Fig 12 - uploaded by Mashallah Rahimi
Content may be subject to copyright.
Schematic of a MLP which shows the forward and backward passes of the error back propagation training algorithm (after Saemi, et al., 2007).

Schematic of a MLP which shows the forward and backward passes of the error back propagation training algorithm (after Saemi, et al., 2007).

Source publication
Article
Full-text available
3D seismic data interpretation plays a key role in identifying Lithofacies and their lateral changes for hydrocarbon reservoirs exploration. Among mathematical analysis techniques, Artificial Neural Network (ANN) offers superior handling over inherent non-linearity of seismic data. Here we applied multi-attribute analysis based on ANN methods and w...

Context in source publication

Context 1
... back-error propagation algorithm is one of the most common- ly used methods in reservoir characterization. Back-error propagation type networks which are also called multi-layer perceptron (MLP) have consisted of neurons with nonlinear activation function (Fig. 12). For classification, the input is the attribute vector and the output would be the classes. The generality of the classifier is directly linked to the number of hidden layers. Hidden layers are internal to the network and have no direct contact with the external environ- ment. Sometimes they are likened to a "black box" within the ...

Similar publications

Article
Full-text available
Data on heterogeneity of porous space and reservoir filtration properties determines the choice of most effective development strategy. The comparative analysis of well-logging and well-test results was carried out to determine the value of filtration heterogeneity. The similarity of the results obtained using different methods makes it possible to...
Poster
Full-text available
Information of subsurface temperature distribution in the Danish area originates from direct measurements in boreholes and from indirect theoretical modelling. “Point observations” of varying quality are available as industrially measured “Bottom Hole Temperatures” from deep exploration boreholes, and accurate continuous equilibrium temperature log...
Article
Full-text available
The study aimed at predicting the porosity of reservoir sands in ‘Arike field’ Niger Delta, Nigeria by converting seismic trace of the interval of interest in the seismic survey into a porosity log to generate a porosity volume. Optimal number of relevant attributes were selected using multi-attribute analysis. The study discovered that three attri...
Article
Full-text available
In this contribution, we describe the formation and evolution of the upconing process in a Mediterranean coastal aquifer. The study area has experienced severe salinization over the last 40 years because of intensive exploitation of groundwater. We used historical and current records of piezometric levels and chloride concentrations to trace the de...
Chapter
Full-text available
The integration of regional 3D seismic, well logs, cores, and biostratigraphic data gathered in a sequence stratigraphic approach provide a basis for interpreting the Oligo-Miocene and Miocene system as a whole. This stratigraphic interval, which contains the most of oil reserves in Campos basin turbidites, was divided into third and fourth order s...

Citations

... Machine learning has been the most popular focus of research in recent years and has shown its powerful data mining capabilities in almost every field, including geosciences (Konate et al. 2015;Saporetti et al. 2018Saporetti et al. , 2019Sun et al. 2019;Asante-Okyere et al. 2020). Many machine learning techniques are applied in the study of automatic lithology identification, such as Adaboost (Han et al. 2021), RF (Ao et al. 2019), SVM (Xie et al. 2018;Bressan et al. 2020;Zheng et al. 2021) and artificial neural networks (Raeesi et al. 2012;Asante-Okyere et al. 2020). Among them, SVM has received attention due to its excellent performance in lithology identification. ...
Article
Full-text available
Lithology identification is an important task in oil and gas exploration. In recent years, machine learning methods have become a powerful tool for intelligent lithology identification. To address the redundancy of conventional logging data and unbalanced distribution among formation lithology classes due to the complexity of depositional environment and inhomogeneity of subsurface space, this paper investigates the affiliation-weighted one-to-one support vector machine (WOVOSVM) lithology identification method based on geochemical logging data. This method uses geochemical logging data, which can directly reflect the formation lithology information, as input, and achieves intelligent and accurate lithology classification under the calculation of WOVOSVM. In this study, Shahezi Formation of Songke 2 Well in Songliao Basin, China is taken as the experimental object, and two data sets with different distribution characteristics are selected as the input. Use WOVOSVM, Adaboost, random forest (RF) and traditional support vector machine (SVM) to identify lithology, and compare and analyze the results. The results are as follows: (1) Accuracy metrics of most of the four classification models were above 60%, indicating the geochemical logging data can effectively reflect the formation lithology information, which is a reliable indicator for the intelligent identification of logging lithology. (2) When the data set has a strong imbalance, the lithology recognition performance of WOVOSVM is better than other methods, the average value of accuracy metrics is more than 72%, F1 value is 8.77% to 14.56% higher than other models, especially in the small sample lithology category recognition, 70% of the samples are correctly classified.
