Probabilistic Evaluation of Hydraulic Fracture Performance Using Ensemble Machine Learning

Landslide disaster risk reduction necessitates the investigation of different geotechnical causal factors for slope failures. Machine learning (ML) techniques have been proposed to study causal factors across many application areas. However, the development of ensemble ML techniques for identifying the geotechnical causal factors for slope failures and their subsequent prediction has lacked in literature. The primary goal of this research is to develop and evaluate novel feature selection methods for identifying causal factors for slope failures and assess the potential of ensemble and individual ML techniques for slope failure prediction. Twenty-one geotechnical causal factors were obtained from 60 sites (both landslide and non-landslide) spread across a landslide-prone area in Mandi, India. Relevant causal factors were evaluated by developing a novel ensemble feature selection method that involved an average of different individual feature selection methods like correlation, information-gain, gain-ratio, OneR, and F-ratio. Furthermore, different ensemble ML techniques (Random Forest (RF), AdaBoost (AB), Bagging, Stacking, and Voting) and individual ML techniques (Bayesian network (BN), decision tree (DT), multilayer perceptron (MLP), and support vector machine (SVM)) were calibrated to 70% of the locations and tested on 30% of the sites. The ensemble feature selection method yielded six major contributing parameters to slope failures: relative compaction, porosity, saturated permeability, slope angle, angle of the internal friction, and in-situ moisture content. Furthermore, the ensemble RF and AB techniques performed the best compared to other ensemble and individual ML techniques on test data. The present study discusses the implications of different causal factors for slope failure prediction.

Accurate modeling of hydrocarbon production is a necessary, yet challenging step for economic exploitation of unconventional resources. One of the main challenges is to model flow to a horizontal well from a complex network of hydraulic and natural fractures. Many unconventional reservoirs comprise well-developed natural fracture networks with multiple orientations and complex hydraulic fracture patterns based on microseismic data. Conventional dual porosity and dual permeability models are not adequate for modeling these complex networks of natural and hydraulic fractures. Also, it is neither practical nor advantageous to model a large number of pre-existing fractures with a discrete fracture model. Therefore, an appropriate approach to model production from low-permeability reservoirs is to perform discrete fracture modeling for hydraulic fractures and employ a dual continuum approach for numerous natural fractures. We have developed a coupled dual continuum and discrete fracture method to simulate production from unconventional oil and gas reservoirs. Large-scale hydraulic fractures (macro-fractures) are modeled explicitly using a discrete fracture model, called EDFM, and numerous small-scale natural fractures (micro-fractures) are modeled using a dual continuum approach. The hybrid model includes three domains: matrix, discrete-fracture, and continuum-fracture domains. A systematic approach is devised to calculate transport parameters between all three domains. Moreover, EDFM allows for not only transverse and longitudinal hydraulic fractures but also macro-fractures of any orientation. Thus, the coupled model provides an effective and reliable environment to improve stimulation designs and completion strategies. We present several examples in this study to show the applicability, robustness, and performance of the hybrid method for the simulation of unconventional oil and gas reservoirs. We examine multi-stage hydraulic fractures with multiple configurations in the presence of numerous pre-existing fractures. Simulations show a noticeable contribution from natural fracture networks on total production. Furthermore, for the tight oil reservoir examined in this study, the stimulation scheme with longer hydraulic fractures improves cumulative oil production compared to the scheme with larger number of shorter hydraulic fractures. We also examine production from a tight gas reservoir wherein hydraulic fractures partially penetrate the formation height. Simulations indicate that inefficient fracture treatment can result in significant loss of production.

