Metal hydride adsorption offers a promising approach to efficient and safe hydrogen storage. However, the kinetics and capacity of hydrogen adsorption are often limited by poor heat and mass transfer within the porous sorbent. The spatial distribution of porosity critically influences these processes by governing local permeability and effective thermal conductivity. In the study, a deep operator network trained on numerical simulation data establishes a quantitative relationship between porosity distribution and crucial physical fields as well as performance metrics in an annular-finned reactor. Then the predictive model is coupled with particle swarm optimization to identify an optimal porosity profile that accelerates hydrogen uptake and minimizes the adsorption duration to reach 80 % of equilibrium capacity. Results demonstrate that the model achieves high prediction accuracy, with average relative root mean square errors of 0.01 %, 0.1 %, 0.3 %, and 2.7 % for temperature, pressure, reaction fraction, and adsorption duration, respectively. The optimized porosity profile exhibits a graded structure, decreasing progressively from the hydrogen inlet toward the heat exchange wall. The superior configuration enhances heat and mass transfer across the main reaction region, leading to a 12.77 % reduction in adsorption time compared to a uniform porosity bed with the same average porosity of 0.5.

Exponential emission of volatile organic compounds (VOCs) from fuel processing and fossil fuel combustion has led to significant environmental challenges. Catalytic conversion using perovskite oxides has emerged as a promising approach for VOCs removal. However, the catalytic performance of perovskites is influenced by complex and nonlinear interactions between composition, structure, and synthesis conditions, which makes predicting catalytic activity and designing high-performance catalysts inefficient. To address this challenge, this paper proposes a machine learning framework that incorporates three components: (i) data-driven modeling for catalytic activity prediction, (ii) feature analysis for mechanistic interpretation and interaction elucidation, and (iii) multiobjective optimization for composition design under stoichiometric constraints. An experimental database was constructed from existing research, focusing on activation energy (E a), T 50%, and T 90% as targets. The extremely randomized trees (extra-trees) model was selected for optimal prediction performance. Feature importance analysis identified Co as the most critical factor, followed by La, Fe, Sr, pore volume, and calcination temperature. Further analysis revealed that Co content, calcination time, and specific surface area exhibit strong positive effects, while Mn content, toluene concentration, and pore diameter exhibit negative effects beyond optimal ranges. Notably, Co content exhibits synergistic interactions with pore volume and calcination time, significantly influencing catalytic performance. Under A-site vacancy and B-site stoichiometric constraints, the NSGA-II generated a Pareto front that reveals the internal competitive relationship between E a, T 50%, and T 90%. Using CRITIC weighting and TOPSIS, the optimal composition La0.84Sr0.08Co0.82Fe0.18O3 was identified. Compared with database averages, this composition reduces E a, T 50%, and T 90% by 37.3%, 3.0%, and 12.1%, respectively. This work introduces a transferable framework combining data modeling, mechanistic explanation, and multiobjective optimization for the rational design of high-performance perovskite catalysts.

The geometric configuration of microchannel reactors significantly impacts inter-channel flow, heat transfer, and species diffusion behaviors, which determine the reaction efficiency and CO preferential oxidation (CO-PROX) performance for hydrogen purification. In the present study, the available energy potential loss caused by heat transfer, mass transfer, and chemical reactions is derived for CO-PROX and then systematically analyzed under different geometric configurations based on computational fluid dynamics. A data-driven deep learning model, integrating microchannel geometry generation, physical field prediction, and performance prediction, is constructed to achieve hierarchical mappings from microchannel geometric parameters to physical fields and performance parameters. Excellent accuracy in microchannel geometry generation is achieved, with an average accuracy of 0.9541. For physical field prediction, the average relative errors of the CO, CO₂, H₂, H₂O, and O₂ concentration fields are 2.47%, 2.97%, 0.04%, 4.05%, and 4.48%, respectively; the average relative errors of the velocity field and pressure field are 3.33% and 3.18%, respectively; and average relative errors of available energy potential loss caused by CO oxidation, H₂ oxidation, heat transfer, mass transfer, and viscous dissipation are 2.58%, 2.71%, 2.15%, 4.93%, and 3.38%, respectively. Optimal microchannel geometries are then identified by maximizing CO conversion rate and minimizing available energy potential loss, respectively. Compared with the original straight channel, the CO conversion rate is increased to 94.91%, while the total energy potential loss is reduced by 74.07% with pressure drop increase kept below 10%.