... Because there are no available methods for the direct measurement of these lithological parameters, indirect measurement through computation from geophysical logs are relied upon to identify different zones of interest from preliminary well log analysis and to integrate these zones with seismic attributes in the region (Chaki, 2015;Paolo et al., 2017;Alnuaimi, 2018;Xie et al., 2018Xie et al., , 2020Dunham et al., 2020b). Well log information and seismic attributes are ideally integrated at reservoir intervals to correlate the good facies and to prepare a reliable reservoir depositional model with improved probabilities of success; however, this mapping between lithological properties and seismic attributes is governed by a nonlinear relationship and more often incorporates uncertainty in heterogeneous facies correlation (Raeesi et al., 2012;Tschannen et al., 2017;Pires de Lima et al., 2019;Kim et al., 2020). Besides, this task of reservoir characterization is laden with high uncertainty due to its association with complex operations such as data mining, data fusion as well as the formulation of knowledge base and inventory (Tschannen et al., 2017;Belozerov et al., 2018;Tian, 2018;Tewari & Dwivedi, 2020). ...
Article
This work critically evaluated the applicability of machine learning methodology applied to automated well log creation towards reliable lithology prediction and subsequent reservoir characterization to overcome the computationally intensive and laborious manual analysis of well logs for improving the cost of exploration, better accuracy of predictions, as well as improved efficiency of production operations. We propose an approach by deploying a correlation tool for parametric correlation between parameters and implementation of machine learning algorithms for understanding hidden meaningful relationships between the feature and target variables for the development of a predictive model. Clustering classification and label extraction followed by dimensionality reduction using principal components analysis were performed in the unsupervised mode. The calibration of the functional relationship between the predictor attributes and target properties by determining the most powerful predictor influencing the target parameters helped to derive interpretations with greater degree of certainty to develop a reliable predictive model using regression analysis in supervised mode. The results were based on a set of five drilled core holes, namely, Rutgers, Somerset, Martinsville, Princeton, and Weston Canal of Newark South Basin, USA, a publicly available set of well logs at the Southern Newark Basin section at the Lamont- Doherty Earth Observatory. The data were split 80:20 into a training set and a test set for evaluation. The evaluation carried out using the R-squared and the root mean squared error to validate the proposed approach revealed values of 0.726 and 0.232, respectively, for the training set and 0.723 and 0.186, respectively, for the testing set. A comparative analysis with state-of-the-art methods such as k-nearest neighbor, support vector regression (SVR), and multivariate regression was also performed. This implementation demonstrated the proposed model�s efficacy well beyond the conventional methods. KEY WORDS: Machine learning, Data visualization, Well logs, Reservoir characterization, Lithology prediction, Petro-physics.
... The growth of 3D seismic attributes has led to the accurate interpretation of sub-surface petroleum systems with regards to lithology, faults, and engineering properties of stratigraphic traps, such as the sandstone-filled incised-valley sandstone (Chen and Sidney, 1997;Li et al., 2011;Raeesi et al., 2012;Asim et al., 2016;Vohs, 2016). Numerous stratigraphic analyses have demonstrated their applications in resolving and characterizing the stratigraphic trap such as the fluvial channel sandstone, point bars, pinch outs, and lateral changes in lithology deriving the channel sandstones of fluvial meandering stream, braided stream, and amalgamated channels (Chen and Sidney, 1997;Carter, 2003;Chopra and Alexeev, 2006;Lazar, 2009;De Matos et al., 2011;Li, 2014;Benard et al., 2014;Fang et al., 2015;Partyka et al., 1999;Marfurt, 2005, 2007;Barbato, 2012;Qi and Castagna, 2013;Li and Lu, 2014;Tayyab et al., 2017;Li et al., 2015;Li et al., 2016). ...
