Gas production from unconventional shale reservoirs is known for rapid declines. Intermittent shut-in production constitutes a technique typically applied to low-production wells during late life stages to maintain economic rates. This technique involves a cyclic process of shutting in the well temporarily to allow it to build up pressure and subsequently switching the well to production. Operators often manage hundreds of wells on intermittent shut-in production; these wells, however, incur different shut-in and production cycle times, thus requiring a complicated management approach. Because every well has a unique production behavior and reservoir characteristics, searching for optimum operational conditions individually is not only technically challenging, but also operationally time-consuming and labor- intensive. Our goal was to use active learning analytic, a type of machine learning deployed on an edge computing platform, to autonomously control and optimize these unconventional gas wells. The field trial results show increased production, reduced liquid loading, decreased manual intervention, and reduced carbon footprint. Our solution utilizes an edge computing platform to deploy the analytic on the wellhead without requiring a stable internet connection. A computing device at the edge connects to controllers on site, processes data, sets system control parameters, and enables automation for operations deploying an optimization algorithm. Active learning algorithms are valuable for use in the optimization of systems that are not mathematically definable. These algorithms are also proven to learn the relationship between the inputs and outputs and use prior knowledge to intelligently search for the optimum settings within the defined operating limits. The low latency of edge computing allows for high-frequency data collection in seconds and a rapid control of the wells. The edge device continuously monitors production and initiates re- optimization as needed when operational conditions change. We developed an analytic that autonomously controls the intermittent production technique where a well is shut-in based on a specified minimum gas production rate and opened when the pressure builds up to the specified target during the shut-in period. The analytic actively learns and measures the ways in which the specified parameters improve production rates. Additionally, the analytic continuously monitors production data and identifies any well liquid loading events. When liquid loading occurs in the wellbore as observed from the production pattern, the analytic automatically shuts in the well to build up pressure and minimizes additional liquid formation. In the field trial, we deployed the edge analytic to monitor gas production and the specified well shut-in and open conditions for 10 different wells in the Haynesville Shale Play. Analyzing each well in the context of approximately 30 intermittent production cycles (shut- in/open), the analytic successfully mapped the surface response, identified the optimal setting for well shut-in/open conditions, and continuously updated the surface response. Overall, the analytic improved production by 4% and reduced the liquid loading occurrences and manual well unloading events by 94%, resulting in an average reduction of approximately 600 tons of CO2 equivalent per well per year. In summary the active learning analytic was developed and deployed on an edge computing platform to 1) optimize intermittent shut-in by searching for the optimum settings that yield the most gas production; 2) automate the optimization process; and 3) monitor the liquid formation for potential loading events. In this paper, we present a use case for an algorithm adapted for the optimization of a dynamic system such as hydrocarbon production from a well.

Inflow Control Devices (ICDs) help reduce the adverse consequences of uneven inflow issues in a lateral completion system. The most common uneven inflow consequences are early water breakthrough and gas coning in water-driven and saturated reservoirs. These issues lead to the dominance of undesired fluid production and consequently, reduced well productivity. Typically, uneven inflow issues are caused by different drivers, including heterogenous permeability, an uneven water saturation profile, and/or complex well completion in a lateral section of a given well. ICDs are placed in permanent positions along the lateral section of a well in order to control zonal production and improve well productivity. The goal of utilizing ICDs is to delay water or gas production and equalize the inflow production from the reservoir to wellbore. However, the uncertainty of reservoir characteristics and operational constraints add complexity to the ICD design and complicate optimization strategies. An optimum ICD design entails identifying the number and size of compartments, packer locations, ICD type, and number of ICDs in each compartment, and the ICD settings such as orifice diameter or flow restriction rating. Extensive reservoir modeling work can be performed to accurately quantify the impact of each ICD design on well production. The intent of this paper is to demonstrate that Bayesian optimization and machine learning techniques can help identify an optimized ICD design in a minimum number of reservoir simulation evaluations. These techniques are implemented into the reservoir simulation workflow to enhance the speed of the analysis and resulting value proposition for the operating customer.Using Gaussian Process Regression as a surrogate, Bayesian optimization makes use of a small number of initial reservoir simulation runs to quantify the uncertainty of the surrogate model in the parameter space. It makes use of an appropriate acquisition function (as determined by the desired exploration-exploitation tradeoff characteristics) to design the next sample (simulation run) to be evaluated. Unlike the ensemble-based optimization algorithms, Bayesian optimization points to the optimum solution sequentially (one evaluation at a time). The proposed workflow automates the optimization process of ICD design evaluation workflow times by 50% in our case studies. The 50% efficiency takes in the time to perform ICD optimization workflow. For instance, the manual iteration ICD design for case study 1 described in this paper was four weeks, which the proposed workflow shortened this time to two weeks.This paper presents two case studies in which the Bayesian optimization technique was used to identify the best ICD completion design. The space parameter in both case studies involves several variables, including the number and location of compartments, the number of ICDs per compartment, and the ICD settings (one such setting, for example, considers orifice diameter size). The goal in the first case study was to find an ICD design that can maximize the net present value over the well lifetime (set to 5 years), while reducing and delaying water production. In this first case study, an 800ft lateral in a horizontal well, with drastic variation of permeability along its lateral length, was considered. In the second case study, 4000ft horizontal length of a well with variations of permeability was analyzed. In this second case, the objective was to extend the life of the well by minimizing the gas-oil ratio and maximizing the oil recovery. The simulation runs stopped after 3 years of production and the best case was chosen based on the aforementioned criteria. In both case studies, the optimization algorithm setup was able to converge to an optimum ICD design within 20 reservoir simulation runs. This alone represents an improvement over the current manual trial and error process in which an expert uses human intuition.

