The oil and gas reservoirs in the Qiketai formation of Middle Jurassic in the Pubei area of Taibei sag, Turpan-Hami basin, are controlled by lithology. Early exploration confirmed that there are thin oil-bearing sand layers with the thickness of 6-15 m at the bottom of the Qiketai formation. It is difficult for conventional inversion methods to predict these sand layers and these methods often yield large errors due to the limitations of the frequency band of seismic data. In order to improve inversion accuracy, a facies-controlled geostatistical inversion method based on low-frequency model optimization was proposed. Combined with the characteristics of large structural relief and greatly varying sedimentary facies in the study area, the low-frequency model was established by combining the compaction trend correction method and the seismic attribute constraint method to obtain the deterministic inversion results. On this basis, a facies-controlled model was established for facies-controlled geostatistical inversion, thus enabling the identification of thin sand layers in the study area. This method effectively complements the low-frequency information missed in seismic signals, and improves the longitudinal resolution of the inversion results. By using this method, a thin sand layer with the thickness of 7 m can be identified, and the inversion result is basically consistent with the actual thickness of sand body, which confirms the effectiveness of this method in predicting thin sand layers in Pubei area.
The reservoirs controlled by deep and large faults are generally thick and deep. Therefore, a well cannot penetrate completely through an entire reservoir. For calculating the oil column height in fault-controlled reservoirs, a physical model of oil column height in fault-controlled reservoir was established. On this basis, the idea of the wellbore temperature profile extrapolation method was discussed, a formula for calculating oil column height with the conversion method of oil-water column pressure coefficient was derived, and the dynamic reserves inverse method considering the cuboid drainage area and the equivalent flow resistance method considering the influence of gravity were proposed. The four methods were applied to two wells drilled into a fault-controlled reservoir in Fuman oilfield of Tarim basin. The results show that the oil column heights calculated by the four methods are consistent, and the average oil column heights of the two wells are 675.39 m and 634.60 m, respectively.
In China, the abundant low-medium maturity shale oil resources present a huge potential for in-situ conversion. To evaluate the stimulation capability of low-medium maturity shale oil reservoir during in-situ conversion, in-situ heating experiments were conducted on the typical low-medium maturity shales from Chang 7 member of Yanchang formation in Ordos basin and Lucaogou formation of Jimsar sag in Junggar basin. By using techniques such as nuclear magnetic resonance testing, vertical optical microscopy observation, computerized tomography scanning, and pulse decay gas permeability measurement, the dynamic changes in nano-scale pores, thermal fractures, porosity and permeability under high-temperature and high-pressure conditions during in-situ conversion were monitored in a real-time manner. The kerogen pyrolysis-induced fractures and the hydrocarbon generation pressurization effect are key factors for significantly improving the microstructure and reservoir properties. Once the temperature exceeds the threshold (400°C), the extension, density, complexity and connectivity of fractures within the shale significantly increase due to kerogen pyrolysis and thermal expansion of hydrocarbons. Secondary pores with diameters ranging from 2 to 50 nm become dominant in the pore structure. Under in-situ stress, the porosities of the two types of shale can be increased by 3.65 and 2.73 times, respectively, while the permeability can be increased by 624.09 and 418.37 times, respectively. Permeability is more stress-sensitive in the high-temperature stage than in the low-temperature stage. Shale reservoir with lower in-situ stress and higher kerogen content exhibit higher stimulation capability and higher thermal fracturing and thermal permeability enhancement capabilities during in-situ conversion.
The production stabilization and water-cut control of multilayer clastic reservoirs have always been a hot topic in oilfield development. At the medium-high water-cut development stage, oilfields usually exhibit obvious decline of production, scattered distribution of remaining oil, and prominent development conflicts between layers. For these oilfields, there is an urgent need for appropriate optimization and control methods to achieve sustained and stable production. Based on the Bayesian posterior probability method and reservoir streamline simulator, by applying a random maximum likelihood function, the history matching problem was solved and a space data set was constructed. Furthermore, by using finite-memory quasi-Newton gradient method, the data space set was inverted to predict the future. The transient flow velocity of the reservoir flow field was characterized by integrating Pollock streamline tracing method. Thus, a reservoir flow field reconstruction method based on data space inversion was proposed. This method allows real-time optimization of the reservoir injection-production parameters without the need for complex and repetitive calculations. It overcomes the limitations of traditional optimization methods in finely describing flow field evolution and fills the gap in the application of data space inversion in flow field optimization. Taking reservoir B in the Bohai oilfield as an example, the proposed method was used to reveal the mechanism of the reservoir injection-production structure optimization and intuitively demonstrate the process of reservoir flow field optimization. The field application results show that the overall water cut of the reservoir is relatively steady, the scattered remaining oil in the target flooding unit is effectively exploited, and the swept area of water flooding expands by 24.85%, indicating a remarkable flow field control effect. These digitalization efforts for reservoirs will provide valuable reference for the development and data-driven flow field control of similar medium-high water-cut oilfields.
