A hybrid learning automata and case-based reasoning for fractured zone detection using petrophysical logs

Abstract

Fractured zone detection and fracture density estimation in oil wells have significant effects on wellbore stability, reservoir modeling, drilling operations, and well production. Although high-resolution image logs facilitate the identification of fracture characteristics, they are not available for a large set of wells as they drilled before development of the enhanced well logging technologies. Petrophysical logs contain valuable information and can be used as an alternative approach in fracture identifications. These logs are almost available in all wells and can be used as useful inputs to shape a predictive model representing fracture characteristics where image logs are not accessible. This paper proposes a generalized case-based reasoning (CBR) method for fractured zone detection via petrophysical logs. To such aim, a set of train wells are used to beget a database composed of both petrophysical data and the image logs. A learning automata-based algorithm is conducted to find the optimal similarity relation between petrophysical logs and manual interpretation of the borehole image logs. Moreover, new decision parameters are introduced to enhance the applicability of the proposed method in real-life projects. The developed model is successfully tested on the Asmari reservoir through several oil wells from, followed by a discussion on results.