidly--the rate simply tied to sophistication of the computer in use. PIA consists of a critical mix of hardware and software which perform four separate functions: 1) image acquisition; 2) image digitization; 3) image segmentation; and 4) image analysis. A sp cial effort has been made to characterize the geometry of the pore complex. Separate spectra related to pore size and pore roughness are generated from each image. In addition, surface area per unit volume of pore can be estimated. Pore spectra can be decomposed and classified using pattern recognition/ classification algorithms or used directly to estimate physical parameters.

Petrography and reservoir physics; I, Objective classification of reservoir porosity

Porosity observed in thin section can be objectively classified using a combination of digital acquisition procedures and pattern recognition algorithms. Pore types are derived from the frequency distributions of sizes and shapes of patches of porosity exposed in thin section. Each pore type is represented by a characteristic distribution of sizes and shapes found in thin section. Most sandstone reservoirs contain fewer than six pore types. Much of the intersample variability in reservoir physics is associated with changes in pore type abundance. The advantages of this approach to porosity classification are (1) the criteria for classification are objectively defined, (2) classification procedure is rapid, accurate, and precise, (3) pore types are understood easily in ter s of conventional genetic classification schemes, and (4) pore type data are related strongly to petrophysical properties.

Petrography and reservoir physics; II, Relating thin section porosity to capillary pressure, the association between pore types and throat size

Porosity in reservoir rocks is configured into a few types of pores whose size and shape are controlled by depositional fabric and postdepositional processes. The size, shape, and abundance of each pore type can be objectively determined from thin section using image analysis and pattern recognition procedures. Each pore type tends to be associated with a limited range of throat sizes. The association between pore type and throat size can be determined using regression procedures linking pore type data obtained from thin section with capillary pressure data. To do so, a set of samples is required wherein the association between pore type and throat size is fixed, but where pore type proportions vary between samples. This condition is met by a sample suite representing res rvoir facies from a single core or, in many cases, from a single field. The relationship between pore type and throat size is an effective means to relate reservoirs in terms of the efficiency of the porous microstructure to multiphase flow. Parameters derived from the relationship can be used to construct accurate physical models that subdivide physical response in terms of the contributions of each pore type.

Petrography and reservoir physics; III, Physical models for permeability and formation factor

Permeability and formation factor are physical properties of porous rocks useful for assessing reservoirs. Neither property varies consistently as porosity varies. The relationship of both properties to porosity is complex, being sensitive to the structure of the porous microstructure, i.e., the sizes of pore throats, the numbers and sizes of pores, and the relationships between pores and throats. Physical models to account for these factors require parameters that describe physically relevant properties of the microstructure. A partial characterization of the relationship between pores and throats is embodied in the relationship between pore type and throat size. This relationship is derived by combining data obtained from thin sections, from which pore types are derived via image analysis, and mercury injection porosimetry, which quantifies throat size information. Parameters derived from such a combination are sufficient to construct simple physical models for permeability and electrical conductivity (inverse formation factor). These models assume a porous medium that has large numbers of flow paths parallel to the potential gradient, such that flow has little tortuosity (i.e., flow parallel to bedding). The contributions of each pore type to permeability and electrical conductivity are computed. Calculated values are close to measurement values. A constant of proportionality is the same for all samples from a reservoir, but can vary between reservoirs, is required, and must have values ranging (for sandstones) from about 2.5 to 3.5 for permeability an 5.0 to 7.0 for conductivity. These values are consistent for an efficiently packed fabric. One result of such modeling is a physical model of Archies cementation exponent m as the ratio of the logarithms of the cross sectional throat area to pore area (per unit area).

Relations between pores, throats and permeability: A petrographic/physical analysis of some carbonate grainstones and packstones

Some physical properties of carbonate rocks are sensitive to mineralogy, others are functions of porosity, and still others are related to the distribution and configuration of porosity. Properties associated with transmissivity (permeability, electrical conductivity) fall in the latter category. Porosity is distributed in rocks as a three-dimensionally interconnected network with large voids (pores) connected at small constrictions (pore throats). Thin sections carry only pore information. A new methodology termed Petrographic Image Analysis consists of a systematic procedure which determines pore/throat relationships. These data are sufficient to construct precise physical models for calculating permeability and formation factor.Petrographic Image Analysis involves: (1) acquisition of digital images of porosity from thin section, (2) measurement of the size and geometric complexity of that porosity, and (3) objective determination of pore types using a pattern-recognition/pore-classification procedure. Once classified, the size, relative proportions and number of pores per unit volume of each pore type can be calculated. Regression procedures provide a first-order model between mercury saturation as a function of pressure and pore type proportion, and thus between pore type and throat size.Analysis of the deviations from regression refine the throat-size assignments on a sample-by-sample basis. From this, permeability and electrical conductivity can be calculated based on Darcy’s and Ohm’s Laws respectively. The models agree with measured properties with high precision. Both models are additive in that the contribution of each pore type is calculated separately. Therefore, the relative mobility of phases occupying different portions of the pore system can be assessed.These principles are illustrated by analysis of the pore/throat system of a suite of carbonate pelletal packstones and peloidal grainstones from a Middle-East limestone reservoir, and the derivation of statistical and physical models estimating the permeability of these rocks. Derivation of formation factor (to be discussed in a later paper) can be done by a similar method due to the close relationship between Darcy’s Law and Ohm’s Law.

Evaluation of strategies for segmentation of blue-dyed pores in thin sections of reservoir rocks

Abstract Petrographic image analysis concerns segmentation and analysis of a rock fabric in the format of a 30 μ thick “thin section.” A major objective of such analysis is to segment pores (voids) from the rock material inasmuch as pore size and geometry control fluid flow. Conventionally, prior to sectioning, the rock is impregnated with blue dyed epoxy. Since few, if any naturally occurring components of reservoir rocks are blue, a segmentation process based upon color is appropriate. Segmentation in this case must accomplish both the correct identification of a pore and the precise definition of its edge. Most of the conventional segmentation techniques, which employ thresholding on histograms, failed in this because of problems associated with high light intensities required for petrographic microscopy and because of gradational boundaries caused by shelving effects. Successful segmentation was accomplished by modeling the digital filters on the human perception of the color of pore pixels. It has been shown that both of the filters developed clearly distinguish pore from nonpore and locate edges with high precision. However, a histogram of hue is still employed to identify the nature of pore boundaries (vertical, shelving, etc.).

A fast and accurate erosion-dilation method suitable for microcomputers

Abstract The concept of image erosion is well known in image processing where it is used as both a smoothing technique [1, 2] and a shape classifier [3]. We, as others, in investigating image quantifications, use image erosion and dilation as an important element in the analysis [4–8]. The purpose of this paper is twofold: (1) to discuss a new improved version of the erosion-dilation method which approximates “round” circles rather than the “diamond” circles or “square” circles which result from the use of the 4-connected or 8-connected erosion-dilation methods, respectivity. (2) Because of the amount of data in a digital image and the amount of time it takes to execute the erosion-dilation process, the process is generally implemented on large mainframes or minicomputers. We discuss a very fast erosion-dilation method which we have implemented using MS-DOS on a PC using only 640K of memory and an EGA board.