A deep kernel method for lithofacies identification using conventional well logs

How to fit a properly nonlinear classification model from conventional well logs to lithofacies is a key problem for machine learning methods.Kernel methods(e.g.,KFD,SVM,MSVM)are effective attempts to solve this issue due to abilities of handling nonlinear features by kernel functions.Deep mining of log features indicating lithofacies still needs to be improved for kernel methods.Hence,this work employs deep neural networks to enhance the kernel principal component analysis(KPCA)method and proposes a deep kernel method(DKM)for lithofacies identification using well logs.DKM includes a feature extractor and a classifier.The feature extractor consists of a series of KPCA models arranged according to residual network structure.A gradient-free optimization method is introduced to automatically optimize parameters and structure in DKM,which can avoid complex tuning of parameters in models.To test the validation of the proposed DKM for lithofacies identification,an open-sourced dataset with seven con-ventional logs(GR,CAL,AC,DEN,CNL,LLD,and LLS)and lithofacies labels from the Daniudi Gas Field in China is used.There are eight lithofacies,namely clastic rocks(pebbly,coarse,medium,and fine sand-stone,siltstone,mudstone),coal,and carbonate rocks.The comparisons between DKM and three commonly used kernel methods(KFD,SVM,MSVM)show that(1)DKM(85.7%)outperforms SVM(77%),KFD(79.5%),and MSVM(82.8%)in accuracy of lithofacies identification;(2)DKM is about twice faster than the multi-kernel method(MSVM)with good accuracy.The blind well test in Well D13 indicates that compared with the other three methods DKM improves about 24%in accuracy,35%in precision,41%in recall,and 40%in F1 score,respectively.In general,DKM is an effective method for complex lithofacies identification.This work also discussed the optimal structure and classifier for DKM.Experimental re-sults show that(m_(1),m_(2),O)is the optimal model structure and linear svM is the optimal classifier.(m_(1),m_(2),O)means there are m KPCAs,and then m2 residual units.A workflow to determine an optimal classifier in DKM for lithofacies identification is proposed,too.

Identification and Evaluation of Low Resistivity Pay Zones by Well Logs and the Petrophysical Research in China

This paper presents an overview of petrophysical research and exploration achievements of low resistivity pay （LRP） zone by well logs in China. It includes geological characteristics and characteristics of well log response of the low resistivity pay zones discovered and evaluated in recent years, as well as the problems in recognizing and evaluating low resistivity pay zones by well logs. The research areas mainly include the Neogene formations in the Bohai Bay Basin, the Triassic formations in the northern Tarim Basin and the Cretaceous formations in the Junggar Basin, The petrophysical research concerning recognition and evaluation of the low resistivity pays, based on their genetic types, is introduced in this paper.

Application of a Method for Calculating the Organic Carbon Content by Well Logs to Faulted Basins

The source rock model used in this project was developed by French Petroleum Research Institute. The total organic carbon content was estimated primarily and directly by using continuous conventional logging curves (such as sonic and resistivity curves), which are calibrated through the laboratory analysis data of organic carbon of cores, cuttings or sidewall cores. Regional evaluations have been carried out in downwarping basins abroad. The Haila′er Basin is a faulted basin and the evaluation of such a basin is a new subject. On the basis of a regional evaluation method for the downwarping basins, a new method suitable to faulted basins was developed. The effect is satisfactory when this new method is applied to the Wu′erxun Sag and the Bei′er Sag.

Study on Calculation Method of Compressional Velocities Based on Field Well Logs

In the past, most of the studies for compressional velocities are based on experimental measurements, which lack the support of field data. The purpose of this study is to estimate the compressional velocities based on well log data of delta front subfacies of Lower Tertiary ages of Ji-Dong oil field, China. At initial stage, we have chosen the well log parameters (effect factors) which strongly influence on compressional velocities and established a new modified equation for compressional velocities, which is based on these effect factors. Then Gardner, De-hua Han and this newly established equation were utilized to calculate the compressional velocities in each well. Finally, Least-square regression was carried out to check the fitting of each equation. Regression results clearly indicate that our purposed equation shows better fitting as compared to Gardner and De-hua Han equations.

