Researchers from ExxonMobil Technology and Engineering Company and ExxonMobil Upstream Company have developed a machine learning based gas-lift optimization workflow for unconventional fields that has demonstrated the potential to increase oil production while improving the efficiency of artificial-lift operations.
The technology was piloted in the Bakken shale play in the United States, where it delivered an average production increase of more than 5% across approximately 30 wells. Following the pilot’s success, the workflow was expanded to more than 200 gas-lift and plunger-assisted gas-lift wells.
Gas lift is an artificial-lift method in which compressed gas is injected into a well to reduce the density of the fluid column and help oil flow to the surface. However, determining the optimum gas-injection rate can be difficult because well conditions change over time, while available compressor capacity must often be shared among multiple wells.
The new workflow uses historical production data and machine-learning models to predict how individual wells are likely to respond to different gas-injection rates. It then applies an optimization system to recommend the most effective allocation of available lift gas across multiple wells.
Unlike conventional gas-lift optimization methods, the approach can operate without requiring extensive downhole instrumentation or new multi-rate well tests. This could reduce operational costs and allow production engineers to optimize larger numbers of wells more frequently.
The system uses automated data pipelines to update production forecasts and refine recommendations as well conditions change. This enables operators to identify wells operating below their optimum performance and adjust gas-injection rates accordingly.
The development is expected to support improved production efficiency, better use of existing compressor capacity, and increased hydrocarbon recovery from unconventional reservoirs. It also demonstrates the growing application of artificial intelligence and data analytics in upstream production operations.
Industry significance
The technology could provide upstream operators with a scalable method for improving production without necessarily drilling additional wells or installing major new infrastructure. By optimizing the distribution of available lift gas, operators may be able to increase output, reduce operating costs, and improve the performance of mature and rapidly declining unconventional wells.