A Review of Permeability-Prediction Methods for Carbonate Reservoirs Using Well-Log Data
- T. Babadagli (U. of Alberta) | S. Al-Salmi (Halliburton)
- Document ID
- Society of Petroleum Engineers
- SPE Reservoir Evaluation & Engineering
- Publication Date
- April 2004
- Document Type
- Journal Paper
- 75 - 88
- 2004. Society of Petroleum Engineers
- 1.6.9 Coring, Fishing, 4.1.2 Separation and Treating, 5.6.2 Core Analysis, 4.3.4 Scale, 4.1.5 Processing Equipment, 5.6.1 Open hole/cased hole log analysis, 5.5.2 Core Analysis, 5.1 Reservoir Characterisation, 2.4.3 Sand/Solids Control, 1.2.3 Rock properties, 2.2.2 Perforating, 5.8.7 Carbonate Reservoir
- 4 in the last 30 days
- 1,978 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 5.00|
|SPE Non-Member Price:||USD 35.00|
The prediction of permeability in heterogeneous carbonates from well-log data represents a difficult and complex problem. Generally, a simple correlation between permeability and porosity cannot be developed, and other well-log parameters need to be embedded into the correlation. The first part of this paper covers an extensive review of the existing correlations in the literature. The use of porosity and other petrophysical properties of rock in permeability prediction is discussed for carbonaceous rocks. This discussion also covers the usefulness of a wide variety of correlations developed using pore-scale (Kozeny-Carman, percolation, and fractal models) to field-scale models (well logs).
In the second part of the paper, a case study is presented. The data are obtained from a complex carbonate field in Oman. Conventional and nonconventional (mainly nuclear magnetic resonance, or NMR) well-log data are evaluated to seek the parameters reflecting a good correlation with permeability. After testing each independent variable against core permeability, the variables yielding the highest correlation coefficient (CC) are included in multiple regression analysis. Data collected from seven wells are used to obtain the permeability correlations for the whole field and for four geological units separately. The test of the correlations is achieved through the comparison of the estimated permeability values to core permeability. Finally, the correlations are compared with the core permeability of the eighth well (data from this well are not included in the development of the correlation) for validation. The correlations are obtained for the four geological units. Two of these units responded well to conventional well-log data; the other two units yielded reasonable correlations only with NMR log data.
Undoubtedly, mapping the permeability field of subsurface reservoirs is one of the most crucial parts of model preparation for performance-estimation studies. Permeability can be obtained truly in laboratory (core analysis) or in reservoir (formation testers and pressure-transient analyses) settings. Pressure-transient analyses provide a single value for the permeability but account for its anisotropic nature. Core analysis and in-situ permeability measurements using wellbore devices (formation testers) rely on the pressure/rate relationship in the estimation of permeability. If these measurements are not available, permeability is estimated indirectly using rock properties acquired through well-log measurement. Because of the abundance of reservoir data from well-log measurements obtainable for every foot or so, correlating these rock properties to permeability has received a great deal of attention.
The models used for the estimation of permeability from rock properties can be categorized as (1) pore level and (2) core/field level. Both techniques relate the permeability primarily to porosity. Starting with the pioneering works by Kozeny1 (K) and Carman2 (C), many different correlations were proposed between porosity and permeability. These equations were found to be suitably applicable for synthetic porous media or unconsolidated sands. Later, modified forms of the K-C equation were proposed for more complex structures.3-5 Permeability also has been correlated to grain and pore characteristics (mainly to the size)6-10 and irreducible water saturation.11,12 Nelson13 and Ahmed et al.14 provided a critical review of these techniques. The NMR technology was used for a better estimation of pore space characteristics to be used in the permeability correlations.15,16
The main issue appears to be the generalization (or nonuniqueness) of the equations proposed. An equation fitting well to the core permeability of a reservoir rock may not be applicable to another one even if it possesses geologically similar properties. Therefore, more parameters are needed to obtain nonunique correlations. These parameters are typically the electrical, radioactive, and sonic properties of the rock obtained from well logs. They were used in permeability correlations.17-21 In these attempts, multivariable regression analysis (MRA) was applied commonly.18-22 As an alternative to the MRA, artificial neural-network technique was proposed and applied successfully.22-24
Relatively newer approaches based on the percolation,25-27 fractal,28-31 and multifractal32 theories were also proposed and tested. Geuguen and Dienes33 provided an extensive review of percolation and other statistical models used for permeability determination.
The first part of this paper reviews these methods. Then, the applicability of the methods to carbonates is discussed. In the second part, a field case is taken, and the MRA is applied to generate permeability correlations for the different units of a reservoir.
Since Kozeny's pioneering work,1 a significant amount of studies regarding the permeability estimation from available core/well data has been reported. All these studies use static information about rock properties, which are readily available in today's technology, whereas permeability is a dynamic parameter that requires an applied pressure throughout the system with respect to flowing rate. The availability of dynamic information throughout the well might be limited because of cost or time restrictions. These restrictions entail the use of easily available static reservoir information to obtain a correlation for permeability. The main advantage of this is the frequency and continuity of well-log data as opposed to core data.
In this section, the techniques proposed so far in the literature are categorized and evaluated. A critical review is performed in terms of their applicability to heterogeneous and complex carbonate systems, along with an exercise. Permeability prediction from static data can be grouped under two categories: (1) pore- (micro) scale data or properties, and (2) field- (core or macro) scale data or properties.
|File Size||7 MB||Number of Pages||14|