The Determination of the Most Effective Petrophysical Logs for Rock typing in a Detrital Reservoir Rock, Using Clustering Methods

The clustering analysis is a method for data classification based on the similarity, so that the most similar data are placed in the same cluster. The cluster analysis of petrophysical data is able to determine the quality of sandstone reservoir based on the different existing lithologies. Since the most sandstone reservoirs have different kinds of interbedded shales as well as different detrital and chemical minerals due to their sedimentation environment, they show a variety of lithologies and consequently different reservoir qualities. The determination of the most effective petrophysical logs based on the environment understudy has a significant effect on the clustering quality, lithology determination, and the reservoir quality evaluation. In this research, to determine the most effective logs in the two stage and k-means methods, the GR, LLD, LLS, MSFL, DT, URAN, THOR, POTA, CALI, NPHI, RHOB logs were chosen as the input in IBM SPSS, Statistic version 24. Based on the obtained results, the GR, LLD, LLS, NPHI, and RHOB logs are the most effective logs for two stage clustering method. This result was in agreement with the k-means method findings. Based on the clustering results, the understudy sandstone formation with a thickness of 222 meters were classified to three lithological groups of carbonate-anhydrite, shaly and clean sandstone with the thickness of 44.40, 73.26, 104.34 meters, respectively. The clean sandstone had the best quality with regards to the average porosity and permeability in compare to the other subsections.