Adsorbent research is crucial in the field of adsorption and separation, so the key to accelerating the development of new adsorption separation technology lies in the screening of porous adsorbents. New porous materials of metal-organic frameworks have received widespread attention in the field of adsorption and separation. The number of them has exploded in recent years, but it has also brought pressure to the screening of adsorbents. Machine learning has brought innovative breakthroughs in the discovery, design and application of porous materials, leading the research of porous adsorbents into a new data-driven paradigm. This article introduced the current status of machine learning research in the field of porous adsorbents in recent years. Through key case studies, it sorted out the progress in the database of porous materials, adsorption performance prediction and other related machine learning works, and analyzed the principles and characteristics of model input in porous material machine learning. Finally, it was concluded that standardized databases, knowledge transfer, bridging the gap between experimental and simulation data and interpretable models were the future development directions of machine learning research on porous adsorbents.The article provided concise resources for researchers who wanted to use machine learning to develop new porous adsorbents.

The adhesion performance of recycled asphalt binder and aggregate interfaces as well as the influence mechanisms of bio-rejuvenators on aged asphalt was investigated through molecular dynamics simulations. Models of aged asphalt and asphalt rejuvenated with 2% cashew shell oil and 2% tall oil were constructed. A limestone aggregate model was further developed and direct tensile simulations were conducted to calculate the interaction energy at the asphalt-aggregate interface, characterizing the adhesion performance. Additionally, mean square displacement and relative concentration were used to explain the diffusion behavior of asphalt molecules at the interface and the improvement mechanisms of bio-rejuvenators. The results showed that cashew shell oil improved the adhesion performance by approximately 32%, primarily through π-π stacking interactions and the polarity of phenolic groups, which reduced the aggregation of aged asphalt. Tall oil enhanced adhesion by about 17%, improving the dispersion of aged asphalt molecules through the depolymerization effect of its molecular branched structure. This study innovatively combined interface tensile simulations with bio-rejuvenators, systematically revealing the influence of bio-rejuvenators on the asphalt-aggregate interface adhesion. The findings provided theoretical support for optimizing the adhesion performance at the asphalt-aggregate interface and for the application of bio-rejuvenators in asphalt, offering important guidance for cross-scale studies on asphalt-aggregate interface performance in pavement engineering.

The structured packing is extensively utilized in gas-liquid separation processes due to its ability to provide a high specific surface area and a low pressure drop. The performance of the structured packing depends on the local gas-liquid distribution, and the perforation has a significant impact on the gas-liquid flow inside the packings. However, the influence of perforation on liquid flow of corrugated plate is often ignored in previous studies. In this study, a gas-liquid two-phase flow CFD model was established under the framework of VOF. The microscale flow of the CO₂ absorbent on the surface of corrugated plate of structured packing in the absorption tower of Huaneng Zhengning 1.5×10⁶t/a carbon capture project was simulated, and the hydrodynamics of typical absorbents on the corrugated plate in the presence of perforations were discussed. The simulations revealed that when the spray density was 20m³/(m²·h), the existence of perforation led to rivulet flow separation and droplet formation, but had little effect on liquid holdup and interfacial area. The liquid maintained a steady rivulet flow along the channels in the absence of perforations, while the liquid flowed in the form of droplets in the presence of perforations. Remarkably, the physical properties of absorbents exerted a significant impact on the liquid flow morphology. The decrease of Ka (i.e., the surface tension decreased or the viscosity increased) resulted in an increase in the time required for the flow to reach the steady state, liquid hold, interfacial area, and wetting area. The influence of contact angle (i.e., the surface texture of the corrugated plate) was effectively studied by modifying the wall boundary conditions. The flow morphology of the 30%MEA on the corrugated plate with perforations transitioned from droplet flow to rivulet flow and finally to film flow as the contact angle decreased. In addition, at a contact angle of 20°, the interfacial area predicted by simulation was consistent with that predicted by the Olujic model, whereas the predicted holdup was marginally lower than that suggested by Billet-Schultes model, yet the overall trend was relatively consistent.

This paper focuses on conducting microscopic-level reaction molecular dynamics simulations of the thermal decomposition and reduction system of trichlorosilane in a hydrogen atmosphere involved in the chemical vapor deposition process of polysilicon production. It qualitatively and quantitatively investigates the influence of the quantity ratio of reactants H₂ and hydrogen radicals H·on the microscopic action mechanism of the reaction process. Additionally, it comparatively analyzes the dynamic evolution and main transformation paths of reactants SiHCl₃, H₂, and H·, as well as intermediates HCl, SiH₂Cl₂, and SiH₄ in different reaction systems, providing fundamental theoretical support for the process improvement of the chemical vapor deposition process of polysilicon. The simulation results indicate that the reactivity of H· in the reaction system is significantly higher than that of H₂. The introduction of H· can noticeably accelerate the thermal decomposition and reduction processes of SiHCl₃ molecules in a hydrogen atmosphere. Specifically, the larger the quantity ratio of H· to H₂ added to the initial reaction system, the greater the number of SiHCl₃ molecules that are transformed when reaching reaction equilibrium. The production amount of intermediate HCl molecules is positively correlated with the quantity of H· added to the initial reaction system. An appropriate amount of H· can prompt SiHCl₃ molecules to form monohydrogenated species, while excessive H· promotes the formation of polyhydrogenated species. When the reaction temperature of the actual reaction system is set at 1000K and the quantity ratio of SiHCl₃ to H₂ is set at 1∶1, it is conducive to the formation of the byproduct SiH₂Cl₂. To obtain the intermediate SiH₄ at a relatively lower reaction temperature (1000K), the quantity ratio of SiHCl₃ to H₂ needs to be at least greater than 1∶1. Increasing the content of H₂ in the reaction system is beneficial for enhancing the yield of the intermediate SiH₄.

Stress softening of rubber materials refers to the phenomenon of stress reduction after multiple cycles of loading (tension-recovery). This phenomenon is a critical factor affecting the effectiveness of related engineering applications and engineering safety. The present study investigated the stress softening phenomenon of styrene-butadiene rubber materials by using all-atoms molecular dynamics simulations. The study explored and summarized the microscopic features and critical factors of stress softening from the molecular scale. The results indicated that the stress softening of styrene-butadiene rubber was determined by interchain interactions. The total kinetic and potential energy distributions of macromolecules were critical for the characterization of interchain interactions. Styrene group units played a dominant role in the total kinetic energy distribution and the total potential energy distribution of SBR,due to the presence in the SBR molecule as side-branched chains and the large size. The butenyl group units exerted a significant influence on the total kinetic energy distribution. However, its impact on the total potential energy distribution was negligible. In contrast, the vinyl group unit played a negligible role in both the total kinetic energy and total potential energy distributions. At the same strain, the free volume in the system increased with the number of cyclic strains and gradually formed concentrated and continuous cavities. These cavities were important microstructural features of the stress softening phenomenon.

Traditional formulated product design relies heavily on experience and extensive experimentation. Computer-aided molecular design (CAMD) transforms this process by identifying chemical substances and their molecular compositions which match the target product's functionality, greatly enhancing design efficiency and accuracy. However, for structural products (e.‍g., detergents, toothpaste), performance is mainly determined by microstructure, and quantifying certain properties (e.‍g., foaming) is challenging due to insufficient data, making it difficult to establish quantitative structure-property relationship (QSPR) models. To address this, this paper proposes a computer-aided formulation design method that combines molecular dynamics (MD) simulation with machine learning. MD simulation is first used to study the microscopic behavior of different ingredients and their formed microstructure in the formulated product, revealing the interaction mechanisms among components. Then, quantifiable MD descriptors are extracted from the simulation results to construct a QSPR model in combination with Bayesian neural network, predicting the relationship between formulation composition and product performance. Finally, a mathematical optimization algorithm is employed to solve for the optimal formulation composition. A toothpaste formulation case study is provided, showing that this method can significantly reduce the formulation design search space and enhance design efficiency and scientificity. It also indicates the direction for formulation design. As technology advances and market demands evolve, computer-aided design methods are expected to play a more important role in the chemical engineering field in the future.

