Search

One of the detrimental issues in upstream oil industry is permeability impairment in oil reservoirs. The permeability may be reduced because of scale deposition originated from injected water incompatibility with the in-situ water. To tackle this issue and monitor permeability impairment owing to scale deposition, an inventive evolutionary approach of Artificial Neural Network (ANN) is utilized which is based on the bio-inspired science. This approach is optimized by various optimization algorithms which provide a high-precision decision-making process with low uncertainty associated with the interconnected weights of the developed neural network model. The constructed intelligent approaches are evaluated using extensive experimental data from the open literature. Moreover, two regression models are developed to highlight the robustness and precision of the addressed techniques. For model validation, the predictions obtained from the smart technique as well as the regression method are compared with the formation damage experimental data. Based on various performance criteria, it was obtained that the predictions from the optimized genetic algorithm have infinitesimal average relative and absolute deviation from the experimental data (< 0.1 %). The results obtained in the current research prove that implication of HGAPSO-ANN in monitoring of formation damage associated with scale deposition in porous media results in reliable estimation of permeability impairment. This can result in designing a more reliable reservoir simulation model that captures the permeability impairment due to scale deposition, hence providing thorough action plans for developing EOR technologies that may cause scale deposition-related formation damage. [ABSTRACT FROM AUTHOR]

Water coning in petroleum reservoirs leads to lower well productivity and higher operational costs. Adequate knowledge of coning phenomena and breakthrough time is essential to overcome this issue. A series of experiments using fractured porous media models were conducted to investigate the effects of production process and pore structure characteristics on water coning. In addition, a hybrid artificial neural network (ANN) with particle swarm optimization (PSO) algorithm was applied to predict breakthrough time of water coning as a function of production rate and physical model properties. Data from the literature combined with experimental data generated in this study were used to develop and verify the ANN-PSO model. A good correlation was found between the predicted and real data sets having an absolute maximum error percentage less than 9%. The developed ANN-PSO model is able to estimate breakthrough time and critical production rate with higher accuracy compared to the conventional or back propagation (BP) ANN (ANN-BP) and common correlations. The presence of vertical fractures was found to accelerate considerably the water coning phenomena during oil production. Results of this study using combined data suggest the potential application of ANN-PSO in predicting the water breakthrough time and critical production rate that are critical in designing and evaluating production strategies for naturally fractured reservoirs. © 2014 American Institute of Chemical Engineers AIChE J, 60: 1905-1919, 2014 [ABSTRACT FROM AUTHOR]

Urea has the highest demand among all solid nitrogenous fertilisers within the agriculture industry. In this paper, a mathematical model and an Artificial Neural Network (ANN) technique are proposed for the simulation and optimisation of the urea plant in an industrial petrochemical company. The developed mathematical model consists of complex vapour-liquid equilibria for the NH3-CO2-H2O-(NH2)2CO system in thermodynamic and reaction frameworks. The smart technique (e.g. ANN) considers the CO2 conversion in terms of temperature and the molar ratios of NH3/CO2 and H2O/CO2 in the liquid phase. The ANN predictions were compared with the real data and results obtained from the mathematical model. An acceptable agreement was attained between deterministic methods. Through implementation of a systematic sensitivity analysis, it was found that a temperature of 191°C, a pressure of 132 atm and a NH3/CO2 ratio of 2.7 are the optimum process conditions for the urea production. It is concluded that the developed ANN (or connectionist) technique is an efficient tool for modelling complex phase equilibria with reaction in the industrial urea plant. [ABSTRACT FROM AUTHOR]

Steam-assistedgravity drainage (SAGD) is an enhanced oil recoverytechnology for heavy (or viscous) oil and bitumen that involves drillingtwo horizontal wells in underground formations. Laboratory work, pilot-plantstudies, and mathematical model development, which are generally costly,difficult, and time-consuming tasks, are taken into account as importantstages in finding an effective and economical method and also predictingthe performance of the SAGD technique for a certain heavy-oil reservoir.Currently, smart techniques as accurate and fairly fast tools arehighly recommended for these purposes. In this work, an experimentalstudy and an artificial neural network (ANN) linked to an optimizationtechnique, called particle swarm optimization (PSO), were employedto obtain performance parameters such as the cumulative steam-to-oilratio (CSOR) and recovery factor (RF) for the SAGD process. The outputsof the developed connectionist modeling (i.e., ANN–PSO) werecompared with actual data, showing an average error lower than 7%,mostly because of the supremacy of the ANN–PSO method comparedto the conventional ANN method and the correlations developed in thisstudy. Furthermore, it is concluded that, among the contributing parameters,reservoir thickness and oil saturation have the most significant impactson RF and CSOR during SAGD operations. The current study confirmsthe potential of hybrid connectionist modeling to screen heavy-oilfractured reservoirs for the SAGD process. [ABSTRACT FROM AUTHOR]