Exploiting the vast quantities of in-place oil in some of China's large sedimentary basins will require understanding and identifying where new technologies and techniques can be applied to maximize production. Exploration in the Ordos basin in north-central China has demonstrated that tight oil reservoirs, such as the Chang 7 formation, are widely distributed; however, they remain relatively underdeveloped. These reservoirs are buried below the conventional oil layers, differ from them by geological composition, and their thin pay zones and low permeability present development challenges. As part of a standard exploration and appraisal program, multiple vertical wells have been drilled to characterize the Chang 7 reservoir quality, and many were completed with only one hydraulic fracturing stage. Unfortunately, average hydrocarbon production was often low, and horizontal well completions with multiple transverse hydraulic fracturing treatments were required to obtain commercial flow rates. Part of the integrated workflow developed for the stimulation program requires the capture of reservoir, geological, and geomechanical properties for use in the design of the stages. The perforations were designed to allow for effective fracturing treatments across the lateral and were placed using the abrasive perforating technique. Additionally, one of these vertical wells was used for real-time microseismic fracturing monitoring to evaluate fracture geometry, fracture orientation, and azimuth. Real-time technology and an integrated workflow effectively increased the understanding of where, when, and how to deploy hydraulic fracturing treatments in the lateral section of the study well. This optimization has resulted in more than an 8-fold increase in production compared to that from vertical wells. Considered a breakthrough, this project has resulted in the planning for additional horizontal wells.

Worm gearboxes are popular across various industrial applications since they offer significant gear ratios in small installation spaces. Despite having multiple advantages, worm gearboxes are subjected to higher friction due to sliding design and are prone to damage and increased transmission error over the period of operation. Delayed diagnosis of worm gearbox degradation can lead to low-quality products and/or unnecessary production line downtime. Using vibration characteristics of worm gearbox, it is possible to determine the fault and transmission error at a given period in time. In this paper, an ensemble machine learning model is trained and deployed to monitor the transmission error of worm gearbox and classify between new, operational and old conditions. 1D CNN (one dimensional convolutional neural network) model is used to automatically extract features in vibration signal of X, Y, and Z axes and predict the relevant state of worm gear. The proposed technique uses ensemble machine learning technique fusion of features extracted by multi-layer 1D CNN for three axes vibration data. The proposed method could achieve 96% accuracy and performs significantly better than traditional sequential and ensemble machine learning models on a dataset of with 7,870 samples with 800 samples labeled as new condition, 4,740 samples as operational and 2,330 as old.

Numerical investigation of limitations and assumptions of analytical transient flow models in tight oil reservoirs

Ensembles are known to reduce the risk of selecting the wrong model by aggregating all candidate models. Ensembles are known to be more accurate than single models. Accuracy has been identified as an important factor in explaining the success of ensembles. Several techniques have been proposed to improve ensemble accuracy. But, until now, no perfect one has been proposed. The focus of this research is on how to create accurate ensemble learning machine (ELM) in the context of classification to deal with supervised data, noisy data, imbalanced data, and semi-supervised data. To deal with mentioned issues, the authors propose a heterogeneous ELM ensemble. The proposed heterogeneous ensemble of ELMs (AELME) for classification has different ELM algorithms, including regularized ELM (RELM) and kernel ELM (KELM). The authors propose new diverse AdaBoost ensemble-based ELM (AELME) for binary and multiclass data classification to deal with the imbalanced data issue.

Flowback rate transient analysis (RTA) is a practical tool for characterizing hydraulic fracture (HF) properties. However, the accuracy of the interpreted results from flowback RTA is challenged by the complexity in two-phase flow in the hydraulic fracture and matrix system. Accordingly, we present a new semianalytical method to characterize HF attributes and dynamics using multi-phase flowback data for tight and ultratight (shale) oil wells. The proposed method includes a two-phase diagnostic plot, a fracture RTA approach for straight-line analysis, and a matrix model capable of characterizing water and oil flow. The RTA approach is based on fracture infinite acting linear flow (IALF) and boundary dominated flow (BDF) solutions, which treats HF as an open tank with a variable production rate at the well and the contribution of water and oil from matrix within the distance of investigation (DOI). The pressure-dependent fluid and geomechanical properties, such as permeability and porosity, are considered in the pseudotime defined in fracture and matrix to reduce the nonlinearity of the system. We tested the accuracy of the proposed method against numerical results obtained from commercial software and verified its applicability by analyzing the flowback and long-term production data from a field example in Eagle Ford shale. The validation results confirm that our method can closely calculate water and oil influx from matrix as well as the average pressure and saturation in the HF and matrix DOI. The accurate estimation of the initial fracture permeability and pore volume demonstrates the applicability of the proposed method in quantifying HF properties from two-phase flowback data exhibiting fracture IALF and BDF regimes. The analysis results show that the estimated initial fracture pore volume shows more accuracy than initial fracture permeability due to the different calculation sources in the straight-line analysis. In short, the proposed method is, to our best knowledge, the first RTA approach incorporating the two-phase water and oil influx from matrix into the inverse analysis of fracture properties and dynamics using straight-line analysis, instead of history matching