Chemical absorption is one of the most widely used CO₂ capture methods, and amine solvent chemical absorption technology is more mature. The amine solvent carbon capture process is a typical chemical process with the characteristics of “three transfers and one inversion”. In this process, the diffusion coefficient is an important parameter for evaluating the mass transfer process, and it also affects the kinetic performance evaluation and the reaction performance evaluation of CO₂ absorption by amine solvents. In this study, the simulation accuracy of the traditional solvent molecular force field model was found to be low. To ensure the accuracy of the simulation, the solvent force-field model was first modified, and the results showed that the modified force-field model could accurately describe the intermolecular interactions and self-diffusion coefficients in the solvent of N-Methyldiethanolamine (MDEA), and the prediction error was less than 3%. Then, based on this model, the self-diffusion coefficients of other monoamine solvents [monoethanolamine (MEA), diethanolamine (DEA), etc.] were calculated, and the quantitative relationship between the self-diffusion coefficients and temperature was deduced. On the other hand, in order to explore the influence of ions generated during the reaction process on the kinetic properties of amine solvent, the simulation systems under different reaction progress were constructed. The findings of the study indicated a gradual decrease of the self-diffusion coefficient during the reaction process, with a concomitant linear relationship between the coefficient and the reaction progress. Furthermore, the simulation of mixed amine solvents (e.g. MEA and MDEA) at varying proportions demonstrated that the self-diffusion coefficient exhibited a linear relationship with the mass fraction of MEA. Through further analysis of the intermolecular interaction energy and radial distribution function, the variation rule of the amine solvent self-diffusion coefficient and its microscopic mechanism were explained. In conclusion, this study modified the force field model and provided important theoretical support and simulation tools for optimizing the CO₂ capture performance of amine solvents.

The support vector machine (SVM) model in the machine learning algorithm was used to predict the size distribution of hydrotalcite, with ultrasonic attenuation signal as the main input variable and the size distribution measured by Malvern laser particle size analyzer as the output. The study firstly extracted and analyzed the features of the ultrasonic attenuation signal, combined with the strong learning ability of SVM, and introduced principal component analysis (PCA) to establish a size distribution prediction model. The model’‍s hyperparameters were optimized using grid search optimization. The experimental results showed that the SVM model could effectively capture the complex relationship between the ultrasonic attenuation signal and the size distribution of hydrotalcite, with the predicted peak shape similar to the measured value. The mean square error (MSE) was 0.1373, the coefficient of determination (R²) was 0.9758, the maximum absolute error of the cumulative size distribution was 0.0405, and the prediction accuracy was significantly higher than traditional statistical methods. The study showed that this method had great advantages in dealing with nonlinear complex data and provided a new idea for the nondestructive detection and real-time monitoring of the size distribution of hydrotalcite. It had high application value and potential for promotion.

This study investigated fluidization experiments of wood particles with five different aspect ratios (AR=1—5) in a rectangular fluidized bed. The total pressure and sectional pressure drops were measured, and the fluidization behavior of different wood cylinders was recorded using high-speed photography. An image processing method was employed to analyze the height and orientation distribution of cylindrical particles at different gas velocities. The results demonstrated that wood cylinders with aspect ratios below 5 could achieve complete fluidization in the rectangular fluidized bed, whereas significant channel flow occured for cylindrical particles with AR=5, making them difficult to fluidize. The study revealed that AR significantly affected both the sectional pressure drop distribution and the minimum fluidization velocity, with the blocking effect of cylindrical particles becoming more pronounced as AR increased. Furthermore, when fully fluidized, wood cylindrical particles with different AR values tended to align their axial direction with the gravity direction. These experimental results and observed channel flow could provide fundamental data for validating and understanding gas-solid systems using computational fluid dynamics simulations.

The double-layer porous media combustor offers advantages of high combustion efficiency, a wide fuel combustion limit, and low pollutant emissions, making it valuable for industrial applications. However, the complex flow-transfer-reaction behaviors inside the reactor leads to challenges of design, control, and optimization. In this work, the pore scale computational fluid dynamics (CFD) was employed to modelling a full-scale double-layer porous media methane combustor. A bridge method was utilized to simulate solid-particle heat conduction, where a porous zone connects with adjacent particles. The effects of various inlet velocities and equivalence ratios on methane combustion behaviors were investigated. The results showed that the flame was dispersed within the pores and stabilized at a certain horizontal position in the combustor. Specifically, the flame was stabilized near the interface of the two porous media layers at low flow velocity (0.3m/s). With the increase of inlet velocity, the flame position is shifted downstream. Conversely, The flame position was shifted to upstream with an increase of equivalence ratio. The flame temperature increased with an increasing of inlet velocity, resulting in a localized temperature distribution, which a high temperatures zone existed in the smaller pore regions. Additionally, the maximum velocities were found to reach 56.0—58.7 times of the inlet velocity. The results also showed that pore scale CFD simulation could reveal the flow-transfer-reaction behaviors of double-layer porous medium combustion reactors.

To enhance the convective heat transfer performance of artificially designed open-cell foams, a novel porous foam model, the Laguerre Voronoi (LV) model, is proposed. This model effectively captures both the periodicity and randomness of real foams, avoiding flow-through and enabling quantitative control over pore blockage issues, while demonstrating excellent convective heat transfer performance. Using a parameterized integrated modeling approach, a series of LV foams with different geometric parameters are reconstructed. Pore-scale numerical simulations are conducted to investigate the influence of inlet velocity, apparent, and pore structure parameters (porosity, hydraulic diameter, cell diameter, and relative strut diameter) on the convective heat transfer performance of LV foams. Experimental measurements are used to validate the reliability of the numerical results, and comparisons are made with existing models, including the Lord-Kelvin (L-K) model, traditional process models, and X-ray computed tomography (X-CT) scan models. Based on this, two types of convective heat transfer correlations are fitted using pore and apparent parameters. The applicability, prediction accuracy, and error distribution characteristics of the correlations are analyzed in detail. The results show that, compared to the L-K model, LV foam exhibits more realistic pore structures and higher convective heat transfer performance. The volumetric heat transfer coefficient increases monotonically with the inlet velocity and follows a parabolic distribution with increasing relative strut diameter (or decreasing porosity), with its symmetry axis related to the Reynolds number. The two convective heat transfer correlations proposed in this paper are applicable to a wide range of geometric parameters and Reynolds numbers, with the maximum relative error less than 20%, showing good prediction accuracy and providing an efficient tool for predicting the convective heat transfer performance of open-cell foams.

The computational fluid dynamics-population balance model (CFD-PBM), which integrates the advantages of CFD in predicting flow field with PBM in calculating bubble size distributions, has been widely applied in numerical simulations of bubble column reactors. However, most existing studies focus on ambient pressure air-water systems, limiting the direct applicability of their findings to industrial slurry reactors operated under high-temperature and high-pressure conditions. This review summarizes recent advances in understanding the effects of pressure, viscosity, surface tension, and particles on hydrodynamic behaviors, with a focus on modeling bubble breakup and coalescence under various operating conditions. Simulation results show that the CFD-PBM coupled model can effectively characterize hydrodynamic behaviors in slurry reactors and has wider applicability than the traditional empirical correlations. While current simulations predominantly address cold-mode laboratory-scale reactors, future investigations should focus on resolving multi-scale coupling challenges in industrial-scale systems. Such advancements are crucial for enabling the translation of the CFD-PBM framework from fundamental research to practical industrial applications.

