Layer | Fill | Outline |
---|
Map layers
Theme | Visible | Selectable | Appearance | Zoom Range (now: 0) |
---|
Fill | Stroke |
---|---|
Collaborating Authors
Li, Kewen
The prevailing methodology in data-driven fault detection leverages synthetic data for training neural networks. However, it grapples with challenges when it comes to generalization in surveys exhibiting complex structures. To enhance the generalization of models trained on limited synthetic datasets to a broader range of real-world data, we introduce FaultSSL, a semi-supervised fault detection framework. This method is based on the classical mean teacher structure, in which its supervised part employs synthetic data and a few 2D labels. The unsupervised component relyies on two meticulously devised proxy tasks, allowing it to incorporate vast unlabeled field data into the training process. The two proxy tasks are PaNning Consistency (PNC) and PaTching Consistency (PTC). PNC emphasizes the feature consistency in overlapping regions between two adjacent views in predicting the model. This allows for the extension of 2D slice labels to the global seismic volume.#xD;PTC emphasizes the spatially consistent nature of faults. It ensures that the predictions for the seismic, whether made on the entire volume or on individual patches, exhibit coherence without any noticeable artifacts at the patch boundaries. While the two proxy tasks serve different objectives, they uniformly contribute to the enhancement of performance. Experiments showcase the exceptional performance of FaultSSL. In surveys where other mainstream methods fail to deliver, we present reliable, continuous, and clear detection results. FaultSSL reveals a promising approach for incorporating large volumes of field data into training and promoting model generalization across broader surveys.
- Reservoir Description and Dynamics > Reservoir Characterization > Seismic processing and interpretation (1.00)
- Data Science & Engineering Analytics > Information Management and Systems > Neural networks (1.00)
- Data Science & Engineering Analytics > Information Management and Systems > Artificial intelligence (1.00)
- Information Technology > Artificial Intelligence > Representation & Reasoning > Diagnosis (0.93)
- Information Technology > Artificial Intelligence > Representation & Reasoning > Expert Systems (0.83)
- Information Technology > Artificial Intelligence > Machine Learning > Neural Networks > Deep Learning (0.68)
TransInver: 3D data-driven seismic inversion based on self-attention
Li, Kewen (China University of Petroleum (East China) Qingdao) | Dou, Yimin (China University of Petroleum (East China) Qingdao) | Xiao, Yuan (China University of Petroleum (East China) Qingdao) | Jing, Ruilin (Shengli Oilfield Company) | Zhu, Jianbing (Shengli Oilfield Company) | Ma, Chengjie (Shengli Oilfield Company)
ABSTRACT Recently, convolutional neural network (CNN)-based deep learning (DL) for impedance inversion has been extended to multiple dimensions. Training multidimensional DL inversion requires extracting supervised information from sparse 1D well-log labels. Fully convolutional networks rely on their parameter sharing mechanism and receptive fields to achieve this, but their perceptual range is limited, making it difficult to capture long-term correlations in seismic data. The transformer is a type of network that is entirely based on self-attention, and it has demonstrated remarkable performance across various tasks and domains. However, its suitability for 3D seismic inversion is restricted by its high computational workload, fixed input size requirement, and inadequate handling of low-level details. The primary goal was to reengineer the self-attention mechanism to optimize its applicability for seismic impedance inversion tasks. The high-dimensional self-attention is decoupled into dual low-dimensional attention paths to reduce the computation of dense connections and matrix dot products. Shared parameters are used instead of full connection, allowing for flexible changes in input sizes by the network. In addition, its local modeling capabilities are enhanced by integrating it with the residual structure of the CNN. We name the resulting structure Self-Attention ResBlock, which is used as the basic unit for constructing TransInver. Comparative experiments indicate that TransInver performs significantly better than 3D methods such as UNet, TransUNet, HRNet, and 1D inversion methods. TransInver produces reliable inversion results using only nine well logs for SEAM Phase I and three well logs for the field data set of the Netherlands F3. This backbone network can deliver excellent inversion performance without depending on any auxiliary means such as low-frequency constraints or semisupervised frameworks.
