Organic waste liquids generated in industrial processes contain organic solvents and high ion concentrations, representing a substantial source of osmotic energy. Ionic osmotic energy conversion is a promising technology for recovering this energy, but its performance is constrained by the trade-off between ion selectivity and permeability. The effects of organic solvents on this balance remain unclear. In this study, aqueous solution was employed as the reference system, and the area-specific resistance, the open-circuit voltage, the short-circuit current, and the power density were measured in methanol, ethanol, and isopropanol systems. The results showed that area-specific resistance increased with solvent molecular size, whereas the power density was highest in methanol and lowest in isopropanol. Notably, the methanol system achieved a power density of 0.71 W·m−2, surpassing the 0.50 W·m−2 obtained in the aqueous system. Molecular dynamics simulations revealed that permeability decreased as solvent molecular size increased. Compared with water, methanol formed larger ion solvation structures and induced stronger electrostatic exclusion, thereby increasing the energy barrier for anions entering the nanochannels. Consequently, the cation transference number increased from 0.533 to 0.583, indicating enhanced ion selectivity. Although ion permeability was reduced, the improvement in selectivity predominated in the methanol system, resulting in superior energy conversion performance. These findings indicate that in water and methanol systems, selectivity plays a dominant role, whereas in ethanol and isopropanol systems, insufficient permeability becomes the primary factor limiting energy conversion despite stronger anion rejection. This work provides theoretical guidance for optimizing osmotic energy recovery from organic liquids.

The fuel cell/battery hybrid powertrain offers a promising solution for fuel cell vehicles by integrating the high energy density of hydrogen fuel cells with the high-power density of batteries. However, real-time energy management of such a multi-source system faces challenges in simultaneously achieving economic efficiency, durability, and adaptability. To address this, this study proposes an online energy management strategy called MOCR-SAC. It incorporates multi-objective constraint rules (including hydrogen consumption, fuel cell degradation, battery degradation, fuel cell optimal efficiency deviation, and battery optimal state of charge deviation) within a Soft Actor-Critic reinforcement learning framework, enabling adaptive and intelligent power allocation. Evaluated on a 12-m fuel cell bus under standard Chinese driving cycles, MOCR-SAC reduces hydrogen consumption by at least 4.28 % and operating costs by 7.32 % compared to conventional SAC (without constraints or using single rules). It also outperforms other online reinforcement learning methods in component degradation, cost, battery SOC regulation, and hydrogen economy. Compared to the global optimum obtained by dynamic programming, its operating cost deviation remains within 4.50 %, while hydrogen consumption is 5.63 % lower. Under both deterministic and uncertain driving cycles, the total operating cost deviates by less than 10 %, demonstrating strong robustness and adaptability. The proposed strategy can be pre-trained offline and deployed online with minimal computational overhead, meeting the real-time requirements of vehicle energy management. In summary, MOCR-SAC significantly enhances the performance, efficiency, and durability of fuel cell hybrid powertrains, offering a practical and scalable solution for sustainable transportation.

Porous transport layer (PTL) in proton exchange membrane water electrolyzer suffers spatial-temporal mismatches in multi-physical transport, which cause high overpotentials and local hotspots. However, current PTL pore-scale models primarily focus on two-phase flow, lacking integrated methods for electro-thermal-mass coupled analysis. In this study, by integrating computational fluid dynamics (CFD) and Python codes, an electro-thermal-mass coupled model for predicting the pore-scale species evolution in the PTL is proposed. Based on this model, the electric conduction, heat transfer, and mass transport mechanisms within the PTL are revealed. The results show that by adjusting the local contact resistance at the interface between the catalyst layer and PTL, directed charge transfer from the bipolar plate to the catalyst layer can be achieved. Meanwhile, local hot spots in the PTL are dominated by oxygen saturation. Enhanced convection preferentially reduces the temperature in low-oxygen regions and has a limited cooling effect on oxygen-rich zones, serving mainly to accelerate oxygen removal and mitigate local overheating. Additionally, due to the larger bubble diameter, the oxygen breakthrough time through the PTL is approximately three times longer than water vapor. This study can provide a guidance method for the structural optimization of PTL and operational adjustment of the electrolysis process.