Convolutional neural network is one of the representative algorithms of deep learning, and has made remarkable achievements in many fields such as computer vision and natural language processing in recent years, among which visual geometry group (VGG) is a widely used convolutional neural network model, and shows great potential in image recognition and classification tasks. However, the application of this algorithm in the data regression prediction scenario is not good, and there are limitations of large number of parameters and large memory consumption. To this end, a new deep VGG architecture (D-VGG) is proposed, which improves the different convolution kernel configurations of block1, block2, block3, block4 and block5 in VGG, adopts a more efficient conv5-32 layout, and adds batch normalization layers in each layer of the block. A Dropout layer is added before the fully connected layer to effectively alleviate overfitting problems and accelerate the convergence of the training process. The improved D-VGG network architecture has demonstrated excellent simulation prediction performance, solving the problem of poor regression prediction performance of convolutional neural networks. The prediction performance of the network is evaluated using the data set of an ethylene cracking furnace in a factory, and compared with other machine learning models CNN, CNN-LSTM, BP neural network, SVR, etc. The experimental results show that the prediction performance of D-VGG model is superior to other models, the test set R² reaches the highest of 0.9748, RMSE decreases by 37.5% compared with VGG16 model, and its MAE, MBE and RMSE error evaluation indexes are the smallest.

Abnormal gas-solid two-phase flow states inside the disengager impair the unit's separation capacity and cause substantial catalyst loss. Abnormal operating conditions of the catalyst loss severely affect the long-term safe and stable operation of the unit. This study employs numerical simulation to investigate the variation laws of gas-solid two-phase flow, heat transfer, and separation processes inside the disengager under conditions of changes in the oil gas flow, changes in catalyst circulation flow, and mechanical failures in the separation system. Additionally, a method for diagnosing abnormal catalyst loss in the disengager is established. The results show that changes in the disengager's operating parameters alter the internal gas-solid content. When the inlet conditions of the top cyclone deviate from the normal operating range, the disengager exhibits intermittent abnormal catalyst loss. Mechanical failures in the disengager cause abnormal gas-solid fluidization states such as gas in-leakage and short-circuit flows, which further disturb the velocity, temperature, and pressure distributions inside. This disruption sharply reduces the top cyclone's separation capacity, leading to sustained abnormal catalyst loss.Based on the catalyst loss mechanism and numerical simulation results, a diagnostic method is established for catalyst loss caused by changes of the operating parameters and equipment mechanical failure. Its accuracy is validated through industrial cases. This study provides theoretical support for preventing and handling catalyst loss in the disengager, facilitating the efficient operation of fluidized catalytic cracking units.

Ammonia cracking for hydrogen production is an emerging method that utilizes the exhaust gas from ammonia combustion as the heat source. A multi-physics coupling numerical model of an ammonia cracking reactor utilizing the waste heat from ammonia-hydrogen combustion was established. In this model, the catalytic region was treated as a porous media. The effects of exhaust gas and ammonia inlet temperature, velocity and flow patterns on ammonia conversion, hydrogen yield, and thermal efficiency were investigated. The results showed that increasing the exhaust gas inlet velocity and temperature, increasing the reactant inlet temperature, and decreasing the reactant inlet velocity could significantly improve the ammonia conversion and hydrogen yield. Additionally, increasing the exhaust gas inlet temperature and reactant inlet velocity, decreasing the exhaust gas inlet velocity and the reactant inlet temperature could be helpful to improve the thermal efficiency of the reactor. Notably, the exhaust gas parameters had a greater influence on thermal efficiency compared to the reactants. These findings contributed to the design and optimization of ammonia cracking reactors integrated with the waste heat from ammonia-hydrogen combustion utilization.

In this study, microchannel heat exchangers were designed using the Gyroid structure, a type of triply periodic minimal surface (TPMS). The heat transfer and fluid flow characteristics of these heat exchangers were analyzed using numerical simulation. The Gyroid structure's mathematical representation and its unique features were discussed in detail. Simulations were conducted on heat exchanger components filled with Gyroid structures of different sizes, and the results were compared and analyzed to develop empirical correlations for heat transfer and fluid flow. To minimize pressure drop, a new design involving anisotropic Gyroid structures was proposed and its effects on heat transfer and fluid flow were investigated. The findings suggested that smaller Gyroid structures enhanced heat transfer but also increased pressure drop. The study identified empirical correlations specifically suited for the Gyroid structure, providing a foundation for designing TPMS-based heat exchangers. In the case of anisotropic Gyroid cells, expanding the size in the direction of fluid flow could improve the overall performance of heat transfer and fluid flow, but it would reduce heat transfer efficiency. Reducing the size perpendicular to the flow hardly affected the overall performance, but it could simultaneously reduce the volume and the pressure drop per unit length, and increase the heat transfer amount per unit volume. It was essential to balance these factors based on the specific design needs.

During the hydrochloric acid activation process of gasification slag, the generation of silica gel affects the overall mass transfer, necessitating the clarification of flow field distribution to optimize the reactor. This study constructs a 3000L reactor model and employs computational fluid dynamics to investigate the effects of impeller structure parameters and operating conditions on flow field characteristics. Using gasification slag activation slurry as the mixing medium, the SST k-ω model and Eulerian model are applied to analyze the effects of anchor frame impeller structure parameters and conditions on the specific power input, effective mixing volume fraction, and mixing time during stirring, through the multiple reference frame method and tracer concentration method. The results show that the anchor frame impeller is suitable for mixing gasification slag slurry, with the staggered anchor frame impeller exhibiting the strongest mixing capability. The changes in impeller diameter and rotation speed significantly affect the flow field distribution, with a 1320mm impeller diameter achieving a good balance between mixing efficiency and energy consumption. Increasing the rotation speed significantly enhances volumetric power density and effective mixing volume fraction, with an optimal operating speed of 20r/min, and the optimal installation height of impellers is 216mm. Under optimal design, rapid mixing of slurry is achieved at a low power consumption of 46.96W/m³.

With the continuous development of manned spaceflight, new requirements such as high speed and long endurance of near-space vehicles continue to emerge. How to accurately simulate the influence of thermochemical non-equilibrium aerodynamic heating on thermal protection system during spacecraft reentry has been a frontier problem of aerospace reentry aerodynamics. According to the macroscopic surface catalytic reaction theory, a finite rate surface catalytic reaction model was constructed on the basis of DSMC. Considering the coupling effect of five-component gas mixture reaction and surface catalytic reaction, the influence of surface catalytic reaction on the surface pressure/heat flow distribution of reentry spacecraft was analyzed by a typical example. The surface heat flux increased by nearly 30% compared with that without considering the surface catalytic reaction. The addition of this model provides a new method for the difficult characterization of thermochemical non-equilibrium surface effects to further improve the hypersonic aerodynamic heating prediction ability of aircraft under reentry environment and support the development of thermal protection design of aircraft towards high thermal load and lightweight.

To enhance the mixing performance of the impinging stream reactor, a novel self-excited oscillating impinging jet reactor was designed, and the flow characteristics and mixing performance within this reactor were numerically simulated. The shear stress transport k-‍ω (SST) turbulence model was adopted to simulate the fluid flow pattern within the nozzle and the flow pattern distribution within the impinging stream reactor. The influences of the Reynolds number on the oscillation frequency of the jet and the mixing performance of the reactor were investigated. The research results showed that the oscillation of the jet at the nozzle outlet was controlled by the repetitive growth process of the circulation bubble within the nozzle cavity, and the flow field distribution within the reactor was affected by the deflection angles of the two jets. The oscillation frequency of the jet increased with the increase of the Reynolds number. High-frequency oscillation accelerated the shear and radial movement velocities of the two fluids. When the Reynolds number was 30000 and the mixing time was 50s, the mixing intensity at the outlet reached 0.962. The smaller the difference in the deflection angles of the two jets on both sides of the impinging stream reactor, the faster the mixing speed. The research results enrich the theoretical understanding of the fluid oscillation characteristics of the impinging stream reactor and provide theoretical references for the development of new reactors.