Miscible gas injection (MGI) processes such as miscible CO2 flooding have been in use as attractive EOR options, especially in conventional oil reserves. Optimal design of MGI is strongly dependent on parameters such as gas-oil minimum miscibility pressure (MMP), which is normally determined through expensive and time-consuming laboratory tests. Thus, developing a fast and reliable technique to predict gas-oil MMP is inevitable. To address this issue, a smart model is developed in this paper to forecast gas-oil MMP on the basis of a feed-forward artificial neural network (FF-ANN) combined with particle swarm optimisation (PSO). The MMP of a reservoir fluid was considered as a function of reservoir temperature and the compositions of oil and injected gas in the proposed model. Results of this study indicate that reservoir temperature among the input parameters selected for the PSO-ANN has the greatest impact on MMP value. The developed PSO-ANN model was examined using experimental data, and a reasonable match was attained showing a good potential for the proposed predictive tools in estimation of gas-oil MMP. Compared with other available methods, the proposed model is capable of forecasting oil-gas MMP more accurately in wide ranges of thermodynamic and process conditions. All predictive models used other than the PSO-ANN model failed in providing a good estimate of the oil-gas MMP of the hydrocarbon mixtures in Azadegan oilfield, Iran. © 2013 Canadian Society for Chemical Engineering [ABSTRACT FROM AUTHOR]

ABSTRACT Reservoir characterization involves describing different reservoir properties quantitatively using various techniques in spatial variability. Nevertheless, the entire reservoir cannot be examined directly and there still exist uncertainties associated with the nature of geological data. Such uncertainties can lead to errors in the estimation of the ultimate recoverable oil. To cope with uncertainties, intelligent mathematical techniques to predict the spatial distribution of reservoir properties appear as strong tools. The goal here is to construct a reservoir model with lower uncertainties and realistic assumptions. Permeability is a petrophysical property that relates the amount of fluids in place and their potential for displacement. This fundamental property is a key factor in selecting proper enhanced oil recovery schemes and reservoir management. In this paper, a soft sensor on the basis of a feed-forward artificial neural network was implemented to forecast permeability of a reservoir. Then, optimization of the neural network-based soft sensor was performed using a hybrid genetic algorithm and particle swarm optimization method. The proposed genetic method was used for initial weighting of the parameters in the neural network. The developed methodology was examined using real field data. Results from the hybrid method-based soft sensor were compared with the results obtained from the conventional artificial neural network. A good agreement between the results was observed, which demonstrates the usefulness of the developed hybrid genetic algorithm and particle swarm optimization in prediction of reservoir permeability. [ABSTRACT FROM AUTHOR]

Wettability alteration of the reservoir rock is considered as one of the promising remedies for the condensate blockage. There are key parameters including wettability state (WS), treatment radius (TR), and treatment time (TT) that considerably influence this treatment process. The objective of this paper is to conduct simultaneous optimization of WS, TR, and TT in a supergiant gas condensate reservoir located in Persian Gulf offshore. For this purpose, two distinct optimization methods are used in this research. The Taguchi design of experiment (DOE) method is first used to find the optimal state with a low number of simulation runs. Afterward, a smart optimization approach is developed through integration of the artificial neural network (ANN) and genetic algorithm (GA) techniques; this hybrid method helps to assess more possible combinations of the three key factors. Also, the compositional commercial reservoir simulator ECLIPSE 300 is used to simulate gas production operation, considering different WS, TR, and TT conditions. It was found that both the Taguchi and the hybrid constructed network (ANN-GA) methods lead to the results in agreement with the simulation results so that the magnitudes of average absolute relative error are 0.6397% and 0.0436% for the Taguchi and ANN-GA, respectively, based on gas recovery factor (GRF) data. However, the ANN-GA method estimates a slightly higher GRF at the optimum state, compared to the Taguchi. Nevertheless, Taguchi DOE with only 25 required data points seems a promising and practical option for modeling and optimization purposes. The ANN-GA optimum condition corresponds to 50.64% GRF with intermediate WS, a TR of 12.7 m, and a TT of about 5 months. • Effective approaches to optimize gas condensate reservoirs are introduced. • Interaction between wettability state, and treatment radius and time is considered. • Taguchi method and hybrid ANN-GA lead to outputs close to simulation results. • The optimal treatment time is 150 days based on Taguchi technique. • Wettability state has a higher impact on condensate recovery. [ABSTRACT FROM AUTHOR]