Porosity Prediction from Well Logs Using Back Propagation Neural Network Optimized by Genetic Algorithm in One Heterogeneous Oil Reservoirs of Ordos Basin, China

A reliable and effective model for reservoir physical property prediction is a key to reservoir characterization and management.At present,using well logging data to estimate reservoir physical parameters is an important means for reservoir evaluation.Based on the characteristics of large quantity and complexity of estimating process,we have attempted to design a nonlinear back propagation neural network model optimized by genetic algorithm(BPNNGA)for reservoir porosity prediction.This model is with the advantages of self-learning and self-adaption of back propagation neural network(BPNN),structural parameters optimizing and global searching optimal solution of genetic algorithm(GA).The model is applied to the Chang 8 oil group tight sandstone of Yanchang Formation in southwestern Ordos Basin.According to the correlations between well logging data and measured core porosity data,5 well logging curves(gamma ray,deep induction,density,acoustic,and compensated neutron)are selected as the input neurons while the measured core porosity is selected as the output neurons.The number of hidden layer neurons is defined as 20 by the method of multiple calibrating optimizations.Modeling results demonstrate that the average relative error of the model output is 10.77%,indicating the excellent predicting effect of the model.The predicting results of the model are compared with the predicting results of conventional multivariate stepwise regression algorithm,and BPNN model.The average relative errors of the above models are 12.83%,12.9%,and 13.47%,respectively.Results show that the predicting results of the BPNNGA model are more accurate than that of the other two,and BPNNGA is a more applicable method to estimate the reservoir porosity parameters in the study area.

Quantitative evaluation of organic richness from correlation of well logs and geochemical data: a case study of the Lower Permian Taiyuan shales in the southern North China Basin

Marine-continental transitional shale is a potential energy component in China and is expected to be a realistic field in terms of increasing reserves and enhancing the natural gas production.However,the complex lithology,constantly changing depositional environment and lithofacies make the quantitative determination of the total organic carbon(TOC)suitable for marine shales not necessarily applicable to transitional shales.Thus,the identification of marine-continental transitional organic-rich shales and the mechanism of organic matter enrichment need to be further studied.As a typical representative of transitional shale,samples from Well MY-1 in the Taiyuan Formation in the southern North China Basin,were selected for TOC prediction using a combination of experimental organic geochemical data and well logging data including natural gamma-ray(GR),density(DEN),acoustic(AC),neutron(CNL)and U spectral gamma-ray(U),and TH spectral gamma-ray(TH).The correlation coefficient,coefficient of determination,standard deviation,mean squared error(MSE)and root mean squared error(RMSE)were selected to conduct the error analysis of the evaluation of different well log-based prediction methods,involving U spectral gamma logging,ΔlogR,and multivariate fitting methods to obtain the optimal TOC prediction method for the Taiyuan transitional shale.The plots of TOC versus the remaining volatile hydrocarbon content and the generation potential from Rock Eval show good to excellent potentials for hydrocarbon generation.The integrated results obtained from the various log-based TOC estimation methods indicate that,the multivariate fitting method of GR-U-DEN-CNL combination is preferable,with the correlation coefficients of 0.78 and 0.97 for the entire and objective interval of the Taiyuan Formation respectively,and with the minimum MSE and RMSE values.Specifically,the U spectral gamma logging method based on single logging parameter is also a better choice for TOC prediction of the high-quality intervals.This study provides a reference for the exploration and development of unconventional shale gas such as transitional shale gas.

