Clustering and classification techniques for blind predictions of reservoir facies

The integration of different data in reservoir understanding and characterization is of prime importance in petroleum geology. The large amount of data for each well and the presence of new unknown wells to be analyzed make this task complex and time consuming. Therefore it is important to develop reliable prediction methods in order to help the geologist reducing the subjectivity and time used in data interpretation. In this paper, we propose a novel prediction method based on the integration of unsupervised and supervised learning techniques. This method uses an unsupervised learning algorithm to evaluate in an objective and fast way a large dataset made of subsurface data from different wells in the same field. Then it uses a supervised learning algorithm to predict and propagate the characterization over new unknown wells. Finally predictions are evaluated using homogeneity indexes with a sort of reference classification, created by an unsupervised algorithm.