Building a predictive statistical model for evaluating the impact of various fracture treatment and well completion designs on production has been of great interest in the oil and gas industry. The objectives of this study were to evaluate the benefits of advanced statistical and machine-learning techniques for predicting production from oil wells, highlight the strengths and weaknesses of these techniques, and gain insight into the relationship between well parameters and production. The predictive models are described through mathematical functions or algorithms that rely on well data (training set). The ongoing dilemma is that these models often result in poor predictions, even if they result in a high R-squared (0.7 or higher). The new perspective that this study brings is the importance of cross-validation with "hold-out" datasets in the workflow to develop reliable statistical models. A database of available completion and production data has been assembled from the North Dakota Industrial Commission (NDIC) and Frac Focus websites and from internal completion documentation. To date, there are at least 6,800 horizontal wells completed in the Middle Bakken formation and 3,600 completed in the Three Forks formation on the North Dakota side of the Williston Basin. Various models such as multiple regression, random forests, and gradient boosting machine were built to predict the cumulative oil production of the Middle Bakken and Three Forks horizontal wells. Model predictive abilities were assessed by cross-validating the root mean squared errors (in cross-validation, a hold-out set was used to assess the modelis predictive ability). The results showed the following conclusions about statistical evaluation techniques: 1) regression models that account for overfitting provided the best predictive ability, 2) gradient boosting model with the highest R-squared value had the worst predictive ability for the specific datasets in this paper— which shows why it is critical to not rely solely on R-squared value to assess a modelis predictive ability, but to also perform cross-validation, and 3) random forests and gradient boosting machine can be used for determining variable importance. Moreover, we observed that there is statistical evidence to support the presence of important interactions among variables in predicting cumulative oil production. For the Middle Bakken and Three Forks wells included in this study, the results showed that water cut, which can be used as a proxy for reservoir quality, is the most important predictor for cumulative oil production. However, the most important completion-related variables for predicting oil production were total frac fluid and proppant pumped. The analysis and results presented in this paper will enable companies to apply the approach to their own data when building production prediction models and analyzing the complex relationships of variables that control well performance.

Summary Selecting appropriate proppants is an important part of hydraulic-fracture completion design. Proppant selection choices have increased in recent years as regional sands have become the proppant of choice in many liquid-rich shale plays. But are these new proppants the best long-term choices to maximize production? Do they provide the best well economics? The paper presents a brief historical perspective on proppant selection followed by various detailed studies of how different proppant types have performed in various unconventional onshore US basins (Williston, Permian, Eagle Ford, and Powder River), along with economic analyses. As the shale revolution pushed into lower-quality reservoirs, the concept of dimensionless conductivity has pushed our industry to use ever lower-quality materials—away from ceramics and resin-coated proppant to white sand in some Rocky Mountain plays, and more recently from white sand to regional sand in the Permian and Eagle Ford plays. Further, we compare early-to-late-time production response and economics in liquid-rich wells where proppant type changed. The performance of various proppant types and mesh sizes is evaluated using a combination of different techniques, including big-data multivariate statistics, laboratory-conductivity testing, detailed fracture and reservoir modeling, as well as direct well-group comparisons. The results of these techniques are then combined with economic analyses to provide a perspective on proppant-selection criteria. The comparisons are anchored to permeability estimates from production history matching and diagnostic fracture injection tests (DFITs) and thousands of wellsite-proppant-conductivity tests to determine dimensionless conductivity estimates that best approach what is obtained in the field. Dimensionless fracture conductivity is the main driver of well performance because it relates to proppant selection thanks to the inclusion of the relationship of fracture conductivity provided by the proppant relative to the actual flow capacity of the rock (the product of permeability and effective fracture length), which is supported by the production analyses in the paper. The paper shows how much fracture conductivity is adequate for a given effective fracture length and reservoir permeability and then looks at the economics of achieving this “just-good-enough” target conductivity, either through less proppant mass with higher-cost proppants or more proppant mass with lower-cost proppants, as well as mesh-size considerations. This paper does not rely on a single technique for proppant selection but uses a combination of various data sources, analysis techniques, and economic criteria to provide a more holistic approach to proppant selection.

Summary In the Williston Central Basin, a well-completion design has a significant effect on well productivity and ultimate recovery. More than 12,000 horizontal wells have been drilled and completed while completion practices continue to vary widely across the basin. Several companies have adopted slickwater-only designs, whereas others have dramatically increased proppant mass. Completion strategies have differed depending on the area in the basin. The objective of this paper is to discuss the effect of various completion changes in the Central Basin and determine which particular change delivers the most “bang for the buck” using a metric of dollars spent per barrel of oil (USD/BO). The approach centered around multivariate analysis (MVA) from an extensive petrophysical/completion/production database to verify what completion and petrophysical parameters independently drive production in different areas. Although MVA has been used by the authors and many others before, statistical models are limited by their ability to provide predictive relationships (mostly simple linear regressions, and unreliable beyond the data range). This paper provides a novel hybrid approach that uses calibrated relationships from physics-based modeling (combination of fracture and numerical reservoir modeling) between completion parameters and production response in combination with statistical MVA results. Specifically, the physics-based model is calibrated or “history matched” to a measured production/completion-parameter response as provided by MVA, thus delivering a constrained and more physically realistic production response to suggested completion changes. This model is then coupled with a completion-cost model to determine which completion method is the most effective to lower USD/BO. Many common completion-parameter changes, such as increasing stage intensity, moving to plug-and-perforate cemented-well designs, increasing injection rate, and increasing proppant mass per lateral foot and fluid volume per lateral foot, have a positive effect on production and are advantageous to lower USD/BO in all areas of the Middle Bakken and Three Forks. The new hybrid MVA approach indicates that pumping slickwater treatments with average proppant concentrations of 1 lbm/gal and treatment sizes from 545 to 750 lbm/ft at pump rates approaching 100 bbl/min through a stage length of 200 ft (50 stages for a 10,000-ft lateral) might be the economic optimum, provided there are no significant well-communication issues.

No abstract

No abstract