Abstract Optimization problems, such as optimal well-spacing or completion design, can be resolved rapidly via surrogate proxy models, and these models can be built using either data-based or physics-based methods. Each approach has its strengths and weaknesses with respect to management of uncertainty, data quality or validation. This paper explores how data- and physics-based proxy models can be used together to create a workflow that combines the strengths of each approach and delivers an improved representation of the overall system. This paper presents use cases that display reduced simulation computational costs and/or reduced uncertainty in the outcomes of the models. A Bayesian calibration technique is used to improve predictability by combining numerical simulations with data regressions. Discrepancies between observations and surrogate outcomes are then observed to calibrate the model and improve the prediction quality and further reduce uncertainty. Furthermore, Gaussian Process Regression is used to locate global minima/maxima, with a minimal number of samples. To demonstrate the methodology, a reservoir model involving two wells in a drill space unit (DSU) in the Bakken Formation was constructed using publicly available data. This reservoir model was tuned by history matching the production data for the two wells. A data-based regression model was constructed based on machine learning technologies using the same dataset. Both models were coupled in a system to build a hybrid model to test the proposed process of data and physics coupling for completion optimization and uncertainty reduction. Subsequently, Gaussian Process Model was used to explore optimization scenarios outside of the data region of confidence and to exploit the hybrid model to further reduce uncertainty and prediction. Overall, both the computation time to identify optimal completion scenarios and uncertainty were reduced. This technique creates a robust framework to improve operational efficiency and drive completion optimization in an optimal timeframe. The hybrid modeling workflow has also been piloted in other applications such as completion design, well placement and optimization, parent-child well interference analysis, and well performance analysis.

Abstract Given limited CO2 supply, operational constraints, and pattern specific reservoir performance, WAG schedule can be customized such that NPV or other metrics are optimized. Depending on the WAG schedule, recovery can fluctuate between 5–15% at the pattern scale due to reservoir heterogeneity causing variations in sweep efficiency. An analytical method was developed to optimize WAG schedules that couples traditional reservoir modeling and simulation with machine learning, enabling the discovery of optimal WAG schedules that increase recovery at the pattern level. A history-matched reservoir model of Chaparral Energy's Farnsworth Field, Ochiltree County, TX was sampled intelligently to perform predictive reservoir flow simulations and artificially build an intelligent reservoir model that samples a broad range of possible WAG scenarios for optimization. The intelligent model generates the next "best" sample to investigate in the numerical simulator and converges on the optima, quickly reducing the number of runs investigated. Results in this paper demonstrate that there can be significant improvements in net present value as well as net utilization rates of CO2 using this analytical technique. The WAG design generated by the intelligent reservoir model should be deployed in the field in early 2016 for validation. It is intended that the intelligent reservoir model will be updated on a regular basis as injection and production data is obtained. This effort represents the beginning of a paradigm shift in the application of modeling and simulation tools for significant improvements in field production operations.

