Hossein Kheirollahi - M.Sc. Student of Petroleum Engineering
Mohammad Chahardowli - Assistant professor of Petroleum Engineering
Mohammad Simjoo - Assistant professor of Petroleum Engineering

In recent decades, data production and gathering is dramatically increased. Therefore, it is very important to process data and discover information from data.Moreover, there is a large amount of data that can be processed in the oil and gas industry, and the output of different the data processing procedures can be applied to extract useful information.In this study, the analysis of production decline curves has been used to describe the important patterns and rules among wells in an oil field.Decline curve analysis is a traditional method for determining and diagnosing problems in well operation, predicting performance and well life based on the real production history. For this purpose, empirical models of Arp’s production decline are used to study different decline types, i.e., exponential decline, harmonic decline, and hyperbolic decline.In this study, we used data of oil wells during different time steps. According to the type of decline curve, we used the modified k-mean algorithm to cluster and classify the used data in three categories. Afterwards, we used the Apriory algorithm to discover and present important rules. In fact, the main objective of this paper is to use a new approach to cluster the oil wells and present very strong rules and clusters. Such an analysis is very useful in reservoir management studies.

Clustering,rule mining,Decline Curve AnalysisK-meansGenetic Algorithm