To determine the fluid saturation under the formation conditions of thin oil reservoirs in the Turpan-Hami basin, based on the data of sealed coring in the Wenmi oilfield and Shanshan oilfield, physical simulation control experiments were performed to simulate the influences of depressurized degassing and evaporation losses on core fluid saturation during coring, and then a fluid saturation correction model suitable for sealed coring of thin oil reservoirs in the Turpan-Hami basin was established. The limit of depressurized degassing loss is clarified, that is, when the initial water saturation is greater than 88% or less than 33%, the depressurized degassing loss is weak and negligible. The new model also takes into account the effects of pore volume change, extraction loss in saturation experiments, and evaporation loss under different flooding conditions on saturation measurement, effectively improving the correction accuracy. The error between the oil saturation derived from the model and that from logging interpretation is 0.17%.
The simulation methods and model precision adopted for layered modeling in 3D geological modeling vary with reservoir characteristics and research purposes at different development stages. From the perspective of 3D geological modeling，the reservoir development can be divided into three stages: reservoir evaluation，new block development，and existing block adjustment. The layered modeling algorithms were analyzed and selected for the 5th fault block in Gangdong district 2. It is proposed that the Kriging algorithm should be used for modeling at the reservoir evaluation stage，with a grid resolution of 100 m × 100 m × 5.0 m；the Kriging or Global B-spline algorithm should be used for modeling at the new block development stage，with a grid resolution of 50 m × 50 m × 1.5 m；and the Local B-spline or Converging average algorithm should be used for modeling at the existing block adjustment stage，with a grid resolution of 10 m × 10 m × 0.5 m. This modeling approach can provide results in more coincidence with actual geological conditions and can meet requirements for reservoir research at each stage.
To determine the collapse mechanism of karst caves in carbonate reservoirs, through stress field simulation, and based on orthogonal two-dimensional sections of the karst caves, a two-dimensional mechanical model was established to simulate the stress distribution characteristics of the carbonate karst caves under negative pressure. By multiple linear regression on controlling variables, a karst cave collapse model coupling with the stress function of the critical fracture point was constructed to predict the relationships among cave collapse and stress, depth and width. It is found that the most important factors influencing cave rock burst and collapse are overlying formation pressure, reservoir compressive strength and flexural strength. For two caves superimposed vertically, when their vertical distance is less than 0.3 times the cave radius, the partition between the two caves breaks, leading to the cave connection. During the collapse and rupture, the cave height changes obviously, while the width changes slightly.
The western Qiulitag structural belt is an important successive area for oil and gas exploration in the Kuqa depression, Tarim basin. Its complex surface and underground geological conditions bring challenges to seismic survey. It has been less explored in the main structural belt, where the high steep structural features and complex fault systems cannot be depicted by using previous 2D (single-line and wide-line) seismic data. To address this problem, a combination of wide-frequency band, wide-azimuth and high-density 3D seismic survey and high-density cable 3D seismic survey is used to increase the effective folds in the complex, high, steep structural areas. The microlog-constrained shallow surface velocity modeling is conducted to improve the static correction accuracy in complex mountainous areas and thick gravel-covered piedmont areas. The six-division method for denoising is adopted to improve the signal-to-noise ratio of seismic data in inter-salt and pre-salt structures. The five-dimensional interpolation and regularization technique is applied to mitigate the impact of irregular observation system in mountainous areas on pre-stack depth migration imaging. Moreover, two-way wave reverse time-migration is performed to increase the migration imaging precision of highly steep formations. Through acquisition-processing integration and seismic-geological fusion, the seismic imaging precision in the western Qiulitag structural belt can be significantly improved.