Horizontal wells and hydraulic fracturing are the key technologies that allow commercial production from tight oil and gas reservoirs. However, rigorous analysis of production data from these reservoirs requires incorporation of the impacts of stress-dependent permeability and multi-phase flow. Changes in the stress state of the system during production may reduce the absolute permeability. Furthermore, gas phase formation and flow in presence of supersaturated oil phase affects fluid dynamics in tight oil reservoirs. This study provides a rigorous methodology for incorporation of the effects of non-static permeability and multi-phase flow in rate transient analysis (RTA) of tight oil and gas reservoirs producing at variable rate/flowing pressures during transient linear flow period. Analytical solutions for the approximate linearized form of the flow equation have been widely used as the basis for RTA tools for conventional reservoirs during transient flow period. However, they lead to considerable error when applied to tight oil and gas reservoirs. In particular, during the transient linear flow period, the slope of the square -root-of-time plot obtained from numerical solution differs from the slope calculated by analytical methods. Efforts have been made by some researchers to obtain a correction factor from the numerical solution of the flow equation to correct the slope of the square-root-of-time plot for single phase flow of gas during transient linear flow period. In this study, an iterative method is used for evaluation of the slope correction factor in the presence of multi-phase flow and non-static permeability for constant-pressure production during transient linear flow period. Further, the correction factor is used for analysis of production data from tight oil and gas reservoirs producing at variable rate/flowing pressures. The correction factor is used in the analysis of different sets of synthetic production data for tight oil and gas reservoirs. The results show that the correction factor can reduce/eliminate the considerable errors associated with the conventional analytical methods in initial permeability estimation. For multi-phase flow cases, the producing gas -oil ratio (GOR) is used to estimate the oil saturation-pressure relationship in the reservoir, which is required for calculation pseudo-pressure and the correction factor. The method developed in this study alleviates the need for using numerical simulation models to generate empirical correlations for the correction factor for the square -root-of-time plot. The easy-to-implement iterative procedure of this method only requires the pressure dependencies of the constituent elements of the hydraulic diffusivity. Therefore, this method is applicable for analysis of production profiles for variety of reservoirs with nonlinear flow equations.

Since cyberbullying impacts both individual victims and entire society, research on abusive language and its detection has attracted attention in recent years. Because social media sites like Facebook, Instagram, Twitter, and others are so widely accessible, hate speech, bullying, sexism, racism, aggressive material, harassment, poisonous comments, and other types of abuse have all substantially increased. Due to the critical requirement to detect, regulate, and limit the spread of harmful content on social networking sites, we conducted this study to automate the detection of offensive language or cyberbullying. We created a new Arabic balanced data set to be used in the offensive detection process because having a balanced data set for a model would result in improved accuracy models. Recently, the performance of single classifiers has been improved using ensemble machine learning. The purpose of this study is to examine the effectiveness of several single and ensemble machine learning algorithms in identifying Arabic text that contains foul language and cyberbullying. Applying them to three Arabic datasets, we have selected three machine learning classifiers and three ensemble models for this aim. Two of them are offensive datasets that are readily accessible in the public, while the third one was created. The results showed that the single learner machine learning strategy is inferior to the ensemble machine learning methodology. Voting performs is the best performing trained ensemble machine learning classifier, outperforming the best single learner classifier (65.1%, 76.2%, and 98%) for the same datasets with accuracy scores of (71.1%, 76.7%, and 98.5%) for each of the three datasets used. Finally, we improve the voting technique’s performance through hyperparameter tuning on the Arabic cyberbullying data set.