Solar-driven hydrogen generation system provides a promising technical way for solar energy harvesting. Here, an off-grid solar-driven hydrogen production system with energy storage based on photovoltaic (PV), osmotic heat engines (OHE), proton exchange membrane electrolysis cell (PEMEC) and lithium-ion battery is established. An energy control strategy, incorporating a sequence operation strategy for the PEMEC array, is designed to manage power allocation and enable all-day energy management. The effects of light conditions, adsorption desalination (AD) cycle time, and minimum state of charge (SOCmin) of lithium-ion battery on the hydrogen production performance and operation stability are systematically analyzed. Under clear summer conditions, the system achieves a maximum daily hydrogen production of 3877.67 L, with PEMEC units operating at rated power for an average of 11.47 h, supported by a nocturnal OHE output of 332.97 W. Longer AD cycle times enhance daytime Gibbs free energy storage, increasing nocturnal power output and hydrogen yield, whereas shorter cycles extend PEMEC operation under fluctuating power. A lower battery SOCmin of 0.4 maximizes daily hydrogen production (3893.38 L) and rated-power operation duration (11.62 h), despite greater power deviation, compared to higher SOCmin settings which accelerate charge-discharge transitions and affect nocturnal OHE output.

Desired working salt solutions can significantly accelerate the energy conversion performance of osmotic heat engines. Here, a robust dynamic model for working aqueous salt solution screening for adsorption-based osmotic heat engines consisting of adsorption desalination (AD) and reverse electrodialysis (RED) for power and cooling cogeneration is developed, which considers impacts of salt solutions on evaporation pressure in the AD module, transmembrane ion and water transportation, as well as electrical resistance in the RED module. A comprehensive analysis is performed to identify optimal physical and chemical working solution characteristics for improving system performance by evaluating 129 kinds of working aqueous salt solutions. Results indicate that working solutions with higher concentrations, osmotic and activity coefficients contribute to electrical efficiency. Higher anion valence has a detrimental effect on electrical efficiency while improving the coefficient of performance (COP). Moreover, larger salt molar mass and elevated osmotic and activity coefficients result in diminished COP. To further validate the simulation results and assess the operational feasibility, an experimental study employing five common salts is conducted, focusing on the RED module. The underlying mechanism for electrical efficiency degradation observed with high-valence anion salt solutions is revealed.

In order to analyze the influence of condensation behavior on the heat transfer process of large heat exchangers under wet conditions, a three-dimensional numerical model of condensation heat transfer of twelve-row wavy fin-and-tube heat exchangers under wet conditions was established. The effects of inlet velocity and relative humidity on heat and mass transfer characteristics were investigated. Through the analysis of the field synergy principle, it is found that the wavy fin can effectively improve the synergy between the velocity field with the temperature field and the concentration field. The wavy fin not only increased the disturbance of fluid but also continuously disrupted the developments of thermal and mass concentration boundary layers, thereby greatly enhancing the heat and mass transfer. The condensation mainly occurs on the first six pipe rows, and the increase of inlet velocity makes the condensation area obviously larger. The inlet relative humidity has little effect on the temperature distribution and sensible heat transfer under wet conditions but has a significant effect on the latent heat transfer and total heat transfer. Finally, the relationship for convective heat transfer coefficients under wet and dry conditions is obtained by multivariate fitting.

