Video: Integrating Downhole Temperature Sensing Datasets and Visual Analytics for Improved Gas Lift Well Surveillance
- Oladele Bello (Baker Hughes, a GE company) | Derek S. Bale (Baker Hughes, a GE company) | Lei Yang (Baker Hughes, a GE company) | Don Yang (Baker Hughes, a GE company) | Ajish Kb (Baker Hughes, a GE company) | Murali Lajith (Baker Hughes, a GE company) | Sony Lazarus (Baker Hughes, a GE company)
- Document ID
- Society of Petroleum Engineers
- Publication Date
- Document Type
- 2018. Copyright is retained by the author. This presentation is distributed by SPE with the permission of the author. Contact the author for permission to use material from this video.
- 6.1.5 Human Resources, Competence and Training, 2.3 Completion Monitoring Systems/Intelligent Wells, 2.1 Completion Selection and Design, 3.1 Artificial Lift Systems, 7.6.6 Artificial Intelligence, 5.6.11 Reservoir monitoring with permanent sensors, 6 Health, Safety, Security, Environment and Social Responsibility, 3 Production and Well Operations, 3.1.6 Gas Lift, 6.1 HSSE & Social Responsibility Management, 4.2 Pipelines, Flowlines and Risers, 7.6 Information Management and Systems, 3 Production and Well Operations, 2 Well completion, 7 Management and Information, 7.6.4 Data Mining, 2.1.3 Completion Equipment
- VISUAL ANALYTICS, IMPROVED GAS LIFT WELL, SURVEILLANCE, DOWNHOLE TEMPERATURE SENSING DATASETS, INTEGRATING
- 1 in the last 30 days
- 2 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 5.00|
|SPE Non-Member Price:||USD 10.00|
Given the near ubiquity of fiber-optic, information and communication technologies in reservoir and well management, there is a significant need for one-stop shop downhole distributed sensing data analysis methods together with machine learning techniques towards autonomous analysis of such data sources. However, traditional approaches of converting distributed temperature sensor (DTS) data to actionable insights for optimizing gas lift well operations management remain dependent on training based on human annotations. Annotation of downhole distributed temperature sensor data is a laborious task that is not feasible in practice to train a big data classification algorithm for accurate and reliable anomaly detection of gas lift valves. Furthermore, even obtaining training examples for event diagnosis is challenging due to the rarity of some gas lift valve problems. In gas lift well surveillance, it is essential to generate real-time results to allow a swift response by an engineer to prevent harmful consequences of gas lift valve failure onsets on well performance. The online learning capabilities, also mean that the data classification model can be continuously updated to accommodate reservoir changes in the well environment. In this paper, we propose a novel online real-time DTS data visual analytics platform for gas lift wells using big data tools. The proposed system combines Apache Kafka for data ingestion, Apache Spark for in-memory data processing and analytics, Apache Cassandra for storing raw data and processed results, and INT geo toolkit for data visualization. Specifically, the data analytics pipeline uses data mining algorithms to statistically learn features from the DTS measurements. The learned features are used as inputs to a k-means algorithm and then use supervised learning to predict the performance status of gas lift valves and raise alarms based on analytics-based intelligent warning system. The performance of the proposed system architecture for detecting gas lift valve anomaly is evaluated under varying deployment scenarios. To the best of our knowledge, DTS data analytics pipeline system has not been used for real-time anomaly detection in gas lift well operations.