Bayesian Network for Wellbore Event Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting adverse well production events in oil and gas fields are not timely, lack robust prioritization, and are prone to inaccuracy due to the reliance on threshold-based systems, leading to false alarms and high time requirements for setting thresholds without confidence limits.
Innovation Solution
A method and system that utilize real-time well production measurements to calculate mathematical derivatives, develop a probability calculator, and display the status of the well, incorporating facilities data and updating calculations with new data to provide accurate and prioritized event detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If threshold-based detection systems are used to identify well production events, then event detection capability is provided, but false alarms increase and confidence limits are lost
Solution Approach 1:
The patent transforms the detection approach by changing from fixed threshold parameters to probabilistic parameters. Instead of using static threshold values that trigger alarms, the system calculates probability distributions of production parameters and determines events based on statistical significance and confidence intervals, thereby providing both detection capability and quantitative confidence measures
Solution Approach 2:
The patent replaces the mechanical threshold-based detection system with a probabilistic statistical model. The rigid on/off threshold mechanism is substituted with continuous probability calculations that incorporate variability and uncertainty, allowing for more reliable event detection with quantified confidence levels
2Reliability
If threshold-based detection systems are used to identify well production events, then event detection is provided, but false alarms increase
Solution Approach 1:
The system changes from fixed threshold parameters to dynamic probabilistic parameters that adapt to the inherent variability of production data. By using statistical confidence intervals and probability distributions, the system distinguishes between normal fluctuations and actual events, significantly reducing false alarm generation while maintaining reliable event detection
Solution Approach 2:
The probabilistic model incorporates feedback from historical production data to continuously refine event detection criteria. The system learns from past events and normal operations to improve its ability to distinguish true events from false alarms, creating a self-improving detection mechanism
3Measurement precision
If engineers manually analyze time-series measurements to discern wellbore events, then professional judgment is applied, but time consumption increases and timeliness decreases
Solution Approach 1:
The system enables self-service event detection by automatically performing the analysis that previously required engineer intervention. The probabilistic model autonomously processes time-series measurements, identifies events, and provides confidence assessments without requiring manual chart review or expert judgment, thereby eliminating time loss while maintaining or improving detection accuracy
Solution Approach 2:
The patent replaces the manual mechanical process of engineer analysis with an automated computational probabilistic model. The system uses algorithms to perform what previously required human expertise, transforming a time-consuming manual process into an instantaneous automated analysis that preserves accuracy while eliminating time loss
4Extent of automation
If basic automated methodology with threshold counters is used to flag events, then automation is provided, but prioritization capability is lacking
Solution Approach 1:
The patent adds a new dimension to automated event detection by incorporating probabilistic confidence levels and event severity rankings. Instead of merely flagging events at a single threshold level, the system provides multi-dimensional information including probability values, confidence intervals, and prioritized rankings that enable intelligent decision-making about which events require immediate attention
Solution Approach 2:
The system transforms the automated methodology by changing from simple threshold counting to probabilistic parameter assessment. The event flagging mechanism is enhanced with probability distributions and confidence metrics that provide prioritization information, allowing the automated system to not only detect events but also rank them by significance and urgency
Data Source
AI summary
A method and system for estimating the status of a production well using a probability calculator and for developing such a probability calculator. The method includes developing a probability calculator, which may be a Bayesian network, utilizing the Bayesian network in a production well event detection system, which may include real-time well measurements, historical measurements, engineering judgment, and facilities data. The system also includes a display to show possible events in descending priority and/or may trigger an alarm in certain cases.