- Geophysics > Seismic Surveying > Seismic Processing (1.00)
- Geophysics > Seismic Surveying > Seismic Modeling > Velocity Modeling > Seismic Inversion (1.00)
Recently, CNN-based deep learning (DL) for impedance inversion has been extended to multiple dimensions. Training multi-dimensional DL inversion requires extracting supervised information from sparse one-dimensional well log labels. Fully convolutional networks rely on their parameter sharing mechanism and receptive fields to achieve this, but their perceptual range is limited, making it difficult to capture long-term correlations in seismic data. Transformer is excelling in several areas because of its self-attention global modeling approach. However, its suitability for 3D seismic inversion is restricted by high computational workload, fixed input size requirement, and inadequate handling of low-level details. In this study, our primary goal was to re-engineer the self-attention to optimize its applicability for seismic impedance inversion tasks. The high-dimensional self-attention was decoupled into dual low-dimensional attention paths to reduce the computation of dense connections and matrix dot products. Shared parameters were used instead of full connection, allowing for flexible changes in input sizes by the network. Additionally, its local modeling capabilities were enhanced by integrating it with the residual structure of the CNN. We named it Self Attention ResBlock (SARB), which is used as the basic unit for constructing TransInver. Comparative experiments show that TransInver performs significantly better than 3D methods such as UNet, TransUNet, HRNet and 1D inversion methods. TransInver produced reliable inversion results using only nine well-logs for SEAM Phase I, and with just three well-logs for the field dataset of the Netherlands F3. This backbone network can deliver excellent inversion performance without depending on any auxiliary means such as low-frequency constraints or semi-supervised frameworks.
- Geophysics > Seismic Surveying > Surface Seismic Acquisition (1.00)
- Geophysics > Seismic Surveying > Seismic Processing (1.00)
- Geophysics > Seismic Surveying > Seismic Modeling > Velocity Modeling > Seismic Inversion (1.00)
Abstract Wettability alteration and interfacial tension (IFT) reduction are two important mechanisms for enhanced oil recovery (EOR). The introduction of nanotechnology from the fields of Biology and Material Science to the application in EOR is emerging because nanoparticles have the potential to alter formation factors like wettability and fluid properties like IFT and viscosity. However, a systematic literature review shows that ambiguity exists regarding whether nanoparticles can change wettability and IFT or not and which component in nanofluid plays a role. In this work, we investigated the effects of bare silica nanoparticles on wettability and IFT using a contact angle goniometer. The results showed that the contact angle measurement on quartz plates had relatively large uncertainty while those on calcite plates showed a clear trend that the smaller the nanoparticle size and the larger the nanofluid concentration, the smaller the contact angle. In addition, silica nanoparticles did not have an effect on IFT. Core flooding experiments showed an increase of 8.7% in oil recovery factor by the use of silica nanoparticles, which support the oil recovery mechanism of wettability alteration.
- North America > United States (0.47)
- Asia > Middle East (0.28)
Abstract Polymer flooding, as one of the EOR (Enhanced Oil Recovery) methods, has been adopted in many oilfileds in China and some other countries. Over 50% oil remains undeveloped in oil reservoirs after polymer flooding. It has been a great challenge to find approaches to further enhancing oil recovery when polymer flooding is over. In this study, a new method was proposed to increase oil production using gas flooding with wettability alteration to gas-wetness when polymer flooding has been completed. The rock wettability was altered from liquid- to gas-wetness during gas flooding. An artificial oil reservoir was constructed and many numerical simulations have been conducted to test the effect of wettability alteration on the oil recovery in reservoirs developed by water flooding and followed by polymer flooding. Production data from different scenarios, water flooding, polymer flooding after water flooding, gas flooding with and without wettability alteration after polymer flooding, were calculated using numerical simulation. The results demonstrate that the wettability alteration to gas-wetness after polymer flooding can significantly enhance oil recovery and reduce water cut effectively. Also studied were the combined effects of wettability alteration and reservoir permeability on oil recovery.
- Asia > China (0.89)
- North America > United States > California (0.28)
- North America > United States > Texas (0.28)
- Asia > China > Hubei > Jianghan Basin > Jianghan Field (0.99)
- Asia > China > Heilongjiang > Songliao Basin > Daqing Field > Yian Formation (0.99)
- Asia > China > Heilongjiang > Songliao Basin > Daqing Field > Mingshui Formation (0.99)
Abstract Pore volumes injected (PVI) has been found to be difficult to represent the longevity and intensity of water flooding in oil reservoirs. In order to find a better parameter to characterize fluid flooding, a new concept referred as to "flushing coefficient" has been proposed, which is a function of time and position. The effects of the viscosity, relative permeability of water phase, pressure gradient, and flooding time are considered in the flushing coefficient. Water flooding experiments were conducted in sandstone rock samples with different permeabilities ranging from about 1.0 to over 1000 md at room temperatures. The experimental data were used to test the new concept.