The high computational cost associated with the highly coupled, nonlinear multi-physics processes and expensive experimental cost of Proton Exchange Membrane Fuel Cells (PEMFCs) pose a significant challenge to membrane electrode assembly design and optimization. To achieve efficient water management and low-Pt catalyst layer (CL) design, this study proposes an Explainable AI-based optimization framework that integrates a data-driven interpretable deep learning surrogate model based on Self-Attention Convolutional Neural Network with a multi-objective optimization algorithm, which enables rapid performance prediction and quantitative analysis of the significance of both CL structural features and operating conditions. The results indicate that CL parameters contribute more significantly to performance and oxygen distribution based on AI prediction model feature importance analysis. Furthermore, temperature and cathode relative humidity exhibit a more pronounced contribution on membrane hydration, necessitating electrode optimization designs with adaptability across operating temperature ranges. Moreover, the dual-objective optimization results indicate that, as temperature increases, the optimal solution sets exhibit reduced membrane water content, although the compromise optimal solutions achieve improved performance, they require higher Pt-loading. Considering the Pt-loading cost constraints, the triple-objective optimization results show that the two compromise optimal solutions improve output power by 8.43% and 0.47% under low Pt condition, respectively. The proposed prediction and optimization framework offers a feasible and promising design strategy for the digital twin system engineering development of high-performance, low-cost PEMFC membrane electrodes.

The direct internally reformed solid oxide fuel cell (DIR-SOFC) has the advantages of wide fuel adaptability and high power generation efficiency. Rapid performance prediction and optimization methods play a very important role in reducing performance improvement of SOFC. In this paper, a DIR-SOFC performance prediction and optimization method based on GA-optimized BP neural network was proposed. Using multi-component fuel as a case, 2060 analysis samples were established by 3D numerical simulation, and the current density and temperature of the DIR-SOFC under different fuel components were predicted and optimized by the proposed method. The results show that this method has the advantages of strong generalization ability, high prediction accuracy and fast calculation speed. Aiming for higher current density and lower maximum temperature gradient, the method is applied to achieve optimization combination of fuel components (H2O, NH3, H2, CO, CH4). At an operating voltage of 0.7 V, the optimal fuel ratio is determined as 0.6% H2O, 25.6% H2, 29% CO, 29.4% CH4 and 15.4% NH3. The current density is 3336 A & sdot;m-2 and the maximum temperature gradient is 169618 K & sdot;m-1. In addition, the weight analysis method was used to study the influence degree of fuel composition on power generation performance. It is found that increasing the volume fraction of H2O and NH3 in the fuel reduces the power generation performance, while increasing the volume fraction of H2, CO and CH4 in the fuel improves the power generation performance. Increasing the volume fraction of H2O decreases the maximum temperature gradient while other gases increase it. These conclusions are consistent with the results obtained by the prediction method, which proves the consistency of the proposed method with the physical mechanism. This study has guiding significance for optimizing the operating conditions of DIR-SOFC.

Solid-liquid transition is characterized by high thermal storage density and near-isothermal endothermic/exothermic processes. Consequently, phase change heat sinks (PCHSs) based on phase change materials hold significant potential for application in heat management of electronic devices under specialized operating conditions. However, existing research still confronts key challenges, including insufficient in-plane thermal conductivity, slow thermal response rates, and inadequate investigation into the effects of encapsulation materials. In this study, a n-octacosane/pi-shaped graphene foam (n-octacosane/pi-GF) phase change composite (PCC) heat sink is proposed. This design enables rapid in-plane heat spreading while facilitating efficient heat transfer and storage along the thickness direction, thereby synergistically optimizing heat spreading, transfer, and storage performance. Comparative analyses were conducted to evaluate the heat transfer characteristics in both in-plane and thickness directions and the thermal management performance of different PCHSs. The integration of aluminum-graphite composite and pi-GF was found to enhance the in-plane heat spreading capability and through-thickness heat transfer performance of the PCHS, mitigate in-plane heat accumulation, accelerate the melting of n-octacosane along the thickness direction, and reduce the heat storage duration of n-octacosane. With increasing heat flux density, the safe operating time initially decreases sharply before entering a phase of gradual decline. Relative to pure n-octacosane, the n-octacosane/pi-GF composite exhibits a 13-fold enhancement in the through-thickness thermal conductivity and a 110-fold improvement in the in-plane thermal conductivity, with the latter reaching 18.8 W & centerdot;m-1 & centerdot;K-1. Furthermore, the thermal management performance of the n-octacosane/pi-GF PCC encapsulated with aluminum-graphite composite is 2.9 times higher than that of the aluminum alloy encapsulated pure n-octacosane PCHS. This work offers meaningful insights for the design of PCHSs with rapid response and high in-plane thermal conductivity, as well as for the rational selection of encapsulation materials.

