Closed-Loop Optimization
Closed-loop optimization is a promising framework when we constantly need to make decisions while accounting for new data. Various cases in energy industry have proven the effectiveness of closed-loop optimization to enhance operations.
In the oil and gas industry, closed-loop optimization has gone through great advancements in recent years. Closed-loop field development (CLFD) optimization is now the state of the art technique for petroleum reservoir optimization which is applicable to both conventional and unconventional resources. There are other processes such as close-loop reservoir management which is typically applied to conventional resources. Companies typically want to drill new wells to increase their production and not lose their share in the market.
In Groundwork Analytics, we can help enhance your reservoir optimization workflow and improve your oil production by as much as 50%. We believe that traditional workflows are inefficient as they are typically based on manual history matching and they don’t incorporate proper optimization techniques. Our process, by contrast, is based on the most efficient techniques for history matching, optimization, selection of representative models, and decision analysis.
History Matching And Model Calibration For Oil and Gas
In Groundwork Analytics, we have been working with various customers, developing advanced tools for statistical model calibration (history matching) of oil and gas simulation models. Production data (oil, gas and water production rates, often reported monthly), geological data, well test and well log data, are all incorporated in the simulation models by our tools. We utilize the best methods for data assimilation (time-dependent data such as production data), geostatistics and Kriging algorithms (for integrating spatial data) and suitable correlations and inference techniques for well-log data interpretation and integration.
The calibrated (history matched) simulation models are much more reliable for use in decision making. We can easily interface with your choice of commercial simulator (CMG, Eclipse, etc). Our goal is to then hybridize classic reservoir simulation methods with new/advanced machine learning tools.
Our product performs local-global history matching (matching field-level data and then well-by-well history matching). We developed methods that we utilize depending on the specific reservoir problem. These include ensemble-smoother methods (commercial software for ensemble-based data assimilation utilizing Eclipse simulator) and ensemble Kalman filter, optimization-based history matching (where efficient derivate-free optimization algorithms are applied to automatically tune the model for matching observed data), PCA-based history matching, and parameterization methods.
In cases, when the simulations take a long time (such as in steam injection / thermal recovery / EOR / CO2 injection), we build accurate proxy models utilizing machine learning and deep neural networks to accelarate computations without sacrificing accuracy.
WMS: Well Management System
Operators deal with a large number of wells and the engineers need to analyze the data of all the wells and deliver analytics that help managers make the right decisions. Decisions may include determining the best candidate wells for workover operations, sweet spots for drilling infill wells, production wells to turn into injectors, changing the recovery process for boosting the production, or finding best performing wells and understanding best practices. In Groundwork Analytics, we have been developing tools and providing services for efficient well data management.