Log-based reservoir characterization is one of the widely used techniques to estimate the reservoir properties and make decisions about future plans for hydrocarbon production. Use of the machine learning tools is becoming a more accessible approach for predictive model development in oil and gas engineering. The objective of this research is to identify and rank the most contributing log variables for estimation of water saturation using the machine learning tools. The multilayer perception artificial neural network (MLP-ANN) and kernel function-based least-squares support vector machine (LS-SVM) techniques are employed to develop predictive models for water saturation. The model can capture the non-linear behaviors and high-dimensional complex relationships among field log data variables. Based on the prediction performance of the models, the Levenberg-Marquardt algorithm-based MLP-ANN and the radial kernel function-based LS-SVM model optimized with coupled simulated annealing optimization technique were found to perform better compared to other models. The MLP-ANN and radial kernel function-based LS-SVM approaches lead to the same feature ranking of predictor variables. It was found that the significance order (higher to lower) of influential log variables are the true resistivity, bulk density, neutron porosity, photoelectric factor, gamma-ray, and sonic travel time based on their relative contribution to the water saturation. The strategy introduced in this study assists to forecast water saturation with a relatively few number of log variables, and thus, reduces the number of necessary logs to run during exploration stage, considerably lowering the exploration costs. • Two deterministic models are proposed to forecast water saturation with aid of log data. • Both ANN and LS-SVM are utilized to rank the log variables contributing to water saturation prediction. • True resistivity and bulk density are the most influential parameters affecting water saturation. • The predictions and feature ranking are in good agreement with real data. • The introduced strategies are useful for formation evaluation, selection of production method, and reservoir management. [ABSTRACT FROM AUTHOR]

• A systematic study is conducted on coronary artery disease (CAD) classification based on Z-Alizadeh Sani dataset. • This study offers an effective visualization tool for CAD classification. • CART methodology is implemented for CAD classification, considering vital input features. • CART offers the highest diagnostic performance, compared to previous models. • Atypical and current smooker features have the maximum and minimum importance in CAD diagnosis. As the most common cardiovascular defect, coronary artery disease (CAD), also called ischemic heart disease, is one of the substantial causes of death globally. Several diagnosis approaches such as baseline electrocardiography, echocardiography, magnetic resonance imaging, and coronary angiography are suggested for screening the suspected patients that may suffer from CAD. However, applying such methods may have health side effects and/or expensive costs. As an alternative to the available diagnosis tools/methods, this research involves a decision tree learning algorithm called classification and regression tree (CART) for a simple and reliable diagnosis of CAD. Several CART models are developed based on the recently CAD dataset published in the literature. Utilizing all the features of the dataset (55 independent parameters), it was found that only 40 independent parameters influence the CAD diagnosis and consequently development of the predictive model. Based on the feature importance obtained from the first CART model, three new CART models are then developed using 18, 10, and 5 selected features. Except for the five-feature CART model, the outcomes of developed CART models demonstrate the maximum achievable accuracy, sensitivity, and specificity for CAD diagnosis (100%), while comparing the predictions with the reported targets. The error analysis reveals that the literature models including sequential minimal optimization (SMO), bagging SMO, Naïve Bayes (NB), artificial neural network (ANN), C4.5, J48, Bagging, and ANN in conjunction with the genetic algorithm (GA) do not outperform the CART methodology in classifying patients as normal or CAD. Hence, the robustness of the tree-based algorithm in accurate and fast predictions is confirmed, implying the proposed classification technique can be successfully utilized to develop a coherent decision-making system for the CAD diagnosis. [ABSTRACT FROM AUTHOR]

