Application of Neural Network Workflow to Facies Modeling in Turbidite Reservoirs, Offshore Nigeria

Abstract

Three dimensional static reservoir modeling is a key aspect of geoscience workflows underpinning development of petroleum reservoirs. It helps accurately characterize reservoirs, plan appraisal and development wells, and achieve high-level Well and Reservoir Management (WRM) over the production life. Facies modeling is a critical step in the static modeling workflow because it allows us model how rock types are distributed in the reservoir and whichis the basis for modeling petrophysical property ranges associated with each facies type. The object-based approach to facies modeling has been well established within the oil and gas industry over the last ten years, however, recent efforts have been made to model reservoir facies using neural networks. This paper reports on the application of the neural network approach to facies modeling as part of the static modeling workflow for several turbidite reservoirs in an offshore Nigeria oil field. The key input data applied to delineate and model the facies were cores, well logs, and seismic. The authors consider this modeling approach to be successful and recommend it for modeling deep water facies with better thin-bed placement, high repeatability, and a shorter turn-around time.