FACIES PREDICTION ON WELL LOGS FROM NAMORADO FIELD, CAMPOS BASIN, OFFSHORE BRAZIL
Lucas Lima de Carvalho1; Laura Lima A. Santos1; Nadege Bize-Forest2; Edmilson Helton Rios1; Paulo Couto1
1 UFRJ; 2 Schlumberger
doi:10.20906/CPS/CILAMCE2017-1085
Resumo
Geo-petrophysical well logging is performed in almost all wells drilled for petroleum exploration. They are important for formation evaluation, well integrity, reservoir surveillance and reserve calculations. Accurate and high resolution identification of geological facies crossed by a well is only possible with the description of rock samples coming from drilling cuts, whole cores or lateral plugs. However, these technics are very expensive, time consuming and limited to specific intervals. This paper describes then how to predict geological facies using conventional wireline logs, such as gamma ray, resistivity, density and neutron and; data mining techniques. Supported Vector Machine and Random Forest algorithms were applied to a raw data set and after a preprocessing workflow. The study is performed in one of the most productive turbidite reservoir in Campos Basin, Southeast Brazil.
Palavras-chave: Machine Learning; Facies Prediction; Campos Basin