Article
Seismic properties can assume a vigorous part in the portrayal of the incised-valley reservoirs the world over. But, accurately optimizing the attributes within the stratigraphically heterogeneous fluvial system is significant, and thus, delineating the porous sandstone is the challenge for prospect delineation. During the forecast of stratigraphic traps, the resolve is to select the frequency band, which can accurately image the thickest part of the reservoirs. The continuous wavelet transforms (CWT) of spectral decomposition (SD) tool is applied for enlightening the incised valley from Miano gas field, SW Pakistan. The seismic and root-mean-square (RMS) amplitudes deliver insufficient reservoir qualities. The CWT technique detects the low-frequency shadow zone at 37 Hz, which indicates the presence of a gas-bearing zone along the fault. The instantaneous frequency model detects a ~35 m thick sandstone lens encased within the erosional sediments, which was corroborated by a 37 Hz tuning cube. The acoustic impedance-based instantaneous spectral thickness model brightens the high-porosity zones within the thick valley-plugged reservoirs of 40 m and 30 m at 37 Hz and 48 Hz.
... Integration of 3D seismic dataset with reservoir information, formation evaluation analyses and geological results are the accurate plans in oil and gas discoveries and production. 3D seismic interpretation along with wireline logs are the two significant elements in describing internal and external complexities of a reservoir and creation of stratigraphic and structural aspects of oil and gas reservoirs (Johann et al. 2001;Raeesi et al. 2012). 3D seismic dataset used in reservoir characterization to determine the distribution of reservoir features including geological structures, sedimentary facies and architectural component (Yao and Chopra 2000;Abreu et al. 2003;Jian and Fanhua 2009;Rezvandehy et al. 2011;Perez-Munoz et al. 2013;Kadkhodaie-Ilkhchi et al. 2014) leads to decrease in the risk factors in oil and gas industry (Omoja and Obiekezie 2019). ...
Article
Full-text available
The Oligo-Miocene-aged Asmari Formation hosts a large volume of hydrocarbon in Iran. In Hendijan Field, the Asmari Formation composed of both siliciclastic and carbonate lithology. The Ghar Member is the most prolific zone within the Asmari Formation which is subdivided into three sandy zones that are separated by thin shaly layers. Petrographic studies revealed that the Ghar interval formed mainly of sub-sorted to well-sorted quartz-arenite, quartz-wack and low amount of dolomite and clay. This study attempts to discriminate the Ghar sand utilizing both petrophysical and seismic dataset and modeling of this interval in terms of facies within the Asmari reservoir using sequential indicator simulation (SIS) algorithm. In addition, assigning the Ghar sand interval to the equal flow units obtained by the Stratigraphic Modified Lorenz Plot method (SMLP). Various cross-plots were employed to detecting the Ghar sand by logs. The consequences of such analyses showed that the Ghar zone was characterized by high value of porosity and, generally, low value of impedance properties. In addition, an accurate pre-stack inversion detects the Ghar sand (using Odegaard-Avseth cross-plot) by lower impedance value from the other high-impedance carbonate parts. Hydraulic flow unit (HFU) analysis also has been conducted based on poro-perm information. The consequences indicated that the HFU-2 and HFU-4 have the best reservoir quality. Eventually, 3D facies modeling of Asmari Formation has been constructed based on SIS method. The results illustrated that the Ghar Member is the best interval of the Asmari Formation and considered as equal to HFU-2 and HFU-4.
... To clarify the distribution rules of buried hill cavities, the following research ideas are designed: the short-time Fourier transform (STFT) is applied to seismic data to obtain a single frequency attribute body of multiple frequencies, combined with along-layer slicing technology, to identify the cavity plane distribution of various time windows (theoretically, the frequency division information can reflect the aperture of the 1/4 wavelength specification) (Raeesi et al., 2012;Tian et al., 2016;Li et al., 2019;Yang et al., 2021). ...
Article
Full-text available
Predicting the distribution of fractured and dissolved reservoirs in granitic buried hills is a prerequisite for finding high-quality hydrocarbon reservoirs and improving the success rate of exploration activities. Granitic buried-hill reservoirs in the Bongor Basin are characterized by strong heterogeneity and poor lateral continuity. Through the analysis of seismic, core, logging and dynamic production data, this study predicts the distribution of fractured reservoirs and fractured-cave reservoirs in the granitic buried hills in the BC Block of the Bongor Basin. In this study, we predict fractured reservoirs using coherence cubes and linearly enhanced attributes and identify fractured-cave reservoirs with single-frequency attribute bodies. Integrated predictions of fractured reservoirs and fractured-cave reservoirs are conducted using attribute fusion techniques. With these methods, good results have been achieved in reservoir prediction in BC buried hills. Furthermore, this study summarizes a set of granitic buried-hill reservoir prediction techniques for densely inverted rift basins in the Central African Rift System.