Adsorption-based battery thermal management (ABTM) offers an efficient passive cooling solution for lithium-ion batteries (LIBs). The thermal performance of ABTM is governed by the heat and mass transfer behaviors during adsorption and desorption. A multi-physics coupling model integrating a pseudo-two-dimensional (P2D) electrochemical battery model with a transient heat and mass transfer model for the adsorption/desorption process is developed. Considering humid air species transport within the porous adsorbent bed, extended dissipation functions and field synergy angles are established to quantify heat and mass transfer performance within the adsorption/desorption processes in ABTM systems. Influencing factors on the system performance are systematically analyzed. Results reveal that a maximum temperature reduction of 7°C is achieved via this adsorption-based battery thermal management system under 1.5 C charging. The synergy angles can directly characterize the temporal adsorption/desorption kinetics as well as the coordination between the flow and temperature fields. Mass transfer dissipation acts as the dominant form of energy loss, accounting for over 70% of the total dissipation. The total mass transfer dissipation peaks at an adsorbent thickness of 2–3 mm. Meanwhile, a moderate porosity range of 0.3–0.4 provides a favorable balance between reduced vapor transport resistance and sufficient adsorbent inventory. In addition, higher relative humidity enhances cooling potential, ensuring sustained evaporative heat extraction under high thermal loads. The heat and mass transfer synergy and dissipation mechanisms revealed could provide a theoretical basis for rational design and optimization of high-efficiency passive battery thermal management systems.

Structural design and optimization are essential for improving the performance of adsorption-based desalination and cooling systems. Moving beyond empirical approaches, the study applies second law analysis to the transient adsorption bed which is the primary source of irreversible loss, quantitatively relating dissipative characteristics with overall system throughput in specific and volumetric terms. The developed formulation delineates entropy generation into irreversibility arising from adsorption kinetics, viscous flow, and heat transfer, which is subsequently used to evaluate triply periodic minimal surface structured beds and enhance system performance. Three-dimensional numerical simulations compare various architectures with differing skeleton and bed porosities, revealing competitive mechanisms under varied conditions. Results show that irreversible features effectively clarify the relationship between complex transport phenomena and overall production. Volumetric performance is optimized when maximizing adsorption irreversibility and fluid flow irreversibility within the adsorbent, and heat transfer irreversibility inside the skeleton. Furthermore, machine learning and genetic algorithms are employed to optimize entropy generation, dissipation, and working capacity. The Diamond-type structure achieves the highest total production, while the optimal bed porosity remains consistent across diverse architectures, indicating strong structural portability.

Battery safety problems in abnormal thermal situations such as operation in cold/hot environment, thermal management defunction and thermal runaway attract addressing attentions. Here from the battery safety perspective, a coupled thermal-electrochemical-mechanical phase-field model is developed for crack propagation and lithium dendrite growth, thus to illustrate the underlying mechanisms of interfacial failure and dendrite evolution in lithium metal all-solid-state batteries (ASSBs) under abnormal thermal situations. The effects of inter-cell temperature distribution direction and magnitude, stacking pressure, and interfacial roughness at on crack propagation and dendrite growth are systematically investigated. Originating from augmented strain energy density by thermal expansion, crack propagation is much accelerated at negative temperature differences (NTD), which provide more space for dendritic growth as well as inducing stronger longitudinal evolution directionality. The fastest dendrite distance at NTD increases much faster than that under positive temperature difference (PTD) for significantly enhanced electrochemical driving force. Under isothermal or PTD conditions, an applied stacking pressure below than 30 MPa can inhibit the crack propagation and fastest dendrite evolution. However, at NTD, any applied stacking pressure contributes to crack propagation and lithium dendrite growth. More initial defects increase the crack region area, meanwhile the crack propagation depth is shortened due to weakened von Mises stress and strain energy density. Considering both crack propagation and fastest dendrite evolution, applying a suitable stacking pressure below 10 MPa to improve the Li/solid electrolyte (SE) interface is desired, thus to reduce the interfacial failure and possibility of short circle. The findings offer an alternative comprehensive perspective to evaluate the battery safety under abnormal thermal conditions, and could provide rational guidance for design and development highly reliable ASSBs.

