This study evaluates subsea pipeline collapse under external hydrostatic pressure by combining 3D nonlinear finite element analysis (FEA) and explainable machine learning. A 3D FE model accounting for initial ovality, dual-defect dimensions, and spatial positions was developed to analyze collapse pressure sensitivity. Based on 234 numerical simulations, a Bayesian Neural Network (BNN) alongside the SHAP (SHapley Additive exPlanations) interpretability framework was applied to deliver probabilistic collapse predictions and quantify feature contributions. Findings reveal that initial ovality and internal corrosion depth are the critical features dominating collapse pressure, with a significant coupling effect where initial ovality noticeably amplifies the negative impact of corrosion on structural stability. Notably, within a specific range of minor defects, the load-carrying capacity of pipelines with double-sided defects is slightly higher than that of single-sided ones due to the geometric compensation effect of the cross-sectional neutral axis. The study clarifies the nonlinear influence of the relative circumferential position of internal and external defects on collapse pressure. Specifically, at low initial ovality, the most critical position is around 60 degrees , while this critical position shifts toward 90 degrees as ovality increases. Under limited samples, the BNN model yields an R 2 of 0.9784, a mean prediction interval width (MPIW) of 2.091 MPa, and a prediction interval coverage probability (PICP) of 0.957. SHAP analysis verifies the dominance of initial ovality and defect depth, aligning with the physical mechanisms. This study emphasizes synergistically considering manufacturing defects and multi-source corrosion damage in pipeline assessment, providing practical engineering guidance for offshore pipelines.

Drill pipe threads, as critical structural components connecting drill pipes, are subjected to harsh downhole environments and complex loads. They are prone to defects such as cracks, galling, wash-out, and other failures, which compromise the safety of drilling operations. To address the inspection requirements for drill pipe threads, this paper proposes a defect inspection system based on alternating current field measurement (ACFM) technology. Using Maxwell simulation software, we analyzed the influence of the bottom geometry of the magnetic core leg on the characteristic magnetic field signals of defects and optimized the magnetic core structure. Furthermore, the influence of varying crack lengths and depths on the detection signals was systematically investigated. The results reveal the mapping relationship between the signal features and the defect dimensions, providing a basis for quantitative inversion. Subsequently, the inspection device was designed, a prototype was manufactured, and the complete system was established. The performance of the system was evaluated using drill pipe joints containing defects at the thread tooth crest, thread tooth side, and thread tooth root. The results demonstrate that the system can effectively identify defects at various positions along the drill pipe thread, verifying its reliability.

In order to address the challenges posed by complex feature correlations, high uncertainty, and insufficient model generalization in predicting the corrosion depth of natural gas pipelines under small sample conditions, this paper proposes a hybrid deep learning framework that integrates physical mechanisms with data-driven approaches. The framework utilizes a Bayesian Network (BN) to identify seven critical features and constructs six interactive features based on physical-electrochemical corrosion mechanisms to enhance physical consistency. The model employs a three-stage architecture: XGBoost serves as the baseline model to learn global nonlinear trends and generate initial predictions. The Kolmogorov-Arnold Network (KAN) is first embedded to perform high-order feature modeling on the residuals of corrosion predictions, enhancing stable representation capabilities. The Gaussian Process (GP) performs residual smoothing correction in the embedded space and outputs a 95% confidence interval. Validation based on 242 sets of sample data collected from excavation sites of buried pipelines in southern Mexico that have been in service for over 50 years.The findings indicate that by employing Bayesian methods for joint hyperparameter adjustment, the model attains a prediction performance of R2 = 0.9613 and a root mean square error (RMSE) of 0.2809 on a dataset comprising 242 groups. This enhancement in prediction accuracy is accompanied by a reduction in RMSE of over 50% when compared to a solitary XGB model. A high R2 value indicates that the model possesses exceptional explanatory power and predictive accuracy, while the 95% confidence interval provides reliable uncertainty boundaries for corrosion risk assessment and safety margin determination in engineering practice. The interpretability of the model was enhanced through the implementation of Shapley Additive Explanations (SHAP) and KAN weight analysis, which facilitated the visualization of both global and local feature contributions. The findings suggest that the water content (wc), dissolved chloride ions (cc), pH, and the interaction feature wc_rp exert a substantial influence on pipeline corrosion. This model achieves a balance between predictive accuracy, interpretability, and uncertainty quantification capabilities, thereby providing a reliable foundation for decision-making processes regarding pipeline corrosion monitoring and maintenance in scenarios involving small sample sizes.