Comparison between double caliper,imaging logs,and array sonic log for determining the in-situ stress direction:A case study from the ultra-deep fractured tight sandstone reservoirs,the Cretaceous Bashijiqike Formation in Keshen8 region of Kuqa depress

The tight sandstone in the Tarim Basin has the characteristics of large burial depth and development of nature fractures due to concentrated in-situ stress. Identifying the present-day in-situ stress orientation is important in hydrocarbon exploration and development, but also a key scientific question in understanding naturally fractured reservoirs. This paper presents a case study where we integrate various methods using wireline and image-log data, to identify present-day in-situ stress direction of ultra-deep fractured tight sandstone reservoirs, in the Kuqa depression. We discuss the formation mechanism of the elliptical borehole, compares the advantages and applicable conditions of the double caliper method,resistivity image logs and array sonic logs method. The well borehole diameter is measured orthogonally,then the ellipse is fitted, and the in-situ stress orientation is identified by the azimuth of the short-axis borehole, but it fails in the borehole expansion section, the fracture development section and the borehole collapse section. The micro-resistivity image logs method reveals the borehole breakouts azimuth, and also the strike of induced fractures, which are used to determine the orientation of in-situ stress. However, under water-based mud conditions, it’s hard to distinguish natural fractures from induced fractures by image logs. Under oil-based mud conditions, the induced fractures are difficult to identify due to the compromised image quality. As for the sonic log, shear waves will split when passing through an anisotropic formation, shear waves will split during propagation, and the azimuth of fast shear waves is consistent with the orientation of in-situ stress. However, it is usually affected by the anisotropy caused by the excessively fast rotation of the well log tools, so that the azimuth of fast shear wave cannot effectively reflect the orientation of the in-situ stress. Based on comprehensive assessment and comparison, in this paper we propose a method integrating various logging data to identify the orientation of in-situ stress. Among various types of logging data, the breakouts azimuth identified by image logs is proved to be the most credible in identifying the orientation of in-situ stress, while using the direction of induced fractures under water-based mud conditions is also viable. However, the azimuth of the fast shear wave is consistent with the orientation of maximum in-situ stress only when the rotation speed of the logging tool is low. The caliper method can be used as a reference for verifying the other two methods. Using this integrated method to study the orientation of in-situ stress in the Keshen8 trap, the results show that faults are an important factor affecting the direction of in-situ stress, while multi-level faults will produce superimposed effects that cause the current direction of in-situ stress to change.

Geological and Engineering‘Sweet Spots'in the Permian Lucaogou Formation,Jimusar Sag,Junggar Basin

Unconventional oil and gas resources require petrophysical logs to answer the question of how best to optimize geological and engineering‘sweet spots'.Therefore,the establishment of a key well with comprehensive descriptions of lithology,reservoir properties,hydrocarbon-bearing properties,electronic well log responses,source rock properties,brittleness,and in situ stress magnitude and direction is important for the effective exploration and production of unconventional hydrocarbon resources.Cores,thin sections,scanning electron microscopy(SEM)and comprehensive well log suites are used to build a key well for the Permian Lucaogou Formation,Jimusar Sag of the Junggar Basin.The results show that there are three main types of lithologies,including siltstone,mudstone and dolostone.Lithologies can be predicted using the combination of conventional well and image logs.The pore spaces consist of interparticle pores,intragranular dissolution pores and micropores.Nuclear Magnetic Resonance(NMR)T_(2)components longer than 1.7 ms are superposed as effective porosity.Permeability is calculated using the Coates model from NMR T_(2)spectra.The ratio of T_(2)components>7.0 ms to T_(2)components>0.3 ms is used to calculate oil saturation.TOC is calculated using theΔlog R method.Brittleness index is calculated using Poisson-Young's method,ranging from 13.42%-70.53%.In situ stress direction is determined,and in situ stress magnitudes(maximum horizontal stress SH_(max),minimum horizontal stress Sh_(min),vertical stress S_(v))are calculated using density and sonic logs.The strike-slip stress type(SH_(max)>S_(v)>Sh_(min))is encountered.The key well which comprehensively includes the above seven properties is established.Geological and engineering(geomechanical)‘sweet spots'are then optimized from the key well by fully analyzing lithology,reservoir property,oilbearing potential,in situ stress magnitude and brittleness.It is hoped that the results support engineers'and geologists'decisions for the future exploitation of unconventional hydrocarbon resources.