Circulating fluidized bed unit is widely used in energy, chemical engineering, and other fields, featuring complex structures such as risers, non-mechanical sealing chamber, cyclone, and dipleg. Under varying operating conditions, on the one hand, there are the characteristics of the coexistence of multiple fluidization regimes in the system. On the other hand, there are also non-uniform mesoscale structures such as granular clusters or bubbles in the system. Reasonable quantification of the flow/transport characteristics in the circulating fluidized bed full-loop system is essential to optimize the operating conditions and reactor structure and find the scale-up law of the full-loop system. Previous work has preliminarily proved that the mesoscience-based structural model based on the idea of multi-scale decomposition in mesoscience can take the influence of non-uniform mesoscale structures into account from the perspective of governing equations and constitutive relations, and it has the ability to accurately simulate complex heterogeneous gas-solid systems under different operation conditions. Due to some critical difficulties such as balancing the pressure drop, mesoscale structures and continuous operation, the fast and accurate simulation of the three-dimensional full-loop system has been restricted. Therefore, this work attempts to use the mesoscience-based structural model to predicted the hydrodynamics of the gas-solid flow in two different types of circulating fluidized bed full-loop systems. The simulation results show that compared with experimental data, the mesoscience-based structural model can accurately predict the hydrodynamics of full looping system such as the axial solid concentration, radial solid concentration, radial velocity, and pressure drop distribution with the high computational efficiency. The model has good applicability and robustness under a wide range of operating conditions with high computational efficiency, which is more suitable to predicting the dense gas-solid flows, and it lays a solid foundation for the optimal design and scale-up of circulating fluidized beds on an industrial scale.

This study utilized numerical simulation techniques to investigate the pyrolysis characteristics of coal with varying moisture content in an externally-heated fixed-bed reactor. The two-fluid model was combined with Zehner-Bauer-Schlunder (ZBS) thermal conductivity model, Breitbach-Barthels (B-B) radiation model, phase transformation model, parallel reaction kinetics model of coal pyrolysis and secondary pyrolysis model of tar, to simulate the heat transfer, mass transfer and pyrolysis processes of coal in a fixed bed reactor. The effects of different moisture content on the generation of coal pyrolysis products were analyzed. The results showed that the temperature approached terminal pyrolysis temperature when the volatilization rate reached maximum at different radial position. The water content affected the temperature distribution, the rate of volatilization and the yield of pyrolysis products in the reactor. With the increase of moisture content, the heating rate of coal seam slowed down, the time for coal seam center to reach the pyrolysis temperature was significantly extended, the pyrolysis process was delayed, and the ^{pyrolysis efficiency was reduced. At the same time, the overall generation rate of pyrolysis products of high-moisture coal also decreased, especially the generation of volatiles was inhibited. In addition, the increase of moisture content affected the yield of each pyrolysis product. The yield of pyrolysis gas and pyrolysis water increased, while the yield of tar and semi-coke decreased. This study obtained the law of the influence of moisture content on coal pyrolysis behavior, which provided a theoretical basis for optimizing coal pyrolysis process for further improving coal conversion efficiency.}

Flow around two cylinders is a fundamental model for a deep understanding of the flow around complex structures, which has been extensively studied by numerous researchers, yet scarcely touches upon the flow at low Reynolds numbers and high Knudsen numbers. In this study, pseudo-particle modeling was employed to simulate the drag and lift forces on the flow around two circular cylinders in both side-by-side and tandem arrangements under the conditions of 0.5≤Re≤5 and 0.1≤Kn≤1.5. The results were as follows: for side-by-side cylinders, the drag coefficient increased with the increase of the distance between them, and once the distance exceeded a critical value, the drag coefficient approached that of the flow around a single cylinder. This critical value decreased with the increase of Re and was independent of Kn; the lift coefficient initially increased and then gradually decreased to zero as the distance widened, with the peak value corresponding distance influenced by both Re and Kn. The minimum distance required for the lift coefficient to approach zero decreased with the increase of Re and Kn. For tandem cylinders, the drag coefficients of both the upstream and downstream cylinders decreased; the drag coefficient of the upstream cylinder first decreased and then increased with the widening of the distance; the drag coefficient of the downstream cylinder increased with the distance and was significantly affected by the upstream cylinder, which was consistent with the situation in the continuous flow region; Kn affected the distance at which the upstream cylinder influenced the downstream cylinder, as the smaller the Kn, the greater the interference distance.

In industrial coal-based Fischer-Tropsch (FT) synthesis, Fe-based catalysts coupled with slurry bed reactors are widely adopted. However, catalyst deactivation necessitates periodic replacement due to inherent limitations. To address this issue, a Matlab-Aspen Plus model was constructed, which integrated the dynamic processes of catalyst deactivation and replacement with steady-state simulation of FT synthesis. Taking an FT synthesis system with an annual capacity of 80×10⁴t/a as a case study, we analyzed the impacts of dynamic catalyst replacement conditions on key performance indicators—Specifically, specific syngas consumption and catalyst consumption per tonne of {\mathrm{C}}_{{\mathrm{3}}^{+}} hydrocarbons. The economically optimal catalyst replacement strategy was to replace 20% catalyst every 132h. Under this condition, the specific syngas consumption and catalyst consumption were 5449m³/t and 1.65kg/t, respectively. This model offers valuable guidance for the process design, performance prediction of FT synthesis units and practical catalyst replacement operations in industrial settings. Furthermore, it provides a practical methodology for simulating process systems involving catalyst deactivation and dynamic replacement, demonstrating broader applicability in similar industrial contexts.

To investigate the gas-liquid flow dynamics in net foam generators, this study employed an Eulerian two-fluid model coupled with the standard k-ε turbulence model to simulate the flow characteristics and mixing behavior. The gas-liquid mixing was quantitatively evaluated through two key parameters of foaming volume and foaming multiple. The impacts of critical factors such as gas flow rate, liquid flow rate, and net size on liquid phase distribution, foaming volume, and foaming multiple were investigated. The results indicated that both the average foam velocity u and the cross-sectional area S of the foam region at the outlet increased with the increase of gas flow rate, leading to enhanced foaming volume and multiple. While higher liquid flow rates increased foaming volume, it simultaneously reduced the foaming multiple, thereby affecting foam quality. The size of the net influenced the process in two ways: larger sizes hindered effective gas-liquid mixing, resulting in insufficient foaming, whereas smaller sizes increased fluid flow resistance, which was adverse to foaming efficiency. These findings suggested that optimal net selection should carefully balance foaming capacity with flow resistance to maximize overall system efficiency.

The gas-liquid interface evolution and solid dispersion in a gas-liquid-solid three-phase stirred tank for energetic material water-suspension granulation were investigated using an Eulerian multi-fluid VOF model. This study systematically examined the effects of rotational speed and liquid-phase viscosity on the free surface morphology, the distributions of the concentration and velocity of solid phase were investigated. The results showed that the model effectively captured the gas-liquid-solid three-phase stirring process involving a free surface. Within the stirring speed range of 250—350r/min, increasing stirring speed reduced the accumulation of solids in the regions near the free surface and the vessel walls and thus enhanced its uniform distribution. When the stirring speed was greater than 350r/min, the entrainment of air was intensified, leading to the existence of large bubbles inside the suspension liquid. Regarding viscosity effects, increasing viscosity could enhance turbulent shear effects within 0.001—0.1Pa·s, which effectively suppressed solid sedimentation and improved suspension homogeneity. When the liquid-phase viscosity increased from 0.1Pa·s to 0.5Pa·s, the reduced solid-phase mobility adversely affected the stability and efficiency of the stirring process. This study provided theoretical insights for the operation and optimization of the water-suspension granulation process.

