Automated well-log interpretation
An unsupervised model on raw well-log data that finds the abrupt changes marking layers of interest, accelerating interpretation and making it consistent.
Challenge
Well-log interpretation was slow and depended on the interpreter. The company wanted machine-learning-assisted insight that accelerates it and keeps it consistent.
Solution
Arundo deployed an unsupervised anomaly-detection model against raw porosity, density, resistivity and caliper logs, detecting abrupt changes in the measurements while distinguishing meaningful kicks from noise and outliers, with a path to a supervised model trained on the customer's confirmations.
Impact
Dramatically faster well-log interpretation with more consistent insight, and a network effect as confirmed anomalies build the supervised model.