Fracture identification of carbonate reservoirs by deep forest model:An example from the D oilfield in Zagros Basin

Identifying fractures along a well trajectory is of immense significance in determining the subsurface fracture network distribution.Typically,conventional logs exhibit responses in fracture zones,and almost all wells have such logs.However,detecting fractures through logging responses can be challenging since the log response intensity is weak and complex.To address this problem,we propose a deep learning model for fracture identification using deep forest,which is based on a cascade structure comprising multi-layer random forests.Deep forest can extract complex nonlinear features of fractures in conventional logs through ensemble learning and deep learning.The proposed approach is tested using a dataset from the Oligocene to Miocene tight carbonate reservoirs in D oilfield,Zagros Basin,Middle East,and eight logs are selected to construct the fracture identification model based on sensitivity analysis of logging curves against fractures.The log package includes the gamma-ray,caliper,density,compensated neutron,acoustic transit time,and shallow,deep,and flushed zone resistivity logs.Experiments have shown that the deep forest obtains high recall and accuracy(>92%).In a blind well test,results from the deep forest learning model have a good correlation with fracture observation from cores.Compared to the random forest method,a widely used ensemble learning method,the proposed deep forest model improves accuracy by approximately 4.6%.

Identification of low-resistivity-low-contrast pay zones in the feature space with a multi-layer perceptron based on conventional well log data

In the early exploration of many oilfields,low-resistivity-low-contrast(LRLC)pay zones are easily overlooked due to the resistivity similarity to the water zones.Existing identification methods are model-driven and cannot yield satisfactory results when the causes of LRLC pay zones are complicated.In this study,after analyzing a large number of core samples,main causes of LRLC pay zones in the study area are discerned,which include complex distribution of formation water salinity,high irreducible water saturation due to micropores,and high shale volume.Moreover,different oil testing layers may have different causes of LRLC pay zones.As a result,in addition to the well log data of oil testing layers,well log data of adjacent shale layers are also added to the original dataset as reference data.The densitybased spatial clustering algorithm with noise(DBSCAN)is used to cluster the original dataset into 49 clusters.A new dataset is ultimately projected into a feature space with 49 dimensions.The new dataset and oil testing results are respectively treated as input and output to train the multi-layer perceptron(MLP).A total of 3192 samples are used for stratified 8-fold cross-validation,and the accuracy of the MLP is found to be 85.53%.

Fracture identification and evaluation using conventional logs in tight sandstones:A case study in the Ordos Basin,China

Fractures are of great significance to tight oil and gas development.Fracture identification using conventional well logs is a feasible way to locate the underground fractures in tight sandstones.However,there are three problems affecting its interpretation accuracy and practical application,namely weak well log responses of fractures,a lack of specific logs for fracture prediction,and relative change omission in log responses.To overcome these problems and improve fracture identification accuracy,a fracture indicating parameter(FIP)method composed of a comprehensive index method(CIM)and a comprehensive fractal method(CFM)is introduced.The CIM tries to handle the first problem by amplifying log responses of fractures.The CFM addresses the third one using fractal dimensions.The flexible weight parameters corresponding to logs in the CIM and CFM make the interpretation possible for wells lacking specific logs.The reconstructed logs in the CIM and CFM try to solve the second problem.It is noted that the FIP method can calculate the probability of fracture development at a certain depth,but cannot show the fracture development degree of a new well compared with other wells.In this study,a formation fracture intensity(FFI)method is also introduced to further evaluate fracture development combined with production data.To test the validity of the FIP and FFI methods,fracture identification experiments are implemented in a tight reservoir in the Ordos Basin.The results are consistent with the data of rock core observation and production,indicating the proposed methods are effective for fracture identification and evaluation.

A formation pressure prediction method based on tectonic overpressure

Traditional formation pressure prediction methods all are based on the formation undercompaction mechanism and the prediction results are obviously low when predicting abnormally high pressure caused by compressional structure overpressure.To eliminate this problem,we propose a new formation pressure prediction method considering compressional structure overpressure as the dominant factor causing abnormally high pressure.First,we establish a model for predicting maximum principal stress,this virtual maximum principal stress is calculated by a double stress field analysis.Then we predict the formation pressure by fitting the maximum principal stress with formation pressure. The real maximum principal stress can be determined by caculating the sum of the virtual maximum principal stresses.Practical application to real data from the A1 and A2 wells in the A gas field shows that this new method has higher accuracy than the traditional equivalent depth method.

The application of machine learning under supervision in identification of shale lamina combination types——A case study of Chang 7_(3)sub-member organic-rich shales in the Triassic Yanchang Formation,Ordos Basin,NW China

Organic rich laminated shale is one type of favorable reservoirs for exploration and development of continental shale oil in China.However,with limited geological data,it is difficult to predict the spatial distribution of laminated shale with great vertical heterogeneity.To solve this problem,taking Chang 73 sub-member in Yanchang Formation of Ordos Basin as an example,an idea of predicting lamina combinations by combining'conventional log data-mineral composition prediction-lamina combination type identification'has been worked out based on machine learning under supervision on the premise of adequate knowledge of characteristics of lamina mineral components.First,the main mineral components of the work area were figured out by analyzing core data,and the log data sensitive to changes of the mineral components was extracted;then machine learning was used to construct the mapping relationship between the two;based on the variations in mineral composition,the lamina combination types in typical wells of the research area were identified to verify the method.The results show the approach of'conventional log data-mineral composition prediction-lamina combination type identification'works well in identifying the types of shale lamina combinations.The approach was applied to Chang 73 sub-member in Yanchang Formation of Ordos Basin to find out planar distribution characteristics of the laminae.