Asphaltenes deposition is considered a serious production problem. The literature does not include enough comprehensive studies on adsorption phenomenon involved in asphaltenes deposition utilizing inhibitors. In addition, effective protocols on handling asphaltenes deposition are still lacking. In this study, three efficient artificial intelligent models including group method of data handling (GMDH), least squares support vector machine (LSSVM), and artificial neural network (ANN) are proposed for estimating asphaltenes adsorption onto NiO/SAPO-5, NiO/ZSM-5, and NiO/AlPO-5 nanocomposites based on a databank of 252 points. Variables influencing asphaltenes adsorption include pH, temperature, amount of nanocomposites over asphaltenes initial concentration (D/C0), and nanocomposites characteristics such as BET surface area and volume of micropores. The models are also optimized using nine optimization techniques, namely coupled simulated annealing (CSA), genetic algorithm (GA), Bayesian regularization (BR), scaled conjugate gradient (SCG), ant colony optimization (ACO), Levenberg–Marquardt (LM), imperialistic competitive algorithm (ICA), conjugate gradient with Fletcher-Reeves updates (CGF), and particle swarm optimization (PSO). According to the statistical analysis, the proposed RBF-ACO and LSSVM-CSA are the most accurate approaches that can predict asphaltenes adsorption with average absolute percent relative errors of 0.892% and 0.94%, respectively. The sensitivity analysis shows that temperature has the most impact on asphaltenes adsorption from model oil solutions. [ABSTRACT FROM AUTHOR]

Carbonated water injection (CWI) is an efficient oil recovery method, which provides solution to the drawbacks of existing related recovery techniques such as water flooding and pure CO 2 injection. The recovery factor achieved from CWI is considerably higher than that for related CO 2 - enhanced oil recovery (EOR) methods, due to effective transport phenomena involved in the displacement process. Additionally, the sequestration of anthropogenic CO 2 makes CWI even more attractive for practical implications. Although CWI has been experimentally proven to be an efficient technique, simulation/mathematical models to capture detailed CWI physics have been unreliable because of the complex recovery mechanisms associated with CWI. A majority of models have been developed based on unrealistic assumptions. Thus, existing models become doubtful and the confidence to apply CWI in the larger scales such as pilot plants becomes low. In this research work, smart methods such as artificial neural network (ANN), least squares support vector machine (LSSVM), and gene expression programming (GEP) are suggested to avoid the impractical and inconclusive assumptions. The connectionist techniques (e.g., ANN, LSSVM, and GEP) relate the recovery factor (RF) to the key input parameters such as pressure, temperature, viscosity, permeability, and injection rate based on pattern recognition without detailed knowledge about the process as well as use of the governing equations. The performance of the deterministic models is evaluated through using statistical parameters such as mean squared error (MSE), maximum absolute percentage error (MAAPE), minimum absolute percentage error (MIAPE), and goodness of fit (R2). The results reveal that the ANN model has the lowest MSE (0.35), MIAPE (0.001), and MAAPE (2.47), and the highest R2 (0.99) in the testing phase. Based on the sensitivity analysis, pressure is recognized as the most important parameter, while temperature has the least rank in terms of significance. The findings of this research study can assist engineers and researchers to provide a reasonable estimation of RF achievable from CWI, which can be an asset in better management and planning of CWI processes toward optimal conditions in terms of technical, economic, and environmental prospects. • Effective deterministic tools are introduced to estimate CWI performance. • The pressure and temperature have the highest and lowest importance. • The models are able to accurately simulate CWI behaviors. • The ANN is a more accurate and reliable model among others. • Sensitivity analysis helps researchers and engineers to determine CWI optimal conditions. [ABSTRACT FROM AUTHOR]

Highlights: [•] Multilayer perceptron (MLP) neural network is used to model calcium chloride dehydration system. [•] A simple correlation is developed to estimate water content of natural gas dried by CaCl2. [•] The model has been developed and tested using 200 series of the data. [•] Statistical and graphical error analyses are presented to show accuracy of the model. [Copyright &y& Elsevier]

