Automated Anomaly Detection of Multi-Metallic Tubulars in Well Integrity Logs Using Signal Mode Decomposition and Physics-Informed Decision Making
Westside Houston
Speaker:
Seminar Date: Sep 11 2025
Registration Opens: Jul 06 2025 - Jul 12 2025
Time: 11:30 AM - 01:00 PM (US CDT)
Admission/Registration Link: None
Donation Link: None
Meeting/Webinar Link: None
Contact: QinShan “Shan” Yang (VP Westside, SPWLA Houston Chapter)
Corresponding: vpwestside@spwla-houston.org
Fees: FREENOTES:
Speaker : Ze Wang (GOWell )
Date : Thursday, Spe 11th , 2025
Time : 11:30 am – 1:00 pm (US CDT)
Venue : SLB, 6350 West Sam Houston Parkway North, Houston, TX 77041
Admission : This activity will include a boxed lunch.
The seminar is sponsored by SLB so there
is no charge for registration,
However, you still need to register using
the applicable links below:
Parking Info : Guest parking is available free of charge.
Upon arrival, please proceed to the front desk to check in
Please register by one day before the event to reserve lunch using the above provided link.
Contact : QinShan (Shan) Yang (SPWLA Houston VP Westside)
Corresponding : vpwestside@spwla-houston.org
ABSTRACT:
Anomaly detection using well integrity logs is crucial for multi-metallic tubular wells, as it helps save operators costly repairs and potential well abandonment. However, features such as collars and artifacts often obscure the signals of outer pipes, making anomaly interpretation particularly challenging. To address this issue, an automated anomaly detection method has been developed that effectively separates collars and corrosion signals from complex log results. This approach significantly enhances analysis accuracy and efficiency in wells with multiple tubulars, up to five layers. The anomaly detection method utilizes cased hole logging images obtained from a pulsed eddy current electromagnetic tool as input. It outputs the location information of collars and anomalies, respectively. The method comprises two steps—signal mode decomposition and the decision-making process. A novel approach, hierarchical multiresolution variational mode decomposition (HMVMD), is introduced to extract both anomaly-related and collar-related signals by decomposing the input into a set of frequency-based modes. The decision-making phase employs a decision tree designed based on Bayes’ theorem, with the process simplified by Markovian modeling. Prior knowledge of cased hole completion is incorporated into the design to further refine results. Field trials in operational wells have been conducted to evaluate the proposed method. By distinguishing the thickness-related signal from raw data, previously obscured anomalies became interpretable. The method excels at denoising the data, effectively reducing noise interference by enhancing signal-to-noise ratio (SNR) up to 29 dB. It saves 90% of the time that log analysts spend manually differentiating collars, traditionally requiring several hours, thereby significantly optimizing the interpretation efficiency. In a five-pipe scenario, the results demonstrate detection accuracy rates of approximately 99% for the inner three pipes. It maintains accuracy rates over 90% and 75% on the fourth and fifth pipes, respectively, where the SNR is low, and the outer-pipe signal is masked by the inner layers. In addition, it maintains high accuracy under complex well scenarios, such as those involving completion equipment and eccentricity. This new approach offers interpretation specialists an efficient and accurate anomaly analysis tool for multi-metallic tubulars.
BIOGRAPHY:
Ze Wang is a research scientist at GOWell, specializing in cased hole well integrity and production logging. His research interests include algorithm development, data-driven solutions, signal processing, and numerical simulation for oil and gas applications. He also has research experience in unconventional reservoirs and carbon utilization and storage. Dr. Wang previously worked as a post-doctoral scholar at Missouri University of Science and Technology, USA. He holds PhD and MS degrees in petroleum engineering from Missouri University of Science and Technology and a BS degree in petroleum engineering from China University of Petroleum, China.