In this paper, a particle-resolved simulation of the solid-liquid suspension in a turbulent stirred tank was presented, where the stirring Reynolds number Re=9800 and the solid particles were in a partially suspended state. The lattice-Boltzmann method was employed to simulate the liquid phase flow directly and the motion of the solid particles was resolved at the particle scale based on the immersed boundary method. In order to validate the results of the numerical simulations, the particle-resolved PIV experiments were performed to investigate the average velocity fields and the distribution of turbulent kinetic energy of the liquid phase under the conditions with particle volume fractions ranging from 1% to 8%. The results demonstrated a good agreement between the particle-resolved simulations and the PIV experimental data. The simulation methodology presented in this paper could offer a novel perspective for modeling the complex interactions among the solid particles and between the particles and the liquid phase flow in the solid-liquid two-phase stirred systems.

As a crucial intermediate in the petrochemical industry, the collaborative optimization of catalyst morphology and operating parameters during the ethylene oxide production process represents the core challenge in enhancing reactor performance. In response to the high cost bottleneck of traditional experimental and simulation methods in analyzing the structure-activity relationship of catalysts, this study integrated particle-resolved computational fluid dynamics (PRCFD) and graph convolutional neural networks (GCN) to construct an intelligent prediction framework for reactor multi-physics fields. A high-fidelity CFD model was established based on the COMSOL platform to investigate the flow-reaction coupling process of cylindrical, single-hole, and five-hole structure catalysts in a randomly packed system. A comprehensive research scenario covering three typical particle morphologies, random packing configurations, and four inlet gas rates was constructed. Compared with real ethylene conversion data, the effectiveness of the simulation parameter settings in COMSOL was verified. The simulations revealed that the effects of catalyst particle shape and inlet gas rate on the ethylene conversion rate and the bed pressure drop exhibited a strong non-linear relationship. Based on the valid simulation data, a graph convolutional neural network was employed to learn the mapping relationships between the geometric shapes of catalyst particles and pressure and concentration. The trained model could rapidly predict the pressure and concentration distributions under different catalysts and inlet gas rates, with a correlation coefficient R² greater than 0.9. This study provided a new paradigm which combines physical interpretability and computational efficiency for the intelligent design of chemical reactors.

The ternary system of tetrahydrofuran (THF), methanol (MeOH), and ethanol (EtOH), characterized by one distillation boundary and two azeotropic points, presents significant challenges for conventional distillation separation. To overcome this limitation, the variation trend of the phase diagram and the movement direction of azeotropic point of the ternary system THF/MeOH/EtOH under different pressures were investigated. Based on thermodynamic topological theories such as distillation boundaries and residual curves, the feasibility of pressure-swing distillation separation of ternary azeotropic mixtures THF/MeOH/EtOH was analyzed. Furthermore, based on thermodynamic topological theories such as residual curves, component equilibrium lines, and distillation boundaries, process conceptual design was carried out for two pressure-swing distillation separation sequences. Through COM technology integration between Aspen Plus and Matlab, a multi-objective particle swarm optimization (MOPSO) algorithm was implemented to optimize both sequences, with economic and safety metrics serving as objective functions. The results indicated that thermodynamic topological theory analysis could quickly realize the conceptual design of pressure-swing distillation separation of ternary azeotropic mixtures. Compared to the EtOH-THF-MeOH separation sequence, the EtOH-MeOH-THF separation sequence exhibited strong advantages in terms of economic and safety performance.

The Agrawal divided-wall column (ADWC) is a modified configuration of the divided-wall column (DWC) that offers economic benefits in terms of capital and operational costs for multi-component separation. This study aims to investigate the dynamic control structure of the ADWC for MEPB systems (methanol, ethanol, n-propanol, and n-butanol) separation. Initially, the steady-state configuration of the ADWC is optimized and analyzed. Composition control structure (CC) can handle feed disturbances of ±20%, while it may not be practical for industrial applications. Temperature control structure (TC) is more commonly used in industry but has a weaker control efficiency of ±10%. Composition-temperature control structure (CTC) is the most promising control structure due to its lower capital cost and adjustment time than CC, as well as its superior control efficiency compared to TC.

Integrating ET, OT, and IT data to build a unique and trustworthy data source for petrochemical enterprises is the data foundation and key to data driven applications for digital transformation. However, ET data has characteristics such as large volume, complex relationships, frequent changes, and unstructured nature. How to achieve data integration and exchange among project stakeholders based on a unified information model faces many challenges. To this end, international organizations such as DEXPI and NAMUR are promoting standardization of asset lifecycle data management in the process industry. This paper firstly outlines the characteristics and management challenges of engineering data, introduces the concept of information modeling, and analyzes the characteristics of process system engineering modeling in terms of functionality, structure, and behavior. It focuses on the four aspects and three types of data structure involved in the integration and exchange of process asset lifecycle data, comprehensively analyzes the relevant standards for data exchange between homogeneous and heterogeneous systems, and summarizes the problems and challenges in existing standardization work. Finally, prospects are made from four aspects, including strengthening the standardization of asset lifecycle data management, developing engineering software, engineering data governance and digital capability building for operation platforms, and asset lifecycle digital twin applications.

A supply chain network of municipal solid waste (MSW) based on waste classification is constructed from the perspective of waste energy and transportation. The selection of treatment methods for MSW and the choice of transportation routes from the waste collection point to the transfer station, as well as from the transfer station to the treatment facility, are comprehensively analyzed. On this basis, a corresponding multi-objective optimization model of the MSW supply chain network is established with the optimization objectives of minimum total cost and minimum total carbon emission, and the ε-constraint method is used to solve it. Then, an example of the supply chain network of MSW in Nanning, Guangxi is taken to illustrate the implementation and effectiveness of this method. The results show that the Pareto frontier of total cost and total carbon emissions of the MSW supply chain network can be obtained using the method proposed in this paper, and the optimal supply chain network and waste energy utilization approaches under different target values can be obtained. When the weight factors are both 0.5, the total cost and total carbon emission of the system are -22034USD/d and 533830kg/d, respectively. In addition, the effects of electricity price changes and waste transfer efficiency coefficients on the Pareto frontier of the MSW supply chain network are explored.

Municipal solid waste incineration (MSWI) is a critical technology for waste management and energy recovery. The process inevitably generates trace amounts of persistent organic pollutant dioxins (DXN). The generation mechanism of this pollutant is still unclear and difficult to directly detect. To gain a comprehensive insight into the full lifecycle mechanism of DXN generation, decomposition, re-generation, adsorption and emission, a full lifecycle prediction model for DXN based on multi-software coupled numerical simulation and fuzzy forest regression was proposed. Firstly, a DXN full lifecycle numerical simulation model was developed using FLIC, Aspen Plus and Aspen Adsorption; subsequently, the DXN simulation mechanism data under multiple working conditions were obtained based on the design and implementation of bi-orthogonal experiments; and finally, a DXN full lifecycle prediction model was established by using the T-S fuzzy forest regression (TSFFR) algorithm. The results showed that the model could obtain the DXN concentration of the full lifecycle, which provided an effective support for the realization of pollution reduction optimal control.

