Well Leak Detection with Distributed Fiber Optic Sensing System

Abstract Well integrity monitoring is a persistent challenge throughout the lifecycle of oil and gas wells, where early detection of downhole leakage is essential for operational safety, environmental protection, and cost control. Distributed Fiber Optic Sensing (DFOS), including Distributed Temperature Sensing (DTS) and Distributed Acoustic Sensing (DAS), provides continuous spatial coverage along the wellbore and offers advantages over conventional point-based logging. However, in field applications, leakage-related responses in fiber-optic data are often weak and masked by slowly varying thermal and acoustic backgrounds, limiting their direct interpretability. This study proposes an engineering-oriented data processing framework aimed at enhancing leakage- and integrity-related anomalies in DTS and DAS measurements. Building on established concepts from conventional acoustic logging, the approach employs raw acoustic energy, ultra-low-frequency (ULF) components, and frequency-band energy (FBE) features to improve anomaly contrast without introducing complex physical models or extensive parameter calibration. The proposed methodology is first evaluated through controlled laboratory-scale physical modeling experiments with known leakage locations and adjustable leakage intensities. The analysis focuses on the spatial distribution and boundary characteristics of energy and temperature responses induced by leakage, establishing practical correspondence between leakage processes and distributed fiber-optic observations. These results provide physically grounded constraints for subsequent field data interpretation. The framework is then applied to real downhole monitoring datasets characterized by strong background variations. The results demonstrate that the combined use of energy-based, ultra-low-frequency, and frequency-band features enables clearer identification of leakage and abnormal flow signatures in both DTS and DAS data, with improved interpretability compared to raw measurements. This study shows that DFOS, when coupled with appropriate and physically motivated data processing strategies, can serve as an effective tool for well integrity monitoring and leakage detection. The proposed approach offers practical value for field deployment and interpretation, particularly in complex downhole environments where traditional point-based measurements or threshold-based methods are insufficient.