Physics-constrained indirect supervised learning

This study proposes a supervised learning method that does not rely on labels.We use variables associated with the label as indirect labels,and construct an indirect physics-constrained loss based on the physical mechanism to train the model.In the training process,the model prediction is mapped to the space of value that conforms to the physical mechanism through the projection matrix,and then the model is trained based on the indirect labels.The final prediction result of the model conforms to the physical mechanism between indirect label and label,and also meets the constraints of the indirect label.The present study also develops projection matrix normalization and prediction covariance analysis to ensure that the model can be fully trained.Finally,the effect of the physics-constrained indirect supervised learning is verified based on a well log generation problem.

Assessment of Hydrocarbon Potential in Owem Field in Niger Delta, Nigeria

Seismic data were integrated with well log to define the subsurface geometry and hydrocarbon trapping potential of Owem field, onshore Niger Delta. The research methodology involved horizon and fault interpretation to produce subsurface structural map. Wireline logs signatures were employed to identify hydrocarbon bearing sand and compute reservoir petrophysical parameters for hydrocarbon reservoir analysis. Two horizons HA and HB were identified and mapped and a time structural map was produced. Two reservoirs R1 and R2 were delineated in the wells A and B. The computed petrophysical parameters for well A showed that the thickness of Reservoir R1 and R2 are 11.5 and 12.5 meters respectively while the porosity and hydrocarbon saturation varies between 0.16 - 0.24 and 0.6 - 0.8 respectively. Similarly the average thickness and porosity of R1 and R2 for well B is about 18.0 meters and 0.12 while the hydrocarbon saturation varies between 0.7 - 0.8. The integration of seismic and well logs data has proved to be a useful tool in the reservoir analysis of hydrocarbon.

Mapping of thin sandstone reservoirs in bisol field, Niger delta, Nigeria using spectral decomposition technique

This study focuses on using spectral decomposition(SD)technique to characterize complicated reservoirs to understand the structural and stratigraphic variations in the interpreted horizons from Bisol field.The purpose of this study is to use geophysical and well logging data sets to map the thin-bedded sandstone reservoirs and prospect zones within the multiple reservoirs in Bisol field,Niger Delta.The interpretation of faults and horizons was carried out on the seismic section,which was further used to produce the structural maps.Seismic attributes such as trace and variance were used to enhance the truncated structures from the seismic section,while the produced spectra were used to delineate the stratigraphy and thickness of the thin-bedded reservoirs.Thin sandstone reservoirs were identified from well logs and consequently mapped on the seismic section.Fast Fourier Transform workflow was successfully used to image the stratigraphic features in the study area.Three horizons(S1T,S2T and S3T)were delineated from the seismic section,and four reservoirs were mapped and correlated across the wells.Frequency analyses from the seismic sectional view revealed some thin pay sandstone reservoirs,which were characterized by high amplitude.Three new probable zones(Prospect A,B and C)of hydrocarbon accumulation were identified using the SD technique.

An intelligent prediction method of fractures in tight carbonate reservoirs

An intelligent prediction method for fractures in tight carbonate reservoir has been established by upgrading single-well fracture identification and interwell fracture trend prediction with artificial intelligence,modifying construction of interwell fracture density model,and modeling fracture network and making fracture property equivalence.This method deeply mines fracture information in multi-source isomerous data of different scales to reduce uncertainties of fracture prediction.Based on conventional fracture indicating parameter method,a prediction method of single-well fractures has been worked out by using 3 kinds of artificial intelligence methods to improve fracture identification accuracy from 3 aspects,small sample classification,multi-scale nonlinear feature extraction,and decreasing variance of the prediction model.Fracture prediction by artificial intelligence using seismic attributes provides many details of inter-well fractures.It is combined with fault-related fracture information predicted by numerical simulation of reservoir geomechanics to improve inter-well fracture trend prediction.An interwell fracture density model for fracture network modeling is built by coupling single-well fracture identification and interwell fracture trend through co-sequential simulation.By taking the tight carbonate reservoir of Oligocene-Miocene AS Formation of A Oilfield in Zagros Basin of the Middle East as an example,the proposed prediction method was applied and verified.The single-well fracture identification improves over 15%compared with the conventional fracture indication parameter method in accuracy rate,and the inter-well fracture prediction improves over 25%compared with the composite seismic attribute prediction.The established fracture network model is well consistent with the fluid production index.

