Since rock plasticity under in-situ conditions poses challenges during fracturing stimulation, extensive research is necessary on deep gas and oil reserves, which will be the primary area of future development. This paper created a competitive, multi-cluster fracture propagation model that considered elastoplastic rock deformation and nonlinear fracture characteristics in deep reservoirs. It also proposed an optimal fracture design of “dense fracture distribution, non-uniform perforation and alternating staged fracturing” based on stress field reconstruction. The findings indicated that suitably reducing the spacing between clusters and increasing the number of perforated clusters minimized local in-situ stress variations through stress interference among fractures. This mitigated the limiting effect of plastic deformation on the propagation of hydraulic fractures, demonstrating a viable approach for enhancing the expansion of fractures in deep reservoirs. The elastoplastic fracture propagation mechanism was examined to elucidate the advantages of close-cutting fracturing technology. The impact of various fracture techniques was analyzed using stress field reconstruction. Alternate fracturing displayed a high degree of stress reconstruction with an extensive propagation range, which facilitated the propagation of multiple fracture clusters in the subsequent fracturing section. The findings offer a theoretical basis for fracture design of deep reservoirs.

The shale oil obtained from the Gulong continental facies in the Songliao Basin represents an important new exploration target in China. However, the strong heterogeneity, low brittleness, and rich paragenesis characteristics of this reservoir lead to poor adaptability of existing fracturing techniques. In this study, a field-calibrated integrated geological engineering model is established using multi-source on-site data, and a productivity model is formulated based on experimentally measured dynamic changes in permeability with pore pressure. The accuracy of the model is verified using microseismic and production data. A multi-parameter collaborative optimization model was developed to examine the non-uniform distribution of shale oil fractures, with the objective of improving production capacity under given constraints. This study presents a design method that, under a constant fracturing scale constraint (i.e., fixed total fracture length and number of fractures) , couples a production prediction model with non-uniform fracture layout optimization. Application to two wells in the Gulong continental shale shows that the optimized non-uniform fracture distribution increases cumulative oil production by 6.3% and 9%, respectively, under the same constant fracturing scale constraint. The optimization results show that the optimal distribution of fracture density and half-length at each position is governed by the input constraints, namely reservoir brittleness, compressibility, permeability, and heterogeneity continuity. Although the experimental data mainly represent the studied well location and target layer section, the proposed workflow that combines production prediction with non-uniform fracture optimization design, after being recalibrated using field data, can provide a transferable reference method for similar continental shale reservoirs.

To address the bottlenecks of water cut rise and oil production decline during the development of high‑water‑cut oilfields， and the challenges of research methods such as the time‑consuming calculation in traditional numerical simulations and the incapability of a single‑output model to consider the multivariable correlation， this study proposes an injection-production closed-loop optimization strategy based on multi-output support vector regression （MSVR） and multi-objective particle swarm optimization （MOPSO）. Following this strategy， MSVR was used to construct a surrogate model for synchronous prediction of multiple variables such as oil production and water cut， which replaces time‑consuming calculation in numerical simulations. Furthermore， an optimization model with the dual objectives of maximizing the cumulative oil production and minimizing the comprehensive water cut was established， and the MOPSO algorithm was introduced for global optimization. This strategy solves the problems in rapid modeling and multi‑objective coordinated optimization of complex accumulation systems， avoids dimensional redundancy and error accumulation in independent modeling with traditional single‑output models， and enables an intelligent decision support to improving oil production while controlling water. The results show that the MSVR surrogate model has an excellent performance in prediction， with the mean absolute error （EMA） reducing greatly from around 6，000 to 1，000 and the determination coefficient （R²） >0.96 for all critical indicators after hyperparameter optimization， significantly outperforming the SVR model in precision and stability. The MOPSO algorithm can effectively address the conflicts among multiple objectives， and capture the Pareto frontier that explicitly uncovers the restrictive relation between oil production and water cut under different injection-production schemes， thereby achieving efficient acquisition of non-dominated solution set. Field application in actual well groups in Block Xing‑10 has demonstrated that the optimized injection-production scheme achieves a decline of 0.582 3% in comprehensive water cut and an increase of 325.91 m3 in cumulative oil production， verifying the reliability of the proposed method in practices. The study provides a replicable， efficient and intelligent solution and technical support for the secondary development of old oilfields in China.