Adsorption desalination offers a promising alternative for freshwater production. Optimizing complex heat and mass transfer structures for the adsorption bed via traditional computational fluid dynamics is challenging due to the "curse of dimension". Present study is dedicated to analyzing the synergetic impact of fin topology and adsorbent porosity distribution on the intricate heat and mass transfer within a heterogeneous porous bed, then identifying the optimal configuration that advances water production. An improved data-free deep learning model based on physics-informed deep operator network is constructed to capture the quantitative relationship between the adsorbent bed topology and heterogeneous adsorbent porosity and the physical fields and performance indexes. The predicted physical fields are compared with data obtained from a commercially available numerical model, showing a high predictive accuracy of 3.7 %, 0.26 %, and 0.63 % in terms of average relative root mean square error. To step further, based on the improved data-free deep learning model, the synergic inverse design for bed topology and porosity distribution is conducted via the particle swarm optimization method, rendering maximum total water production. Synergically optimized bed topology and adsorbent porosity distribution configurations under different adsorption durations are achieved, which results in an 18.2 % improvement in total uptake.

The growth behavior of lithium dendrites significantly impacts the lifespan and safety of lithium-metal batteries. Here a non-linear phase-field model coupling the separator phase and heat transfer characteristics is established to investigate the evolution mechanism of lithium dendrites. Impacts of nonuniform temperature configuration, separator thermal conductivity and structure on the lithium dendrite growth and morphology are systematically analyzed. Results reveal that under non-isothermal configurations, larger temperature difference contributes to lithium deposition amount and longer maximum dendrite height. Lithium dendrite growth under the negative temperature difference is much faster than that under the positive one. Lithium deposition amount and maximum dendrite height both increases phenomenally with increasing separator thermal conductivities, reaches the maximum value, then decreases. The impacts of separator structures on the lithium dendrite deposition are strongly related to the direction of the temperature difference. Under the positive temperature difference, larger upper layer sizes contribute to the lithium dendrite growth, while smaller upper layer size hinders lithium dendrite growth. Large or small lower layer sizes both inhibit the lithium dendrite growth. Under the negative temperature difference, ununiform bilayer separator structure contributes to the growth of lithium dendrites.

Thermal and electric inhomogeneity in lithium-ion battery packs can significantly impact the lifespan and the safety of the unit cell and the whole battery pack. Here, to reveal the inhomogeneity, and explore the interaction between the thermal inhomogeneity and electrical inhomogeneity, the electrochemical-thermal coupled model is employed to study the thermal and electric inhomogeneity of 4s3p lithium-ion battery packs. The effects of ambient temperature, coolant inlet flow rate, initial state of charge (SoC), and driving cycle conditions on the standard deviations of thermal and electrical performance indicators of unit cells in the battery pack are analyzed. Results reveal that the batteries present obvious thermal and electrical inhomogeneity at high discharge rates. At 20 degrees C, the standard deviations of voltage and temperature at a discharge rate of 2C were 24.9 times and 7.75 times that at a discharge rate of 0.5C, respectively. The standard deviation of cell current and heat production increases with increasing discharge time, reach their maximum values at about 75 % DoD, and then decrease. At the discharge rate of 2C, the maximum standard deviation of current reaches 0.05A. Increasing ambient temperature contributes to the standard deviation of cell current and heat production. Increasing the coolant inlet flow rate decreases the standard deviation of cell voltage. Ununiform initial SoC distributions have a larger impact on standard deviations of electrical performance indicators. For homogeneous initial SoC, the standard deviation of voltage is less than 0.001, while that of voltage reaches 0.01 similar to 0.03 V for non-uniform initial SoC. The standard deviation of current is increased by about 6 times, while that of heat production rate is increased by about 10 times. This paper may offer rational insights to rational design of lithium-ion battery packs and thermal management strategies.

