Search

For an oil well to be said to have been successfully and conclusively drilled, the drilling fluid lies at the heart of the solution. Therefore, the guarantee to solving issues in oil well drilling is to contrive an optimal drilling fluid. However, there is usually a complex interplay of factors involved during drilling fluid formulation, property determination, its performance in the well and its relationship with other wellbore drilling parameters. This is so because drilling muds exhibit time dependent properties. This time dependency is the direct product of the synergy among the various active additives that make up the mud and the characteristic of each additive especially at downhole conditions where the effects of temperature and pressure are well pronounced. These additives are more often than not diverse in size, chemical activity, density and surface energy. Deriving knowledge from the data from these parameters in order to develop a functional relationship between them is a challenging task requiring advanced modelling techniques as well as human intuition and experience. The dependence on human intuition and on the experiential knowledge of professional mud engineers lays bare the shortcomings of traditional mud design techniques. Artificial intelligence techniques have been shown to alleviate this challenge. Exploiting the abundant literature on the various applications of artificial intelligence in oil and gas operations, several works that show how and what artificial intelligence techniques are used in the drilling fluid industry, and what have been achieved due to their use have been selected. In this paper, a review of existing artificial intelligence techniques and their applications in drilling fluid engineering is given. This paper also dug up and analyzed the strengths and pitfalls of each artificial intelligence technique. The examination of the strengths and deficiencies was done using the following virtues as the basic criteria: robustness against noise, self-organization, generalization ability, data volume requirements and the convergence speed. The artificial intelligence techniques presented in this paper include: artificial neural networks (ANNs), fuzzy logic, support vector machines (SVM), hybrid intelligent systems (HIS), genetic algorithms (GA), case based reasoning (CBR) and particle swarm algorithm (PSA). An overview of the applications of classical artificial intelligence in drilling fluid engineering is also presented. From the review, it was gathered that the ANN technique is the most widely applied in drilling fluid engineering accounting for over 54% of the papers reviewed; while lost circulation problem was the most predicted well problem related to drilling fluids accounting for over 17% of the mud problems predicted. It was also observed that a blend of AI techniques performed better than when each one of the AI techniques was used singly. Finally, judging the AI techniques on the criteria mentioned above, ANN was found to meet all the listed criteria except for its slow speed of convergence while ANN, GA, SVM and fuzzy logic were all found to be robust against noise. [ABSTRACT FROM AUTHOR]

Multiphase flowmeters are important to monitor oil wells, as they allow operators to obtain a real-time estimate of the production. However, due to its high installation cost, uncertainty, and reading errors, the oil and gas industry started to invest in virtual meters, mathematical models that can obtain the oil and gas flow forecast using the instrumentation already available in the well, such as pressure and temperature sensors. The purpose of this article is to develop a virtual multiphase flow predictor using artificial intelligence models such as Long Short-Term Memory (LSTM), Gate Recurrent Units (GRU), Multi-Layer Perceptron (MLP) neural networks and deep learning. As input data, this paper uses a public database available on the Oil and Gas Authority (UK) to train the models and demonstrate the possibility of predicting multiphase flow with a reasonable error margin, proving that the method is efficient in estimating well production rates. • In the oil industry, it is important to measure the production from each well. • A Virtual Meter is a low-cost alternative. It can measure multiphase flow without any additional equipment. • This paper proposes a Virtual Meter using neural networks such as MLP, LSTM, GRU and deep learning combinations. • Prediction performance was evaluated using the Oil and Gas Authority (OGA) public database and it shows promising results. • Experiments using a combination of multiple layers shows this type of prediction takes advantage of deep learning techniques. [ABSTRACT FROM AUTHOR]

Shear wave velocity is a critical parameter for the characterization of hydrocarbon reservoirs. Compared with compressional wave velocity which almost exist in every well, shear wave velocity hasn't been recorded in those days for the older wells due to the lack of logging equipment or limited funds. Furthermore, measuring shear wave velocity is fairly time- and money-consuming as it can only be gained by the analysis of core samples conducted in the lab or from the dipole sonic imager (DSI). To cope with the above puzzles, a new methodology, by integrating extreme learning machines (ELM) and technique of mean impact value (MIV), is proposed in this paper with the support of log data collected from two wells located in unconventional shale gas reservoir in Ordos Basin, China. Based on mean impact value, a well-trained ELM model is taken to identify optimal well logs and five well logs are identified to provide the significant and valid information for the estimation of shear wave velocity. 3201 data points collected from well L4 are used in constructing the model. Compared with artificial neural network used in the former studies Levenberg-Marquardt algorithm (ANN-LM), the ELM model's prediction accuracy has been evaluated, the results indicates that the ELM model outperforms ANN-LM model with faster calculating speed and better performance. Additionally, the efficiency of the model proposed here is well investigated by using another well with 3201 data points. Through the comparison among ELM, support vector regression (SVR), convolutional neural network (CNN) and four widely known empirical formulas, what can be concluded is that ELM model is more efficient in fast calculation and high precision. From the result, it can be demonstrated that the use of ELM model, combined with the analysis of mean impact value (MIV), is a more efficient and promising method for shear wave velocity estimation process from conventional well log data, which can be recognized as a promising tool with an extended application. And this research can be applied into a software system for rapid acquisition of shear wave velocity logs. Image 1 • A powerful tool is proposed for shear wave velocity estimation. • The improvement in Vs estimation using AI methods. • It is proved that each intelligence model with selected well log inputs has higher accuracy. • ELM model is more efficient in fast calculation and high precision. [ABSTRACT FROM AUTHOR]