A large number of fractures of different scales are developed within the shale oil reservoir space. It is of great significance for the design and operation of hydraulic fracturing to study the multiscale fracture propagation morphology and the influence of inter-fracture interference on the fracture propagation. In this study， an extended finite element model （XFEM） of multiscale fracture propagation in hydraulic fracturing was established and validated. Using this model， the multiscale fracture propagation morphology and inter-fracture interference during propagation were analyzed， and the influences of different fracturing conditions on fracture parameters （e.g. initiation pressure， width， and length） were identified. The results show that， during synchronous fracturing， the inward multiscale fracture propagation was suppressed to an increasing extent with the increase of one of the fracture scales； and during asynchronous fracturing， the first fracture closed and extended after the expansion of the second fracture was completed. The perforation spacing had impacts on fracture initiation pressure， fracture width and fracture length， and the larger the fracture scale， the greater the fracture spacing required for full expansion. When the angle between the fracture and the direction of maximum principal stress reached 45°， the inward fracture propagation was most strongly inhibited， the fracture initiation pressure reached the maximum， and the fracture width reached the minimum. The research results provide theoretical guidance and reference for the design and operation of hydraulic fracturing， and contribute new ideas for understanding how interference between multiscale hydraulic fractures during propagation affects fracture complexity， which is a sensitive factor for productivity.

In the extraction of shale gas and oil, the vibration characteristics of the drill string significantly influence wellbore quality, potentially leading to wellbore instability, excessive tool wear, and diminished drilling efficiency. This study tackles the challenges associated with drill string vibrations by developing an integrated technical framework of multi-field coupled dynamic modeling, Sobol-based key parameter identification, and NSGA-II-driven multi-objective structural optimization, and proposes a synergistic vibration suppression strategy combining structural parameter adjustment and hydraulic damper configuration based on multibody dynamics and finite element analysis. Initially, a dynamic model that accounts for the coupling between the wellbore and the drill string is developed to scrutinize the impact of various vibration modes on wellbore quality. Subsequently, detrimental vibrations are mitigated through the optimization of structural parameters, including but not limited to stiffness distribution and the strategic placement of vibration absorbers. Finally, the efficacy of the optimized design is substantiated through numerical simulations and field experiments. The results demonstrate that the optimized drill string achieves a simulation average reduction of 30% in lateral vibration amplitude across the rotational speed range of 60-120 RPM and a simulation average improvement of 25% in the attenuation of axial vibration energy. These enhancements notably bolster drilling stability and elevate wellbore quality. This research furnishes both theoretical and technical underpinnings for the efficient development of shale gas and oil resources.

As oilfields progress into the middle and late stages of production, insufficient liquid supply leads to frequent start-stop cycles in wells, which significantly increase energy consumption. PV has emerged as a new approach for enhancing energy efficiency during oil production. However, balancing the inherent inconsistencies and volatility of PV generation with the stability requirements of oil well energy consumption remains a challenge for optimizing intermittent production during PV-grid complementary power supply. To address this issue, this study proposes a two-stage intermittent production optimization method based on prescheduling and dynamic adjustment for PV-grid complementary systems. A prediction module incorporating spatiotemporal attention was developed to monitor PV fluctuations and generate a prescheduling strategy for intermittent production, thereby improving PV consumption. To minimize the impact of PV forecasting errors on the optimization results, a reference-point-weight adaptive adjustment mechanism was introduced to enhance the NSGA-III algorithm. Considering multiple optimization objectives, including stable production and enhanced PV utilization, the prescheduling strategy was dynamically adjusted, and the optimal intermittent production scheme for well groups using PV-grid complementary power supply systems was obtained using the improved NSGA-III algorithm. The experimental results demonstrated that the proposed method reduced operational costs by 35.18%, achieved a PV utilization rate of 78.34%, and attained an electricity-saving rate of 33.91%, compared to the conventional intermittent production scheme.

