Daniel Asante OtchereTarek Omar Arbi GanatJude Oghenerurie Ojero
11页
查看更多>>摘要：Feature Selection,a critical data preprocessing step in machine learning,is an effective way in removing irrelevant variables,thus reducing the dimensionality of input features.Removing uninformative or,even worse,misinformative input columns helps train a machine learning model on a more generalised data with better performances on new and unseen data.In this paper,eight feature selection techniques paired with the gradient boosting regressor model were evaluated based on the statistical comparison of their prediction errors and computational efficiency in characterising a shallow marine reservoir.Analysis of the results shows that the best technique in selecting relevant logs for permeability,porosity and water saturation prediction was the Random Forest,SelectKBest and Lasso regularisation methods,respectively.These techniques did not only reduce the features of the high dimensional dataset but also achieved low prediction errors based on MAE and RMSE and improved computational efficiency.This indicates that the Random Forest,SelectKBest,and Lasso regularisation can identify the best input features for permeability,porosity and water saturation predictions,respectively.
查看更多>>摘要：Complex lithology association along vertical direction results in thin layers with different interbed patterns of the Lower Triassic Jialingjiang Formation(T]j),Sichuan Basin.The prediction of thin tight reservoirs in the Jia-lingjiang Formation is important for further oil and gas exploration.However,seismic data in the eastern Sichuan Basin has poor lateral continuity of seismic events because of strong interference noises,which makes it difficult to pick horizons.Moreover,low dominant frequency(<25 Hz),low seismic resolution,and weak response characteristics of strong transverse heterogeneity on logging parameters make it challenging for the thin reservoirs prediction.To tackle these issues,in this study,we propose a target-oriented workflow by integrating seismic data processing with interpretation to delineate the thin reservoirs and corresponding hydrocarbon production capacity.In the processing workflow,we first remove the noise interferences using the structure-oriented filtering(SOF),which can improve the effective information of seismic events and also preserve the features of fault structures.Second,we propose a novel mediod based on spectral shaping and sparse inversion to enhance the seismic resolution.Tests on filed data demonstrate diat the effective information and resolution of processed seismic data are significantly enhanced.Then,in the seismic interpretation,fluid mobility and spectral decomposition are first used to delineate the spatial distribution of favorable reservoirs.Furthermore,multi-attribute integrated interpretation involving impedance analysis,reservoir tiiickness estimation and porosity inversion is used to evaluate the storage capacity and productivity of the tiiin reservoirs.The estimated impedance,reservoir tiiickness and porosity show good agreement with well information,demonstrating the feasibility of the proposed mediod for accurately predicting thin reservoirs in tiiis area.Multi-attribute integrated interpretation revealed that the favorable tiiin reservoirs are mainly developed from the nortiieast to the southwest in the study area,which can provide effective guidelines for future production well drilling and distribution delineations of a favorable reservoir in the Eastern Sichuan Basin.
查看更多>>摘要：In history matching,the calibration of a prior reservoir model is computationally expensive because many forward reservoir simulation runs are required.Multiple posterior(or calibrated)reservoir models need to be sampled to consider high reservoir uncertainty,which increases the computational cost significantly.In this study,we propose a novel deep-learning-based history matching method that efficiently samples posterior reservoir models for fluvial channel reservoirs.Three convolution-based neural networks(NNs)are used in the proposed method to sample posterior models quickly witiiout conventional calibration processes:convolutional autoencoder(CAE),convolutional neural network(CNN),and convolutional denoising autoencoder(CDAE).First,low-dimensional latent features are extracted from prior models using CAE because the dimensionality of static data is too high to find the relation between the prior models and corresponding simulated dynamic(production)data.Next,CNN is used to find the relation between the latent features of the prior models and the corresponding production data,which are the output and input data of CNN,respectively.The CNN output is refined using CDAE to improve the geological connectivity of the posterior models.The performance of the proposed mediod is compared with non-convolution-based mediods that combine fully-connected NN structures(multi-layer perceptron(MLP))and dimension-reduction techniques(principal component analysis(PCA)and stacked autoencoder(SAE))in the benchmark egg model.The proposed mediod outperforms the other methods(MLP-PCA and MLP-SAE)in terms of geological constraints for fluvial channels and the computational cost of sampling posterior models.
