A new intelligent model that successfully learns from high-dimensional data and effectively identifies high-production areas and optimum lateral-re-entry candidates is presented. The model is entirely data driven and uses Wang-Mendel (WM) rules extraction, fuzzy logic (FL), pattern recognition, and Voronoi mapping. The authors applied their model to a large field with thousands of wells and multiple production layers, finding that it outperformed previous methodologies significantly.
Introduction
This paper demonstrates the application of data-driven-modeling technology to a relatively new domain, petroleum engineering—in particular, to production-opportunity identification and well-work optimization.
The Kern River field is the single largest producing onshore heavy-oil asset in North America. The structure is homoclinal, dipping southwest into the basin, with nine distinctive productive formations.