In response to the problems such as complex near-wellbore fractures, difficult far-wellbore fracture propagation, and limited stimulated reservoir volume (SRV) caused by the "thousand-layer thin pancakes" configuration of the Guolong shale oil reservoir in the Songliao Basin, China, triaxial mechanical and fracture visualization experiments were conducted on shale samples. Combined with digital image correlation technology and laser pulse ultrafast resolution technology, the micro-scale deformation and supersonic-scale fracture expansion characteristics of the Guolong shale were captured in real time. A constitutive model reflecting the flexible deformation and anisotropy of the Guolong shale and a mechanical model considering competitive fracture initiation-propagation from multiple perforation holes under the coupling of stress interference and flow distribution were established to reveal the control mechanisms of pore density, pore number, and pore distribution on fracture propagation. The results show that by reducing the number of holes and increasing the perforation density, the stress interference between multiple perforation holes can be effectively mitigated, and combined with the extreme limited entry (ELE), the fracturing fluid can be evenly distributed. Compared with the high-density perforation (8 holes per cluster), the low-density perforation (6 holes per cluster) yields an increased opening rate by approximately 45 percentage points. Compared with spiral perforation, the 30 degrees phase angle conjugate directional perforation enables both stress interference reduction and enhanced longitudinal/transverse reservoir connectivity, and it can easily form vertical energy concentration, as indicated by stress field, to drive fracture expansion across layers. The directional perforation + ELE fracturing mode has been verified through field practice. After changing the perforation method from 60 degrees-180 degrees phase angle spiral perforation to 30 degrees phase angle conjugate directional perforation, and reducing the number of perforations from 12-16 holes per cluster to 5-7 holes per cluster, the SRV increased by 17.4% and 48.9%, respectively.

Shale reservoirs often exhibit interbedded lithologies that enhance vertical mechanical heterogeneity, inhibiting hydraulic fracture propagation and reducing vertical reservoir utilization. To address this, appropriate fracturing techniques are needed to improve vertical fracture growth. This study employs an extended finite element method combined with a cohesive zone model to simulate hydraulic fracture propagation in vertically heterogeneous shale. The model investigates the effects of cluster number, cluster spacing, vertical stress difference, clay content, fluid viscosity, and injection rate on fracture height and width. Response surface methodology (RSM) is used to evaluate parameter sensitivity and coupling effects, identifying optimal combinations. Results show that fewer clusters, smaller spacing, higher stress contrast, and lower clay content promote vertical fracture penetration. In contrast, more clusters, larger spacing, higher injection rate, and increased viscosity enhance fracture width. Cluster number and spacing are the most influential parameters; reducing both simultaneously significantly improves fracture height and width. Fluid viscosity and clay content have weaker effects. The optimal cluster number is 3-4 with spacing of 6-11.77 m. Fluid viscosity should be maintained within the range of 24.56-mPa s. The injection rate should be dynamically optimized between 9.09 and 21.19 m(3)/min. These findings offer theoretical guidance for fracturing design in heterogeneous shale reservoirs.

Accurate production forecasting is the foundation for formulating efficient development strategies. In waterflooding oilfield development, well production not only follows its own historical trends but is also influenced by surrounding oil and water wells. Consequently, developing production forecasting models requires the simultaneous incorporation of temporal and spatial dependencies. However, existing spatiotemporal models struggle to effectively capture the complex nonlinear relationship between oil production and water injection rates, leading to suboptimal predictive performance. To address this, this paper proposes an enhanced model named MR-STGormer that incorporates three key improvements to the STGormer architecture: 1) Designing a temporal-sequenceaware hierarchical progressive masking strategy to guide the model in learning spatio-temporal representations of injection and production while generating augmented samples; 2) Proposing an injection-production-aware masked reconstruction module, enabling the model to implicitly learn the joint dynamic patterns between water injection and oil production; 3) Constructing a dual-effect loss function integrating Huber loss and reconstruction loss to achieve an effective balance between accurately capturing production trends and resisting abnormal disturbances. Experiments on real block data demonstrate that MR-STGormer outperforms other mainstream models across multiple evaluation metrics in both single-well and block-level oil production forecasting, achieving approximately 6 % higher prediction accuracy than the best baseline model. This validates its practical value for deployment in actual oilfields.

The laminated structure and lithological alternation characteristics of continental shale restrict the height propagation of hydraulic fractures, thereby increasing the challenges in fracturing design. Based on scratch tests, indentation experiments and seepage-stress coupling model, this study systematically characterizes fracture height propagation patterns in monolithologic reservoir formations. The results demonstrate that argillaceous shale tends to develop short and wide fractures, whereas felsic shale predominantly generates long and narrow fractures. Within the multilithology fracture propagation framework, we establish a fracture height growth coefficient through analytic hierarchy process by integrating three key mechanical parameters: brittleness index, tensile strength, and fracture toughness. The cross-layer propagation capability of fractures is characterized using a cross-lithology gradient cumulative value. A negative correlation is observed between the fracture height growth coefficient and the achieved fracture height. As the cross-lithology gradient cumulative value increases, fracture height exhibits a decreasing trend. Consequently, the fracturing initiation point should preferentially be selected at locations minimizing the vertical cross-lithology gradient cumulative value. This study provides a theoretical basis for optimizing fracturing locations. By enhancing fracture height growth, it improves vertical reservoir utilization efficiency and contributes to the efficient development of shale oil resources.