At present, the decision-making process for crude oil procurement and mixing processing plans relies heavily on manual experience or mathematical planning programming methods, which suffer from issues such as long solution times and inability to consider the overall context. To address these issues, we proposed a mixed-integer nonlinear programming model incorporated the “P model” concept, specifically tailoring for typical mixing processes and crude oil procurement needs in refining scenarios. Additionally, an iterative solution algorithm based on p-norm and interior point method was designed based on the unique characteristics of integer variables. The results showed that, under the optimization background of 10 crude oils, 54 physical properties, and 58 processing units, our method outperformed commercial solvers in terms of both speed and economic efficiency. Furthermore, our approach exhibited superior robustness across multiple scenarios.

In response to the high carbon emissions, high energy consumption, and low utilization efficiency of coal resources in traditional coal to methanol processes, two new green hydrogen efficient coupling processes (process Ⅰ and Ⅱ) were proposed, which introduced carbon dioxide hydrogenation technology and dry reforming of methane technology, respectively. Taking the traditional coal to methanol route with an annual production capacity of 3×10⁵t as a case study, the material changes and energy consumption of the green hydrogen coupling processes were systematically analyzed through theoretical analysis and Aspen simulation. A comprehensive techno-economic analysis was conducted, comparing the new processes with the traditional coal to methanol process from multiple dimensions, including energy consumption, carbon emission intensity, carbon utilization efficiency, investment costs, and production costs. The results showed that compared with the traditional coal to methanol process, the carbon element utilization rate of the new process Ⅰ and new process Ⅱ had increased from 38.74% to 84.56% and 67.60%, the coal consumption per ton of methanol had decreased from 1.42t to 0.65t, and the carbon emission intensity per unit of methanol had decreased by 62.84% and 56.42%, respectively. Through the analysis of investment and production costs, it was found that due to the influence of hydrogen production scale, the investment of new process Ⅰ was relatively high, while the total investment of new process Ⅱ was comparable to that of the traditional process. Presently, owing to the high cost of hydrogen, the unit production costs of methanol for the two new processes were 1.84 times and 1.51 times of the traditional process, respectively. However, with the implementation of increasing carbon taxes and decreasing hydrogen production costs, the economic advantages of the new processes would become increasingly apparent. Both processes significantly reduced carbon emissions while increasing methanol production capacity, offering advantages in terms of energy efficiency and economic performance, and demonstrating promising application prospects.

The optimization of industrial plant layout can significantly enhance production efficiency while reducing both production costs and operational risks. However, existing optimization methods often result in complex and disorganized layouts that are difficult to implement in practice. To address this issue, a structured and flexible plant layout optimization strategy was proposed, incorporating a flexible bay structure and obstacle-avoiding pipeline routing constraints. The approach was based on an improved obstacle-avoiding minimum spanning tree algorithm, which optimized traditional pipeline layouts by generating efficient routes that avoided plant obstacles. Additionally, employing an improved flexible bay structure for modeling allowed for a more adaptable and orderly plant layout, simultaneously enhancing space utilization while ensuring the simplicity and operational ease of the plant arrangement. Furthermore, the economic impact of pipeline design, land use, and safety risks was integrated into the objective function to guarantee an economically viable, safe, and practical layout solution. Experimental results demonstrated that this method delivered plant layouts with significant advantages in terms of economic efficiency, safety, and structural orderliness. This approach offered an effective reference for enterprises aiming to achieve more efficient plant management and production, showcasing strong practicality and potential for wider application.

Polyarylester (PAR) is a high-performance specialty engineering plastic widely used in various industries. The mainstream synthesis method for PAR is interfacial polycondensation, favored for its mild reaction conditions, mature process and superior product performance. However, PAR produced by this method encounters significant post-processing challenges, especially during washing, which involves high water consumption and product losses. Therefore, optimizing the washing process to minimize water usage and product waste is critical. This study employed process simulation software to develop a PAR washing model. As the washing process involved two phases and multiple components, the built-in modules were inadequate for simulating the complex phase separation. To address this, a custom Fortran-based module was developed and integrated into the washing separation system, enabling phase separation calculations based on rigorous phase equilibrium mechanisms and facilitating the analysis of different washing methods' impacts on water consumption. This study proposed multi-stage crossflow and countercurrent washing methods, expanding on the traditional single-stage washing process. Simulation results identified the three-stage countercurrent washing method as the most cost-effective approach, reducing water consumption by 78.05% compared to the two-stage crossflow washing process.

One major challenge of hydrogen network integration in refinery is the optimization of hydrogen network with complex process flows. To address this issue, a collaborative optimization method based on evolutionary response surface method is proposed to simultaneously optimize the methanol production process and the hydrogen network. This method establishes mechanistic models for the reactor, flash tank, and distillation column in the CO₂ hydrogenation to methanol process. Based on these, this method constructs corresponding evolutionary surrogate models, which are validated and refined using mechanistic model results. An efficient optimization framework is proposed to enhance model accuracy and algorithmic iteration efficiency. This method is applied to the hydrogen network integration optimization of a refinery, considering factors such as methanol yield, equipment manufacturing costs, and hydrogen consumption. Results demonstrate that the method effectively increases methanol production while optimizing hydrogen network within the refinery, reducing equipment manufacturing and hydrogen consumption costs, and achieving an annual economic benefit of more than 7 million CNY. The method is computationally efficient and significantly improves the accuracy of optimization results compared to traditional approaches, providing an effective solution for the optimization of refinery hydrogen network with methanol production processes.

Supercritical/dense-phase CO₂ pipeline transportation is the safest and most economical way to realize long-distance and large-scale carbon transportation in the world. The hydraulic-thermal calculation model for supercritical/dense-phase CO₂ can provide theoretical support for the selection of CO₂ pipeline process design parameters and program selection. Aiming at the deficiencies of the existing models, this paper takes the pipeline steady-state continuity equation, equation of motion and energy conservation equation considering the elevation difference as the basis, adopts the PR equation of state to calculate the physical parameters of CO₂, combines the thermodynamic equations of enthalpy, entropy, temperature, pressure, and density, and deduces and establishes a one-dimensional non-isothermal supercritical/dense-phase CO₂ pipeline steady-state transport hydraulic-thermal calculation model by putting forward reasonable assumptions, and comprehensively associating the equations of motion and energy equations. The model is verified by experiments and OLGA software in various aspects with high accuracy. It is simple and easy to use, which avoids the complicated and time-consuming numerical solution process and greatly reduces the computational difficulty. It can provide theoretical support and technical support for the localization of the simulation software for supercritical/dense-phase CO₂ pipeline transport process.

Through thermodynamic, combustion dynamics, and gas dynamics analyses, a physical and mathematical model is refined. The CFD method is employed to simulate the hydrogen/air deflagration process in hydrogen refueling stations under three constraint configurations. Validation against previous experimental results confirm the effectiveness of the mathematical model. This study investigates the impact of hydrogen leakage explosion accidents in hydrogen refueling stations with different constraint types. The results indicate that the flame shape is significantly influenced by the constraints. The development of the overpressure field can be divided into an overpressure accumulation phase and a release phase. A mechanism for the formation of combined deflagration pressure waves is proposed: the flame burns layer by layer, with the energy produced accumulating layer by layer. The superimposed energy is released outward in the form of overpressure shock waves, which under the induction of the top constraint, form a mesh-like structure of combined overpressure shock waves. Risk assessment based on overpressure criteria demonstrates that detonation intensity peaks under full constraint, with overpressure shock waves exceeding 200kPa. This hazardous condition poses fatal risks to personnel within 0—8.29 m.