Experimental Study on Calculating Capillary Pressure from Resistivity
Hou, Binchi (Research Institute of Shaanxi Yanchang Petroleum (Group) CO., LTD.) | Liu, Hongliang (China Petroleum Logging TuHa Business Division) | Bian, Huiyuan (Xi'an University of Science and Technology) | Wang, Chengrong (China Petroleum Logging TuHa Business Division) | Xie, Ronghua (Daqing Oilfield CO.LTD., PetroChina) | Li, Kewen (China University of Geosciences(Beijing)/Stanford University)
Abstract Capillary pressure and resistivity in porous rocks are both functions of wetting phase saturation. Theoretically, there should be a relationship between the two parameters. However, few studies have been made regarding this issue. Capillary pressure may be neglected in high permeability reservoirs but not in low permeability reservoirs. It is more difficult to measure capillary pressure than resistivity. It would be useful to infer capillary pressure from resistivity well logging data if a reliable relationship between capillary pressure and resistivity can be found. To confirm the previous study of a power law correlation between capillary pressure and resistivity index and develop a mathematical model with a better accuracy, a series of experiments for simultaneously measuring gas-water capillary pressure and resistivity data at a room temperature in 16 core samples from 2 wells in an oil reservoir were conducted. The permeability of the core samples ranged from 9 to 974 md. The gas-water capillary pressure data were measured with confining pressures using a semi-porous plate technique. We developed the specific experimental apparatus to measure gas-water capillary pressure and resistivity simultaneously. The results demonstrated that the previous power law model correlating capillary pressure and resistivity works well in many cases studied. A more general relationship between the exponent of the power law model and the rock permeability was developed and verified using the experimental data.
- Asia > China (1.00)
- North America > United States > Texas (0.29)
- Research Report > New Finding (0.65)
- Research Report > Experimental Study (0.41)
- Asia > China > Heilongjiang > Songliao Basin > Daqing Field > Yian Formation (0.99)
- Asia > China > Heilongjiang > Songliao Basin > Daqing Field > Mingshui Formation (0.99)
New Mathematical Models for Calculating Proppant Embedment and Fracture Conductivity
Li, Kewen (China University of Geosciences (Beijing) and Stanford University) | Gao, Yuanping (Petroleum Exploration & Production Research Institute) | Lyu, Youchang (Sinopec) | Wang, Man (China Pingmei Shenma Group)
Summary Proppant embedment plays a significant role in decreasing fracture aperture and conductivity, especially for weakly consolidated sandstones, shale (oil and gas) rock, and coalbeds. Empirical and semiempirical models were usually used to calculate the embedment of proppants. However, the accuracy of matching or predicting the proppant embedment with these existing models may not be satisfactory in some cases. On the other hand, it is difficult to determine the coefficients of these models. In this study, analytical models were derived to compute proppant embedment and fracture conductivity. One can use these new models to calculate the proppant embedment, proppant deformation, the change in fracture aperture, and fracture conductivity in the ideal or experimental situations of either single-layer or multilayer patterns in the fractures under closure pressures. The new models showed that the proppant embedment and fracture conductivity are affected by the factors, for example, of closure pressure, fracture aperture, the elastic modulus of proppant and coalbed, the size of proppant, and the concentration of proppant-paving. Experimental data of proppant embedment in fractures and fracture conductivity of different proppants at different closure pressures were used to test the models derived in this study. The results from matching the experimental data with the new and the existing models were compared. The results showed that the new models, especially the revised new models, could match the experimental data in all the cases studied. The new models for calculating the proppant embedment and fracture conductivity with a better accuracy are of great significance in selecting proppants, which is helpful to achieve high fracture conductivity and then high oil or gas production of conventional and (especially) unconventional resources such as shale oil, shale gas, and coalbed methane.
