Downhole Leak Detection: Introducing A New Wireline Array Noise Tool
- Qinshan Yang (GOWell international, LLC) | Jinsong Zhao (GOWell international, LLC) | Marvin Rourke (GOWell international, LLC)
- Document ID
- Society of Petroleum Engineers
- SPE/ICoTA Well Intervention Conference and Exhibition, 26-27 March, The Woodlands, Texas, USA
- Publication Date
- Document Type
- Conference Paper
- 2019. Society of Petroleum Engineers
- 5.5 Reservoir Simulation, 1.6.12 Plugging and Abandonment, 2.10 Well Integrity, 5 Reservoir Desciption & Dynamics, 2.1.3 Completion Equipment, 1.6 Drilling Operations, 7.6.6 Artificial Intelligence, 3 Production and Well Operations, 2 Well completion
- Leak Detection, Sensor Matrix, Well Interity, Machine Learning, Bayesian
- 11 in the last 30 days
- 198 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 9.50|
|SPE Non-Member Price:||USD 28.00|
This paper will introduce a new generation wireline Array Noise Tool (ANT). This tool is used to detect downhole acoustic / vibration activities originating from fluid-structure friction flow. One of main applications in Well Integrity (WI) and Plug & Abandonment (P&A) for ANT is to locate leak sources in well completions and tubulars. The innovative sensor matrix and system configuration together with three novel data processing methods are studied and developed to address the following primary challenges;
Tiny acoustic leakage signals (-30dB to -60dB), for example, the minor leaks behind pipes or even inside the formation matrix,
Strong road-noise acoustic signal contamination from tool motion while dynamic logging,
Nonstationary and/or nonlinear signal distortions because of tool flexural vibrations, and
Downhole seismic noise.
The tool can be operated both in stationary logging and in dynamic logging.
The wide-band sensor matrix is designed with a unique configurable technique to form different measurement arrays. As a result, the tool can simultaneously acquire absolute and differential acoustic signals. By using this sensor matrix we are able to improve Signal-to-Noise Ratio (SNR) by up to 20 to 30dB. From the acquired data, we employ a multi-dimensional machine learning (ML) classification module, cascaded with cluster iteration to separate real leak signatures from other unwanted noise signals. After a data conditioning process, the wave velocity-domain decomposition method is utilized to further distinguish the leak signal propagation characteristics against other noise propagations to enhance overall SNR for leak detectability. Lastly we use a Bayesian likelihood analysis to identify the leak depth locations with a confidence index based on the information contained in both signal energy and signal velocity. We are able to achieve 15dB to 20dB SNR improvement from this data processing methodology. The system design goal is to eliminate unwanted acoustic noise that is not associated with leaks, while maintaining sufficient sensitivity to pick up minor leaks.
The tool has been logged commercially in the US, Middle East, East Asia, and Latin America. The tool performance has been validated through simulation, lab tests, and field logs. Field logging examples are demonstrating a leak detection success rate above 95%. Field cases include multi-annulus, low flow rate, and gas well field examples. Field results will be presented in this paper.
ANT instrument technology and the associated advanced processing methods are a new solution for detecting the leak source locations and monitor leak paths, especially, in Well Integrity (WI), Plug and Abandonment (P&A), and many other well applications.
|File Size||1 MB||Number of Pages||16|