This paper proposes a dynamic reliability and system resilience assessment method for the leg structures of a mobile offshore production unit (MOPU) that is subjected to long-term service in complex marine environments. A fault tree of the leg system is first constructed and mapped to a Bayesian network (BN). Time slices are then introduced to expand the network into a dynamic Bayesian network (DBN). To capture history-dependent cumulative effects, a non-homogeneous state transition mechanism driven by physical damage laws is embedded within the DBN framework, within which the Melchers power-law corrosion model and the Palmgren-Miner fatigue model are dynamically modelled. A restoration sub-network comprising diagnosis capability, resource accessibility and maintenance capability is built, and a semi-Markov process is adopted to describe the transition probabilities of degradation and recovery. A quantitative resilience indicator is defined based on the “resilience triangle” theory. Numerical simulations are carried out for four typical scenarios: normal operation, frequent jacking operations, fishing-vessel collision, and blowout. The results show that fatigue degradation is the dominant factor affecting reliability; welded joints are the most sensitive nodes, and under collision/blowout conditions the sensitivity of stress-concentration nodes rises to the second highest; bottom-side collision combined with blowout.

This study proposes a robust control framework that integrates sliding mode control (SMC) with a novel hybrid observer (UKF-LSTM in series) to stabilize separator level and pressure. The stability of the control system is ensured by the Lyapunov method. A significant innovation is a hybrid observer that combines an Unscented Kalman Filter (UKF) and a Long Short-Term Memory (LSTM) network in series to accurately estimate the unmeasurable multiphase inflow. In OLGA plug flow simulations, the framework reduced flow estimation Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) by 73.9 % and 64.7 % over the baseline. The Control tests showed Integral of Squared Error (ISE), Integral of Absolute Error (IAE), and Integral of Time-weighted Absolute Error (ITAE) were 49.8 %, 24.8 %, and 18.0 %, with convergence accelerated by at least 250 s. Results demonstrate that the method achieves a practical balance between accuracy, robustness, and computational efficiency, making it suitable for real-time industrial separator control under variable conditions.

Waste drilling fluid is an important material in oil production, and solid-liquid separation through flocculation is primarily used to treat waste drilling fluids. However, the flocculation effect is currently assessed manually based on experience, which is inefficient and yields low recognition accuracy. Moreover, there is a lack of effective methods to monitor the flocculation effect of waste drilling fluids. In this paper, we propose a lightweight network architecture, incorporating the MobileNetV3 network and a bilinear structure, to classify flocculation images of waste drilling fluids. Flocculation images were collected through beaker experiments, and texture features were extracted to classify the flocculation process into four stages: 10%, 40%, 70%, and 100%. Data augmentation was used to enhance the diversity of samples, and the BS-MobileNetV3 network model outperformed AlexNet, VGG16, GoogLeNet, ResNet18, ResNet50, ResNet101, MobileNetV2 and MobileNetV3 across evaluation metrics such as accuracy, precision, recall, specificity and F1 score. Experimental results indicate that the improved BS-MobileNetV3 achieves an accuracy of 99.4% and a recall of 98.6%. Our BS-MobileNetV3 shows great potential for recognizing the flocculation effect of waste drilling fluids, making it a highly effective identification method.

The paper propose a new lightweight Inception-DVS model for distributed fiber optic vibration warning of longdistance oil and gas pipelines. This model extends the advantages of the Inception network, designs a new model structure to adapt to complex environments, and effectively reduces the computational load and parameters of the model by optimizing the convolution kernel and hierarchical design. The experimental results show that the model maintains 91 % detection accuracy while the model size is only 3.519 MB and the detection speed reaches 3.2 s, which significantly improves the real-time and efficiency. Compared to traditional models, the Inception DVS model performs well in resource-constrained edge computing environments, accurately identifying multiple pipeline threat events such as manual mining, external intrusion, and vehicle travel, while significantly reducing maintenance costs. In addition, the paper also verified the application of the model in the actual pipeline, and the results show that our model can respond quickly and issue accurate early warnings, effectively reduce the workload of manual inspection, and improve the safety and reliability of pipeline operation.

