Auckland university of technology, auckland, new zealand fields of specialization. Accelerating physicsbased simulations using neural network. Available well logs and cores were used as inputs to the hybrid model. In this paper we propose an empirical analysis of deep recurrent neural networks rnns with stacked layers. Optimal operation of multi reservoir system using dynamic programming and neural network h. Reservoir computing with untrained convolutional neural. Machine learning applied to 3d reservoir simulation. We develop a proxy model based on deep learning methods to accel erate the simulations. The online version of the book is now complete and will remain available online for free.
Reservoir simulation is an area of reservoir engineering that, combining physics, mathematics, and computer programming to a reservoir model allows the analysis and the prediction of the fluid behavior in the reservoir over time. The training of the neural network is done using a supervised learning approach with the back propagation algorithm. Machine learning in reservoir production simulation and. Simulating reservoir operation using a recurrent neural. Introduction webster defines simulate as to assume the appearance of without reality. Artificial neural network for 4d reservoir modeling system.
Reservoir computing by felix grezes a dissertation submitted to the graduate faculty in computer science in partial ful llment of the requirements for the degree of doctor of philosophy, the city university of new york. It can be simply considered as the process of mimicking the behavior of fluid flow in a. Machine learning models to support reservoir production optimization. The architecture of the artificial neural network that is used in gros. Optimal operation of multireservoir system using dynamic. Datadriven reservoir modeling society of petroleum engineers. After the input signal is fed into the reservoir, which is treated as a black box, a simple readout mechanism is trained to read the. Artificial neural network and inverse solution method for. The paper gives a brief overview of neural networks and describes a feedforward, backpropagation network for geostatistical simulation. Geochemical equilibrium determination using an artificial neural network in compositional reservoir flow simulation. Each link has a weight, which determines the strength of one nodes influence on another. Preliminary concepts by the asce task committee on application of arti. In the suggested model, multi reservoir operating rules are derived using a neural network from the results of simulation. Such tools offer the user to obtain precise simulations of a given computational paradigm, as well as publishable.
Reservoir parameter estimation using a hybrid neural. Reservoir computing has attracted much attention for its easy training process as well as its ability to deal with temporal data. We have demonstrated the benefits of committee neural networks where predictions are redundantly combined. A system and method for modeling technology to predict accurately wateroil relative permeability uses a type of artificial neural network ann known as a generalized regression neural network grnn the ann models of relative permeability are developed using experimental data from waterflood core test samples collected from carbonate reservoirs of arabian oil fields three. Hydrologic applications by the asce task committee on application of arti. An artificial neural network consists of a collection of simulated neurons. For developing the ann models, three alternative networks i. A neural network based general reservoir operation scheme.
Machine learning in reservoir production simulation and forecast serge a. The book inspires geoscientists entrenched in first principles and engineering concepts to think. Neural networks is the archival journal of the worlds three oldest neural modeling societies. A sequence of 25 normalized 5 min rainfalls was applied as inputs to predict the runoff. Multiple linear regression and artificial neural networks. Additionally, a reasonable and effective reservoir operating plan is essential for realizing reservoir function. To explore the application of a deep learning algorithm on the field of reservoir operations, a recurrent neural network rnn, long shortterm memory. Reservoir properties from well logs using neural networks. Then, a suitable neural network architecture is selected and trained using input and. A new approach to reservoir characterization using deep. In the algorithm, a few simulation runs of different reservoir realizations are first made using 3level fractional factorial design.
This textbook is only one of the tools to teach the reservoir simulation techniques at the university and in post graduate courses efficiently. One is the pantai pakam timur field, located in northern sumatra, indonesia, where the data from only two wells were available and the other is iwafune oki field, located in the sea of japan, eastern japan, where wells were concentrated in the. Hjelmfelt and wang 1993ac developed a neural network based on the unit hydrograph theory. Simulation of petroleum reservoir performance refers to the construction and operation of a model whose behavior assumes the appearance of actual reservoir behavior. Highly inspired from natural computing in the brain and recent advances in neurosciences, they derive their strength and interest from an ac. The operation of the network is illustrated with two simple onedimensional examples which can be followed through with hand calculations to give an insight into the operation of the network. Basic applied reservoir simulation, textbook series request pdf. Artificial neural networks learn the nature of the dependency between input and output variables. The deep learning textbook can now be ordered on amazon. To explore the application of a deep learning algorithm on the field of reservoir operations, a recurrent neural network rnn, long shortterm memory lstm, and. Neural networkbased simulationoptimization model for.