In order to study the fracability evaluation method for low-permeability tight reservoirs, experiments were conducted on six core samples from Well Y301 and Well Y3 in the Yongjin oilfield, Shawan sag, Junggar basin, and the parameters such as rock mineral composition, porosity, stress-strain curves, P-wave velocity, and S-wave velocity were obtained. The experiment results agreed well with logging data, and an empirical rock mechanical model was established for the study area. Meanwhile, based on the equivalent medium model, a new model considering mineral composition and pore structure characteristics was developed for calculating rock brittleness index. Then, a method for constructing the rock mechanical parameter profile of low-permeability tight reservoirs based on logging data was established and applied in Well Y301. The application results show that the Qigu formation in Well Y301 has good fracability, which lays a foundation for the comprehensive evaluation of fracability of tight sandstone reservoirs.
Fault-controlled carbonate reservoirs are highly heterogeneous, with interweaving development of pores, fractures, and vugs of various sizes. For this kind of reservoirs, the dynamic reserves calculated using conventional material balance methods may be larger than the static reserves. By incorporating water-oil ratio and considering rock compressibility coefficients for different pore-fracture-vug media, a comprehensive compressibility coefficient suitable for the fault-controlled reservoirs was derived. On this basis, a new flow material balance equation was established for the fault-karst reservoir, and its accuracy and applicability were verified using numerical simulation. The research results show that the dynamic reserves calculated by the new equation have an error of only 0.1099% with the static reserves obtained from numerical simulation, confirming the new equation’s reliability and accuracy. In the Halahatang area, the relative error between the dynamic reserves calculated using the new equation and the static reserves derived from geological modeling for multiple wells ranged from -4.82% to -0.15%, which is significantly lower than that calculated using the conventional material balance equation. The results obtained from the new equation are closer to actual conditions, making it more suitable for calculating the reserves of the fault-controlled carbonate reservoirs in the Halahatang area.
Considering the varying lithofacies and lithology of the proximal glutenites in the Dongying sag，a three-dimensional geological model of the glutenites was established for wide-azimuth seismic forward modeling. Using the simulated data cube，and through azimuthal stacking of gathers in OVT-domain，the effects of azimuth variation on parameters such as seismic travel time and amplitude were analyzed，and the relationships between azimuth/amplitude and favorable reservoirs were established. The results show that the variation in the sedimentary direction of the glutenites causes azimuth differences in seismic wave propagation，leading to azimuthal anisotropy in seismic reflections. The data cube obtained from azimuthal stacking at the azimuth perpendicular to the sedimentary boundaries is more sensitive to the responses of the top and internal boundaries of the glutenite，with stronger amplitudes. It more effectively reveals the contacts between glutenites of different periods，thereby facilitating the accurate identification of glutenite and fine prediction of favorable reservoir distribution. Wide-azimuth OVT-domain seismic data are proved effective in glutenite prediction，and have been successfully applied in predicting glutenite reservoirs in the steep slope zone of the northern Dongying sag，with the prediction results in good agreement with actual drilling results.
The continental shale reservoirs in the Lucaogou formation of the Jimsar sag in the Junggar basin are lithologically composed of dolomite, argillaceous dolomite, dolomitic mudstone, dolomitic siltstone, and siltstone. The mechanical properties and energy evolution of the continental shale were investigated through laboratory mechanical experiments. The results show that there are significant differences in the rock mechanical properties of different lithologies within the shale reservoir. The compressive strengths of dolomite, dolomitic siltstone, siltstone, argillaceous dolomite and dolomitic mudstone are 112.09 MPa, 98.20 MPa, 85.98 MPa, 81.28 MPa and 58.30 MPa, respectively. With the increase of confining pressure, the brittleness of the rock samples of the continental shale reservoirs decreases and the ductility increases. The rock samples with different lithologies have different energy levels at the peak strength, indicating strong heterogeneity. Furthermore, the total energy, elastic energy and dissipated energy of the rocks with same lithologies under triaxial compression are higher than those under uniaxial compression.