The noise intensity of drilling mud pulse signal is large. It is difficult to recognize signal at once because of the low signal to noise ratio (SNR) of the detected signals. Moreover, the SNR has a low accuracy. A method which is called stacked wavelet autoencoder (SWAE) for recognizing mud pulse signals based on deep learning is proposed in this paper. The model is composed of wavelet neural network and autoencoder, which are trained for mud pulse signal classification specifically. Combining with the drilling mud pulse signal, the recognition performance of the typical data set is analyzed and tested. SWAE enhances the SNR of output signal by using signal detection method. And then the output detected signal is considered as a signal classification attribute. Finally, experimental results show that SWAE is suitable for mud pulse signal recognition, and it has strong ability to extract features from samples automatically and robustness. Experimental results confirm that the performance of SWAE is better than the methods in the state of the art under the same SNR. [ABSTRACT FROM AUTHOR]

This paper develops a novel AI and data-driven predictive model to analyze and forecast energy markets, and tests it for gasoline demand of Saudi Arabia. The AI model is based on a genetic algorithm (GA), artificial neural network (ANN), and data mining (DM) approach for time-series (TS) analysis, referred to as GANNATS. The GANNATS predictive model was successfully designed, trained, validated, and tested using real historical market data. Results show that the model yields accurate predictions with robust key performance indicators. A double cross-validation of the model verified that Saudi Arabia's gasoline demand declined by 2.5% in 2017 from its 2016 level. The model forecasts that Saudi gasoline demand will maintain a mild growth over the short-term outlook. Variables impact and screening analysis was performed to identify the influencing factors driving the gasoline demand. The recent decline in Saudi gasoline demand is primarily attributed to the improvements in vehicle efficiency, lifting of fuel price subsidies, declining population growth, and changes in consumer behavior. This paper enriches existing knowledge of best practices for forecasting domestic and global gasoline demand. In addition, the methodology presented improves on traditional econometric models and enhances the predictability and accuracy of forecasts of gasoline demand. • Presents a novel predictive model for energy markets. • Provides GANNATS modeling development framework for time series applications. • Applies GANNATS model to a case study of modeling, analyzing and predicting gasoline demand in Saudi Arabia. • Verifies the predictive model with a cross-validation process for Saudi gasoline demand. • Identifies the influencing and impactful factors driving the gasoline demand of Saudi Arabia. • Provides forecast and analyses for Saudi gasoline demand. [ABSTRACT FROM AUTHOR]

Cyclic steam stimulation (CSS) is one of the more popular EOR techniques due to the existence of giant heavy oil reserves existing in different parts of the world. Numerical reservoir simulation plays a critical role in investigating the mechanisms and optimizing field development strategies of CSS processes. An artificial neural network (ANN) based model is considered to be a powerful subsidiary tool of high fidelity models for its fast computational speed and reliable prediction capability. This work focuses on the development of a robust surrogate model as a screening/design tool for cyclic steam injection processes using artificial neural network technology. The major contribution of this work includes training of the ANN model using a network topology optimization workflow to help the ANN better understand the complex data structures that are encountered in such processes. The developed ANN model successfully incorporates rock-fluid properties such as relative permeability and temperature dependent viscosity as input parameters together with the other relevant data. Last but not least, the network model utilizes a hybrid structure to adapt to the automatic cycle switching scheme that can be encountered in cyclic steam injection processes. The paper shows that the ANN model can be employed both as a classification tool and a nonlinear regression tool. The model is validated via extensive blind tests against high fidelity simulation models and can be used as a powerful screening and process design tool in global optimization of the process. [ABSTRACT FROM AUTHOR]

