Document Type : Original Research Paper
Authors
1 Ph.D. Student, Faculty of Mining and Metallurgical Engineering, Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran
2 Ph.D. Student, Faculty of Engineering, University of Tehran, Tehran, Iran
3 Associate Professor, Faculty of Engineering, University of Tehran, Tehran, Iran
4 Professor, Faculty of Engineering, University of Tehran, Tehran, Iran
5 Ph.D., Faculty of Engineering, University of Tehran, Tehran, Iran
Abstract
In evaluating the quality of the reservoir sandstone facies, clay usually has significant effect in reducing reservoir effective porosity, permeability as well as calculation accuracy of formation fluids saturation. There are several methods for identifying and measuring the amount of clay. In sandstone reservoirs, diversity of type and amount of clay minerals may change the capacity of cation exchange (CEC) measured in the reservoir rocks. The last parameter (i.e. CEC) can be an important criterion for zoning of reservoir based on the type of clay minerals. Cation Exchange Capacity (CEC) measurement is used as one of the subsidiary clay typing methods. This parameter is the ability of clay to absorb and release of cations in the surrounding solution, which has a specified range for each clay mineral. In cases of clay mixtures, CEC values tend toward the range of the dominant clay type of sample. In this study, cation exchange capacity of the clay minerals has been calculated in two wells of the Gonbadli Gas Field in the Shurijeh sandstone reservoir. First, CEC of 20 samples has been measured using Bower method and employing intelligent estimator based on neural network as well. Based on the petrophysical logs and laboratory results, an appropriate model was fitted to estimate this parameter in well interval. According to the CEC values of clay minerals, existing data classified into five categories including clean zone and zones of clay containing kaolinite, chlorite-illite, halloysite with two water molecules and montmorillonite. For this purpose Bayesian, Parzn and K- nearest neighbor (KNN) classifiers were used. Finally, the obtained results in comparison with the results of X-ray diffraction experiments (XRD) showed good agreement.
Keywords
)