The compound distillation sequence (CDS) combines simple distillation columns with dividing-wall columns into an integrated system for multi-component mixture separation. This integration not only reduces capital investment costs and operational energy consumption, but also facilitates industrial implementation. However, the inherent nonlinear characteristics of dividing-wall columns and strong coupling effects from material flow connections between distillation units pose significant challenges for stable operation and effective control of CDS. To overcome these challenges, three temperature inferential control schemes were developed, including a temperature control scheme, a temperature difference control scheme, and a double temperature difference control scheme. Closed-loop simulation demonstrated that all three schemes could ensure the stable operation and effective control of the CDS. Furthermore, as the control scheme evolved from the simpler temperature control to the more complex double temperature difference control, although the complexity of control scheme increased, the steady-state deviations in product purity were significantly reduced. The implementation of these control schemes addressed the operational and control challenges of the CDS, providing a solid foundation for their industrial application.

Xylene isomer are crucial chemical intermediates, but traditional production methods and yield increasing strategies result in significant differences in isomer content. The content of para-xylene (PX) can be changed from the thermodynamic equilibrium (about 23.8%) to more than 90%, which poses a great challenge for low-carbon and efficient separation strategies. Addressing variations in isomer content from different process sources, this paper focuses on the separation efficiency, energy consumption, and techno-economic feasibility of three separation strategies such as simulated moving bed (SMB) adsorption, cryogenic crystallization and reactive distillation. As the PX content in xylene isomers increases to over 90% from thermodynamic equilibrium compositions, the energy consumption, equipment costs, and utility requirements of the SMB adsorption process significantly increase, leading to decreased economic viability. Conversely, reactive distillation and cryogenic crystallization processes exhibit opposite trends. SMB adsorption demonstrates significant techno-economic feasibility when PX content is low, whereas reactive distillation achieves higher separation efficiency at higher PX contents. Cryogenic crystallization shows lower energy consumption and superior techno-economic performance when PX content ranges among 45.49%—72.57%.

In recent years, the frequent occurrence of safety accidents in VOCs treatment projects is due to the lack of targeted safety theory and method guidance, and the safety evaluation of related industries need to build. The industrial coating industry is one of the main emission sources of industrial VOCs. Therefore, according to the characteristics of VOCs treatment and emission in industrial coating industry, a VOCs treatment engineering safety evaluation system based on process simulation was proposed, including safety check list, HAZOP/risk matrix/LOPA analysis, shock wave overpressure calculation, and post-leakage analysis (Gaussian model +MATLAB). On the one hand, the introduction of Aspen process simulation reduced the evaluator's dependence on experience through "actual + simulation" HAZOP analysis, and also provided simulated data of risk boundary values and standard specified values to support quantitative analysis. On the other hand, the instantaneous response value of Aspen dynamic simulation parameter fluctuation to key indicators and the dynamic change rate over time could quantify the risk deviation and determine the reasonable range of system operation, which provided a basis for formulating emergency response plan and accident prevention measures. The proposed safety evaluation system was used to evaluate and apply typical process cases, and the results showed that the safety evaluation system could effectively identify potential safety hazards in VOCs treatment projects of industrial coating industry, and provide improvement suggestions to reduce the need for the number of evaluation personnel and experience. With the help of simulation results, the influence range of explosion and leakage accidents could be visually presented, which provided a basis for setting alarm value and determining reasonable operating range.

The acid gas quenching tower in high-sulfur natural gas desulfurization units exhibits significant vulnerability to condensation-induced erosion, potentially leading to hazardous medium leakage and consequent unplanned shutdowns or safety incidents. To reveal the multi-field interactive damage processes of heat transfer, condensation, mass transfer, and corrosion within it, and guide early prediction of erosion damage and risk prevention in industrial facilities, this paper employs the Euler-Lagrange method, coupling an inhomogeneous interphase drag model with an acid gas condensation phase change model and a dissolution chemical equilibrium equation. A coupled prediction model for erosion and corrosion rates is embedded to establish a simulation method for multi-field interactive damage caused by acid gas condensation erosion. This method simulates the effects of various parameters such as droplet phase mass fraction, gas-phase condensation rate, droplet size, as well as concentrations of carbon dioxide and hydrogen sulfide. The results indicate that this simulation method effectively reveals the multi-field interactive damage process; compared with industrial measurement data, the prediction error is within 10%. Parametric analysis reveals the following hierarchical influence on erosion rate: droplet phase mass fractio>droplet size>gas-phase condensation rate >CO₂ concentration >H₂S concentration. Factors affecting thinning locations due to erosion and their weights are: droplet phase mass fraction>droplet size>gas-phase condensation rate >CO₂ concentration >H₂S concentration.

In the early stages of hazardous gas leakage accidents, improper handling can lead to secondary incidents such as combustion and explosion. Thus it is crucial to develop a gas tracing method for rapid localization of leak source. Gas tracing and localization, an inverse problem of gas diffusion, remains challenging in scientific research and engineering applications. Artificial neural networks, integrated with tracing and localization schemes offer a promising solution for this inverse problem, enabling rapid and accurate source localization. The dynamic source localization datasets are extracted from computational fluid dynamics simulation results. A long short-term memory neural network model for real-time prediction of leakage source locations based on sensor data sequence is established and optimized. Results show that the artificial neural network-based localization model can accurately predict the leakage source, with predicted points within 20m of the actual source location, achieving an accuracy of 97.49%. After a set of sequential concentration data is input, the preliminary location of the leakage source can be predicted within 0.04737s, which is significantly faster than traditional source localization methods.

Methanol-to-olefins (MTO) is a successful non-petroleum route for producing light olefins such as ethylene and propylene, and process conditions significantly influence product distribution. To accurately simulate and optimize the product distribution, a molecular-level model for the MTO reaction-regeneration process based on industrial-scale data was developed. The model included sub-models for MTO kinetics, reactor, coke-burning kinetics, and regenerator. The reaction pathway was determined by the hydrocarbon pool mechanism, and the reaction network was developed. Then, the kinetics model was coupled with a fluidized bed reactor model to describe the conversion behavior of various components in the reactor, and the regenerator model was constructed using the same approach. The developed reactor model and regenerator model were hence coupled to simulate the reaction-regeneration process. The model was validated using industrial production data through single-point calibration and long-term prediction. The results demonstrated that the model could accurately predict the distribution and yield of key products such as ethylene, propylene, and C₄. On this basis, a quantitative relationship between process conditions and key product distribution was obtained by sensitivity analysis and optimization algorithm. The optimization results demonstrated that the developed MTO model had good prediction and optimization capabilities. This work provided an accurate computational tool for process simulation and optimization of industrial MTO plants.

Battery thermal management is critical to ensure the thermal safety of lithium-ion batteries. Although traditional finite element analysis methods have been widely applied in battery thermal management research, they have limitations, such as low computational efficiency and complex parameter settings. This paper presented a machine learning model that combines feature engineering with finite element analysis results. By employing an orthogonal design method, the required finite element simulation data volume was effectively reduced. The support vector regression (SVR) model was used to accurately predict the temperature characteristics of a hybrid battery pack. The non-dominated sorting genetic algorithm Ⅱ (NSGA-Ⅱ) was applied to systematically analyze the synergistic optimization relationship between battery structural parameters and cooling strategies, proposing an optimal solution that balanced heat dissipation performance and energy consumption efficiency. Compared with traditional methods, the proposed approach significantly enhanced computational efficiency while maintaining prediction accuracy, providing a novel approach for the intelligent design of battery thermal management systems. The “feature extraction-machine learning modeling-multi-objective optimization” framework constructed in this study not only accurately predicteed battery temperature characteristics but also provided decision support for optimizing thermal management solutions in various application scenarios. This method has significant engineering application value in fields such as electric vehicles and energy storage systems, contributing to the improvement of battery system safety and energy efficiency.