Drilling wells in challenging oil/gas environments implies in large capital expenditure on wellbore's construction. In order to optimize the drilling related operation, real-time decisions making have been put in place, so that prediction of rate of penetration (ROP) with accuracy is essential. Despite many efforts (theoretical and experimental) throughout the years, modeling the ROP as a mathematical function of some key variables is not so trivial, due to the highly non-linearity behavior experienced. Therefore, several researches in the recent years have been proposing to use data-driven models from artificial intelligence field for ROP prediction and optimization. This paper presents an extensive review of the literature on ROP prediction, especially, with machine learning techniques, as well as how these models can be used to optimize the drilling activities. The ROP models are classified as traditional models (based on physics-models), statistical models (e.g. multiple regression), or machine learning methods. This review enables to see that machine learning techniques can potentially outperform in terms of ROP-prediction accuracy on top of traditional or statistical models. Throughout this work, an extensive analysis of different ways of obtaining ROP models is carried out, concluding with different strategies adopted in literature to perform data-driven model optimization. Despite the saving potential which can be achieved with real-time optimization based on data-driven ROP models, it is noticeable that there is a lack of implementation of those techniques in the industry, as per literature review. To take a step forward in real implementations, the petroleum industry must be aware that yet no rule of thumb already exists on this specific area, but still, good and very reasonable results can be achieved by following the best practices identified in this review. In addition, the modern practices of machine learning provide promising guidelines for implementing projects in oil and gas industry. • An extensive review on ROP models is performed, focusing on machine learning methods. • Machine learning algorithms offers a flexible and accurate way to predict ROP. • Several strategies for data-driven modeling are covered, including their application. • This papers investigates also how ROP models can be used for drilling optimization. [ABSTRACT FROM AUTHOR]

Interwell Stratigraphic Correlations Detection (ISCD) guides reservoir modeling and oil development. Many existing AI (artificial intelligence) methods have been proposed for ISCD. However, it is difficult to generate labels for large-scale geological data, which leads to the problem of small samples. In this paper, we propose a few-shot learning-based approach to detect stratigraphic correlations for overcoming this challenge. Specifically, we design a Knowledge Enhanced Few-shot Transformer ISCD model (KEFT-ISCD) to enhance reservoir sample features. We design a dynamically balanced marginal softmax (d b m - s o f t m a x) to further optimize the model loss for identifying edge features, which improves the stratigraphic matching effects. In addition, we design a bi-window co-sliding approach to address the cross-matching problem in practical stratigraphic matching. To the best of our knowledge, this is the first work to use few-shot learning for the ISCD. We evaluate the proposed method with different well sections in a pair of adjacent wells from a real-world well logging dataset. Experimental results indicate that the proposed KEFT-ISCD performs well and achieves a detection accuracy of 91.12%. We also conduct experiments on different wells and blocks. The results further demonstrate the generalizability of the proposed approach. • A Few-shot learning for interwell stratigraphic correlation detection is developed. • A knowledge-enhanced strategy is proposed to provide similar layer sample dimensions and enhance features for further geological studies. • A dynamically balanced marginal softmax is designed to identify edge features. It identifies fine-grained features of the reservoir. • Results on real-world data show that our method improves the accuracy of stratigraphic correlation detection in small sample cases. [ABSTRACT FROM AUTHOR]

This paper (Part II) reports on the development of a Formation Damage Expert System (FODEX) for aiding in the preliminary diagnosis of formation damage types that consist of emulsions, water blocks, wettability reversal and the deposition of asphaltenes and wax in hydrocarbon reservoirs. The reasoning process is based on compiled rules that have been arranged in decision trees constructed using knowledge derived from standard industrial practices integrated with empirical models to assure vigorous expert opinion. The structured approach for formation damage diagnosis has been detailed in Part I of this work. FODEX automates the reasoning process embedded in the decision trees using logic blocks represented as four modules. Fuzzy logic has been used in handling incomplete and conflicting knowledge encountered in determining the likelihood of asphaltene and wax deposition. The developed formation damage expert system has been tested with three documented field cases of producing wells in the Rangely, Typhoon, and Magwa–Marrat oil reservoirs. FODEX decisions regarding the type of damage inflicted on these fields have been validated with compositional PVT simulation software. The expert system made a thorough diagnosis of damage types, in agreement with PVT simulations, field observations, PVT cell testing and special core analysis. [ABSTRACT FROM AUTHOR]

Conventional models and artificial intelligence (AI)-based models have been widely applied for energy consumption forecasting over the past decades. This paper reviews conventional models and AI-based models in energy consumption forecasting, and discusses the models in the aspects of forecasting horizon, applied areas, type of model, and forecasting accuracy. The conventional models are categorized into time series models, regression models, and gray models. The AI-based models are grouped into artificial neural network-based models and support vector regression machine-based models. Additionally, to the best of our knowledge, the evaluation of the models' performance in different forecasting horizons is a critical issue that has not been solved. Thus, for better evaluate the performance of forecasting models, a detailed reference range of mean absolute percentage error (MAPE) in energy consumption forecasting will also been proposed. The review results show that conventional models are preferred for the yearly energy consumption forecasting in national level. Among them, nonlinear regression models can not only explicitly describe the relationship between consumption data and influencing factors but also obtain the lowest average MAPE (1.79%) for long-term energy consumption forecasting. AI-based models are robust and full-scale in all applied areas and forecasting horizons. This paper provides valuable suggestions for researchers in model selection and serves as an initial study of the evaluation benchmark construction for energy consumption forecasting. • The published papers of energy consumption forecasting are reviewed. • The models applied in energy consumption forecasting are analyzed. • The accuracy of the models in different forecasting horizons is investigated. • The reference range of MAPE is summarized. • The further prospects of energy consumption forecasting are highlighted. [ABSTRACT FROM AUTHOR]

