Casing Degradation Prediction for Well Integrity

Challenges

Casing degradation is a critical concern in well integrity management; unplanned failures result in significant downtime, costly workovers and safety risks.
Casing integrity is critical to structural integrity and zonal isolation, yet it is rarely monitored at sufficient frequency.
Degradation is multi-factor — corrosion, mechanical stress and downhole chemistry — while the baseline approach focuses mainly on metal loss and treats well age as the primary risk driver.

Approach

Degradation logs, production streams, pressure build-up measurements and fluid chemistry are combined into aligned time-series and static features, enriched with domain-driven indicators.

An ML ensemble flags the wells with a high probability of casing degradation, and feature importance and sensitivity analysis translate the model's signals into actionable hypotheses about what is driving it.

Key features

Data collection and preprocessing across logs, production, pressure and chemistry
ML ensemble flagging wells at high probability of degradation
Model parameters tuned for predictive performance
Degradation drivers explained through feature importance and sensitivity analysis

Impact

33%

Reduction in unnecessary workovers caused by risk overestimation in the baseline approach

Wells flagged before failure, not after

As stated in the company deck

Degradation drivers explained per well