The gas flow distribution in solid oxide fuel cell (SOFC) stacks has a significant impact on the electro-thermal-mass transport performance and safety. In this study, an external manifold structure incorporating a spider-web-inspired perforated sheet is designed for a planar SOFC stack. Numerical simulations are performed to investigate the effects of key geometric parameters (the angle of the sector-shaped holes (θ1), manifold depth (H), and posterior diameter of the flared inlet tube (Doutlet)) on flow uniformity and pressure drop. Compared with a configuration without a perforated sheet, the non-uniformity index (NUI) decreases from 2.26% to 0.16% when θ1 =2°. To obtain the optimal geometric configuration, dimensionless treatment is applied to θ1, Doutlet, and H. The Taguchi method is used to derive the optimal parameter combination. The NUI of the SOFC stack after optimization is further reduced to 0.0658%, and the stack pressure drop decreases from 561 Pa to 412 Pa. Compared with a conventional round hole perforated sheet, the stack employing the spider-web perforated sheet achieves a more uniform gas flow rate to each cell, resulting in further enhancement in overall flow distribution.

Manifold microchannels exhibit excellent performance in electronic device thermal management. By incorporating topology optimization and designing enhanced heat transfer ribs within the channels, the cooling performance can be further improved. However, existing research on manifold microchannels topology optimization mainly focuses on individual channel based on the assumption of uniform flow distribution. Additionally, few studies have conducted topology optimization on the heated surface of manifold microchannels. This paper employs a partitioned strategy to conduct topology optimization on the heated surface of manifold microchannel heat sink considering the uneven flow distribution characteristics within individual channels. Based on a typical Z-type manifold microchannel, two topology-optimized rib structures were developed by employing average temperature minimization as the optimization objective under different pressure drop constraints. Subsequently, a comparison was performed between the topology-optimized structures and existing inline pin fin structures with geometrically optimized configuration. The study found that both topology-optimized manifold microchannels configurations outperformed the pin fins manifold microchannels in terms of overall performance, average temperature of the heated surface, temperature uniformity of the heated surface, thermal resistance, and pumping power consumption. Specifically, the temperature uniformity of the heated surface improved by an average of 12 % and 11.2 %, total thermal resistance decreased by 7.9 % and 6.7 % on average, and when the average temperature of the heated surface was maintained at 65 degrees C, the required pumping power was reduced by 12.4 % and 5.2 %, respectively. The performance evaluation criterion values of both topology-optimized structures are higher than the corresponding inline pin fin structures, reaching up to 1.52, demonstrating superior overall performance. This study provides a new perspective for the structural design of manifold microchannels flow passages.

Addressing renewable energy challenges through hydrogen requires the advancement of proton exchange membrane fuel cells (PEMFCs) for efficient hydrogen utilization and large-scale deployment. As the core of a PEMFC, the catalyst layer (CL) faces significant performance degradation challenges, particularly under low platinum (Pt) loading conditions. Therefore, advancing low-Pt CL designs through multi-scale to cross-scale optimization is crucial for improving both output performance and durability. This review systematically examines CL development from the perspective of multiscale design: molecular-scale manipulation of triple-phase interfaces, pore-scale optimization of water-gas transport in agglomerate structures, and macro-scale performance evaluation coupled with multi-physics fields. The chemical and mechanical degradation mechanisms of CL under dynamic operating and freeze-thaw cycles conditions are additionally reviewed. Importantly, this review highlights the applications of artificial intelligence for cross-scale bridging and synergistic reinforcement in durable low-Pt CL design, and prospect hybrid physics-informed and data-driven framework for degradation prediction. The aim of this review is to provide a novel perspective for the advancement of low-Pt PEMFC CL design, which can also offer guidance for other electrochemical renewable energy conversion technologies.