Foam has many applications in the oil and gas industry, either in hydraulic fracturing, enhanced oil recovery, or drilling operations. The success of these operations depends largely on understanding the behavior of foam rheology, which is complex. The literature contains many models used to estimate the effective bulk foam viscosity; most were based on fitting parameters estimated from limited-experimental data. Nevertheless, the fitting parameters are not valid at different operating conditions such as temperature, pressure, and shear rate. This results in models with limited applicability as the laboratory conditions are hardly replicated. In this study, we generated 360 data points of effective bulk foam viscosity using the high pressure high temperature (HPHT) foam rheometer device. A wide range of conditions was examined, such as temperature, pressure, shear rate, foam quality, and composition. The gas-phase consisted of either CO 2 or N 2 , while four types of water representing different salinities were used in the liquid phase. The foam was generated using seven different commercial surfactants at different concentrations. Also, low pH chelating agent and corrosion inhibitor were added in some experiments. The data pool was analyzed using four machine learning techniques: Artificial Neural Network (ANN), Decision Trees (DT), Random Forest Regressor (RFR), and K-Nearest Neighbor (KNN). ANN showed the highest accuracy with R2 of 0.972 and 0.985 on the training and testing datasets, respectively. Also, the relative importance of features was examined using Pearson, Spearman, and Kendall correlation coefficients. The most significant parameters in reducing foam viscosity were temperature, corrosion inhibitor, and shear rate, respectively. On the contrary, foam quality positively impacted the foam viscosity, where 80% foam quality was the maximum tested condition. The impact of pressure, surfactant concentration, water type, and chelating agents were complex. This paper provides a simplified ANN-based model which can be used on the fly to predict the effective bulk foam viscosity in both laboratory and field conditions. • This research generated 360 data points of effective foam viscosity utilizing the HPHT foam rheometer. • The data pool was analyzed using four machine learning techniques. • The performance of ANN was much better than the other three techniques. • The sensitivity analysis with the proposed ANN model showed that it can predict the physics-based behavior. • An empirical model was provided in this study to predict the effective foam viscosity. [ABSTRACT FROM AUTHOR]

Permeability is one of the critical properties of reservoir rocks that is used to describe the ability in conducting fluids through pore spaces. This parameter cannot be simply predicted since there are nonlinear and unknown relationships between permeability and other reservoir properties. To obtain information about permeability, core samples are analyzed or well tests are performed conventionally. These are, however, very expensive and time-consuming to perform. Well log data is another source of information which is always available and much cheaper than core sample and well testing analysis. Thus establishing a relationship between reservoir permeability and well log data can be very helpful in estimation of this vital parameter. However, establishing relationship between well logs and permeability is not a simple task and cannot be done using a simple linear or nonlinear method. Relevance Vector Regression (RVR) is one of the robust artificial intelligence algorithms proved to be very successful in recognition of relationships between input and output parameters. The aim of this paper is to show the application of RVR in prediction of permeability in three wells located in a carbonate reservoir in the south part of Iran. To do this, genetic algorithm (GA) was used as an optimizer to find the best logs for prediction of permeability. Comparing the results of RVR with that of a Support Vector Regression (SVR) indicated more accuracy of RVR in prediction of permeability. However, SVR can still be considered as a second option for prediction of petrophysical properties due to its reliable efficiency. However, it should be noticed that all of the predictions using well logs data are limited to the intervals where logs are available. Thus more studies are still required to propose alternative methods whose results can be used for the entire reservoir. [ABSTRACT FROM AUTHOR]

Abstract: Viscosity of crude oil is an important physical property that controls and influences the flow of oil through rock pores and eventually dictating oil recovery. Prediction of crude oil viscosity is one of the major challenges faced by petroleum engineers in production planning to optimize reservoir production and maximize ultimate recovery. This paper presents prediction of the complete viscosity curve as a function of pressure using artificial intelligence (AI) techniques. The viscosity curve predicted using artificial intelligence techniques derived from gas compositions of Canadian oil fields closely replicated the experimental viscosity curve above and below bubble point pressure when compared with correlations of its class. Functional Networks with Forward Selection (FNFS) outperformed all the AI techniques followed by Support Vector Machine (SVM). [Copyright &y& Elsevier]