- Geology > Geological Subdiscipline > Geomechanics (1.00)
- Geology > Rock Type > Sedimentary Rock > Clastic Rock > Mudrock > Shale (0.74)
A Modified Method and Experimental Verification for Estimating Relative Permeability from Resistivity Logging Data
Bian, Huiyuan (China U. of Geosciences) | Li, Kewen (China U. of Geosciences and Stanford U.) | Yang, Jinghai (Logging & Testing Services Company,Daqing Oilfield CO.LTD.) | Pei, Jianya (Daqing Oilfield Co. Ltd.) | Lao, Pengcheng (Daqing Oilfield Co. Ltd.) | Li, Xiaowei (Daqing Oilfield Co. Ltd.)
Abstract The existing method to infer relative permeability from resistivity data was modified by including more parameters such as residual oil saturation. Both oil-water relative permeability and resistivity were measured simultaneously in the same core sample at a room temperature in order to verify the modified model. Altogether 16 core samples in 2 wells from Daqing oil field, China have been tested. The permeability ranged from about 10 to 800 md. The oil-water relative permeability data were measured using a dynamic displacement technique. Oil-water relative permeability data were inferred from the resistivity data measured in the laboratory and logged from the well using the modified model. The model data were then compared to the experimental data. We demonstrated that the relative permeability of both oil and water calculated from the resistivity data measured in the same core samples and logged from the same wells were close to the experimental data measured using a dynamic displacement approach. The modified model had a greater accuracy compared with the existing models. Using the modified model, it would be possible to obtain the different distribution of relative permeability characteristics in different kinds of formations in a reservoir. It may also be feasible to infer relative permeability data while drilling if resistivity well logging is being taken. Introduction Relative permeability is one of the important parameters controlling multiphase fluid flow in porous media. These data are traditionally obtained with experimental measurements. However, relative permeability is expensive, difficult, and time-consuming to measure in the laboratory, especially for the rocks from unconventional oil and gas reservoirs such as shale plays, tight sands, and extremely low permeability reservoirs. It is also difficult to maintain exact reservoir conditions in taking a core or a fluid sample from the reservoir and bringing it to surface and it is almost impossible to conduct the measurements in real time. Consequently, there has been a decades-long research effort to develop methods and procedures to infer relative permeability using network modeling. Recently, the industry has been researching new methods to extract relative permeability in-situ including the utilization of specially designed permanent downhole electric resistivity array, pressure, and flow rate measurements. Relative permeability can also be derived from other parameters such as capillary pressure data. Mahmoud et al. (2013) predicted the capillary pressure from well logging data in carbonate reservoir and sandstone reservoir. Purcell (1949) reported a mathematical model to calculate the relative permeability from capillary pressure data. From then on, many researchers worked on this area. Li (2005, 2007 and 2010), Li and Horne (2006) and Li and Williams (2006) have made a lot of contribution for estimating the relative permeability using resistivity well logging data. Based on the reaserch on the interrelation between capillary pressure, resistivity and relative permeability reported by Li (2010), Alex et al. (2012) considered to modify the model in double porosity systems. They developed a method to calculate relative permeability and caplillary pressure from resistivity well logging data in naturally fractured reservoirs.
- North America > United States (0.88)
- Asia > China > Heilongjiang Province > Daqing (0.25)
- Asia > China > Heilongjiang > Songliao Basin > Daqing Field > Yian Formation (0.99)
- Asia > China > Heilongjiang > Songliao Basin > Daqing Field > Mingshui Formation (0.99)
- Asia > China > Heilongjiang > Songliao Basin > Saertu Field (0.93)
Abstract One of the difficulties of characterizing polymer flooding comes from the non-Newtonian and time-dependent properties of polymer solutions, and shear rates must be determined to calculate the viscosity of polymer solutions flowing in porous media. Although there have been many models to do so, it is still difficult to choose which shear rate model to conduct the calculations. In this study, several polymer flooding experiments were designed and conducted in rock samples with different permeabilities to evaluate the models for determining shear rates of polymer flooding in porous media. Experiments were also conducted in simple glass capillary tubes to reduce the effect of polymer adsorption on the evaluation. Four frequently-used shear rate models for porous media were chosen to be evaluated. Using the experimental data, the values of shear rate of the four models were computed respectively, as well as the pseudopermeability of the core samples under single-phase flow but different flow rates. The best shear rate model among the four frequently-used models was determined, and the rule of making the judgments was proposed and verified, that is, the pseudopermeability at the intrinsic shear rate should be equal or close to the absolute permeability of the sample in the cases of capillary tubes.
- Asia > China (0.30)
- North America > United States (0.28)