Determination of porosity and permeability in reservoir intervals by artificial neural network modelling, offshore Eastern Canada

A quantitative integration of porosity/permeability measurements and well log data from the major reservoir intervals throughout the basin is carried out using a back-propagation artificial neural network (BP-ANN), modified with the Marquardt algorithm. After data preprocessing and training/supervising example preparation, a model for the relationship among porosity, permeability and well log responses was established with the BP-ANN technique. The BP-ANN model was then used to construct profiles of porosity and permeability in both cored and uncored wells for the Avalon, Hibernia and Jeanne d'Arc formations from well logs. The BP-ANN derived porosity and permeability curves provide a basis for further reservoir studies, such as inter-well permeable units recognition and correlation, and basin-wide reservoir quality evaluation.