Abstract: Naturally fractured reservoirs are generally complex and highly heterogeneous. There are usually scarce data sources in these reservoirs that may be found over a wide range of scales. Therefore, fracture characterization is a complicated task and an integrated method is required to effectively overcome this multi-scaled problem and to combine data obtained from as many tools as possible. One sophisticated tool for this purpose is artificial intelligence. Recently, Fuzzy Logic and Neural Network have been used to obtain a 2D fracture intensity map in Hassi Messaoud field. The aim of this paper is to show the applicability of Artificial Neural Network and Fuzzy Logic in characterizing Parsi naturally fractured reservoir. First, using Fuzzy Inference System (FIS), Fracture Index (Fracture Intensity Index or FI) is calculated along the wellbore. For decreasing the uncertainty, using FIS, static and dynamic data (log, well test and core data) are coupled which results in a more reliable Fracture Index. Moreover, use of log data in FIS makes it possible to calculate the fractured index in those blocks where wells are perforated. Hence, a 3D characterization becomes possible. Then, for calculating the spatial distribution of the Fracture Index, two types of Neural Networks, Multi-Layer Perceptron (MLP) and Radial Basis Function (RBF), are used. The inputs of these networks are some geological and geomechanical drivers (e.g. shale volume, porosity, permeability, bed thickness, proximity to faults, slopes and curvatures of the structure), and the output is the Fracture Index. Because of the overfitting problem in MLP network, fuzzy ranking method is used for selecting only parts of the mentioned drivers as inputs of MLP network, and also a partially connected MLP network are used. RBF network produces more reasonable Fracture Index. Next, a 3D fracture intensity map in Parsi oil field is developed. According to this map, the high production rate of some wells can be explained in this field. Also, this 3D fracture intensity map can reduce the uncertainty in reservoir simulation. Therefore, a better field development scenario can be designed. The results are promising and can be easily extended in other naturally fractured reservoir. [Copyright &y& Elsevier]

Oil and gas metering is primarily used as the basis for evaluating the economic viability of oil wells. Owing to the economic implications of oil and gas metering, the subject of oil and gas flow rate measurement has witnessed a sustained interest by the oil and gas community and the academia. To the best of the authors' knowledge, despite the growing number of published articles on this subject, there is yet no comprehensive critical review on it. The objective of this paper is to provide a broad overview of models and modelling techniques applied to the estimation of oil and gas flow rate through chokes while also critically evaluating them. For the sake of simplicity and ease of reference, the outcomes of the review are presented in tables in an integrated and concise manner. The articles for this review were extracted from many subject areas. For the theoretical pieces related to oil and gas flow rate in general, the authors relied heavily upon several key drilling fluid texts. For operational and field studies, the authors relied on conference proceedings from the society of petroleum engineers. These sources were supplemented with articles in peer reviewed journals in order to contextualize the subject in terms of current practices. This review is interspersed with critiques of the models while the areas requiring improvement were also outlined. Findings from the bibliometric analysis indicate that there is no universal model for all flow situations despite the huge efforts in this direction. Furthermore, a broad survey of literature on recent flow models reveals that researchers are gravitating towards the field of artificial intelligence due to the tremendous promises it offers. This review constitutes the first critical compilation on a broad range of models applied to predicting oil and gas flow rates through chokes. • Oil and gas flow rates models have been reviewed. • No universal model for oil and gas flow rate prediction exist. • A critique of the extant oil and gas flow rate models has been done. • Machine learning techniques provide fluid flow rate models with higher accuracy. [ABSTRACT FROM AUTHOR]

In recent years, the increasing energy demand has led the oil and gas industry to explore unconventional reservoirs. The hydraulic fracturing technique (fracking) has been adopted in order to increase the reservoir drainage area. Nevertheless, there is an environmental concern about the contamination of aquifers due to this technique. The operation design requires predicting the induced fracture geometry to avoid hazards related to fracking. Hydraulic fracturing changes the state of stress at crack tip leading to more uncertainties in the definition of crack geometry, especially in naturally fractured formations. For such, analytical solutions and numerical simulations have been employed in recent decades. Nevertheless, the numerical models require high computational effort. This paper proposes an artificial neural network (ANN) to predict the interaction between hydraulic fracture and natural fractures. We performed over 800 simulations to build the training database varying the rock mechanical properties and model parameters, such as the approach angle between hydraulic fracture and natural fracture, in-situ stress magnitudes, friction angle, and fracture energy. The ANN results are compared against analytical solutions and numerical models, showing excellent agreement. These results show that the trained neural network can predict fracture interaction accurately. They also suggest that the most sensible parameters were taken into account in the proposed ANN. • Considerable number of fracture interaction simulations to database. • Application of Artificial Intelligence Technique. • Verification of the results with analytical solutions and simulations using other meshes. • Applied case study. [ABSTRACT FROM AUTHOR]

