Carbonate Reservoir Characterization

Technology Today Series articles are general, descriptive representations that summarize the state of the art in an area of technology by describing recent developments for readers who are not specialists in the topics discussed. Written by individuals recognized as experts in the area, these articles provide key references to more definitive work and present specific details only to illustrate the technology. Purpose: to inform the general readership of recent advances in various areas of petroleum engineering. During the past 15 years, methods and technology for predicting the performance of carbonate reservoirs have improved dramatically. This advance occurred in response to the realization that more than half of the oil that could be swept by waterflooding is not contacted and remains in the reservoir. Attempts to simulate this performance and locate the remaining oil by use of simple reservoir models and newly developed flow-simulation computer programs have failed mainly because of the extreme heterogeneity that characterizes carbonate reservoirs. It is, in fact, the extreme geologic and petrophysical heterogeneity typical of carbonate reservoirs that distinguishes carbonate from siliciclastic reservoirs. Research programs focused on understanding the nature of the heterogeneity and developing methods to characterize carbonate reservoirs, together with improvements in computer capability and simulation programs, have led to more reliable predictions of reservoir performance and to methods for locating volumes of unswept oil in reservoirs under waterflood. Reservoir characterization encompasses the understanding and methods used to characterize reservoir heterogeneity. It can be defined as the construction of realistic 3D images of petrophysical properties used to predict reservoir performance, and it is a multidisciplinary, integrated task involving expertise in reservoir geology, geophysics, petrophysics, well logging, geostatistics, and reservoir engineering. Three-dimensional images are obtained from geological models constructed with core, wireline-log, and geophysical data. Petrophysical properties, obtained from core, wireline-log, and production data, are distributed within the geological model by linking petrophysical properties to geologic fabrics and by use of advanced geostatistical and geophysical methods. Finally, the model is put into a numerical simulator for testing and predicting future performance. The extreme petrophysical heterogeneity found in carbonate reservoirs is clearly demonstrated by the wide variability observed in porosity-permeability crossplots of core-analysis data. Research has shown that basic rock fabrics dominate control of petrophysical heterogeneity; within a rock-fabric facies, porosity and permeability have little spatial correlation and are widely variable at the scale of inches and feet. Permeability, in particular, can vary by a factor of 10 or more at the small scale and is nearly randomly distributed (Fig. 1).1,2 This result suggests that much of the variability observed in core-analysis data is spatial noise and can be averaged within rock-fabric facies for the purposes of constructing a reservoir model. Only rock fabrics, not pore-throat size, permeability/porosity ratio, or flow-zone indicators, have vertical and lateral continuity. Therefore, rock-fabric facies are the basic elements for characterizing a carbonate reservoir.3 Rock fabrics are geologic descriptors that characterize pore size according to particle size and sorting, interparticle porosity, and various types of vuggy porosity. The main limestone rock fabrics are grainstone, grain-dominated packstone, and mud-dominated fabrics. Dolostone rock fabrics are similar but require a description of dolomite crystal size in mud-dominated dolostones as an added control on pore size. These basic fabrics are modified by the amount of fabric-selective vuggy pore space,4 and many rock fabrics can be linked directly to depositional facies. However, some rock fabrics cannot be linked to depositional facies because extensive modifications to the fabric have occurred since deposition through a geologic process known as diagenesis. Constructing models of reservoirs comprising complex fabrics, such as fracture and karst fabrics, is difficult and is the subject of current research. Each rock fabric has a specific porosity-permeability transform, and the vertical stacking of rock-fabric facies, together with interparticle porosity, provides the basis for estimating permeability in uncored wells. The wireline-log problem is determining interparticle porosity, as well as total porosity and rock fabric, for input into a general porosity-permeability transform. Porosity, acoustic velocity, resistivity, and saturation logs are used for this task, and calibrating wireline-log responses to core descriptions of rock fabrics is key to developing useful algorithms.5 Because rock-fabric and petrophysical data obtained from cores and wireline logs are one-dimensional, a geological framework is required to distribute the data in 3D space. In the past, geological models were constructed by identifying depositional facies from core data, then distributing those facies by use of depositional models on the basis of modern carbonate depositional processes. However, reservoirs described in this manner do not contain sufficient detail to capture basic reservoir heterogeneities, and facies correlations from well to well often are highly uncertain. Development of sequence stratigraphic methods greatly enhanced the accuracy of well-to-well correlations and provided the means of capturing basic scales of reservoir heterogeneity.