Alkaline water electrolysis (AWE) is a mature and economically viable technology for green hydrogen production, playing a critical role in integrating renewable energy and transitioning to sustainable systems. However, its widespread application faces multi-level challenges. At the component level, these include an incomplete understanding of electrocatalytic mechanisms, limited durability of diaphragm materials, and complex multi-physics coupling within cells. At the equipment level, obstacles involve non-uniform heat and mass transfer, inefficient gas-liquid separation under variable loads, inadequate design tools for large-scale stacks. System-level challenges comprise dynamic response limitations, management with intermittent power, and operational maintenance complexities, hindering stable integration renewables like wind and solar. Digital-intelligent technologies provide a transformative approach to address challenges through deep integration across the AWE hydrogen production chain. This review systematically analyzes recent advancements in these technologies. For core components, data-driven methods, particularly machine learning, revolutionized materials development. AI-powered prediction models and automated synthesis platforms accelerate screening and design of high-performance, non-precious electrocatalysts for both hydrogen and oxygen evolution reactions. These techniques also enable performance prediction and screening of advanced diaphragm materials, enhancing ionic conductivity and durability capabilities. Deep learning-based surrogate models facilitate rapid optimization of field structures and operational parameters, overcoming computational bottlenecks of traditional simulations. At equipment level, the focus shifts to managing intricate thermal-mass transport phenomena and enabling rapid design. Innovative modeling strategies, including physics-based model reduction and equivalent network frameworks, overcome limitations of high-fidelity 3D simulations. These methods effectively characterize critical issues in electrolysis stacks, such as localized hot spots, bubble coverage, shunt currents, and flow distribution non-uniformity, guiding the design efficient, stable large-scale stacks. For gas-liquid separators, optimization of internal components and flow fields improves separation efficiency, crucial for system safety and operational flexibility under dynamic conditions. On the system managing dynamic response characteristics is essential for renewable energy assimilation. Research has evolved experimental characterization of transient behaviors to developing sophisticated dynamic models and multi-modal control strategies. Advanced control algorithms, including Model Predictive Control, coordinate load, temperature, and pressure management, extending operational flexibility while ensuring safety constraints like hydrogen-in-oxygen concentration limits. Intelligent operation and maintenance concepts leverage digital twin technology and data analytics for real-time monitoring and predictive maintenance, enhancing system reliability and economic performance. Explorations of hydrogen industry-specific AI agents suggest potential for decentralized, autonomous optimization in large-scale operations. conclusion, digital-intelligent integration presents a comprehensive framework for overcoming AWE's limitations. enabling cross-level integration from materials to system operation, it facilitates high efficiency, longevity, flexibility, cost-competitive green hydrogen production at scale. Future efforts should prioritize understanding degradation mechanisms under extreme conditions, developing physics-data integrated cross-scale models, validating technologies megawatt-scale demonstrations, and establishing a robust "component-equipment-system" intelligent ecosystem.

Proton exchange membrane water electrolysis constitutes a key pathway for renewable energy utilization. Given the coexistence of gas and liquid in core components of electrolyzers, including catalyst layer, transport porous layer, and flow channel, revealing the cross-component gas-liquid two-phase transport process is crucial. However, existing numerical methods for gas-liquid two-phase flow struggle to balance cross-component gas-liquid interface capture, computational scale, and computational efficiency. In this study, a continuum capillary pressure model is proposed. Combined with the VOF method, the model enables phase interface capture within the representative elementary volume of porous media, significantly improving the computational efficiency of two-phase simulation in porous media. It bridges the gap for simulating the coupled gas-liquid twophase transport processes between the pore scale (catalyst layer, porous layer) and macroscale (flow channel). Based on the proposed method, the cross-component gas-liquid two-phase interface evolution process in proton exchange membrane electrolyzers is obtained. Compared to parameters such as flow field structure, water flow rate, and two-phase flow regime, oxygen saturation in the porous transport layer provides a more direct and effective characterization of mass transfer performance. Accordingly, a Transport Performance Index integrating the opening ratio and flow uniformity has been established to quantitatively evaluate the gas-liquid transport performance of flow fields. The simulation model and evaluation index developed in this study offer a new approach for the optimization of two-phase transport in proton exchange membrane electrolyzers.