[Display omitted] • AI and GM methods are integrated to consider catalyst effect on gasified yields. • The delta term is determined using characteristics of biomass and catalysts. • The hybridized GM improves the hydrogen yield prediction. • Feed concentration has a key impact on the hydrogen yield. • There is a very good match between the real data and predictions. Supercritical water gasification (SCWG) of biomass offers a promising pathway for sustainable energy production. This study presents a novel approach that combines an artificial intelligence (AI) technique with the Gibbs minimization (GM) method to enhance the GM approach by involving the catalyst effect on the gasified yield. The proposed strategy incorporates a modified Gibbs energy calculation, incorporating a delta term that quantifies the influence of the catalyst's properties on the Gibbs free energy. The delta term is determined using experimental data, including characteristics of biomass and the employed catalysts. An artificial neural network (ANN) model is developed to predict the delta value and is integrated with the GM method to improve the estimation of gasified product yields in the presence of a catalyst. The ANN model establishes correlations between the catalyst specifications such as pore volume, specific surface area, and pore diameter, and delta value. The trained model can predict the delta term, allowing the catalyst's role to be considered within the GM technique. This integration addresses a limitation of the conventional thermodynamic model, which typically ignores the catalyst's influence. The developed ANN model exhibits high accuracy, with an R2 value of 0.92 and a mean squared error (MSE) of 0.042. The hybridized GM approach substantially improves the hydrogen yield prediction compared to the basic model, enhancing (reducing) the MSE and mean absolute error by 98% and 93%, respectively. Furthermore, sensitivity analysis reveals that increasing temperature and reducing pressure contribute to improved hydrogen production. Additionally, lower biomass concentrations are shown to be beneficial for enhancing the hydrogen yield. [ABSTRACT FROM AUTHOR]

Abstract: Appropriate production method selection for Viscous Oil (e.g., Heavy Oil, Extra Heavy Oil, and Bitumen) Naturally Fractured Carbonate Reservoirs (VO NFCRs) mostly depends on the quality of the fluid and reservoir properties. Selection of a particular production method for a reservoir is generally evaluated through an exhaustive experimental, field pilot, and mathematical modeling approach. In the absence of robust and quick predictive tools, using connectionist techniques for performance prediction of a particular production method can be a valuable asset. In this study, a new screening tool is developed based on Artificial Neural Networks (ANNs) optimized with Particle Swarm Optimization (PSO) to assess the performance of steamflooding in VO NFCRs. As expected, Recovery Factor (RF) and Cumulative Steam to Oil Ratio (CSOR) during steamflooding are highly affected by the magnitudes of oil saturation and viscosity. The developed PSO–ANN model, conventional ANN and statistical correlations were examined using real data. Comparison of the predictions and real data implies the superiority of the proposed PSO–ANN model with an absolute average error percentage<6.5%, a determination coefficient (R 2)>0.98, and Mean Squared Error (MSE)<0.06, in contrast with conventional ANN model and empirical correlations for prediction of RF and CSOR. This indicates a great potential for application of hybrid PSO–ANN models to screen Viscous Oil carbonate reservoirs for steamflooding. [Copyright &y& Elsevier]

Directional drilling has become popular in recent decades in both onshore and offshore operations due to reduced drilling costs and improved recovery. In a directional well drilling operation, drill cuttings tend to settle down at the lower side of the inclined annular section. If the generated cuttings are not removed from the hole section properly, it causes a cuttings bed formation in the annular section. Different drilling related problem such as poor rate of penetration, excessive torque and drag, increase differential sticking often associated with a poor hole cleaning which eventually lead to increase in non-productive time (NPT). Therefore, the success of inclined well drilling operation largely depends on effective cleaning of drill cuttings from the annular section. A variety of parameters, including the fluid rheology, mud velocity, cuttings size, drill pipe rotation and drill pipe inclination generally influence the cuttings transport performance. Optimization of these drilling parameters is crucial to ensure proper hole cleaning. In this study, a Computational Fluid Dynamics (CFD) model for the inclined well section is used to investigate the cuttings transport efficiency (CTE). An Eulerian-Eulerian multiphase flow model is proposed and validated with lab scale experimental data. The experiments were performed in a 6.16-m-long annular test section having an outer pipe diameter of 4.5-inch and an inner pipe diameter of 2.5-inch. The setup is equipped with Electrical Resistance Tomography (ERT) system to measure the instantaneous solid volume fraction and a visualization section. A non-Newtonian Heschel-Bulkley (HB) fluid was used as drilling mud and solid glass beads of 2.50 mm–3.00 mm were used as cuttings in the experiment. This study shows a good agreement in visualization of mechanistic three-layer model of cuttings transport in terms of ERT data from experiment and CFD simulation. The validated CFD model is used to perform 5- Factors factorial design and analysis of variance (ANOVA) study. ANOVA shows that the interaction effects of mud velocity-cuttings size, mud velocity-inclination are statistically significant. Finally, this study proposed a statistical model to estimate the CTE of an inclined well considering the two factor and three factor interactions among the variables. Also, the model shows that drill pipe rotation has negligible effect in improving cuttings transport efficiency in the inclined well. The proposed model also reveals that cuttings size and fluid velocity account for 78% contribution in the transport efficiency for an inclined well. Furthermore, an Artificial Neural Network (ANN) method is used to verify the contribution of lower order interaction in the statistical model. Though empirical model shows few lower order two factor interactions (fluid rheology -cuttings size, fluid rheology-inclination), and three factor interactions (velocity-cuttings size-inclination, fluid rheology-cuttings size-inclination); ANN model shows that lower order interaction are significant in model prediction and should not be ignored. The findings of this study can help in better understanding the interaction behaviour among drilling parameters and optimized cuttings transport efficiency in the inclined drilling operation for a wide variety drilling parameters. • A statistical model describes the cuttings transport efficiency with interaction. • ANOVA analysis shows significant interaction among drillings parameters. • ANN shows the significance of lower order interaction in statistical model. • Cuttings bed distribution analysis shows similar profile for ERT and simulation. • Inclination upto 60° from horizontal is more prone to cuttings accumulation. [ABSTRACT FROM AUTHOR]

