SparkBeyond Helps Equinor Solve 30-Year Industry Puzzle

To remain competitive, oil and gas companies are striving to transform their operations, improving the reliability and availability of their assets while reducing costs and carbon emissions. 

Advances in data analytics, AI, and the IoT have helped leading energy players to optimize complex processes; track down elusive sources of loss and inefficiency; and respond more effectively to volatility, shocks, and disruptions.

As the cost of drilling and completion is the major component of most global oil and gas capital-expenditure developments, there has been one elusive challenge that had perplexed the industry for decades: an accurate way to distinguish between oil and gas when drilling.

Offshore projects designed around certain oil-production targets can face harsh realities when they end up producing more associated gas than expected. It’s the difference between drilling an underperforming and problematic well, and one that can pay out hundreds of millions of dollars over its lifetime. 

A new award-winning approach developed by petrotechnical engineers at Equinor uses SparkBeyond’s technology to significantly improve its well placement and production performance. “Using partial information to predict entire oil and gas reservoirs…has been tried for 30 years," Equinor’s senior reservoir technology specialist Tao Yang explained. Instead, using inexpensive and underused mud gas data, Equinor has developed a digital twin pseudo-log, describing it as a  “reservoir fluid identification system”. 

An industry objective for decades, the improvement in reservoir fluid typing while drilling has become a reality through Equinor’s digital innovation. Unlike other advancements made on this front, Yang explained that this new approach is the first of its kind to combine such a large database of PVT data with a machine-learning model “common to any well.” That means “we do not need to know where this well is located” to make a GOR prediction, said Yang.

Working closely with SparkBeyond’s AI-analytics technology, the Equinor team compared a database of more than 4,000 reservoir samples with real-time analysis of the mud gas that flows up a well as it’s drilled. Drawing upon countless data points from various hardware and software components, the machine-learning model predicts the gas-to-oil ratio (GOR) of the drilled-through rock, on completion of production. The proposed work approach integrates information from multiple disciplines and makes the real-time fluid identification task much more reliable for operational decisions.

Executed in real-time, the model powers an alert system that indicates when drillers encounter uneconomic production zones. Such real-time fluid identification minimizes risk by giving drillers the information they need to direct their drilling where the odds of finding higher proportions of oil or condensates are better.