Travis Wiens

Reliable prediction of interfacial tension (IFT) is critical for optimizing engineering processes like hydrogen separation, high-density energy storage and fluid displacement in underground oil reservoir systems. The goals of this study are: (a) to assess the efficacy of optimized ensemble machine learning (EML) models and (b) parametric sensitivity analysis to interpret feature importance while predicting interfacial tension (IFT) across an extensive range of thermodynamic conditions. The dataset comprises 498 experimentally measured IFT records collected under diverse conditions of temperature (T), pressure (P), and gas compositions of hydrogen (H₂), carbon dioxide (CO₂), and methane (CH₄) in complex reservoir brine systems. The EML algorithms of extreme gradient boosting (XGB), random forest (RF), decision tree (DT), light gradient boosting (LGB) and adaptive boosting (ADB) are applied to construct the data-driven computational models to obtain IFT. The model’s effectiveness and robustness are evaluated using statistical performance indices. SHapley Additive exPlanations (SHAP) analysis is adopted to interpret and rank feature importance, thereby ensuring the model’s transparency and interpretability. The XGB model showed excellent performance, with high determination of coefficient (R², 83%) and least root mean square error (RMSE, 4.81 mN/m) compared to other models of RF (81, 5.06), DT (79.94, 5.23), LGB (77.52, 5.54) and ADB (69.32, 6.46) accordingly. Based on SHAP scores, it is revealed that T is the most significant parameter to control IFT, and we ranked them higher to lower weighty order, like T, H₂, P, CO₂ and CH₄. This study demonstrates the efficacy of EML and physics-guided computational models, coupled with feature selection techniques, in capturing IFT phenomena relevant to a subsurface hydrogen–brine geological system. It also establishes an interpretable artificial intelligent framework to uphold the project design, risk evaluation, and optimization of underground hydrogen storage systems in saline aquifers and other geologically complex formations for energy transition.

Accurate prediction of the drilling rate of penetration (ROP) is crucial to optimizing drilling performance, wellbore stability, enhancing operational efficiency, and reducing the overall cost of hydrocarbon exploration. This study focused on evaluating and comparing different machine learning models for ROP prediction and identifying the most influential operational drilling and gas‐derived parameters affecting ROP performance for clastic sedimentary formation. To achieve these objectives, four intelligent models of bidirectional long short‐term memory (LSTM), random forest (RF), categorical boosting (CTB), and extreme gradient boosting (XGB) are implemented for assessing the combined effect of drilling parameters and gas‐derived indices on ROP prediction. The results demonstrated that the optimized deep learning model with LSTM achieved superior predictive accuracy than the ensemble models of RF, CTB, and XGB, with an excellent coefficient of determination ( R ² ) of 85%, with a root mean square error (RMSE) value of 2.58 m/h. Interpretable SHapley Additive exPlanations (SHAP) showed that the most dominant factors are hook load and bit depth for ROP estimation. It is found that SHAP‐based contributions of gas‐derived factors and heavy gas index are also significant, exhibiting the robustness and adaptability of LSTM for assessing the combined effect of drilling mechanical parameters and gas‐derived indices on ROP prediction. This study contributes to scientific transparency and operational excellence by developing an interpretable intelligent framework for reliable prediction of subsurface drilling rates in hydrocarbon production. The proposed framework provides reliable, data‐driven insights that empower petroleum geologists and engineers to make safer drilling decisions, enhance operational efficiency, and promote economically sustainable subsurface resource development.