Tight glutenite reservoirs are known for strong heterogeneity, complex wettability, and challenging development. Gas-Assisted Gravity Drainage (GAGD) technology has the potential to significantly improve recovery efficiency in glutenite reservoir. However, there is currently limited research on GAGD processes specifically designed for glutenite reservoirs, and there is a lack of relevant dimensionless numbers for predicting recovery efficiency. In this study, we developed a theoretical model based on the characteristics of glutenite reservoirs and used phase-field method to track the oil–gas interface for numerical simulations of dynamic GAGD processes. To explore the factors influencing gas-driven recovery, we simulated the effects of strong heterogeneity and dynamic wettability on the construction process under gravity assistance. Additionally, we introduced multiple dimensionless numbers (including capillary number, viscosity ratio, and Bond number) and conducted a series of numerical simulations. The results demonstrate that gravity enhances the stability of the oil–gas interface but causes unstable pressure fluctuations when passing through different-sized throat regions, particularly leading to front advancement in smaller throats. Although strong heterogeneity has negative impacts on GAGD, they can be mitigated by reducing injection velocity. Increasing oil-wettability promotes oil displacement by overcoming capillary forces, particularly in narrower pores, allowing residual oils to be expelled. Among the dimensionless numbers, the recovery efficiency is directly proportional to the Bond number and inversely proportional to the capillary number and viscosity ratio. Through sensitivity analysis of the dimensionless numbers’ impact on the recovery efficiency, a new dimensionless NGlu considering heterogeneity is proposed to accurately predict GAGD recovery of tight glutenite reservoirs.

The structural design of the ball valve significantly impacts the maximum pressure-holding capability of pressure-retaining coring tools. In this study, the pressure-bearing structure of the ball valve was optimized, and a theoretical model for its pressure resistance was established. Through numerical simulation, the maximum von Mises stress [Formula: see text] and effective seal width S were established as evaluation indicators for the valve's pressure retention performance. Based on a sensitivity analysis of the ball valve's structural dimensions, three key design parameters were identified: the valve body inner diameter [Formula: see text], the sealing surface adjustment amount [Formula: see text], and the pressure surface adjustment amount [Formula: see text]. Using response surface methodology (RSM) and central composite design (CCD), a regression model was developed to correlate [Formula: see text], [Formula: see text], and [Formula: see text] with [Formula: see text] and S. The Non-dominated Sorting Genetic Algorithm II (NSGA-II) was then applied for multi-objective optimization, yielding optimal parameters: [Formula: see text] = 60 mm, [Formula: see text] = 37 mm, and [Formula: see text] = 35 mm, the corresponding values of [Formula: see text] and S are 806.67 MPa and 11.02 mm, respectively. The optimized results were compared with numerical simulations, showing errors of 3.53% for [Formula: see text] and 6.9% for S, thereby validating the accuracy of the predictive model. Compared to the initial design, the optimized configuration reduced [Formula: see text] by 8.1% and increased S by 118.2%, significantly enhancing the pressure-bearing strength and sealing performance of the ball valve. This research proposes a novel approach to enhance the pressure-holding capacity of ball valves, providing certain theoretical guidance for improving the performance of pressure-retaining coring equipment.

The flow mechanism of CO2 flooding serves as the theoretical foundation for examining the synergic integration of oil recovery and CO2 storage. Immiscible CO2 flooding has attracted considerable research attention due to its simplicity and cost-efficiency. However, minimal studies are available regarding the flow characteristics and EOR mechanism of immiscible CO2 flooding in in-situ temperature-pressure coupling conditions at the pore scale. Therefore, this study employed a modified high-temperature, high-pressure microfluidic system to simulate the in-situ CO2 and water injection processes in a combined temperature-pressure environment and analyze the multiphase flow characteristics in the pores. The injection rate, displacement pressure difference, displacement efficiency, and residual oil distribution were quantitatively analyzed at different pressures. The results indicated higher residual oil clustering after water flooding at the same injection rate. CO2 flooding significantly reduced residual oil clustering and enhanced the oil flooding effect. The multiphase flow dynamics, type of remaining oil in different injection conditions, and flow characteristics of immiscible CO2 flooding were determined. A higher confining pressure interrupted the CO2 flow, which destabilized the displacement front increased the recovery efficiency by 12.9%. Furthermore, a higher injection rate and displacement pressure increased the recovery efficiency by 24.9% and 6.1%, respectively.