Reservoir parameter estimation using a hybrid neural network. International journal of modeling and simulation for the petroleum industry, 9 1. Performing reservoir simulation with neural network enhanced data. To overcome this problem, in this study, a backpropagation neural network is trained to approximate the simulation model developed for the chennai city water supply problem. If a random recurrent neural network rnn possesses certain algebraic properties, training only a linear readout from it is often sufficient to. Pdf a neural network based general reservoir operation. Cascade, elman and feedforward back propagation were evaluated. Falguni parekh2 pg student, water resources engineering and management institute, faculty of technology and engineering, the maharaja sayajirao university of baroda. Soft computing for reservoir characterization and modeling. A neural network approach to geostatistical simulation. Ebook kalman filtering and neural networks as pdf download. Performing reservoir simulation with neural network enhanced. This model was then used to predict porosity and permeability for the reservoir and these values were then included in a reservoir simulation model. Current state of reservoir simulation and modeling of shale.
For instances, permeability prediction with artificial neural network modeling from well logs 2, reservoir parameter estimation using a hybrid neural network 3, application of fuzzy logic and. It introduces the basic concepts of soft computing techniques including neural networks, fuzzy logic and evolutionary computing applied to reservoir characterization. We employed matlabs neural network fitting toolbox to train a proxy neural network model. Stateoftheart coverage of kalman filter methods for the design of neural networks this selfcontained book consists of seven chapters by expert contributors that discuss kalman filtering as applied to the training and use of neural networks. Performing reservoir simulation with neural network. Predicting reservoir water level using artificial neural network. Read the latest articles of journal of petroleum science and engineering at, elseviers. Location of the study area and the rain gauge stations network. Optimal design of the neural network modules and the size of the training set. These optimizations are very demanding computationally due. Those of you who are up for learning by doing andor have. Reduced order reservoir simulation with neuralnetwork based. This course has a supplemental book located in our spe bookstore entitled datadriven reservoir modeling. Falguni parekh2 pg student, water resources engineering and management institute, faculty of technology and engineering, the maharaja sayajirao university of baroda, samiala391410, vadodara, gujarat, india1 offg.
Although the traditional approach to the subject is almost always linear, this. Oil reservoir simulation, artificial neural networks. Reservoir systems operation model using simulation and neural. Saptono 35 mapping the gas column in an aquifer gas. Levenbergmarquardt training algorithm was used for training a neural network architecture with one hidden layer and thirty hidden neurons. Monte carlo simulation and artificial neural network are applied to two areas for predicting the distribution of reservoirs. Pdf artificial neural networks for predicting petroleum quality. Mohaghegh 2000 noted pattern recognition as one of the neural networks strengths. The available porosity and permeability data needed to build a reservoir simulation model are old and sparse. The arti cial neural network paradigm is a major area of research within a.
Softcomputingforreservoircharacterizationandmodeling. This paper presents a study aimed at forecasting water level of reservoir using neural network approaches. The readers have to work with a sophisticated reservoir simu lator to deepen their theoretical knowledge too. An introduction naveen kuppuswamy, phd candidate, a. Applying machine learning algorithms to oil reservoir. Journal of petroleum science and engineering neural network. A system and method for modeling technology to predict accurately wateroil relative permeability uses a type of artificial neural network ann known as a generalized regression neural network grnn the ann models of relative permeability are developed using experimental data from waterflood core test samples collected from carbonate reservoirs of arabian oil fields three groups of data sets. In the suggested model, multireservoir operating rules are derived using a neural network from the results of simulation. Applying machine learning algorithms to oil reservoir production optimization mehrdad gharib shirangi stanford university abstract in well control optimization for an oil reservoir described by a set of geological models, the expectation of net present value npv is optimized. Professor shahab mohaghegh, being one of the most innovative and experienced thought leaders in the field of datadriven modeling in the upstream, has written a comprehensive and readable book that finally puts to bed the persistent complaints in the industry. Prediction of reservoir properties by monte carlo simulation. This study opened up several possibilities for rainfallrunoff application using neural networks. Application of artificial neural networks for calibration. Pdf a neural network based general reservoir operation scheme.
Abstract the use of artificial neural networks ann for reservoir analysis now makes it possible to predict important reservoir properties from combinations of data such as well logs, production data, seismic data, etc. Reservoir computing emerges as a solution, o ering a generic. Stochastic reservoir simulation using neural networks. Mohaghegh 2000 noted pattern recognition as one of the neural network s strengths. A critical analysis claudio gallicchio and alessio micheli department of computer science, university of pisa largo bruno pontecorvo 3 56127 pisa, italy abstract. Graupe, 2007, and are being successfully applied across. Predicting reservoir water level using artificial neural network shilpi rani1, dr. Application of artificial neural networks for calibration of. A neural network approach to geostatistical simulation pdf. Application of artificial neural networks for reservoir.
Optimal operation of multireservoir system using dynamic programming and neural network h. Deltav neural gives you a practical way to create virtual sensors for measurements previously available only through the use of lab analysis or online analyzers. Recurrent networks o er more biological plausibility and theoretical computing power, but exacerbate the aws of feedforward nets. Both models offer avenues for predicting reservoir capacity at gauged sites without the expense of timeseries based simulation alternatives. In quantitative geological modeling with an artificial neural network approach, time information can be considered as input variable to better describe dynamic evolution patterns of reservoir parameters. This paper forms the second part of the series on application of arti. Machine learning in reservoir production simulation and forecast. Reservoir systems operation model using simulation and.