In recent years, high-temperature and high-pressure (HPHT) oil and gas wells have been extensively exploited due to the ever-increasing demand for oil and gas resources. Blowout preventers in oil and gas wells are subjected to high temperature, high pressure, vibration, and corrosion from CO2, leading to accelerated degradation of the blowout preventers. In severe cases, this can result in blowout preventer failures, triggering blowouts and other accidents that cause casualties and property losses. Therefore, the dynamic performance evaluation of the blowout preventer stack under harsh environments is necessary to prevent accidents. This paper proposes an evaluation method for the performance degradation of blowout preventers in different scenarios based on dynamic Bayesian networks (DBNs). First, single-factor models are established for the blowout preventer under different scenarios, considering the influence of various factors. These single-factor models are then coupled using Miner's rule. Second, the coupled model is mapped to a Bayesian Network (BN) to obtain the failure rate. Then, the multiple error shock model is used to analyze the impact of common cause failures on the blowout preventer stack. Finally, considering the influence of multiple degradation states and different maintenance methods, a DBN model for the blowout preventer stack is established to analyze its performance. The model is then validated. The results show that this method can evaluate the degradation of the blowout preventer stack under normal operating conditions as well as the rapid degradation under the influence of blowouts.

Tribo-corrosion-fatigue coupling damage greatly affects the service safety of wire rope. A self-made test rig was employed to investigate the influences of fatigue load characteristics (amplitude, stress ratio, maximum) on bending tribo-corrosion-fatigue damage of wire ropes. The tribological properties and electrochemical corrosion of the wire ropes, and the coupling damage and mechanisms were discussed in detail. Results indicate a lower friction coefficient between wire ropes in artificial seawater. The main wear mechanisms in seawater are adhesive wear, fatigue wear, oxidation wear and corrosion wear as compared to abrasive wear, adhesive wear, oxidation wear and fatigue wear in pure water. Lower stress ratios amplify corrosion and adhesive wear. Fatigue fractures are mainly ductile, with stress ratio being a significant factor. Tribo-corrosion-fatigue coupling exacerbates damage in contact ropes compared to fatigue ropes. Seawater increases fatigue rope damage but reduces contact rope damage. Higher fatigue load amplitude and maximum value, and lower stress ratios, increase friction coefficient, deformation, volume loss, wire breakage, and decrease corrosion resistance. These findings provide essential data for assessing wire rope service life under seawater corrosion conditions and ensuring offshore drilling safety.

In this study, we propose a reliability framework that combines the GO method with fault tree analysis and integrates human factor reliability analysis. Taking the application of remotely operated vehicle (ROV) in underwater oil and gas production as an example, a reliability model of ROV operation process is established. The human reliability in the ROV operation process is analyzed on the basis of Cognitive Reliability and Error Analysis Method (CREAM) extension approach. The influence of human factor reliability and equipment reliability on the system is compared and analyzed. The importance of each operation step is also compared and analyzed. A sensitivity analysis of the ROV during operation is performed. This study provides a theoretical basis for the improvement and maintenance of the operational reliability of ROV. The proposed reliability modeling approach solves the reliability analysis problem of large and complex operating systems.

The high integrity pressure protection system (HIPPS) on the Floating Production Storage and Offloading (FPSO) unit is essential for handling various emergencies. However, if the location of the fault cannot be accurately identified, proper measures may not be taken to isolate the hazard. This paper presents a fault diagnosis method for HIPPS on FPSO units based on Dynamic Bayesian network (DBN). The method considers the influence of sensor and system equipment degradation on the diagnosis results and avoids the problem of overdiagnosis in static diagnosis networks. Six fault diagnosis cases of the system are analyzed and discussed to verify the accuracy and effectiveness of the proposed method. By changing the failure rate of the faulty component, it is determined that the posterior probability of the faulty component increases with the increase of the failure rate at the same time.