Ice-induced scour is a serious challenge for the subsea pipelines in the Arctic shallow waters. Estimation of the maximum pipeline deformation and its minimum burial depth can guarantee the operational integrity of these structures in the ice-prone regions. The pipeline buried below the ice keel is still threatened by subgouge soil deformation that is extended down the ice tip due to the shear resistance of the seabed soil. Determining the subgouge soil deformations is a challenging process that needs costly experimental and numerical simulations. In this paper, an alternative and cost-effective methodology has been proposed using a robust neural network-based method titled "generalized structure of group method of data handling" (GS-GMDH) for the first time to simulate the horizontal and vertical subgouge soil deformation profiles in clay. Using the parameters governing the subgouge soil deformations, nine GS-GMDH models were defined. The premium GS-GMDH models and the most influencing input parameters comprising the soil depth and the gouge geometry were introduced by performing a sensitivity analysis. Subsequently, results of the best GS-GMDH models were compared with the classical group method of data handling (GMDH), the artificial neural network (ANN), and the empirical approaches. An uncertainty analysis showed that the GS-GMDH slightly overestimated the horizontal and underestimates the vertical subgouge soil deformations. A partial derivative sensitivity analysis (PDSA) was also performed to assess the influence of the input parameters on the subgouge soil deformations. Lastly, a set of GS-GMDH-based equations were proposed for fast estimation of the subgouge soil deformations in clay. [Display omitted] • Arctic subsea pipelines are buried for physical protection against the ice scour. • Determining the minimum burial depth needs accurate modelling of the ice-seabed interaction. • Costly analysis and experiments are usually required for assessment of the ice-seabed interaction. • An artificial intelligence model was developed to simulate the ice-seabed interaction in clay. • GS-GMDH could be an accurate and cost-effective alternative for prediction of the ice-induced subgouge deformations. [ABSTRACT FROM AUTHOR]

Oil and gas companies use the reservoir simulation approach to history match and forecast reservoir production. While this strategy is easy and often effective, it has inherent setbacks, including a limited ability to capture temporal and spatial variability of the data, uncertainty in the data and reservoir models, and most importantly, impacts of these factors/parameters on the overall production risk profile. This paper presents a hybrid method to forecast the production rate and risks attributed to oilfield development. The hybrid connectionist strategy integrates a data-driven model to capture variability with a probabilistic model to capture uncertainties and associated risks. The multilayer perceptron artificial neural network (ANN) is built on the basis of geological realizations and used as a substitute for the commercial simulator (e.g., CMG) to model reservoir production behaviour for the effective facilitation of the production prediction. The model has a generalization capability and captures the temporal-spatial dependency and non-linear complex relationships. The Bayesian network model is developed to assess the production risks. It employs the concept of the early warning index system. The results show that the predicted oil production profile closely matches the history matched data from the simulator, and the assessed dynamic risks conform with the field reality. The hybrid modeling strategy leads to the minimum and average percentage errors of 0.02% and 5.28% respectively in the reservoir production forecast. The application of the proposed approach would assist in effective reservoir management decision making, enabling a risk-based optimal field performance of reservoir. • A new hybrid method is proposed to forecast production rate and risks attributed to oilfield development. • The strategy integrates a data-driven model to capture variability with a probabilistic model to capture uncertainties/risks. • The ANN model captures the temporal-spatial dependency and non-linear complex relationship of the reservoir flow behaviour. • The Bayesian network model is developed to evaluate the production risks using early warning index. • The proposed approach can assist in effective reservoir management decision making. [ABSTRACT FROM AUTHOR]