With rapid development of artificial intelligence (AI) technology and the reduction in application costs, AI has penetrated many traditional industries, driving changes in the industrial landscape. The chemical industry, an important part of the global economy, has long faced challenges such as high energy consumption and environmental pollution. It is confronted with a series of "neck-breaking" problems, including complex process optimization and system scheduling, low catalyst R&D efficiency, difficulty in fault diagnosis and inaccurate product prediction. Artificial neural network (ANN), with its powerful nonlinear mapping, self-organization and adaptive learning and big data-driven characteristics, has been gradually integrated into basic chemical research and production processes, providing new opportunities to solve these problems. This paper reviewed the current status of ANN applications in the chemical industry, including catalyst design and selection, reaction condition optimization, chemical product analysis and prediction, process system optimization, and environmental monitoring and management. It discussed breakthrough paths and specific cases of ANN-driven chemical industries, analyzed the shortcomings and challenges of existing ANN applications in the chemical industry, and finally proposed directions for future applications in the chemical industry.

Artificial intelligence has driven rapid advancements in the design of complex chemical engineering processes with a new data-driven model, serving as a powerful force behind transformative developments in the chemical industry and holding significant implications for the evolution of research paradigms, new technologies, and industrial processes. This paper focuses on the progress of intelligent algorithms in the design of complex chemical processes, and systematically outlines the applications in molecular structure and property prediction, recommendation of reaction and separation pathways, and intelligent optimization of process parameters. The paper also summarizes the performance of intelligent algorithms in big datasets collection and cleaning, pattern recognition, and trend prediction. It provides an in-depth analysis of the challenges faced by intelligent algorithms in chemical process design, including issues related to the lack of professional feature data quality and the insufficient interpretability of models. The paper proposes the practical need for the development of comprehensive multi-level chemical big datasets, the continuous exploration of the relationship between intelligent algorithm structures and chemical process node information, and further improvements in the interpretability and structural stability of intelligent models. These efforts aim to construct a large-scale model framework for intelligent chemical process design, from molecular structure recognition to process design, and ultimately realize intelligent design in the chemical industry.

Construction of smart factory is an important channel to improve the level of intelligent manufacturing in manufacturing industry, and it is also the main direction for realizing the strategy of “Made in China 2025”. After nearly ten years of development, driven by the national industrial policy and the enterprises' own development needs, a large number cases of intelligent manufacturing pilot demonstrations, new models of intelligent manufacturing and digital workshops have been built in various regions. The manufacturing industry has made significant improvements in automation, informatization enhancement and intelligent leadership. However, as intelligent manufacturing enters the deep water zone, problems such as data islands, integration difficulties, personal demand, and value creation gradually become prominent, which has become the bottleneck restricting the construction of a smart factory. Thus, there is still a long way to build a “real” smart factory widely recognized by enterprises. The continuous innovation and development of industrial internet, artificial intelligence, automated control, industrial software and other technologies has provided more possibilities for the construction of smart factory. This paper proposed a new architecture for process industry smart factory construction: 1+2+N (1 factory operating system +2 automations +N industrial APPs), in order to help enterprises solve the bottleneck problems encountered in the construction of smart factories, guide enterprises to make better use of new technology and business integration innovation, and achieve a new paradigm of enterprise production independent operation and enterprise management excellent operation of smart factory. At the same time, this paper verified this new architecture in a chemical enterprise smart factory construction project. Looking forward to the future, with the development of technology and the continuous maturity of products, we hope that this architecture can be more widely recognized and adopted by the industry, and become a new standard of the construction architecture for smart factory in process industry.

'Molecular refining' is a technology that describes the reaction process at the molecular level and achieves an accurate description of the specific composition and quality of oil products. It is one of the digital modeling methods for petrochemical production processes. Based on the diesel hydrofining unit of a petrochemical enterprise in China, this paper constructs a diesel molecular reconstruction model and a diesel hydrofining reaction kinetic model to meet the needs of intelligent factory construction. Firstly, the physical property library of diesel molecules including 1352 molecules was established, and the molecular type-homologous (MTHS) matrix was used to construct the diesel molecular reconstruction model. The maximum relative error between the simulated value and the real value was 4.83%, which proved that the model was reliable. Secondly, combined with the characteristics of the device, a reaction network containing 246 reactions and 282 molecules was constructed by screening and determining the reaction molecules and establishing the reaction rules of diesel hydrofining. Based on the influence degree of the alkyl side chain on the chemical reaction rate constant, the reaction rate constant influence factor was set up, and the reaction rate constant correlation model was established. The 492 reaction parameters were reduced to 116, which greatly reduced the number of variables, and the kinetic correlation model of diesel hydrofining reaction at the molecular level was constructed. The results showed that under 398℃, 9MPa and space velocity of 1h^-1, the relative error between the calculated value and the real value of the sulfur content of the product after hydrodesulfurization was less than 10%, and the model had good stability, which proved that the model was reliable. Based on the established model, the prediction of product composition under different operating conditions could be realized. This study can provide research ideas for the construction of intelligent factory models in petrochemical enterprises.

Ionic liquids (ILs) possess the merits of high stability, excellent solubility, designability, facile recovery, and environmental friendliness. Particularly, they exhibit remarkable potential in carbon capture on account of their high solubility for CO₂. However, due to the wide variety and expensive price of ILs, it is time-consuming and laborious to rely on experimental research, and it is crucial to construct a thermodynamic prediction model for ILs systems. Since the UNIFAC predictive model was of great value for ILs selection and convenient for subsequent selection of gas separation tasks and process design and optimization, the UNIFAC model was proposed to predict the activity coefficient of the ILs-CO₂ system. In this work, the experimental data of CO₂ absorption by ILs were collected, the activity coefficient was calculated by combining the phase equilibrium, and the activity coefficient database of ILs-CO₂ system was established. The parameters of UNIFAC model (R_k and Q_k ) were calculated by COSMO method and van der Waals rule, respectively. Based on the experimental values, the UNIFAC interaction parameters were fitted. The prediction performance of the UNIFAC model established by the two methods was compared by the average relative error (AARD). The results showed that the UNIFAC model established by COSMO method (AARD=7.68%) could reduce the prediction error of activity coefficient of ILs-CO₂ system by 4.89 percentage points compared with van der Waals method (AARD=12.57%). On this basis, the activity coefficient UNIFAC model of ILs-CO₂ system was established, and a database of interaction parameters for nearly 100 pairs of groups was obtained. In addition, based on the group contribution characteristics of the UNIFAC model, all group fragments contained in the interaction parameter database could be extended to new ionic liquid systems, so that the established UNIFAC model could be used to predict the activity coefficients of new kinds of ILs-CO₂ systems. Due to the group contribution feature of the UNIFAC model, the UNIFAC model developed in this work could predict the activity coefficients of ILs-CO₂ binary system composed of the groups included in the database, providing a solid basis for the molecular design of the subsequent ILs method for gas absorption.

With the deepening research on gas hydrates in carbon capture and oil-gas transportation, the study to their phase equilibrium is significantly important. The gas forming hydrates can be either pure components or mixtures. Hydrate formation involves three- or four-phase equilibria, including I-H-V (ice-hydrate-vapor), L_w-H-V (liquid water-hydrate-vapor), L_w-H-L_HC (liquid water-hydrate-liquid hydrocarbon), and L_w-H-V-L_HC (liquid water-hydrate-vapor-liquid hydrocarbon). The phase diagrams of hydrates formed by different gas compositions exhibit notable differences, particularly in the number of four-phase points. This paper focuses on methane, nitrogen, carbon dioxide, propane and natural gas mixture systems, discussing numerical calculation methods for hydrate phase diagrams containing one or two four-phase points and four-phase lines. Numerical simulations are conducted using the independently developed SimTech Simulator, and the results are compared with those from the commercial software Pro/Ⅱ, validating the reliability of the Simulator.