Although overpressure burst of pipeline with pressure-protection systems occurs infrequently in case of blockage, it endangers personnel safety and causes serious economic loss when it happens. Therefore, assessing the dynamic probabilistic risk of overpressure burst of pipeline with pressure-protection systems in case of blockage is necessary to prevent accidents. This study proposes a method based on a dynamic Bayesian network (DBN) for assessing the dynamic probabilistic risk of overpressure burst of pipeline with pressure-protection systems in case of blockage. First, a fault tree (FT) model of pipeline overpressure burst is constructed. The constructed FT model is then mapped into the DBN model to solve the uncertainty of the model. Second, the leaky Noisy-OR gate is used to define the uncertain logical relationships of relevant nodes in the DBN model. The effect of common cause failures on a pipeline's redundant pressure-protection system is analyzed using the multiple error shock model. Finally, a natural gas export pipeline is presented as an example to demonstrate the proposed method. Results show that the method can effectively assess the dynamic probabilistic risk of overpressure burst of pipeline with pressure-protection systems in case of blockage.

In order to solve the separation efficiency problem of the vertical centrifuge. First, understand the working principle and dynamic theory of the centrifuge, preliminarily design the technical parameters from the experience and experimental data of the prototype, solid works model, specify the overall structure scheme, and complete the virtual assembly. Secondly, the work performance is verified. The production capacity is calculated with power to meet the design requirements. Then, to verify the design strength and deformation meet the design requirements. Then, the finite element numerical analysis of the internal flow field. The FLUENT software is used to analyze the internal flow field velocity distribution and the turbulent kinetic energy of the centrifuge. Finally, a solid phase distribution rule analysis was performed to investigate the effect of particle diameter size on the separation performance at a suitable particle volume fraction.

The plugging and abandonment (P&A) operation is required when an offshore well is abandoned. This study proposes an analysis method based on dynamic Bayesian network (DBN) to assess the dynamic risk of riserless well intervention systems performing the P&A operation. First, each phase of P&A operation is depicted by a GO model. The fault tree (FT) model of each mission execution system is constructed to analyze the equipment factors leading to the failure of the mission. The mirror DBN model is established from the GO and FT models according to the mapping algorithm and the parameters of the DBN model are obtained by mapping according to the GO and FT model parameters. A complete risk assessment model is established. Next, the standardized plant analysis risk-human reliability analysis method is used to quantify the probabilities of human factors in the DBN model. Finally, the dynamic risk of the operation process is assessed. The developed DBN is verified using a method based on three axioms. Results show that the reservoir abandonment phase has the greatest effect on P&A operation process risk. Sensitivity analysis is performed to identify critical equipment of the system. The proposed method can be a useful tool to assess the dynamic risk of P&A operation process.

The simulation and control of the severe slugging flow in the subsea multiphase pipeline is the focus of research in the production and exploitation of oil companies. Severe slug flow results in severe fluctuations of pressure and flow rate at both the wells end and the receiving host processing facilities, causing safety and shutdown risks. To prevent the severe slugging flow regime in multiphase transport pipelines, an ODE model is established by using the mass conservation law for individual phases in the pipeline and the riser sections. Then, the proposed model is compared to the results from the OLGA simulation. A comparative study of different slugging flow control solutions is conducted. Extended Kalman Filter (EKF), Back Propagation Neural Network (BPNN) and EKF&BPNN are used for state estimation and combined with PI controller. The EKF and BPNN are good nonlinear filters. However, when the nominal choke opening is increased, they work unsatisfying. The EKF&BPNN observer shows slightly better results than EKF and BPNN when the system has high input disturbance.

The Riserless Well Intervention (RLWI) system that performs complex offshore oil well Plugging and Abandonment (P&A) operations is a typical Multi-Mission Phased-Mission System (MM-PMS), which requires multiple missions to be completed within a phase. P&A processes involve complex operations and equipment that can contaminate local marine ecosystems if they fail. Therefore, it is necessary to evaluate the reliability of the RLWI system. This paper proposes a dynamic reliability evaluation model for analyzing the RLWI MM-PMS. The GO model of the phase operation process and the Fault Tree (FT) model used to analyze the failure of each mission were established, and a Dynamic Bayesian Network (DBN) model based on the GO model and the FT model was developed for reliability evaluation. The established model can analyze the changes in the reliability of the RLWI MM-PMS more comprehensively, and can also clarify the importance of different missions and different system components. In addition, considering the impact of the marine environment on operators, the Standardized Plant Analysis Risk-Human (SPAR-H) reliability analysis is used for quantification. These findings can guide the improvement of the reliability of the RLWI system and the success rate of P&A operations.