In the field of thermal management for electronic devices, transient conjugate heat transfer (TCHT) poses significant challenges due to alternating heat sources, necessitating cold plate designs that achieve high heat transfer efficiency, low flow resistance, and lightweight construction. Conventional topology optimization (TO) for conjugate heat transfer (CHT) employs discrete adjoint methods optimized for steady-state scenarios, whereas TCHT applications often resort to simplified governing equations to reduce computational burdens, at the expense of accuracy. To address this, we introduce a finite-volume-based Continuous Adjoint-based Transient Topology Optimization for Conjugate Heat Transfer (CATTO-CHT) framework, which derives adjoint equations directly from the continuous Navier-Stokes and energy equations, preserving transient fidelity while enhancing efficiency via semi-analytical sensitivity analysis and checkpointing for time-dependent adjoints. Implemented on the OpenFOAM platform, CATTO-CHT integrates density-based TO with a Darcy porosity model for fluid-solid variation, alongside transient adjoint solvers that minimize time-averaged temperatures under power dissipation constraints. Applied to water-cooled plates under constant, square-wave, and sine-wave heat sources, the method yields vein-like biomimetic flow channels, with parametric analyses revealing how heating duration and oscillation periods influence branching patterns to improve flow distribution and boundary layer disruption. Compared to straight-channel baselines under equivalent mass and pump power constraints, the optimized designs enhance equivalent heat transfer coefficients by factors of 1.78, 1.65, and 1.55 for the respective heat source cases, underscoring CATTO-CHT's potential for advanced, lightweight solutions in dynamic thermal applications.

Reversible solid oxide fuel cells(RSOCs)possess both fuel cell and electrolysis cell operating modes,with their electrolysis mode capable of H2O-CO2 co-electrolysis,serving as a novel carbon conversion technology.However,research on their application in distributed energy systems remains insufficient,and their potential for active carbon reduction has yet to be fully explored.This research proposes an integrated en-ergy system coupling RSOCs with solar energy,establishes a system simulation model that includes combined cooling,heating,and power as well as active carbon emission reduction,and investigates the system's opera-tional characteristics across different seasons,fuel utilization rates,gas sources,and price fluctuations.Addi-tionally,a comprehensive evaluation index considering energy efficiency,environmental benefits,economic performance,and sustainability is proposed.Taking the energy supply of a 2 000 m2 office building as an ex-ample,the maximum daily carbon emission reduction can reach 53 kg in summer when solar radiation is abun-dant.Furthermore,parametric studies indicate that fuel utilization rates and methane proportions are positively correlated with energy efficiency and exergy efficiency,while the system's economic performance is most sen-sitive to electricity prices.The active carbon reduction energy system proposed in this study differs from pas-sive reduction systems by achieving on-site active carbon emission absorption and can flexibly couple with tra-ditional energy systems to meet the low-carbon emission requirements.

The graphene units, with the advantages of excellent thermal conductivity, good toughness, low density, and high connectivity, play a role in enhancing heat conduction. The porous characteristics of the graphene substrates have a significant impact on the heat transfer performance of the composite materials intended for energy storage and heat dissipation applications. Nevertheless, the solid-liquid phase change mechanisms within the generally used graphene substrates, such as two-dimensional (2D) ordered configuration and a three-dimensional (3D) random foam-like arrangement, remain inadequately understood. In this study, two stochastic reconstruction methods were proposed to reconstruct the realistic structure of the 2D and 3D graphene, respectively. Then, the 3D pore scale Lattice Boltzmann method (LBM) based on the total enthalpy method was developed to explore the melting characteristic in both 2D and 3D graphene structures. The numerical results indicate that the vertical thermal conductivity of the 2D structure is markedly superior to that of the 3D foam structure, especially in the case of high porosity (0.95), which is six times that of the 3D structure. For the 3D foam structure, the phase change process exhibits consistency in both horizontal and vertical orientations. The process can be categorized into three distinct regions: an initial rapid phase change region dominated by conduction, a phase change retardation region, and a slow phase change region dominated by natural convection. For the 2D ordered configuration structure, the melting behavior is quite different. When heated vertically, the melted phase change materials show a radial distribution, and the distribution of the phase interface is consistent when the porosity ranges from 0.8 to 0.9. When heated horizontally, the phase interface is perpendicular to the bottom surface. In particular, as porosity reached 0.9, the melting time required was 12.6 times longer compared to that in other porosities. This study contributes to a deeper understanding of how the structural characteristics of graphene substrates influence the internal phase transformation processes in composite materials, thereby providing valuable insights for the optimal design of composite graphite-based materials.

