Data-Driven Multivariate Monitoring of Gas Well Integrity

Maintaining the integrity of gas wells is crucial for safe and efficient production, as failures can result in significant environmental and operational consequences. This research presents a multivariate statistical process monitoring framework designed to identify deviations that indicate potential well integrity problems. The methodology uses Principal Component Analysis (PCA) on observed well data to capture the main sources of variation across various pressure and flow metrics. The derived PCA scores and residuals are statistically monitored using Hotelling's T² and Squared Prediction Error (SPE) statistics to pinpoint abnormalities. To enhance fault detection and interpretation, we investigate a combined monitoring index that merges the T² and SPE statistics, yielding a balanced measure of deviations in well integrity. We apply this framework to simulated oil well data and to observed gas well data. The suggested framework is shown to provide a data-driven solution for well monitoring, aiding in the identification of abnormal behavior in gas well operations.

Keywords: Well integrity, process monitoring, principal components, multivariate statistics.