... Neural network is an example of such methods that has been used to capture the implicit complex patterns and relations in data sets and use them to derive predictions. They are designed to imitate the communication of neurons in the brain, and have been applied to solve various problems in the petroleum industry, including history matching (Razak and Jafarpour 2020; Mohd Razak and Jafarpour 2020a), enhanced oil recovery (Parada and Ertekin 2012;Amirian et al. 2013), reservoir characterization (Raeesi et al. 2012;Alizadeh et al. 2012), fracture characterization (Malallah and Nashawi 2005), and reservoir management and optimization (Mohaghegh et al. 2014;Zhong et al. 2020). ...
Article
The flow and transport processes that take place during hydrocarbon production from hydraulically fractured unconventional reservoirs are not well understood. As a result, current simulators cannot provide reliable predictions of the production behavior in the field. In addition to imperfect physics, the prediction errors can be caused by the inability to conveniently integrate important field data, such as well logs, drilling, and completion parameters, into existing physical models. A neural network (NN) model is developed to learn the (residual) errors in simulation-based production prediction as a funcation of input parameters of an unconventional well. Once trained, the NN model augments the physics-based predictions by adding the learned reiodual to predict the production response of a new well. To learn the discrepancy between the simulated and observed production data, the NN model is trained using a labeled dataset consisting of the prediction errors (as labels) and the corresponding input parameters (features), such as formation, completion, and fluid properties. During training, a mapping is identified from the input parameters to their respective prediction errors. To facilitate the residual learning, first a convolutional autoencoder architecture is used to map the simulated and observed production responses to a low-dimensional latent space. This step is followed by a regression model that learns the mapping between the collected field parameters and the corresponding latent space representation of the prediction errors. The two steps are included in a single NN architecture and trained simultaneously. The proposed residual learning method is designed to compensate for prediction errors originating from a combination of imperfect representation of the physics and inaccurate simulation inputs, including uncertain descriptions of the reservoir and fracture properties. The performance of the proposed residual learning approach is evaluated using synthetic data as well as a field case study from the Bakken play in North Dakota.
... Neural network Algorithms have nonlinear, adaptive learning capabilities that allow them to handle complex, multidimensional, and noisy logging data. Raeesi et al. (2012) used BP neural network to determine the petrography of their study area based on seismic data, logging data, and geological data. Puskarczyk (2019) compared artificial neural networks and related improved algorithms with commonly used machine learning algorithms. ...
Article
Full-text available
Traditional lithology identification left the problems of low accuracy, recognition efficiency and generalization ability. Facing the logging data with outliers, unbalance and high complexity, we propose a lithology identification method based on an improved neighborhood rough set and AdaBoost. On the basis of the classical neighborhood rough set, the selection of the neighborhood radius and the running time are optimized. The redundant information in logging data is then effectively eliminated. Thus more sensitive logging curves are selected without changing the physical meaning of logging attributes. Then the selected data are input into the AdaBoost model to construct a lithology identification model. About 54,000 samples from 5 boreholes are tested in the study area. The accuracy of classification on the test set is about 98.42%. Compared with BP neural network and random forest algorithm, the proposed method owns advantages in recognition accuracy and generalization ability. It can provide help for complex lithology recognition in the study area.
... Logging curves are commonly considered to be the most efficient and accurate method of quantitatively evaluating reservoir heterogeneity (Raeesi et al., 2012;Amoura et al., 2019;Yarmohammadi et al., 2020). However, most methods focus on using conventional types of logs (i.e., gamma, resistivity, sonic, density, etc), while image log data is mostly ignored even though it can directly reflect the sedimentary characteristics of reservoirs. ...