Data driven reservoir modeling, also known as topdown model tdm, is an alternative to the. Can we upscale original problem of reservoir simulation to the level of functional dependence between observed outputs e. Such approach requires independent programming to implements 4d reservoir modeling systems and thus introduces a new research area with great development potential. Multiple linear regression and artificial neural networks models for generalized reservoir storageyieldreliability function for reservoir planning. Reservoir simulation process reservoir simulation is briefly. Abstract a combined approach of a dynamic programming algorithm and artificial.
Application of machine learning and artificial intelligence in. Reservoir modeling uses all available information which includes at a minimum logs data, and fluid and rock properties. It contains stateoftheart techniques to be applied in reservoir geophysics, well logging, reservoir geology, and reservoir engineering. A neural net can be learned to collect multiple point statistics. Seismic characterization prediction of reservoir properties by monte carlo simulation and artificial neural network in the exploration stage k. One is the pantai pakam timur field, located in northern sumatra, indonesia, where the data from only two wells were available and the other is iwafune oki field, located in the sea of japan, eastern japan, where wells were concentrated in the central part of. Reservoir computing, recurrent neural network learning architectures, agent architectures, machine learning applications. The book describes how to utilize machinelearningbased algorithmic protocols to reduce large quantities of difficulttounderstand data down to. The reservoir is an important hydraulic engineering measure for human utilization and management of water resources. In this twopart series, the writers investigate the role of arti. Artificial intelligence and in particular artificial neural networks ann. Development and application of reservoir models and. Physicsbased models and data models introduction neural computations such as artificial neural networks ann have aroused considerable interest over the last decades e. Reservoir computing is a framework for computation derived from recurrent neural network theory that maps input signals into higher dimensional computational spaces through the dynamics of a fixed, nonlinear system called a reservoir.
Terekhov neurok techsoft, llc, moscow, russia email. In this study, a deep learning neural network was developed to estimate the petrophysical characteristics required building a full field earth model for a large reservoir. The most common formulation used in reservoir simulation is the. The deep learning textbook is a resource intended to help students and practitioners enter the field of machine learning in general and deep learning in particular. Caudill presented a comprehensive description of neural networks in a series of papers caudill, 1987, 1988, 1989. A schematic diagram of this process is shown in figure 1. Shahabs book mohaghegh, datadriven reservoir modeling, 2017, the topdown model. Spe member price usd 120 datadriven reservoir modeling introduces new technology and protocols intelligent systems that teach the reader how to apply data analytics to solve realworld, reservoir engineering problems. This paper presents an artificial neural network ann approach for forecasting of reservoir water level using ten daily data of inflow, water level and release. The performance is analyzed using a simulation model for the. Neural networks can discover highly complex relationships between several variables that are presented to the network. Predicting reservoir water level using artificial neural. A subscription to the journal is included with membership in each of these societies.
Machine learning applied to 3d reservoir simulation marco a. This paper considers the development of a computationally fast model for simulation of. Reduced order reservoir simulation with neuralnetwork based hybrid model. Modeling and simulation, computational systems biology, bioinformatics. Pdf deep neural networks predicting oil movement in a. A neural network is characterized by its architecture that represents the pattern of connection between nodes, its method of determining the connection weights, and the activation function fausett 1994. While the larger chapters should provide profound insight into a paradigm of neural networks e.
Hou 15 application of neural networks in determining petrophysical properties from seismic survey b. Datadriven reservoir modeling reservoir analytics is defined as the. In many cases, complex simulation models are available, but direct incorporation of them into an optimization framework is computationally prohibitive. Cardoso 1 introduction the optimization of subsurface. This allowed direct simulation of the trained neural network to obtain an updated reservoir parameters. A reservoir computing system consists of a reservoir part represented as a sparsely connected recurrent neural network and a readout part represented as a simple regression model. Mathematical model computer codes numerical model physical model figure 1. For multireservoir operating rules, a simulationbased neural network model is developed in this study. Novel connectionist learning methods, evolving connectionist systems, neurofuzzy systems, computational neurogenetic modeling, eeg data analysis, bioinformatics, gene data analysis, quantum neurocomputation, spiking neural networks, multimodal information processing in the brain, multimodal neural network. For porosity prediction we have made a study initially with a single neural network and then by the cm approach. Deltav neural is easy to understand and use, allowing process engineers to produce extremely accurate results even without prior knowledge of neural network theory. The better solutions found by the ga were tested with.
1421 1545 741 802 263 642 982 1133 1167 589 1376 49 715 96 933 1278 1308 189 1313 876 775 587 913 129 1384 199 102 1147 469 1162 1070 404 1235 1304 591 461 263 462 887 427 1353 1369 570