Recent advances in well design and production techniques have brought considerable attention to exploitation of tight (low permeability, absolute permeability <1 mD) oil resources. Drilling of long horizontal wells and deployment of hydraulic fractures along these wells (multi-fractured horizontal wells) can substantially improve the primary production rates from such reservoirs. Nevertheless, the low effective permeability of these reservoirs to oil hinders the sustainability of favorable oil rates and at some point applying some EOR technique becomes inevitable. In the current study, CO2 miscible flooding and WAG process in a tight oil reservoir are investigated. Although several studies have investigated different aspects of the process in conventional oil plays, the design of an effective scheme in tight oil formations is more complex. These complexities are related to the proper design of the fractures (half-length, conductivity, orientation (transverse vs. longitudinal), etc.) and their relative placement along producers and injectors and the operational constraints on each well or segment of the well. In this work, we utilize an EOR scheme design where multi-fractured horizontal wells are used for both injection and production, and the hydraulic fracturing stages are staggered to delay breakthrough and improve sweep efficiency. For a set of defined parameters, compositional simulations are conducted to investigate the effect of the CO2 slug size, WAG ratio and cycle length on the recovery efficiency of the model. The recovery from the aforementioned EOR process is then compared with its corresponding base case in which the reservoir has gone through periods of primary and water-flooding stages. The results of this study show that the incremental oil recovery from WAG process in tight formation can reach as high as 20%.

A review of the current status of induced seismicity monitoring for hydraulic fracturing in unconventional tight oil and gas reservoirs

As a result of current commodity price differentials, North American resource development has shifted towards unconventional liquids-rich and light tight oil plays. Due to the low permeability of these plays, extensive hydraulic fracturing is commonly required for commercial development. Operators are seeking new methods to characterize hydraulic fractures, particularly early in the well life. One such method is to utilize regularly gathered high frequency (hourly or greater) fluid production and flowing pressures to model the flowback process. Previous studies have suggested that this data can be quantitatively analyzed to estimate hydraulic fracture half-length and other fracture properties. Flowback of multi-fractured horizontal tight oil wells, stimulated with water-based fluids, commonly exhibit two distinct segments: a) a short period of single-phase water flow, which continues until breakthrough of formation fluids; and b) multi-phase flow (water, oil and gas) following breakthrough. The first flow regime observed in data collected at common frequencies typically consists of fracture storage/depletion of fracture fluid. This flow regime is followed by the breakthrough of hydrocarbon and formation water which results in a deviation from the depletion signature. The analysis procedure used for analyzing this data builds upon the analytical history-matching methodology presented by Clarkson and Williams-Kovacs (2013b). Consistent with the previous work we model the first segment as a single phase depletion of the fracture pore volume, from which a pre-breakthrough estimate of fracture permeability and half-length can be determined. The second segment is modelled by assuming transient linear flow of oil and formation water to the fracture, under the assumption of perfect displacement of frac water by formation fluids. From the second segment we are able to estimate long-term effective fracture permeability and half-length. However, as pointed out by Clarkson and Williams-Kovacs (2013b) there is a large degree of uncertainty in this type of analysis as a result of the number of unknowns which are being adjusted to provide an adequate history-match. To better understand the uncertainty and the impact of each parameter, stochastic simulation will be used to provide a range of parameter values, which provide an adequate fit of the data, and to determine which parameters have the greatest impact on the match. Additional improvements over the previous work include the consideration of different fracture geometries, the use of several fracture models to estimate fracture permeability, modeling produced water salinity to track the contribution of formation water and salt dissolution and additional constraints on relative permeability curve selection. The field case presented by Clarkson and Williams-Kovacs (2013b) for a prolific light tight oil reservoir is reanalyzed, along with a second well from the same pad for proof of concept and demonstration of the developed techniques.