Nowadays, the pre-CO2 fracturing-gas flooding-sequestration synergistic technology (PCFS) has shown a high application potential in improving oil recovery of glutenite reservoirs. However, the influence of CO2 on the physical properties of reservoir rock and its diffusion behavior within the reservoir matrix have not been systematically elucidated in the application of PCFS. In this work, QEMSCAN, SEM, CT scanning, and NMR testing were used to measurement the evolution of pore structure in the glutenite under CO2 exposure. Also, using the VOF method, a two-phase flow numerical simulation was conducted to evaluate the displacement behavior of CO2 within the glutenite matrix. The results demonstrate that the CO2-water-glutenite interaction causes to the dissolution of the pore structure, exhibiting different dissolution phenomena at multiple scales. The most significant dissolution occurred at the millimeter scale, characterized by an increase in pitting, expansion of grain margin fractures, and the increment of porosity of 90.86 %. Following exposure to CO2, both the content and range of saturated water in the pores increased, particularly in larger pores (with a fluid relaxation time of T-2 > 10 ms). The overall permeability of the 3D Representative Elementary Volume (REV) model increases, improving seepage capacity and recovery efficiency. However, some larger mineral particles resulting from mineral decomposition may become stuck in the pores, and secondary mineral precipitation and swelling in the lower flow channels can block CO2 clusters, impeding flow. The findings underscore the complexities of CO2 interactions in glutenite reservoirs and provide insights for optimizing PCFS technology.

The oil and gas industry is increasingly focusing on exploring and developing resources in deep earth layers. At high temperatures, confining pressures, and geostress differences, rock has the mechanical characteristics of plastic enhancement, which leads to the unclear mechanism of hydraulic fracture expansion. The current fracturing model and construction design lack pertinence, and the fracturing reform is difficult to achieve the expected effect. This paper established a model of elastoplastic hydraulic fracture propagation in deep reservoirs. It considered the enhancement of plasticity by examining the elastoplastic deformation and nonlinear fracturing characteristics of the rock. The results confirmed that the hydraulic fractures in deep reservoirs propagated due to plastic energy dissipation after fracture tip passivation, while the stress concentration declined, which increased propagation resistance. The relationship between geology, engineering factors, degree of plasticity, and fracture propagation is discussed, while the conditions that promote fracture propagation are analyzed to provide theoretical support for deep reservoir fracturing design.

To address the characteristic of frequent lithological alternations in the continental shale of the Songliao Basin in China and meet the refined requirements of reservoir modeling, it is necessary to establish a higher-precision lithology identification method. This study conducted scratch tests on shale reservoir cores from the 2360m-2409m interval of the Qingshankou Formation in the Songliao Basin, Jilin, obtaining nine mechanical characteristic parameters, including hardness, compressive strength, and Poisson's ratio. By integrating convolutional neural network (CNN) and auto-encode network (AE), a novel lithology identification method based on scratch data was proposed. The optimal lithology identification scale was selected, and the performance of this method was compared with that of other neural network approaches. The results demonstrate that when the identification scale is set at 20 × 9, the test dataset achieves an accuracy of 89.58%, with recall rates exceeding 84% across all lithology recognitions, outperforming other identification scales. The convolutional autoencoder network (CAE) exhibits superior accuracy and recall rates in lithology identification compared to other neural networks, enabling a more precise representation of the actual lithological characteristics. This study provides a novel methodological approach for reservoir lithology identification and lays a foundation for modeling fracture propagation in heterogeneous shale reservoirs.