Machine learning-guided predictive models are attractive in rock modeling for different scholars to obtain continuous profiles of rock compressive strength in rock engineering. The major objectives of the study are to assess the implications of machine learning (ML)-based connectionist models to obtain the unconfined compressive strength (UCS) of rock, to perform parametric sensitivity analysis on petrophysical parameters, and to develop an improved correlation for UCS prediction. The least-squares support vector machine (LSSVM) is applied to develop data-driven models for the prediction of UCS. Additionally, the random forest (RF) algorithm is applied to verify the effectiveness of predictive models. A database containing well-logging data is processed and utilized to construct connectionist models to obtain UCS. For the efficacy of predictive models, statistical performance indicators such as the coefficient of determination (CC), average percentage relative error, and maximum average percentage error are utilized in the study. It is revealed that the RF- and LSSVM-based models for predicting UCS perform excellently with high precision. Considering the parametric sensitivity analysis in the predictive models for UCS, the formation compressional wave velocity and formation gamma-ray are the most strongly contributing predictor variables rather than other input variables such as the modulus of elasticity, acoustic shear wave velocity, and rock bulk density. The improved correlation for predicting UCS shows high precision, achieving a CC of 96% and root mean squared error of 0.54 MPa. This systematic research workflow is significant and can be utilized for connectionist robust model development and variable selections in the petroleum and mining fields, such as predicting reservoir properties, the drilling rate of penetration, sanding potentiality of hydrocarbon reservoir rocks, and for the practical implications of boring and geotechnical engineering projects.

The modeling of transient flows of liquids through tubes is required for studies in water hammer, switched inertance hydraulic converters, and noise reduction in hydraulic equipment. While 3D gridded computational fluid dynamics (CFD) methods exist for the prediction of dynamic flows and pressures in these applications, they are computationally costly, and it is more common to use 1D methods such as the method of characteristics (MOC), transmission line method (TLM), or frequency domain methods. These 1D methods give good approximations of results but require many orders of magnitude less computation time. While these tubes are typically curved or coiled in practical applications, existing 1D solution methods assume straight tubes, often with unknown deviation from the curved tube solution. This paper uses CFD simulations to determine the correction factors that can be used for existing 1D methods with curved tubes. The paper also presents information that can be used to help evaluate the expected errors resulting from this approximation.

Low-cost small-scale (<100 W) electrohydrostatic actuators (EHAs) are not available on the market, largely due to a lack of suitable components. Utilizing plastic 3D printing, a novel inverse shuttle valve has been produced which, when assembled with emerging small-scale hydraulic pumps and cylinders from the radio-controlled hobby industry, forms a low-cost and high-performance miniature EHA. This paper presents experimental test results that characterize such a system and highlight its steady, dynamic, and thermal performance capabilities. The results indicate that the constructed EHA has good hydraulic efficiency downstream of the pump and good dynamic response but is limited by the efficiency of the pump and the associated heat generated from the pump’s losses. The findings presented in this paper validate the use of a 3D printed plastic inverse shuttle valve in the construction of a low-cost miniature EHA system.

To acquire a linear load pressure with the potential of energy-saving compared with conventional throttling hydraulic systems, a switched hydraulic circuit (referred to as a flow-dependent inertia hydraulic converter) featuring a flow-dependent inertial effect was proposed. In this device, the flowing fluid modulated by an equivalent fast switching valve allows the fluid inertance at the connection pipe and rotation inertia incorporated in the flywheel to be combined. The theoretical and simulation models of the flow-dependent inertia hydraulic converter were established, and then the flow-dependent inertia was discussed and quantified. The predictions of the load pressure, flywheel speed and system efficiency were confirmed by experimental measurements. Results indicated that the theoretical and simulation predictions match well with the experiments for this configuration. The pressure wave propagation effect within the connection pipe is responsible for the variation in the transient load pressure and flywheel speed that is initially modulated by the rotation inertia theoretically, which seriously questions the predecessor's hypothesis of using a compressible volume to model the upstream connection pipe. The parabolic mean suction flow induced by the flow-dependent inertia draw our attention to the available supply flowrate and duty cycle. Finally, the throttling loss and system efficiency that varies with the desired characteristics is superior to that of conventional throttling hydraulic systems theoretically, and an acceptable system efficiency was obtained under the premise of ignoring friction experimentally.