查看更多>>摘要：The middle Permian organic matter-rich mudstone in the South Yellow Sea Basin has been considered the most important source rock for Paleozoic reservoirs in the basin.However,information on the variation in deposi-tional environment within the Qixia Formation to lower Longtan Formation interval is lacking,and the formation mechanism of these organic matter-rich mudstones is unclear.Based on a series of geochemical analyses of borehole CSDP-2,the paleoclimatic conditions,nature of the watermass,paleoproductivity and terrestrial detrital influx of the different intervals in the middle Permian of the South Yellow Sea Basin were assessed to determine the main factors controlling organic matter enrichment.All three intervals(from the Qixia Formation to the lower Longtan Formation)were deposited under a relatively warm and humid paleoclimate with moderate terrestrial detrital input.The Qixia Formation was deposited under oxic paleoredox conditions in a normal marine environment.During deposition of the Gufeng Formation,the paleoredox conditions changed to suboxic with intermittent anoxia associated with paleoproductivity higher than that in the other studied intervals.The lower Longtan Formation was characterized by a brackish watermass,suboxic paleoredox conditions and low paleoproductivity.The covariant relationships among paleoproductivity,paleoredox conditions,watermass salinity indicators and TOC content indicate that paleoproductivity was the critical factor controlling organic matter abundance.High primary productivity not only provided organic matter but also enhanced organic matter preservation by consuming oxygen in the watermass,which resulted in the formation of the organic matter-rich source rock in the Gufeng Formation.
查看更多>>摘要：Reservoired liquid petroleum has limited potential to be preserved under high temperature conditions due to thermal cracking and potentially thermochemical sulfate reduction,and thus exploration targets in ultradeep strata are mainly gas and condensate gas.However,a giant ultradeep liquid petroleum accumulation,with an average depth of>6500 m,was discovered recently in the Hadexun area,Tarim Basin.In this study,integrated geochemical analyses were conducted on the black oil samples from the Hadexun area.Intact terpanes and steranes and low concentration of diamondoids were detected,indicating that these oils were generated at peak oil window and were barely altered by secondary geochemical process(thermal cracking or others)despite their depth.The associated wet gas with light isotopic profiles was classified as mature oil-derived gas co-generated with the oil.Petroleum accumulation analyses suggested that the favorable carbonate reservoir bodies in the Middle Ordovician Yijianfang Formation were attributed to karstification and weathering,and mainly distributed along the major strike-slip fault belts.Together with the thick mudstones in the overlying Upper Ordovician Sangtamu Formation,they formed a favorable reservoir-seal assemblage in the Hadexun area.Massive volumes of liquid petroleum were generated and expelled from the Lower Cambrian black shales in the Late Hercynian and migrated into the Ordovician carbonate reservoirs through the strike-slip fault system,and subsequently were exposed to stable tectonic conditions.Due to the rapid subsidence since the Neogene,insufficient temperature-time exposure eliminated the impact of thermal cracking;therefore,the petroleum accumulation was preserved in the liquid phase,and further indicates that huge petroleum resource potential remains in ultradeep strata in the Tarim Basin.
查看更多>>摘要：Micro-CT analysis is employed for the first time to evaluate the effects of bioturbation on porosity distribution in contourite fades,namely:i)sandy clastic contourites from the Late Miocene Moroccan Rifian Corridor;and ii)dominant-calcareous contourites from Eocene-Middle Miocene Cyprus paleoslope.Porosity distribution is affected by bioturbation due to trace maker behavior and burrow infilling,but there is no clear,single relationship.Macaronichnus,especially M.segregate,located in sandy clastic contourite facies would have the greatest impact,increasing porosity due to the grain-selective deposit-feeding behavior of the trace maker.Porosity values are up to three times higher in the burrow rim than in the infilling material,which favors a dual-porosity flow medium.Samples with M.s.degiberti record higher porosity values in the host sediment than in the burrow fill,but any increase of porosity in the burrow rim is not certain.This variable influence on porosity distribution(M.segregate vs.M.s.degiberti)might be associated with different trace maker behaviors during feeding activities,or even with variable producers.Chondrites located in muddy chalk contourite facies has a neutral effect on porosity distribution,porosity data of the host sediment and the filling material being similar,in agreement with a comparable grain size.Thalassinoides located inside calcarenitic contourite facies has a minor effect on porosity distribution,showing similarities between passively infilled material and the host sediment.Our results also reveal the importance of secondary diagenetic overprints when homogenizing primary differences.Accordingly,understanding the impact of bioturbation on porosity distribution in contourite facies appears to be a key factor for evaluating the real potential of these unconventional reservoirs.This study represents a first step forward in discerning the relationship between contourite facies,ichnological features,petrophysical properties and reservoir geology.
查看更多>>摘要：Shear wave velocity(S-wave velocity)has great significance for reservoir characterization and can effectively reduce the ambiguity in seismic interpretation.However,owning to its high cost and technical difficulties,it is usually difficult to obtain for the whole region.Therefore,the prediction of S-wave velocity currently has turned into an urgent task.Thus,the S-wave velocity was predicted using ID convolutional neural network(ID-CNN)with the well logs.In the first step,the traditional Xu-White and Xu-Payne model were used to carry out the inversion of bulk modulus,shear modulus and density,and illustrated the phenomenon for the traditional rock-physics model.Then,ID-CNN,including one input layer,four convolutional layers,three fully connected layers and one output layer,was proposed,in which three types of well logs(conventional well logs,array induction well logs and gamma ray spectral well logs)sensitive to the S-wave velocity were determined through the well logs analysis.Finally,based on the established ID-CNN model,the influence of different well logs combination on the S-wave velocity prediction was analyzed,indicating that the addition of different well logs can improve the accuracy of S-wave velocity.And different machine learning methods(MLs),including back-propagation(BP)and support vector regression(SVR),were also compared,showing that the predicted accuracy of 1D-CNN has been improved by 4.2% on average.Two cases for sandstone and carbonate reservoirs also showed that ID-CNN proposed can achieve higher accuracy and better performance than traditional MLs and rock-physics models.Furthermore,the proposed method can be applied to the other oil and gas exploration fields,improving the exploration accuracy and increasing the production of oil and gas.