Abstract As one of the largest unconventional resources in the world, the Bakken petroleum system has approximately 7.4 billion barrels of recoverable oil as estimated by the U.S. Geological Survey in 2013. These resources lie in a 200,000-square-mile area in North America. However, primary recovery rates are low, typically 3%–5%. The use of carbon dioxide (CO2) as an injection fluid may reduce the hydrocarbon viscosities in the reservoir and allow additional trapped oil to be produced. Another benefit of CO2 injection is to mitigate climate change. In this study, detailed petrophysical and geological field models focusing on the Middle and Lower Members of the Bakken Formation in the Bailey and Grenora areas of western North Dakota were developed based on field characterization, well log interpretation, and laboratory core analysis by scanning electron microscopy, ultraviolet fluorescence, and standard optical microscopy techniques. A fine-scale model with a pair of horizontal wells was extracted from the initial Bailey model to examine the potential effectiveness of CO2-based oil recovery techniques. Based on characterization results for the simulation investigation, two three-stage hydraulic fractures were incorporated into the model for CO2 injection and oil production wells, respectively, as well as the characterized oil properties such as swelling test and minimum miscibility pressure measurement from selected reservoirs. A total of four cases for simulation were designed to enhance oil recovery by CO2 injection. These cases were run on a single porosity–permeability model to test the basic concepts and later incorporate these results into a dual porosity–permeability model. The results show that CO2 injection may play a significant role in increasing oil production. Results of this study indicate that production can be enhanced by 43% to 58% compared to cases without CO2 injection. Since a single-porosity, singlepermeability model was used, relative permeability curves that represent the fractures and matrix were both tested to determine the effects of this variable on the outcome. The results indicated that CO2 enhanced oil recovery (EOR) is very sensitive to the relative permeability, and that result varied over 100% between the cases. This case study for CO2 storage in the Bakken Formation provides a basic guideline to address CO2-based EOR. Specifically, the efforts provide modeling input for future investigations that could lead to the design and implementation of a possible pilot-scale field test.

Little work has been done to examine the ability of tight hydrocarbon-bearing formations to serve as targets for CO2 storage and utilization. The Bakken Formation is one of the largest tight oil formations in North America. Laboratory experiments using Bakken rocks and fluids, evaluation of reservoir characterization data, and geomodeling activities were conducted to understand the potential for CO2 utilization and storage in tight oil formations. Results suggest that tight oil formations may serve as targets for geologic storage of CO2 and that the injection of CO2 may be effective in enhancing the productivity of oil from those formations.

The Plains CO2 Reduction (PCOR) Partnership performed a case study on the feasibility of underground carbon dioxide (CO2) storage in the basal saline system of central North America. The calculated volumetric CO2 storage resource potential in this system is 373 Gt. Two dynamic modeling scenarios were designed to address the dynamic CO2 storage capacity and pressure transient. Various strategies were tested including injection well location and spacing, injection optimization, and water extraction during CO2 injection. This study underscores the potential difference in CO2 storage potential between estimates made with volumetric approaches and those made with dynamic methodologies.

A binational effort, between the United States and Canada, characterized the lowermost saline system in the Williston and Alberta Basins of the northern Great Plains–Prairie region of North America in the United States and Canada. This 3-year project was conducted with the goal of determining the potential for geologic storage of carbon dioxide (CO2 ) in rock formations of the 517,000 sq mi Cambro-Ordovician Saline System (COSS). This project was led on the U.S. side by the Energy & Environmental Research Center (EERC) through the Plains CO2 Reduction (PCOR) Partnership and on the Canadian side by Alberta Innovates Technology Futures (AITF). The project characterized the COSS using well log and core data from three states and three provinces and determined its storage potential by creating a heterogeneous 3-D model and determined the effects of CO2 storage in this system using dynamic simulation. The area underlain by the COSS includes several large CO2 sources that each emits more than 1 Mt CO2 /year. Assuming that each of these sources will target the COSS for the storage of their CO2, the primary questions addressed by this study are 1) what is the CO2 storage resource of the COSS, 2) how many years of current CO2 emissions will it be capable of storing, and 3) what will be required and what will be the effect of injecting 104 Mt/yr of CO2 into the COSS?

The Plains CO2 Reduction (PCOR) Partnership, through the Energy & Environmental Research Center, is collaborating with Petroleum Technology Research Centre in site characterization; risk assessment; public outreach; and monitoring, verification, and accounting activities at the Aquistore project. The PCOR Partnership constructed a static geological model to assess the potential volumetric storage capacity of the Aquistore site and provide the foundation for dynamic simulation for the dynamic CO2 storage capacity. Results of the predictive simulations will be used in the risk assessment process to define an overall monitoring plan and assure stakeholders that the injected CO2 will remain safely stored.

A workflow was developed to properly assess the CO2 storage resource potential of a deep saline formation using the methodology proposed by the U.S. Department of Energy. To illustrate the workflow, the Minnelusa Formation of the Powder River Basin is used as an example of how a CO2 storage resource methodology could be applied to a saline formation given varying levels of information. It is important to accurately estimate the effective volumetric CO2 storage resource potential of a target formation, and new storage efficiency values are presented with the workflow.

The focus of this report is to evaluate and discuss geochemical modeling and laboratory studies performed by the Energy & Environmental Research Center to determine potential chemical reactions between CO2, brine, and rock on the portion of the Cambro-Ordovician saline system that occurs in North Dakota, Montana, and South Dakota.