The lifting and sinking motion generated by the drilling platform causes large fluctuations in the suspension tension of the large hook, which affects the stability of the drilling pressure acting on the rock at the bottom of the well, greatly reduces the efficiency of drilling and decreases the service life of the drill bit and the drill column under the action of cyclically varying loads, causing an increase in the overall drilling cost. In order to avoid the above situation, a nonlinear model of a semi-active lift-sink compensation system is designed and a method to improve the compensation efficiency of the system is proposed. Firstly, the relevant physical model of the hydraulic system is established, and then its simulation model is built according to the working principle of the semi-active lift and sink compensation system. Then the effects of different factors are added to the simulation model of semi-active lifting and sinking compensation, and finally the effects of different factors on the compensation efficiency of the active lifting and sinking system are compared. The simulation results show that a regular sinusoidal wave with a peak value of 6.5 m and a period of 10 s as the ship's lifting and sinking displacement input will reduce the compensation efficiency of the semi-active lifting and sinking compensation system, while the smaller the stiffness of the wire rope, the better the compensation effect of the drill column and the higher the compensation accuracy. The proposed method can effectively improve the compensation efficiency of semi-active lifting and sinking compensation system.

The stability of the subsea oil and gas production system is heavily influenced by slug flow. One successful method of managing slug flow is to use top valve control based on subsea pipeline pressure. However, the complexity of production makes it difficult to measure the pressure of subsea pipelines, and measured values are not always accessible in real-time. The research introduces a technique for integrating Unscented Kalman Filter (UKF) and Wavelet Neural Network (WNN) to estimate the state of subsea pipeline pressure using historical data and a state model. The proposed method treats multiphase flow transport as a nonlinear model, with a dynamic WNN serving as the state observer. To achieve real-time state estimation, the WNN is included into the UKF algorithm to create a WNN-based UKF state equation. Integrate WNN and UKF in a novel way to predict system state accurately. The simulated results show that the approach can efficiently predict the inlet pressure and manage the slug flow in real-time using the riser's top pressure, outlet flow and valve opening. This method of estimate can significantly increase the control effect.

: This work aims to investigate the microstructure and the mechanical behavior of dissimilar API X80-AISI 4130 steel welded joints of coiled tubing. The microstructure at the weld interface of the AISI 4130 steel side results in a relatively lower toughness. The HAZ of AISI 4130 steel has an obvious coarsening phenomenon. The tensile tests show that the deformation process of the dissimilar weld joint is mainly controlled by the two base materials: AISI 4130 steel at the beginning of the deformation and API X80 steel at its end. Charpy impact toughness tests conducted on dissimilar welded specimens led us to conclude that the higher Ni content accounts for a more gradual decline in impact toughness for AISI 4130 steel.

During the drilling process, the complicated geological environment makes drilling operations more difficult as the drilling depth increases, leading to a greater susceptibility to drilling incidents. The parameters obtained from a drilling incident are usually incomplete and the sample size is small, which is difficult to be used for incident analysis. This paper proposes a new method for the diagnosis of downhole drilling incidents. The drilling data is generated based on an Auxiliary Classifier Generative Adversarial Networks (ACGAN) and an incident diagnosis model is built using the Bayesian network (BN). The effectiveness and practicality of the proposed method are proved by the actual case study. Based on historical data, data augmentation is performed using the ACGAN model, and then parameter learning of BN is conducted. The established BN model based on large data samples can be used for the diagnosis of downhole incidents. The precision and F1-score of diagnosis are above 80%. Root cause diagnosis of downhole incidents can be performed by backward inference of Bayesian methods. It can prevent the occurrence of downhole incidents. The results prove the proposed method can diagnose downhole incidents in real time and obtain the causes of downhole incidents.