The mathematical models for production cycle of oil and gas reservoirs were reviewed systematically. On this basis, the generalized functional production decline formula was introduced to replace the decline function in the generalized whole-process mathematical model for production cycle, and it is no longer necessary to determine the decline function according to the driving type and flow characteristics of the oil and gas reservoirs. Meanwhile, considering that the generalized whole-process mathematical model for production cycle can be integrated, the expressions of its increasing function items were summarized to form three types of increasing function expressions. When the composite time and undetermined parameters take different values, the new generalized whole-process mathematical model for production cycle can not only be converted into the basic mathematical models for various production cycles, but also form other new mathematical models for production cycle, possessing the general formula and extensibility in the whole-process mathematical model for production cycle. In order to reduce the difficulty in solving the undetermined parameters, five simplest and most common methods for solving the composite time formula and functional mathematical model for production cycle are given. The satisfactory application results verify that the new model is worthy of promotion in other oil and gas reservoirs.
Bottomhole flowing pressure (BHFP) is a key factor determining the rational production system of coalbed methane (CBM) wells for purpose of long-term stable production. The constant mass model (CMM) is not applicable to the wells with double-layer commingled production, since it does not consider the acceleration pressure drop (APD) in the reservoir interval and the mass variation in well sections. Additionally, the BHFP in the lower reservoir is taken as a control parameter for the two intervals, which does not meet the adjustment requirements of the upper reservoirs. In this paper, the APD expression was decomposed and derived, the relationship between APD and the radial flow rate per unit length was established, and the pressure drop formula for the reservoir interval with radial inflow was derived. The reservoir was divided into multiple intervals, and the pressure drop calculation method for each interval was established. Based on the gas/water flow rates in each well section, the corresponding equations for calculating gas/water phase velocities were derived. Combining the above equations, a variable mass model (VMM) was established. The production data were input into the VMM and CMM for comparative verification. The results show that when gas and water are co-produced, the error of the VMM is 2.75%-6.58%, while the error of the CMM is 7.15%-15.18%, indicating that the VMM is more accurate. The BHFP differs significantly in the two reservoir intervals, with the maximum difference of 47.3%. Therefore, it is necessary to adjust the production system depending upon the respective BHFP of the two reservoirs. The VMM can accurately provide BHFP for each commingled interval, so it agrees more with the field conditions. It also avoids the problem of using the same BHFP for both intervals, which hinders precise adjustment of the production system. Thus, the new model provides a technical support for developing optimal production strategies and achieving high and stable production.
The seismic frequency enhancement processing method based on multi-layer residual network combines high-frequency well logging information with seismic data through an intelligent network. This method effectively improves vertical resolution while maintaining lateral continuity，facilitating the identification of thin reservoir beds. In the AMH area，the seismic data processed by conventional techniques enable only the identification of carbonates thicker than 30 m，but not of thinner beds. The seismic frequency enhancement processing method based on multi-layer residual network was proposed for application in this area. First，a training was performed using the multi-layer residual network，a deep learning network，with the near-wellbore seismic amplitudes as training data and the relative wave impedance data from well logging as training labels. Thus，a predictive model for relative wave impedance curve was obtained. By using seismic data as input，the deep network training model was solved to obtain a relative wave impedance data cube，and then a data cube of reflection coefficient corresponding to the frequency-enhanced seismic data cube was obtained. After analyzing the geological conditions of the target area，appropriate wide-frequency wavelet was extracted after calibration，and then convolved with the reflection coefficient cube，so that a frequency-enhanced seismic data cube was obtained. Reservoir inversion was performed using the frequency-enhanced seismic data cube. The inversion results are of high resolution vertically，well matching the main target beds，and can be identifiable and traceable laterally. Ultimately，the identification of thin beds in the AMH area was realized through the application of high-resolution seismic inversion results. The seismic frequency enhancement processing based on multi-layer residual network together with the corresponding high-resolution model inversion can identify beds thicker than 10 m in the AMH area. This method effectively addresses the problem of infeasible thin bed identification using low-resolution seismic data，and improves the accuracy in predicting thin beds. It is referential for identifying similar thin beds.
Faults and fractures are developed in the marine carbonate gas reservoirs on the right bank of the Amu Darya basin. Water is active locally, which leads to severe water invasion during development. Through high-temperature and high-pressure displacement experiments on full-diameter core samples from complex fractured reservoirs, the influences of fracture permeability, fracture penetration degree and water volume multiple on water invasion in gas reservoirs were analyzed. The water invasion patterns in different fractured core samples were investigated by considering the dynamic changes in the water-gas ratio (WGR). The results indicate that as the fracture penetration degree, fracture permeability, and water volume multiple increase, the slope of the WGR curve under the corresponding water invasion pattern increases, suggesting more severe water invasion and channeling. The areas with incomplete fracture penetration can effectively restrain any sudden water invasion. Accordingly, the characteristics of water invasion patterns were further analyzed by using the water invasion diagnosis curves, and the index chart for diagnosis of water invasion in the study area was optimized.
