Autonomous Virtual Flow Metering for Offshore Oil & Gas Production
Client Overview
Client: <Confidential> Super-Major Oil and Gas Company
Industry: Offshore Oil & Gas Production in the North Sea
Project Duration: October 2012 – June 2013
The Challenge
A major offshore production platform required real-time estimates of oil, gas, and water production for individual wells operating between scheduled well tests. Traditional methods relied heavily on frequent well testing, but operational constraints—including a 70-90% production turn-down, mechanical issues affecting lift gas systems, and limited testing opportunities—made continuous monitoring difficult. The operator needed an autonomous system that could:
– Provide accurate production estimates without constant well testing
– Self-calibrate and maintain accuracy during changing operational conditions
– Operate reliably through system outages and data interruptions
– Deliver standardized quality metrics with each estimate
The Solution
IntelliDynamics deployed its Data-Driven Virtual Flow Meter (DD-VFM) system—an autonomous, self-calibrating solution providing real-time production estimates for oil, gas, and water across 11 producing wells. The system operated on a datacenter server, integrating with multiple OSISoft PI data historians to access both historical and real-time data.
Key Capabilities:
– Autonomous well state and test detection
– Real-time production estimation with confidence intervals
– Automated model recalibration and remodeling
– Well test validation and quality assessment
– Intelligent “catch-up” processing after outages
– Automated production allocation
Implementation Highlights
Scope:
– 11 producing wells monitored simultaneously
– 3 phases per well (oil, gas, water)
– 33 virtual meters deployed
– Fully autonomous operations achieved
Technical Features Implemented:
– Intelligent Well Test Recommendation: System automatically identified when wells operated outside historical test conditions, flagging when new tests were needed
– Autonomous Remodeling: Detected failed sensors and rebuilt models automatically to maintain operations
– Outage Recovery: “Catch-up” functionality automatically processed missed data after network, server, or PI system outages
– Confidence Metrics: Converted quality indicators into units-based confidence intervals (±bbl/day, mmscfd)
– Dual Adaptation: Self-calibrated using both well test data and separator meter totals
Results & Impact
DD-VFM vs. Theoretical Models – Dramatic Performance Advantage:
A direct comparison was conducted between DD-VFM and the platform’s existing physics based rate-and-phase models throughout the evaluation period, revealing substantial performance differences:
Relative Accuracy Comparison:
– Oil Estimation: DD-VFM achieved 97% relative accuracy vs. separator meters compared to 64% for physics based models—a 51% improvement
– Gas Estimation: DD-VFM achieved 97% relative accuracy vs. separator meters compared to 72% for physics based models—a 35% improvement
Key Advantages:
– Autonomous Calibration: Unlike theoretical models requiring manual recalibration, DD-VFM maintained superior accuracy autonomously through:
– Automatic adaptation to separator meters every 2 hours
– Self-calibration on available well tests
– Intelligent quality monitoring and automated remodeling
– Intuitiveness Enhancement: DD-VFM’s congruence-based adaptation algorithm ensured estimates remained physically intuitive (responding correctly to choke and lift gas changes), addressing a key limitation where both model types can exhibit counter-intuitive behavior during heavy adaptation periods
– Robust Performance Under Stress: DD-VFM maintained high accuracy even during extreme operational challenges (70-90% production turn-down, no lift gas, minimal well testing)
The comparison validated that DD-VFM dramatically outperformed theoretical models for both oil and gas production—all while eliminating manual intervention requirements, delivering significant operational and economic advantages.
Operational Achievements:
– Full autonomous operation maintained for 45+ days (March 1 – April 15, 2013)
– High system availability despite challenging platform conditions
– Successful operation during extreme turn-down (70-90% below normal production)
– Zero manual intervention required during evaluation period
– Robust outage handling with automatic data recovery demonstrated multiple times
System Capacity:
– Operated at only 7.72% CPU utilization with 11 wells
– Estimated capacity: 100+ wells on single server instance
Innovations & Best Practices
The project resulted in several industry-advancing capabilities:
1. Congruence-Based Adaptation Algorithm: A new approach ensuring meter-based adjustments never contradicted physical well responses, preserving intuitiveness while maintaining accuracy
2. Multi-Separator Testing: Successfully handled parallel testing on multiple separators simultaneously—a first for DD-VFM deployments
3. Unified Well State Detection: Centralized detection algorithms using temperature thresholds combined with valve positions, proving more reliable than choke-position-only methods
4. Production Insights from Startups: Developed methodology to extract individual well production data from differential startup sequences when well tests were unavailable
5. Interactive Data Environment (i2e): Prototype tool deployed to 20 users, providing trends, histograms, scatter plots, and data export capabilities
Business Value
Despite operating under challenged conditions with minimal well test feedback, the DD-VFM system demonstrated:
– Reduced operational uncertainty through continuous production monitoring
– Maintained accuracy approaching “Class B” standards (>90%) without regular well testing
– Eliminated manual effort for production estimation and allocation
– Provided actionable insights through automated well test recommendations
– Ensured business continuity through intelligent outage recovery
– Delivered 97% relative accuracy for both oil and gas—dramatically exceeding theoretical model performance (64% oil, 72% gas) while operating fully autonomously without manual recalibration
The field trial validated DD-VFM’s capability to operate autonomously in demanding offshore environments, delivering reliable production estimates even under sub-optimal conditions while establishing new industry best practices for adaptive virtual metering.
About IntelliDynamics
IntelliDynamics provides intelligent, autonomous data-driven solutions for all industries, including oil and gas, specializing in virtual flow metering, production allocation, production optimization and real-time asset production control.
This case study demonstrates DD-VFM performance under real-world operational constraints. For optimal “Class A” performance (>95% accuracy), multi-rate well testing and regular calibration are recommended.
Interested in Autonomous Virtual Metering for Your Operations?
Email us at info@intellidynamics.net