The rate of penetration (ROP) accounts for a substantial portion of the overall drilling cost. The drilling optimization process, which mostly involves the adjustment of the mechanical drilling parameters, is therefore of prime importance in ensuring efficient drilling. However, drilling formations with assorted types of lithology necessitate the involvement of more parameters to reduce uncertainty and enhance confidence when predicting the ROP. The objective of this paper is to introduce an ensemble model based on random forest (RF), in which artificial neural network (ANN), and adaptive neuro-fuzzy inference system (ANFIS) are the base learner models, to predict the ROP across different lithological formations In this study, two types of actual field data of Well-1 were employed to build the model: (i) mechanical drilling parameters collected from real-time sensors allocated at the rig site, and (ii) petrophysical properties obtained from conventional well logs. Well-2 with more than 2300 unseen data points was used to compare the capability of the base learners (i.e., ANN and ANFIS), standalone RF, and RF-meta model in predicting the ROP with two of the earliest published ROP empirical models (Maurer's and Bingham's models). The results showed that the RF-meta model outperformed the base learners and Maurer's and Bingham's empirical models in predicting the ROP in Well-2 with a low absolute average percentage error (AAPE) of 7.8 % and a high coefficient of determination (R 2) of 0.94. • Prediction of the rate of penetration (ROP) while drilling different formation lithologies. • ROP prediction using an ensemble learning model based on RF, in which ANN, and ANFIS were served as base learner models. • Investigation of the relevance importance for the input parameters by employing principal component analysis (PCA). [ABSTRACT FROM AUTHOR]

Abstract: Field development strategies are at the forefront of common engineering practices in the oil and gas industry. Reservoir simulation is the most commonly applied methodology to generate an optimum field development plan. However, reservoir simulation can be an energy and cost intensive method that often relies on rather subjective assumption of input parameters, due to lack of accurate field data. In this paper, a new approach using Artificial Neural Network (ANN) technology is proposed to predict individual well performances and accordingly develop infill drilling strategies. Due to its predictive capabilities, ANN is used as a tool to construct a correlation for production prediction. Seismic attributes, which capture heterogeneity of the reservoir geology, and completion information are used as network inputs. In calculating the interference effects, the geometry of the flow system under consideration was used together with the geometric location and the starting production schedule of each well within the system. The method was successfully implemented on a case study of the 19N 94W Township of the Wamsutter field in Wyoming using actual seismic attributes, completion information, well configuration, and production data. Production predictions were generated by the network for all locations at which seismic attributes were available. More promising locations were then selected for infill drilling purposes based on predicted productions at these locations. The predicted initial rate and 10-yr cumulative production were considered in the selection of infill drilling locations with high productivity potential. Results from this work show that the ANN was able to map the relationship between production, completion information, interference effects, and reservoir characteristics captured in seismic attributes. The proposed methodology allowed the construction of spatial maps of gas production, revealing new sweet spots which could not be identified from the existing production history alone. The production maps derived from the ANN predictions contain important heterogeneous features associated with reservoir properties reflected in seismic data. Even though well interference was initially thought to have a limited effect on well performance for the case study presented, the incorporation of well interference parameters in the network design improved production predictions, suggesting that well interference has a more significant impact on well performance than originally anticipated. [Copyright &y& Elsevier]

Abstract: The production performance of a heterogeneous reservoir cannot be realistically predicted without accurate reservoir characterization. Lack of detailed permeability data often hampers the reservoir characterization efforts. In this study, artificial intelligence techniques were successfully utilized to predict the missing information. More specifically, the objective of this study was to develop an accurate reservoir description with the aid of artificial neural networks (ANN) utilizing available geophysical well log data and limited core data. The superior ability of ANN for pattern recognition makes it a prime candidate for permeability prediction from well log data. The most complete reservoir description is generally provided through identification and characterization of the flow units. The presence of flow units suggests that separate permeability-log data relationship exists for each unit. The flow units in the wells with core data were first defined using self optimizing (Kohonen) neural network to cluster core-log data in a systematic manner. Utilizing the flow unit definition, a back-propagation artificial neural network to predict the flow units in the wells without core data was trained and tested. A second back-propagation artificial neural network was then trained and tested to predict the permeability profile within each flow unit. However, the use of limited permeability measurements to train and test the second network resulted in inconsistent predictions due to sensitivity to the arrangement of the data. An innovative approach for training and testing of the ANN was then developed to overcome the problem and to provide consistent and reliable predictions. The two networks which predicted the flow units and their attributes were utilized to develop the description of the reservoir. Primary and secondary production data were utilized to evaluate the reservoir description. The reservoir description substantially improved the simulation of the secondary recovery performance. The simulation results confirmed the presence of heterogeneities which have had profound impact on the performance of the reservoir. The methodology presented in this paper can serve as a guideline for the characterization of reservoirs with limited data. [Copyright &y& Elsevier]

