GEO 2010 · 2010 · 14 citations · 0 references
EngineeringComputer ScienceEarth SciencesData MiningMachine LearningWell LoggingGeological ModelingLithology IdentificationPorosity EvaluationReservoir GeologySupport Vector MachinesWell Log InterpretationCarbonate FaciesCarbonate Facies AnalysisGeological Facies AnalysisMachine Learning For Reservoir CharacterizationMachine Learning In PetrophysicsPermeability Prediction From Well LogsPorosity From Well LogsReservoir Data
This research investigate the capabilities of data mining in identifying carbonate litho-facies from well<br>logs based on extreme learning and support vector machines. Formation facies usually influence the<br>hydrocarbon movement and distribution. Identifying geological formation facies is critical for economic<br>successes of reservoir management and development. The identification of various facies, however, is<br>a very complex problem due to the fact that most reservoirs show different degree of heterogeneity.<br>Last decade, there has been an intense interest in the use of both computational intelligence and<br>softcomputing learning schemes in the field of oil and gas: exploration and production to identify and<br>predict permeability and porosity, identify flow regimes, and predict reservoir characteristics. However,<br>most of these learning schemes suffer from numerous of important shortcoming. This paper explores<br>the use both extreme learning and support vector machines systems to identify geological formation<br>facies from well logs. Comparative studies are carried out to compare the performance of both extreme<br>learning and support vector machines with the most common empirical and statistical predictive<br>modeling schemes using both real-world industry databases and simulation study. We discuss how the<br>new approach is reliable, efficient, outperforms, and more economic than the conventional method.