Metal hydride (MH) reactors suffer from low thermal conductivity, limiting hydrogen storage performance. This study employs topology optimization in a steady-state model to design fin structures for embedded multi-tube MH reactors, minimizing average temperature. Results demonstrate that topology-optimized fins outperform both genetic algorithm (GA)-optimized and straight fins. At a 0.10 fin volume ratio, the topology-optimized reactor reached 90% saturation 23.8% and 19.2% faster than straight-fin and GA-optimized reactors, respectively. Performance gains increased progressively with higher fin ratios: as the ratio rose from 0.10 to 0.25, the reaction rate advantage over straight fins expanded from 150 mg & centerdot;s(-1) to 770 mg & centerdot;s(-1). Furthermore, adjusting cooling tube-to-fin volume ratios under consistent total high thermal conductivity materials volume further enhances performance. While increasing the tube volume ratio from 5% to 10% at low total volumes (<20%), reaction rates initially decrease. However, at 20% total volume, reactors with 10% tube ratio exhibit 6.4% higher reaction rates versus initial configurations. This advantage increases to 19% at 25% total volume, confirming that optimal heat element distribution can critically boost reactor efficiency.

Developing fast and efficient predictive models for metal hydride (MH) reactors is crucial for large-scale reactor design, performance prediction, and control system application. Based on the lumped-body concept and conservation laws, this study proposes a novel 2D + 1D multi-physics network model for metal hydride (MH) reactors. This model is developed to tackle the challenges of high computational cost in traditional CFD models and the poor adaptability of traditional reduced-order models, especially when the reactors have complex heat transfer configurations. The model couples heat transfer, mass transfer and chemical reactions with axially dividing the reactor into layers (1D component) and treating them as a series of interconnected sub-reactors. The cross-sectional MH reactors structure within each layer is adaptively constructed using an image recognition method, accurately and conveniently capturing the reactor's structural features. The structural features are stored in the reactor characteristic matrix. Each element represents a 3D cube interconnected via thermal and flow resistance networks. The proposed model is validated under varying operation conditions for several different MH reactors. For all test cases, the maximum absolute error in hydrogen saturation is 4 × 10−2, and the relative error in the average bed temperature is 2.2%. Compared to 3D CFD simulations the proposed model improves the computational efficiency by about 15–40 times with maintaining fidelity under the test cases. The proposed model framework supports expansion across multiple application scenarios, such as thermal integration systems coupled with fuel cells, and thus possesses certain practical engineering value.

Accurate prediction of the metal hydride (MH) bed's effective thermal conductivity (ETC) is key for precise MH reactor design. Existing ETC models often do not fully account for multiple heat transfer pathways and neglect the dynamic variations in gas thermal conductivity and interparticle contact state, resulting in large deviations between predicted hydrogen absorption and desorption rates and experiments. This study develops two ETC prediction models for MH beds, both established by simultaneously considering three heat transfer pathways and the Smoluchowski effect in the pore space, which apply to non-hydrogen conditions without reaction (TCI model) and to hydrogen conditions with reactions (TCI-HAD model), respectively. The TCI model accounts for gas thermal conductivity changes and multiple heat transfer paths, with a validation deviation of 2-8 % under both atmospheric and dynamic gas pressure. Compared with other ETC models that do not couple reactions, the TCI model exhibits the best comprehensive prediction accuracy and strong applicability. Then, based on the TCI model, the TCI-HAD model for the hydrogen atmosphere is constructed by further considering the change of the interparticle contact state with the reaction fraction. Experimental verification results show that the mean absolute percentage error (MAPE) of the TCI-HAD model is 6.15 %. Compared with other ETC models that couple reactions, the TCI-HAD model achieves the highest prediction accuracy, with MAPEs of 4-10 %. Finally, based on the TCI-HAD model, the effects of particle size, MH bed temperature, and MH bed porosity on ETC were investigated during the hydrogen absorption process, respectively. Results indicate that the MH bed temperature and porosity are the main factors affecting ETC. The ETC models proposed in this study can provide theoretical support for investigations of coupled heat and mass transfer in MH beds, as well as for the determination of ETC in numerical simulations of MH reactors.