查看更多>>摘要：This article has been retracted:please see Elsevier Policy on Article Withdrawal(https://www.elsevier.com/about/our-business/policies/article-withdrawal).This article has been retracted at the request of the Editor-in-Chief.The authors have plagiarized part of a paper that had already appeared in Chemical Engineering Science,Volume 232,15 March 2021.https://doi.org/10.1016/j.ces.2020.116352.
查看更多>>摘要：Structural analysis and temporal evolution of the Dehdasht Structural Basin in the Central Zagros through the constructed cross-sections and 2D sequential restoration have defined the timing of trap formation in the basin.At the level of the competent layer(Early Cretaceous to Oligo-Miocene),the basin is deeper toward the center where it is covered by the thicker Miocene evaporites and syn-tectonic clastic deposits.The evolutionary history of the basin based on the sequential restoration proposes that the anticlinal traps are formed since the Pabdeh deposition at Paleocene-Eocene on the boundaries and during the middle Miocene Mishan deposition within the basin.The proposed structural evolution and the suggested oil window using the burial depth and the present thermal gradient for the Dehdasht Basin give insight into its hydrocarbon potential.Accordingly,the Cretaceous Kazhdumi source rock has started oil generation since Paleocene-Eocene(deposition of the Pabdeh Formation)in the deeper central part until the present in the subsurface relatively large-amplitude anticlines close to the boundaries.We also suggest tiiat the Paleocene-Eocene Pabdeh Formation contained oil mainly in the subsurface low-amplitude anticlines in the central part of the basin during deposition of the Gachsaran and Mishan formations at early-middle Miocene.The older oil generation in the basin is attributed to its thicker deposits in the Paleocene-Eocene and early Miocene resulting in deeper burial depth for the source rocks.The obtained results propose that the subsurface relatively large-amplitude anticlines near the basin margins with the possible hydrocarbon(mainly generated from the Cretaceous Kazhdumi shale)are the most likely targets for future oil exploration.Additionally,we obtained tiiat the boundary fore-thrusts and the caprock erosion due to the lateral anticlinal growth control the oil and gas seepages on the Dehdasht Basin boundaries.
查看更多>>摘要：The East Bokaro Basin of the Damodar valley is a potentially prospective CBM(coalbed methane)play having significant cumulative coal seam thickness,in-situ gas content,vitrinite percentage,and adequate thermal maturity.Successful CBM recovery needs a detailed understanding of the organic content,pore structures/networks,storage properties and gas flow mechanism.The present work attempts to systematically investigate East Bokaro coal for organo-petrographic controls on gas content and generation,variations in sorption capacity and saturation,pore mechanisms,cleat intensity,cleat aperture distribution and spacing.The values of in-situ gas,sorption capacity and methane concentration(C1)vary from 3.52 to 30.93 cc/g(dry ash-free basis),15.40-32.40 cc/g(dry ash-free basis),and 66-93 vol%,respectively.The atomic ratios H/C and O/C indicate that thermally matured coal seams contain type III-IV kerogen positioned in the dry gas window.The decrease of hydrogen-containing liptinite with increasing depth reveals the function of thermal gradient on the cracking of liptinitic compounds with successive evolution of hydrocarbons and the development of a carbon-rich pore matrix.The H/C ratio is also influenced by the increasing content of vitrinite and reflectance values of deeper coal.More than 63 % of desorbed gas was determined from desorption measurement and low sorption time(τ,mainly<10 days).This demonstrates good diffusion characteristics of the studied coal.It shows the tendency of desorbed gas diffusion from pores reaches to cleat-fractures with negligible influence of secondary mineral infillings.The high-pressure sorption studies of methane on various samples indicate substantial open-pore characteristics,supporting adsorption,diffusion and gas release.The relationship of C3/C1 and C2/C1 ratios demonstrates that the hydrocarbons in coal primarily originated from the thermogenic transformation of organic matter.Such an assessment is also supported by the stable isotope(8~(13)C1)value that ranges between-22.70 %o and-57.30 ‰.However,some of the lighter isotope values(<-50 ‰)indicate a mixed origin of gases,which may be due to the influx of fresh-water to coal associated aquifers carrying bacteria received from local drainage.Geochemically and thermally altered dissolved and partially filled pores,shown by SEM photographs,negligibly influence gas sorption,diffusion,and flow mechanism in coalbeds.The pore network model signifying that the studied coal seams are microstructurally different comprises a lateral difference in pore and cleat/fracture.
NSTL