Unsupervised Machine Learning for Asphaltene Anomaly Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Asphaltene deposition in deepwater wells leads to operational issues such as unplanned shut-ins and costly production curtailments due to the inability of existing methods to accurately predict and prevent asphaltene-related anomalies in oil and gas production equipment.
Innovation Solution
A scalable unsupervised machine-learning model that utilizes historical and real-time sensor data to predict asphaltene anomalies by reconstructing operation characteristics and determining an anomaly score, allowing for early detection and prevention of potential failures before actual shut-ins occur.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional monitoring methods are used to detect asphaltene deposition, then equipment can be monitored for operational parameters, but early prediction and prevention of asphaltene anomalies cannot be achieved
Solution Approach 1:
The machine learning model performs preliminary analysis of operational parameters to predict asphaltene anomalies before they occur. The system reconstructs operational characteristics and generates anomaly scores in advance, enabling proactive intervention rather than reactive response to actual deposition events.
Solution Approach 2:
The patent introduces an intermediary machine learning model that acts as a mediator between raw operational data and actual asphaltene deposition detection. The model reconstructs operational characteristics and generates anomaly scores, serving as an intermediate layer that enhances prediction capability without requiring direct complex sensing of asphaltene formation.
2Loss of time
If machine learning models are implemented to predict asphaltene anomalies, then early detection capability is improved, but computational requirements and model training complexity increase
Solution Approach 1:
The machine learning model is trained in advance using historical operational data to learn patterns associated with asphaltene anomalies. This preliminary training enables the model to rapidly process real-time data and generate anomaly scores without requiring complex computations during actual detection, thus reducing detection time while managing training complexity.
Solution Approach 2:
The system creates a virtual model of normal operational characteristics through the machine learning model. This copied representation of normal operations serves as a reference against which actual operational data is compared, enabling rapid anomaly detection without requiring complex real-time analysis of all possible deposition scenarios.
3Loss of information
If continuous monitoring of operational parameters is performed, then more data is available for analysis, but data processing requirements and computational resources increase
Solution Approach 1:
The machine learning model extracts only the most relevant features and patterns from the continuous stream of operational parameters. By focusing on key indicators that correlate with asphaltene anomalies, the system processes comprehensive information without requiring computational resources to analyze every detail of the raw data, thus balancing information completeness with energy efficiency.
Data Source
AI summary
An unsupervised machine-learning model is trained using historical operation characteristics of a well. Operation characteristics of the well for a duration of time is reconstructed by the unsupervised machine-learning model. Whether an asphaltene anomaly will occur in the future at the well is predicted based on the difference between the operation characteristics of the well and the reconstructed operation characteristics of the well.