Rock physics

Clustering well-log data into electrofacies using machine learning with a purely unsupervised approach, cluster-number optimization, and outlier removal based on spatial autocorrelation.

Predicting rock properties while drilling a well, especially if at several tens of meters ahead of the drill bit, can be key to reduce the drilling risks and their associated costs

The representation of geological objects in micromodels is often insufficiently realistic

Well logs interpretation is often rather subjective and highly time-consuming

It is often burdensome to handle large amounts of wells files