Prediction of new perforation intervals in a depleted reservoir to achieve the maximum productivity: A case study of PNN logging in a cased-well of an Iranian oil reservoir

Pulsed neutron-neutron(PNN)logging is based on emitting neutrons into the near-wellbore zone and computing the neutron count decay due to scattering and capturing.The main application of this logging tool is to determine the current oil saturation and to detect channeling in perforated and non-perforated intervals behind the casing.Correct interpretation of the results obtained from PNN logging enables engineers to predict new perforation intervals in depleted reservoirs.This study examines the application of PNN logging in a well located in one of Iranian oil reservoirs.The interpretation procedure is described step by step.The principle of the PNN logging and the specifications of the tool are discussed and the applications of PNN logging in evaluation of oil saturation,identification of water flooded zones and prediction of potential perforating zones are described.Channeling is also investigated between all layers,good and poor oil zones are characterized based on the calculated oil saturations and new perforation intervals are suggested with the aim to boost oil production from the reservoir.The results of this study show that zones 1 to 5 having low oil saturations,are interpreted as depleted oil zones.Zones 6 to 8 are interpreted as good oil zones having high potential to produce oil.Zone 9 is interpreted as a water zone.

Fracture permeability estimation utilizing conventional well logs and flow zone indicator

Characteristics of the natural open fractures on the oil and gas reservoirs is crucial in drilling and production planning. Direct methods of fractures studies such as core analysis and image log interpretation are usually not performed in all drilled wells in a field. Therefore, in absence of these data, the indirect methods can play an important role. In this study, an integrated algorithm is introduced to identify the fractures and estimate its permeability employing conventional well logs. First, open fractures were identified and their properties including density, aperture, porosity and permeability were estimated using FMI log. Subsequently, the fracture index log (FR_Index) was estimated utilizing conventional logs including density, micro-resistivity, sonic (compressional, shear and stoneley slownesses), and caliper logs. After that, the fracture index permeability was estimated by improving the FZI permeability equation. The coherence coefficient between two estimated fracture permeability logs is 0.66. A good correlation is observed on the high permeability zones, but the lower correlation on the low permeability zones. It is notified that, in the high fracture permeability zones, the conventional logs are heavily impacted by fracture permeability. However, due to lower vertical resolution of conventional logs compared with the image logs, the conventional logs are less influenced by less dense fracture zones. However, this algorithm can be used with acceptable accuracy in all uncored and image log wells.

A New Empirical Method for Constructing Capillary Pressure Curves from Conventional Logs in Low-Permeability Sandstones

Pore structure reflected from capillary pressure curves plays an important role in low-permeability formation evaluation. It is a common way to construct capillary pressure curves by Nuclear Magnetic Resonance（NMR） log. However, the method＇s efficiency will be severely affected if there is no NMR log data or it cannot reflect pore structure well. Therefore, on the basis of J function and diagenetic facies classification, a new empirical model for constructing capillary pressure curves from conventional logs is proposed here as a solution to the problem. This model includes porosity and the relative value of natural gamma rays as independent variables and the saturation of mercury injection as a dependent variable. According to the 51 core experimental data sets of three diagenetic facies from the bottom of the Upper Triassic in the western Ordos Basin, China, the model＇s parameters in each diagenetic facies are calibrated. Both self-checking and extrapolation tests show a positive effect, which demonstrates the high reliability of the proposed capillary pressure curve construction model. Based on the constructed capillary pressure curves, NMR T_2 spectra under fully brine-saturated conditions are mapped by a piecewise power function. A field study is then presented. Agreement can be seen between the mapped NMR T_2 spectra and the MRIL-Plog data in the location of the major peak, right boundary, distribution characteristics and T_2 logarithmic mean value. In addition, the capillary pressure curve construction model proposed in this paper is not affected by special log data or formation condition. It is of great importance in evaluating pore structure, predicting oil production and identifying oil layers through NMR log data in low-permeability sandstones.

Applicability of deep neural networks for lithofacies classification from conventional well logs: An integrated approach

Parametric understanding for specifying formation characteristics can be perceived through conven-tional approaches.Significantly,attributes of reservoir lithology are practiced for hydrocarbon explora-tion.Well logging is conventional approach which is applicable to predict lithology efficiently as compared to geophysical modeling and petrophysical analysis due to cost effectiveness and suitable interpretation time.However,manual interpretation of lithology identification through well logging data requires domain expertise with an extended length of time for measurement.Therefore,in this study,Deep Neural Network(DNN)has been deployed to automate the lithology identification process from well logging data which would provide support by increasing time-effective for monitoring lithology.DNN model has been developed for predicting formation lithology leading to the optimization of the model through the thorough evaluation of the best parameters and hyperparameters including the number of neurons,number of layers,optimizer,learning rate,dropout values,and activation functions.Accuracy of the model is examined by utilizing different evaluation metrics through the division of the dataset into the subdomains of training,validation and testing.Additionally,an attempt is contributed to remove interception for formation lithology prediction while addressing the imbalanced nature of the associated dataset as well in the training process using class weight.It is assessed that accuracy is not a true and only reliable metric to evaluate the lithology classification model.The model with class weight recognizes all the classes but has low accuracy as well as a low F1-score while LSTM based model has high accuracyas well as a high F1-score.