Document Type : Research Paper
Department of Petroleum Engineering, Kish International Campus, University of Tehran
Institute of Petroleum Engineering, University of Tehran
Department of Earth Sciences, University of Tabriz
NMR log data are used extensively to obtain reservoir parameters. Cluster analysis is a viable technique to segregate different rock types in order to increase the accuracy of permeability models. Performed cluster analysis on T2 distribution data is not necessarily consistent with core derived data. In this research we tried to integrate the reliable parameters extracted from T2 distribution data by applying peak analysis and inserting into cluster analysis. Results indicate that TCMR, peak reading, prominences, T2Lm and width are the best permeability indicators. In cluster analysis, a fundamental problem is to determine the best estimate of the number of clusters, which is usually taken as a prior in most clustering algorithms. Accordingly, NMR log data distribution values versus number of clusters were used to obtain the optimal number of clusters. This has been done by means of the knee method that finds the “knee” in a number of clusters vs. clustering evaluation graph. The optimal number of clusters in this case was five. Then, the best fitted values of the coefficients of well-known SDR model for each cluster were determined. Results show that calculated permeability using cluster analysis shows higher correlation with core derived permeability. Since this is the core part of the group attempt to use extracted T2 distribution features in permeability estimation in carbonate reservoirs, so more investigation is required to attempt satisfactory results to standardize the value of the coefficient of the permeability models in carbonate rocks with different petrophysical properties.