Automate the verification of well log data and minimize laboratory observation.
Oil Well Log Analysis
Challenge
Well log data processing requires expert knowledge as geological data contains many
uncertainties, making logging interpretation a very time-consuming process.
Solution
Application of ML models to automate the well log data processing
was recommended.
The data set was taken from a real oil field in Russia to develop ML models.
Gradient tree boosting and Convolutional neural network algorithms were used to
reduce uncertainties in data interpretation.
Prediction by interval of depths using CNN turned out to be more
successful than the prediction by sample using gradient tree boosting algorithm.
Project Info
Category :Well Logging
Client : Oil & Gas
Location : Russia
Completed Date : 2020
Value-Addition
The process of well logging analysis and identification of reservoir location was
partially automated.
Time taken to interpret and verify well log data was reduced by 30%
The segmentation approach by ML algorithms was proven to be more successful than the traditional approach.