Article
Full-text available
The low permeability sandstone reservoir in the Ordos Basin displays heterogeneity with sedimentation and tectonic origins, which is mainly manifest by interbedding of sandstone and mudstone, bedding, and fractures (). There is a clear difference between this type of heterogeneity and pore heterogeneity and diagenetic heterogeneity. At present, academia pays less attention to this kind of heterogeneity and lacks a quantitative evaluation method. The imaging log can describe this kind of heterogeneity directly. The Tamura texture features (TTF) method was used to calculate the roughness of different heterogeneous intervals. It is found that the fracture has the largest roughness, followed by the oblique bedding and the horizontal bedding section, and the massive bedding has the smallest roughness. The GR curve roughness calculated by EMD is consistent with that calculated by TTF. Therefore, TTF can be used to quantitatively evaluate the heterogeneity of low permeability sandstone reservoirs based on the imaging log when the imaging log has the same size. The roughness of the imaging log calculated by the TTF method has a strong coupling with the sedimentary cycle. This method is accurate, objective, and easy to understand. This is another important application of TTF in addition to quantitative evaluation of the heterogeneity of low permeability sandstone reservoirs.
... Most researchers tend to predict the reservoir properties, which can then classify the net reservoir or pay. Such research includes studies by Raeesi et al. [13], where seismic attributes, wireline logs, and artificial neural networks were used to investigate the identification and classification of reservoir lithofacies and their inherent heterogeneities. Their results concluded that not all seismic attributes are relevant for facies classification and that the attributes need to be related to the reservoir's geological processes and petrophysical properties. ...
Article
Accurate net pay classification is essential in hydrocarbon resource volumetric calculation. However, there is no universal methodology developed for its evaluation hence the existence of many incongruent views on its application since it is data-driven and differs for each reservoir. This research incorporates machine learning and data analytics in predicting net pay, intending to reduce uncertainties associated with the net-pay classification. Log analysis was performed to determine the cut-offs for sonic, neutron, density, and gamma-ray using unsupervised learning and data analytics. The log cut-offs were calculated with petrophysical properties; shale volume, water saturation, permeability, and porosity. A Bayesian Optimised Extreme Gradient Boosting (Bayes Opt-XGBoost) model was applied to predict the petrophysical properties using five wireline logs. The model’s performance and a computational function in classifying net reservoir resulted in an accuracy of 0.93, a combined precision of 0.94, a combined recall of 0.92, and a combined F1-score of 0.93. The model and methodology were deployed on a new well for validation. The classification of net reservoir zones via the proposed data analytics method, Bayes Opt-XGBoost predicted petrophysical properties, and computational function code matched mobility drawdown test data for the well. These results indicate that the developed methodology and machine learning model can work for other reservoirs since the additional computational function code can be manipulated for any data-driven estimated cut-offs. This developed approach can determine net reservoir and net pay zones in any sandstone reservoir.
... Most researchers tend to predict the reservoir properties, which can then be used to classify the net reservoir or pay. Such research includes studies by Raeesi et al. [11], where seismic attributes, wireline logs and artificial neural networks were used to investigate the identification and classification of reservoir lithofacies and their inherent heterogeneities. Their results concluded that not all seismic attributes are relevant for facies classification and that the attributes need to be related to the reservoir's geological processes and petrophysical properties. ...
Preprint
Accurate classification of net pay is of high importance in hydrocarbon resource volumetric calculation. However, there is no universal methodology developed for its evaluation hence the existence of many incongruent views on how it is applied since it is data-driven and differs for each reservoir. This research incorporates machine learning and data analytics in predicting net pay, intending to reduce uncertainties associated with the net-pay classification. Log analysis was performed to determine the cutoffs for sonic, neutron, density and gamma ray using unsupervised learning and data analytics. Petrophysical properties; shale volume, water saturation, permeability and porosity, were calculated using the log cutoffs. A Bayesian Optimised Extreme Gradient Boosting (Bayes Opt-XGBoost) model was applied to predict the petrophysical properties using five wireline logs. The performance of the model along with a computational function in classifying net reservoir resulted in an accuracy of 0.93, a combined precision of 0.94, a combined recall of 0.92 and a combined F1-score of 0.93. The model and methodology were deployed on a new well for validation. The classification of net reservoir zones via the Bayes Opt-XGBoost predicted petrophysical properties and computational function code matched mobility drawdown test data for the well. This indicates that the developed methodology and machine learning model can work for other reservoirs