Graphical abstract Highlights • Various connectionist tools are developed to predict the membrane fouling resistance in MBRs. • The LSSVM outperforms other models in estimating the fouling resistance. • GEP technique leads to development of a mathematical expression. • This research investigation helps to make better operating and policy decisions while employing MBRs. • The importance of the input parameters is ranked as Flux > TMP > Temperature > MLSS. Abstract Membrane fouling as a major concern in development and optimization of membrane bioreactor (MBR) technologies has been the focus of numerous engineering and research investigations. Considering the complexity of membrane fouling occurrence, mathematical modelling techniques have been progressively proposed to forecast this phenomenon for optimizing MBR performance. A majority of the models are not reliable and accurate enough in terms of theoretical and practical prospects. In this research work, smart methods including artificial neural network (ANN), gene expression programming (GEP), and least square support vector machine (LSSVM) are suggested to avoid utilization of complex modelling methodologies and costly and time-consuming measurements. The developed models relate fouling resistance to key parameters such as permeate flux, temperature, and transmembrane pressure. To enhance the performance of conventional connectionist tools, particle swarm optimization (PSO) algorithm with global optima is utilized. This study aims to simulate the MBR efficiency by calculating membrane fouling resistance. The performance of the smart models is evaluated based on the mean squared error (MSE), maximum absolute percentage error (MAAPE), minimum absolute percentage error (MIAPE), and coefficient of determination (R2). The results reveal that the developed LSSVM tool has the lowest MSE (0.0002), MAAPE (3.18), and MIAPE (0.01), and the highest R2 (0.99) in the testing phase. The transmembrane pressure and permeate flux are the most important parameters affecting the membrane fouling resistance. This study can help to obtain a better understanding of membrane fouling process to achieve optimal conditions for MBR systems in terms of design, operation, and optimization prospects. [ABSTRACT FROM AUTHOR]

• A hybrid smart model is employed for optimization of gas injection. • A novel objective function is introduced for optimal gas injection scenario. • A systematic sensitivity analysis is conducted to identify vital factors. • Maximizing cumulative oil production is not the best decision. • There is a good match between the ANN predictions and simulation/real data. A considerable volume of oil remains in the reservoir after terminating the primary and secondary oil recovery stages. It is the target of enhanced oil recovery (EOR) processes to increase the overall oil displacement efficiency. Gas injection process is the most commonly used technique in EOR. There has always been a demand for optimizing the gas injection process through modifying the determining factors such as injection/production well placement, injection rate, composition of the injected gas phase, to name a few. In this study, a combination of artificial neural networks (ANN) and genetic algorithm (GA) was used to discover and then optimize the connections between different properties and variables of a reservoir which is undergoing a gas injection process. A commercial reservoir simulator was used to simulate the production performance of the target reservoir. A new objective function was also developed to optimize the gas injection process based on the final profits gained. According to the results, production maximization was not the best decision since it was not the most profitable alternative. However, a profit-based optimization considering the costs of gas injection and the oil price was proved to be a better solution to the optimization problem. For a five-year production period, employing the profit-based optimization for well productivity index calculation resulted in an earned profit of 5.7 × 109 USD, achieved through injection of 203,600 MMSCFD of gas. On the other hand, production maximization approach led to a lesser earned profit of 5.39 × 109 USD due to injection of much greater volume of gas (310,000 MMSCFD). The results of the ANN modelling demonstrated the ability of the hybrid smart model in predicting production performance of the target reservoir under gas injection process. In addition, the ANN model was capable of accurately predicting noise-containing data which makes it a highly reliable candidate to apply on real field data. [ABSTRACT FROM AUTHOR]