Timely removal of liquid water and the supply of the reaction gas in the gas diffusion layer (GDL) plays a critical role in improving the performance of polymer electrolyte membrane fuel cells (PEMFCs). Modifying the design of the GDL structure is an effective strategy for regulating the percolation process of liquid water and the supply of reaction gas. In this study, several GDLs with in-plane nonuniformly distributed pore sizes were designed to construct an ordered liquid water transport pathway. Two pore-size patterns with a "V" shape and an inverted "V" shape were designed through the orientation control of fiber distribution. In the inverted V-shaped pattern, the pore size exhibited a wave crest distribution along the in-plane direction, whereas, in the V-shaped pattern structure, the pore size was troughed along the in-plane direction. The three-dimensional (3D) multiphase Lattice Boltzmann method (LBM) and 3D diffusion LBM were used to investigate the liquid water percolation process and the reaction gas transport process in the GDL, respectively. The numerical results indicated that liquid water tends to concentrate in layers with macropores in the nonuniform GDL. Compared with the uniformly distributed GDL, these two pore size patterns can accelerate the drainage velocity and lower the water content. The reversed V-shaped pattern was further optimized to obtain the optimal width of the layers with macrospores. The results showed that a length of 96 mu m is recommended to balance the concentrated effect and low-concentration areas. Under dry conditions, the gas transport capacity was insensitive to pore size distribution, whereas, under partially saturated conditions, both the V-shaped and inverted V-shaped structures of a nonuniform design weakened the impeding effect of liquid water on the gas supply. Moreover, the effective gas diffusion coefficient of the nonuniform study can reach up to 3.85 times of the uniform structure. This work promotes the understanding of different in-plane distributed pore size styles on the water percolation behavior in the GDL, thereby contributing to the optimal design of the GDL and PEMFCs.

Oxygen transport resistance (OTR) severely limits proton exchange membrane fuel cell (PEMFC) performance with low Platinum (Pt) loading lower than 0.10 mg cm- 2, prompting the design of order-structured catalyst layer (OCL) with efficient multi-physical transport. In this work, a cross-dimensional model coupling a 2D PEMFC model with a 1D analytical model for OCL is developed to analyze oxygen transport in OCL. A Support Vector Regression (SVR) model combined with Sobol global sensitivity analysis is proposed for quantifying impacts of structural parameters on oxygen transport and output performance. It is shown that decreasing Pt loading from 0.10 to 0.01 mg cm- 2, the OTR in OCL is increased by 193 %, contributing 71 % of the total OTR in cathode with ultra-low-Pt loading lower than 0.01 mg cm- 2. The distance between the center of carbon nanotubes (CNTs) is the second most important factor influencing OTR and output performance revealed by SVR-Sobol sensitivity analysis, followed by the length of CNTs, the radius of CNTs and ionomer-to-carbon mass ratios. It is indicated that a well-optimized OCL can decrease OTR in OCL by about 20 % and boost normal peak power density by nearly 7 % compared with the original OCL with Pt loading of 0.10 mg cm- 2.

The rapid removal of condensate droplets is important for achieving stable dropwise condensation, thus improving the thermal efficiency of condensing equipment. The microstructure on the condensing surface can facilitate the departure of droplets. However, most existing microstructures are arrays of a single structure, which can only generate a single directional Laplace pressure gradient. To further improve the droplet-repelling performance of the microstructure, we propose a hierarchical superhydrophobic surface composed of opposed wedge bumps and diverging microgrooves in between, which is capable of generating multidirectional Laplace pressure gradients to trigger the self-jumping and collision-induced jumping behaviors to improve the jumping ability of condensate droplets. Through the application of a three-dimensional multiphase simulation model to condensate droplets on the microstructure, the mechanism of droplet's spontaneous movements is clarified and the optimal surface for departing droplets is determined. Through laser direct writing, chemical etching, and self-assembled monolayers, the optimal microstructure is fabricated on a copper plate. Wet air condensation visualization experiments have shown that the hierarchical superhydrophobic surface is able to realize the rapid removal of large droplets under the multidirectional Laplace pressure gradients. The droplet number density of the hierarchical superhydrophobic surface can reach 9.67 × 108 m-2, which is 106% higher than that of the plain superhydrophobic surface. The surface coverage of the droplet is reduced by 15% compared to the plain superhydrophobic surface, showing excellent potential for enhanced droplet removal.