Abstract: Fracture gradient constitutes an essential parameter in the pre-design stage of drilling operations, reservoir exploitation and stimulation. Over the years, fracture gradient has been acknowledged to be one of the most important key element that needs to be accurately determined in order to achieve optimum operation results. Several calculation methods have been presented in the literature for different regions of the world. All of these techniques were based on either parametric or empirical correlations, which require a prior knowledge of the functional form. Moreover, these methods are only valid at the area where they were developed. This paper presents an artificial neural network model to predict the fracture gradient for the Middle East region. The data set used in the study consists of more than 21,000 points with fracture gradient measurements taken from 16 wells drilled in seven geological prospects covering more than 200 sq mi. The excellent results obtained from the developed model establish a new simple mechanism for fracture gradient prediction that is based on readily obtainable parameters. The proposed neural network model predicts the fracture gradient as a function of pore pressure, depth and rock density. A detailed comparison between the results predicted by this method and those predicted by other techniques is presented. [Copyright &y& Elsevier]

Abstract: Neuro-simulation, which conjoins hard computing protocols with soft computing protocols, has been adapted to various applications in petroleum industry. This approach offers the advantage of optimizing number of model/simulation studies that need to be conducted. Simulation/model studies, in this context, refer to various modes of operations through which data are collected and results generated. Thus, experimental protocols followed in laboratories, field tests and operations conducted, and analyses carried using numerical and/or mathematical models are all considered as components of these studies. In this paper, an example implementation of neuro-simulation in the optimization of number of experiments that need to be conducted in measuring relative permeabilities is illustrated. [Copyright &y& Elsevier]

Abstract: During the last 10 years, the application of neural networks to solve oilfield problems has advanced. More recently, fuzzy logic proved beneficial when selecting variables to serve as neural network inputs. This paper presents a background to establish that neural networks are more than a “black box” and summarizes three applications of artificial intelligence technology to predict the secondary-to-primary ratio of a waterflood candidate using public domain information, the potential gas-producing rate of a behind pipe interval given only gamma ray and density logs, and the performance of single-well chemical imbibition treatments. Artificial intelligence used in this manner is essentially a statistical analysis, and as such the size of the dataset and the extent of the domain are important. Ignoring these important factors could result in overtraining of the neural network. Special care should be taken to avoid the pitfalls associated with overtraining. The neural network architectures were designed using a trial-and-error technique. Initially, a constructive design was employed by adding complexity to the architecture in terms of increasing the number of input variables as well as the number of hidden layers and nodes. A technique based on conventional statistical parameters was developed to numerically describe the patterns observed in log crossplots. These numerical descriptions were then prioritized and used as neural network inputs to be correlated with known production response. [Copyright &y& Elsevier]

Abstract: In highly heterogeneous reservoirs classical characterization methods often fail to detect the location and orientation of the fractures. Recent applications of Artificial Intelligence to the area of reservoir characterization have made this challenge a possible practice. Such a practice consists of seeking the complex relationship between the fracture index and some geological and geomechanical drivers (facies, porosity, permeability, bed thickness, proximity to faults, slopes and curvatures of the structure) in order to obtain a fracture intensity map using Fuzzy Logic and Neural Network. This paper shows the successful application of Artificial Intelligence tools such as Artificial Neural Network and Fuzzy Logic to characterize naturally fractured reservoirs. A 2D fracture intensity map and fracture network map in a large block of Hassi Messaoud field have been developed using Artificial Neural Network and Fuzzy Logic. This was achieved by first building the geological model of the permeability, porosity and shale volume using stochastic conditional simulation. Then by applying some geomechanical concepts first and second structure directional derivatives, distance to the nearest fault, and bed thickness were calculated throughout the entire area of interest. Two methods were then used to select the appropriate fracture intensity index. In the first method well performance was used as a fracture index. In the second method a Fuzzy Inference System (FIS) was built. Using this FIS, static and dynamic data were coupled to reduce the uncertainty, which resulted in a more reliable Fracture Index. The different geological and geomechanical drivers were ranked with the corresponding fracture index for both methods using a Fuzzy Ranking algorithm. Only important and measurable data were selected to be mapped with the appropriate fracture index using a feed forward Back Propagation Neural Network (BPNN). The neural network was then used to obtain a fracture intensity maps throughout the entire area of interest. A mathematical model based on “the weighting method” was then applied to obtain fracture network maps, which led to a better description of the major fracture trends. The obtained maps were compared and the results show that the proposed approach is a feasible and practical methodology to map the fracture network. [Copyright &y& Elsevier]

Abstract: An evolution of intelligent detection methods applied in the oil industry, using geological and geophysical data will be given. The most challenging problems of exploration and production may require application of advanced technologies. Our focus in this paper will be to highlight applications of soft computing and artificial intelligence. Specifically, the strength and weakness of human intelligence versus machine intelligence and the need for combining human and machine intelligence is visited. The role soft computing methods (neural networks, fuzzy logic and evolutionary computing) that can play in establishing “hybrid” intelligence for addressing E&P problems is emphasized. [Copyright &y& Elsevier]

