Reservoir Characterization Clustering Analysis to Identify Rock Type Using KMEANS Method in South West Iranian Oil Field

Determination of rock types is of special importance in the construction of static and dynamic models of hydrocarbon reservoirs. Estimating the properties of reservoir rocks increases the accuracy in predicting the amount of reservoir storage and its performance. Numerous models have been proposed by experts to determine the types of reservoir rocks. But most of the proposed models are based on conventional methods based on engineering and geology of carbonate reservoir rocks. Therefore, using a machine learning method to determine rock species in comparison with previous methods and comparing its efficiency and performance with other methods seems necessary. In this study, core and log data in maroon oil reservoir after preparation were match using Dynamic Time Series (DTW) technique for depth matching. The brain data were then clustered by the non-supervised machine learning method. The kernel data clustering process was also performed by conventional model-based methods such as flow zone index method (FZI) and Winland. Then, the clustering results were validated and compared with kmeans, FZI and Winland methods by having the lithology information of the logs. The kmeans method with a 93.5% accuracy criterion succeeded in performing the highest cluster resolution, which showed that the kmeans data-based machine learning method is a suitable alternative to conventional model-based methods for clustering rock typing.