Hydrocarbon Play Assessment of “Oswil” Field, Onshore Niger Delta Region

Hydrocarbon play assessment of any field involves the evaluation of the production capacity of hydrocarbon reservoir unit in the field.This involves detail study of the reservoir petrophysical properties and geological interpretation of structures suitable for hydrocarbon accumulation in the field as observed from seismic reflection images.This study details the assessment of hydrocarbon play in OSWIL field onshore in Niger Delta,with the intent of appraising its productivity using a combination of seismic,well logs,petrophysical parameters and volumetric estimation using proven techniques which involves an integrated methodology.Two reservoir windows“R1”and“R2”were defined from five wells OSWIL-02,04,06,07 and 12.The top and base of each reservoir window was delineated from the wells.Structural interpretation for inline 6975 revealed two horizons(X and Y)and eight faults labelled(F1,F2,F6,F8,F10,F16,F17 and F18).Five faults(F1,F6,F10,F17 and F18)were identified as synthetic faults and dip basin wards while three faults(F2,F8 and F16)were identified as antithetic faults and dips landwards.Time-depth structural map at top of reservoirs R1 and R2 revealed structural highs and closures.These observations are characteristics of growth structures(faults)which depicts the tectonic style of the Niger Delta.Results of petrophysical evaluation for reservoirs“R1”and“R2”across the five wells were analysed.For reservoir“R1”effective porosity values of 27%,26%,23%,20%and 22%were obtained for wells OSWIL-04,12,07,06 and 02 respectively with an average of 23.6%,while for reservoir“R2”effective porosity values of 26%,22%,21%,24%and 23%for wells OSWIL-04,12,07,06 and 02 were obtained respectively with an average of 23.2%.This porosity values correspond with the already established porosity range of 28-32%within the Agbada formation of the Niger Delta.Permeability index of the order(K>100mD)were obtained for both reservoirs across the five wells and is rated very good.Hydrocarbon saturation(Shc)across the five wells averages at 61.6%for reservoir“R1”and 67.4%for reservoir“R2”.Result of petrophysical model for porosity,permeability and water saturation reveal that the reservoir system in R1 and R2 is fault assisted and fluid flow within both reservoirs is aided by presence of effective porosity and faulting.Volumetric estimation for both reservoirs showed that reservoir R1 contains an estimate of 455×106 STB of hydrocarbon in place,while reservoir R2 contains an estimate of 683×106 STB of hydrocarbon in place.These findings impact positively on hydrocarbon production in the field and affirm that the two reservoirs R1 and R2 are highly prospective.

Comparison of machine learning methods for estimating permeability and porosity of oil reservoirs via petro-physical logs

This paper deals with the comparison of models for predicting porosity and permeability of oil reservoirs by coupling a machine learning concept and petrophysical logs.Different machine learning methods including conventional artificial neural network,genetic algorithm,fuzzy decision tree,the imperialist competitive algorithm(ICA),particle swarm optimization(PSO),and a hybrid of those ones are employed to have a comprehensive comparison.The machine learning approach was constructed and tested via data samples recorded from northern Persian Gulf oil reservoirs.The results gained from the machine learning models used in this paper are compared to the relevant real petrophysical data and the outputs achieved by other methods employed in our previous studies.The average relative absolute deviation between the approach estimations and the relevant actual data is found to be less than 1%for the hybridized approaches.The results reported in this paper indicate that implication of hybridized machine learning methods in porosity and permeability estimations can lead to the construction of more reliable static reservoir models in simulation plans.