Conventional pre-stack depth migration imaging techniques separate the migration imaging and static correction processing. In the processing of seismic data from complex mountainous areas, due to the factors such as complex surface condition, drastic lateral velocity variation near the surface, and exposure of high-velocity interval, the static correction based on the assumption of surface consistency may cause wave field distortion. This distortion leads to big errors in calculation of travel time, affecting the effects of depth migration imaging. To solve this problem, a full-depth modeling and imaging technique based on true surface migration was proposed. This technique starts with velocity modeling and travel time calculation from the surface elevation, and addresses static correction implicitly in the migration imaging process. It has been satisfactorily applied in the processing of complex mountainous data.
Fractured reservoirs are important exploration and development targets for increasing oil and gas reserves and production. The conventional post-stack seismic data can not meet the needs of fractured reservoir identification. This paper presents a method for the relative separation of reflected and scattered seismic wave fields based on diffusion filtering. This method predicts fractured reservoirs by adjusting diffusion coefficient and the number of iterations to separate the post-stack seismic data cube into 2 data cubes of reflection wave field and scattered wave field，and then extracting seismic attributes such as coherence，curvature，and ant tracking from the scattered wave field data cube. The method was applied to predict fractures in the weathered crust fractured bedrock reservoirs in the Songliao basin and the fractured reservoirs in the Maokou formation in the Sichuan basin. The results show that the method has higher resolution in predicting fractured reservoirs. The method was verified by the actual drilling data of two wells in the weathered-crust fractured reservoirs in the Daqing exploration area，showing a coincidence rate of 67%. This method can provide references for exploration，development，and well deployment in similar reservoirs.
Based on the configuration theory and the analytic hierarchy process (AHP), the configuration shapes of the combined conventional mercury intrusion-constant rate mercury intrusion (CMI-CRMI) curves were classified, and a universal three-segment configuration pattern for the curves was established. The indicative significance of this pattern to the pore-throat systems and their wetting hysteresis was clarified. The results show that the combined CMI-CRMI curves consist of three configuration segments: a, b, and c, which are interconnected but exhibit distinct shapes. The segment a displays an overlapping shape, indicative of a macro-pore-throat system, where the combined CMI-CRMI curve shows no wetting hysteresis. The segment b demonstrates a separated shape and can be subdivided into subsegments b1 and b2. Subsegment b1 indicates a meso-pore-throat system, where the CRMI intrusion curve shows no wetting hysteresis, but the CMI curve does. Subsegment b2 also indicates a meso-pore-throat system, where the combined CMI-CRMI curve shows wetting hysteresis. The segment c exhibits an overlapping shape, representing a micro-pore-throat system, where both the CMI and CRMI curves exhibit equal wetting hysteresis. The deformation of the mercury meniscus during CMI is concentrated in the segments b and c, while the deformation of the mercury meniscus during CRMI is concentrated in the segments b2 and c. Subsegment b1 in both the CMI and CRMI curves can be used for contact angle correction. This three-segment configuration pattern of the combined CMI-CRMI curves provides a significant guidance for segmental contact angle correction and pore-throat distribution characterization.
In the application of reservoir lithology identification, the efficiency, accuracy and effective information integration ability of machine learning algorithm have been fully verified, especially in unconventional reservoirs with strong heterogeneity such as shale. Based on the optimal selection of parameters such as natural gamma, T2 geometric mean, structural index, skeleton density index, density, and deep lateral resistivity, and using a random forest algorithm combined with recursive feature elimination (RF-RFE), major lithologies of the shale reservoirs in the Middle Permian Lucaogou formation in the Junggar basin were identified. Lithology prediction was conducted on the same dataset using conventional RF and support vector machine (SVM) algorithms, and the results were compared with those obtained from thin-section identifications. It is found that RF-RFE yields better results with only half of the logging parameters, and the parameters defined by optimal selection help reduce the algorithm’s running time. Thus, the use of RF-RFE algorithm can realize optimal selection of characteristic logging parameters, more accurate identification of shale lithology, and reduction of running time. The algorithm provides a new approach for complex lithology identification and multi-parameter selection.