This paper develops a rigorous and advanced data-driven model to describe, analyze, and forecast the global crude oil demand. The study deploys a hybrid approach of artificial intelligence techniques, namely, the genetic-algorithm, neural-network, and data-mining approach for time-series models (GANNATS). The GANNATS was developed and applied to two country cases, including one for a high oil producer (Saudi Arabia) and one for a high oil consumer (China), to develop crude oil demand forecasts. The input variables of the neural network models include gross domestic product (GDP), the country's population, oil prices, gas prices, and transport data, in addition to transformed variables and functional links. The artificial intelligence predictive models of oil demand were successfully developed, trained, validated, and tested using historical oil-market data, yielding excellent oil demand predictions. The performance of the intelligent models for Saudi Arabia and China was examined using rigorous indicators of generalizability, predictability, and accuracy. The GANNATS forecasting models show that the crude oil demand for both Saudi Arabia and China will continue to increase over the forecast period but with a mildly declining growth, particularly for Saudi Arabia. This decreasing growth in the demand for oil can be attributed to increased energy efficiency, fuel switching, conversion of power plants from crude oil to gas-based plants, and increased utilization of renewable energy, such as solar and wind for electricity generation and water desalination. In this study, the feature engineering of variables selection techniques has been applied to identify and understand significant factors that impact and drive the crude oil demand. The proposed GANNATS methodology optimizes and upgrades the conventional process of developing oil demand forecasts. It also improves and enhances the predictability and accuracy of the current oil demand forecasting models. • Developed GANNATS models for modeling, analyzing and predicting the crude oil demand. • Identified and discussed the influential factors driving the crude oil demand of Saudi Arabia and China. • Predicted crude oil demand for Saudi Arabia and China for the short-term outlook. • Provided a guide for AI/ML generalization attributes. • Provided golden rules and tips for modeling GANNATS. [ABSTRACT FROM AUTHOR]

By defining an optimum injection/production strategy in the water flooding process, the water front movement is controlled and an early breakthrough is avoided, and as a result, the sweep efficiency is increased. The most important part of injection/production planning is determining the injection fluid front position. However, using a commercial reservoir simulator comes up with a considerable time and CPU effort, particularly in the case of large and complex reservoirs. Several proxy models, either physics-based or Data-Driven, have been developed to predict water front movement with low computational time and cost, each offering some advantages and shortcomings such as error accumulation and short-term predictions. In this paper, we use a hybrid view in developing a classification based smart proxy model at the grid-block level (CSPMG) for front prediction to benefit from the advantages of both physics-based and Data-Driven proxies. The idea is to formulate the problem by using the physical principles underlying the problem (physical-view) and then use machine learning classification models to capture the pattern between inputs and the target feature (Data-Driven view). Based on the Buckley-Leverett theory, which is a physics-based method for front advancement in porous media, water front prediction was formulated as a classification problem in which the grid-blocks behind and ahead of the front were considered as separate classes and a label was assigned to each class. Then, artificial neural networks (ANNs) were trained on a training database to predict the class label of each grid block. The water front was considered as the boundary between two adjacent classes. A binary and a ternary classification problem were formulated and two proxy models were developed. A blind test was carried out to compare their results with each other, with a Data-Driven regression model, and with those of a reservoir simulator. The results showed that the CSPMG matches the reservoir simulator results and outperforms the regression model and makes longer predictions. •A hybrid view is adopted to construct a reinforced proxy by including physics for front advancement prediction. •The problem is formulated by using the Buckley-Leveret theory (physical view) as a classification problem. •The problem is modeled with artificial neural networks as the machine learning classification tool (Data-Driven view). •The proposed approach outperforms the existing regression-based approaches in both precision and prediction time interval. •The simulation time is reduced in comparison with the commercial reservoir simulator. [ABSTRACT FROM AUTHOR]

Fluid loss to subsurface formations is a challenging aspect during drilling operations in petroleum industry. Several other drilling issues such as fluid influx and pipe sticking can be triggered in such scenarios, posturing a significant risk to rig personnel, environment, and economical drilling. Therefore, prediction and early detection of lost circulation events are required for safe and economic drilling operation. Several theoretical studies have been performed to detect and predict fluid loss event during hydrocarbon extraction. This paper reviews the existing conventional and intelligent models developed for early detection and prediction of lost circulation events. These predictive and detecting models comprise of Artificial Intelligence (AI) algorithms that require improvements for data reduction, universal prediction and compatibility. The review also covers several sensor-based techniques, different geostatistical-based models and Pressure-While-Drilling (PWD) tools for their applications in early loss circulation detection. In addition, loss circulation zones types, severity level, scenario and common preventive measures are also included in this review. This study aims to provide a systematic review of the published literature from the last forty years on the developed conventional and intelligent models for detection and prediction of fluid loss events and emphasizes on increasing AI involvement for precise results. • Fluid loss early detection and prediction improves drilling operational decision making. • Limited studies on loss circulation predictive modelling compare to its detection. • Updated PWD equipment provides quick and improved decision during real time usage. • Hybrid ML and CBR are recommended loss circulation predictive methods. • Predictive models require improvements for data reduction, and compatibility. [ABSTRACT FROM AUTHOR]