Conformance control study by micrometer sized dispersed particle gel in three-dimensional tight oil fracture network

Summary Shale oilgains much attention as an unconventional resource, which has a potential of significant oil production. Moreover, the technology of horizontal drilling and hydraulic fracturing has encouraged the exploration and development of shale oil reservoirs. An optimal hydraulic fracturing design is essential to achieve a high oil production rate. However, industry is still faced with many challenging issues in hydraulic fracturing optimization, especially under geological and economic uncertainties. These uncertainties can result in a significant increase in capital costs and negative effects on project revenue. This paper presents innovative hydraulic fracturing robust optimization fordevelopment of unconventional reservoirs to reduce the current challenges and minimize the future investment expenses. We first address a sensitivity analysis for creatingmultiple scenerios of a simulation model, consideringuncertainties of geology and reservoir properties. Most of the previous hydraulic fracturing optimization workflows often ignoredthese uncertainty parameters; therefore, the conventional optimal solution obtained based on a single realization may deviate severely from the actual optimality. To capture geological and reservoir uncertainties, a large number of geological reservoir realizations must be generated honoring geological constraints. Hydraulic fracturing parameters (e.g., fracture half-length, fracture spacing, and fracture height), injection composition and operating conditions are optimized simultaneously based on these multiple geological realizations. The sensitivity analysis results indicate that the four parameters that strongly affect the shale oil production include matrix permeability, fracture half-length, fracture spacing and rock compressibility. To maximize oil recovery, a series of physics-based optimizations have been performed for a hydraulic fracturing design by applying the Partical Swarm Optimization algorithm in a robust optimizer. Both the conventional optimization, based on a single geological realization and robust optimization, based on multiple geological realizations have been conducted. After the conventional optimization process, the ultimate oil recovery yields about 18.63% OOIP. Then the robust optimization procedure is applied with 210 realizations. Due to the high computational cost of shale oil simulations, it is impractical to use all generated models in the robust optimization workflow. Thus we propose a method for the robust optimization of hydraulic fracturing (HF) and gas injection composition based on a small set of geological realizations that represent the overall uncertainty of a reservoir. This is achieved by ranking all realizations according to the performance of each realization in terms of cumulative oil production. The comparison shows that the robust optimization increases all the NPV percentile values. This indicates the increased robustness of the optimal solution under uncertainties and thus an adoption of the robust optimization procedure can significantly reduce the project risk. This research is to present an innovative optimization approach which is fast and robust and can be used to efficiently capture the uncertainties of the key factors including geology, reservoir properties, and economic variation.

ABSTRACTDeep tight sandstone oil and gas reservoirs are exerting an increasingly crucial role in the augmentation of fossil energy reserves and the provision of energy. On account of the intricate geological conditions and the deficiency of a comprehensive set of exploration and development engineering technologies as well as supporting processes, the present development of deep tight sandstone oil and gas reservoirs remains in its nascent stage. Through the analysis and generalization of the horizontal well development technology for deep tight sandstone oil and gas reservoirs, a series of technologies have been established, encompassing reservoir geological evaluation and modeling, horizontal well development reservoir engineering validation, horizontal well geological design, and enhanced oil recovery processes. By taking the C 6 reservoir in Ordos basin, China as the research subject, in light of the research outcomes regarding the damage mechanisms and potential damage factors of tight sandstone oil and gas reservoirs, a geological evaluation approach based on the well log response characteristics was constructed, clarifying the porosity and permeability features of the C 6 reservoir, establishing the numerical model of the oil reservoir, and further elaborating the methods for dividing the development layers, selecting the development well pattern, and determining the development well density. The design parameters of horizontal well‐segmented hydraulic fracturing were meticulously optimized, resulting in a minimum cluster spacing of 7 m and a maximum cluster spacing of 20 m. Given the influence of horizontal stress differences, the optimum fracturing density was ascertained to be 16 perforations per meter, and the optimal fracturing fluid volume was identified through simulation to range from 12 to 25 m3/m. The crucial technologies for the development of tight sandstone oil and gas reservoirs in horizontal wells have been clearly identified, offering theoretical direction for the efficient exploitation of deep tight sandstone oil and gas reservoirs.