CO2 miscible flooding provides dual advantages in enhancing oil recovery and facilitating geological sequestration, and has become a key technical approach for developing low-permeability oil reservoirs and carbon emission reduction. The pore-scale flow mechanisms governing CO2 behavior during miscible flooding are crucial for achieving efficient oil recovery and secure geological storage of CO2. In this study, pore-scale two-phase flow simulations of CO2 miscible flooding in porous media are performed using a coupled laminar-flow and diluted-species-transport framework. The model captures the effects of diffusion, concentration distribution, and pore structure on the behavior of CO2 miscible displacement. The results indicate that: (1) during miscible flooding, CO2 preferentially displaces oil in larger pore throats and subsequently invades smaller throats, significantly improving the mobilization of oil trapped in small pores; (2) increasing the injection velocity accelerates the displacement front and improves oil utilization in dead-end and trailing regions, but a “velocity saturation effect” is observed—when the inject velocity exceeds 0.02 m/s, the displacement pattern stabilizes and further gains in ultimate recovery become limited; (3) higher injected CO2 concentration accelerates CO2 accumulation within the pores, enlarges the miscible sweep area, promotes a more uniform concentration field, leads to a smoother displacement front, and reduces high-gradient regions, thereby suppressing local instabilities, and improves displacement efficiency, although its effect on overall recovery remains modest; (4) CO2 dynamic viscosity strongly influences flow stability: low-viscosity conditions promote viscous fingering and severe local bypassing, whereas higher viscosity stabilizes flow but increases injection pressure drop and energy consumption, indicating a necessary trade-off between flow stability and operational efficiency.

Deep reservoir in-situ environments enhance rock plasticity. Most current experiments that combine temperature and pressure are difficult to observe the crack tip directly, impeding the understanding of deep rock fracture properties. This paper used a digital image correlation method and the principle of fracture mechanics to establish an innovative visual experimental rock fracture system with in-situ environment. Crack tip initiation and propagation were captured dynamically in real-time temperature and pressure coupling conditions, and the influence on shale fracture properties was analyzed, providing a feasible method for examining the fracture behavior of deep reservoir rocks. The results showed that the fracture characteristics of deep shale should be reflected and characterized by jointly considering temperature and confining pressure. Shale plasticity produced crack tip passivation in combined in-situ temperature and pressure conditions, and reduced the local stress concentration effect. The post-peak softening relationship showed a nonlinear “convex” trend, indicating nonlinear fracture characteristics with increased toughness. The exponential coefficient representing the degree of nonlinearity rose as the buried depth increased. The exponential coefficient related to the buried depth was introduced to modify the constitutive model, which was used to effectively characterize the nonlinear fracture behavior of deep reservoirs and provide a basis for deep reservoir fracturing reform research.

The complex vibration phenomenon occurs in the downhole environment of the gas-liquid hydrocyclone, which affects the flow field in the hydrocyclone. In order to study the influence of vibration on hydrocyclone separation, the characteristics of the flow field in the downhole gas-liquid hydrocyclone were analyzed and studied under the condition of vibration coupling. Based on Computational Fluid Dynamics (CFD), Computational Solid Mechanics Method (CSM) and fluid-solid coupling method, a fluid-solid coupling mechanical model of a gas-liquid cyclone is established. It is found that under the condition of vibration coupling, the velocity components in the three directions of the hydrocyclone flow field change obviously. The peak values of tangential velocity and axial velocity decrease, and the asymmetry of radial velocity increases. The distribution regularity of vorticity and turbulence intensity in the overflow pipe becomes worse. Among them, the vorticity intensity of the overflow pipe is obviously enhanced, and the higher turbulence intensity near the wall occupies more area distribution range. The gas-liquid separation efficiency of the hydrocyclone will decrease with the increase of the rotational speed of the screw pump, and the degree of reduction can reach more than 10%. However, this effect will decrease with the increase of the rotational speed of the screw pump, so the excitation effect caused by the rotational speed has a maximum limit on the flow field.

The rough structure in natural fractures will have impact on the fluid flow performance. It is difficult to understand its mechanism through direct research because of complex and disordered structure. In this context, a digital roughness crack model and introducing artificial roughness elements are constructed. And a crack simulation program based on the Lattice Boltzmann Method was written and verified to study the mechanism of the effect of crack roughness on seepage. The results show that: (1) Artificial elements with different shapes have different effects on fluid flow in the fracture, which are reflected in the local flow pattern. The rectangular element has the greatest obstruction effect on the fluid, and has the most developed vortex structure locally. The semicircular element and the triangular element have a decreasing effect on the fluid flow, and the degree of development of the local vortex is also reduced. (2) It is found that the interaction effect between height and distance disappears when the height is the same and the ratio of the distance and height between adjacent roughness elements is greater than 5 and two opposite vortices between adjacent rough elements merge into one. (3) The existence of roughness elements in rough fractures will have a strong viscous effect on the fluid. And with the change of the shape of the roughness element, the resistance strength of the viscous effect to the fluid is also different. The research results help to understand the seepage law of fluid flowing in rough fractures, and have certain reference significance for industrial applications such as oil production.