Adsorption desalination offers a promising way for freshwater production. Heterogeneous distribution of adsorbent significantly impacts the heat and mass transfer characteristics in the adsorbent bed, thus the water production performance. Here we developed a novel data-free physics-informed deep operator network with domain decomposition to learn solution operators for domain-dependent transient and complex partial differential equations in the adsorption desalination process. The transient spatial temperature, pressure, and uptake fields under spatially heterogeneous adsorbent porosity distribution are predicted via the proposed multi-domain physics-informed deep operator network. The generalization error is evaluated using calculated data via computational fluid dynamics simulation, demonstrating an average relative root mean squared error of less than 1.1% in the uptake, meanwhile, the calculation speed is accelerated up to about 1000 times than computational fluid dynamics simulation. Furthermore, a data-free self-validated framework for inverse design is proposed for heterogeneous adsorbent porosity optimization, which is difficult to achieve using traditional numerical methods. A significant improvement of 8.78% in total uptake is achieved compared to the uniform porosity configuration. Present study conquers the heterogeneous porosity optimization problem in the adsorption process and may serve as a way for heterogeneous structure and material property optimization in other engineering disciplines.

Heat source field inversion and detection (HSFID) has drawn increasing attention as the exponentially growing application for integrated circuits, which offers promising way for determining the system's unnormal operation condition. In the HSFID, embodying the physical constraints in the neural networks could significantly reduce the data demand for training and offer higher reconstruction accuracy. In present study, the physics-informed neural network (PINN) is employed to achieve the goal of HSFID. The problem of reconstructing the heat source field is transformed into the challenge of temperature field reconstruction. The PINN is employed to conduct the HSFID with various locations, shapes, sizes and power densities under multi-heat source configurations. For the twosource configuration, the heat source shape and position similarity (HSSPS) for detecting triangular heat sources is 98.9 %, meanwhile for four heat source configurations, the HSSPS is 93.5 %. In complex heat source systems where the location, shape, size and power density change randomly and simultaneously, the maximum temperature mean absolute error (TMAE) value is around 0.003 K, the maximum value of the temperature absolute error (M-TAE) value fluctuates in the range of 0.02 K, and the HSSPS is not less than 92 %.

Efficient solar-to-hydrogen system can substantially accelerate the achievement of the carbon neutrality commitment. Here, a novel solar powered hydrogen production system with energy storage is proposed. It comprises a solar energy collector, an adsorption desalination (AD) module, solution storage devices, a reverse electrodialysis (RED) module, a DC-DC converter module, and a proton exchange membrane (PEM) electrolyzer module. The impacts of solar radiation intensity, RED working voltage, and AD adsorption time on the hydrogen production power, hydrogen production rate, and overall hydrogen efficiency are systematically investigated. The optimal system performance under the direct coupling mode and energy storage mode in a day are identified via machine learning and particle swarm optimization. When the system operating under the direct coupling mode, the optimal total amount of hydrogen produced reaches 126.29 L per day. When the system operating under the energy storage mode, the optimal total hydrogen production presents a maximum value of 140.84 L per day at the RED working time of 13h. Compared with the direct coupling mode, the total hydrogen production of the system under the energy storage mode is increased by 11.5%.

Mitigating the hydrogen leakage risk during the refilling process contributes to promoting the hydrogen energy commercialization. In this study, the high-pressure hydrogen leakage and dilution characteristics were analyzed with fans installed on the hydrogen refueling station canopy. Two fan conditions are involved: blowing upwards condition, and blowing downwards condition. The role of leakage pressure, atmospheric wind velocity, and fan conditions on the spatiotemporal evolutionary characteristics of flammable gas clouds (FGC) were discussed. Results showed that the hydrogen distribution is significantly impacted by wind velocity and fan blowing conditions when the hydrogen-powered vehicle is refueled. At a small leakage pressure, under windless conditions, employing fans contribute to the FGC dissipation. At windy conditions the fan blowing upwards does not accelerate the FGC dissipation, but rather increases the its volume and slows down its dilution. At a large leakage pressure, at small wind velocities, fan blowing upwards contributes to FGC dilution. However, at large wind velocities, fan blowing upwards is not effective in accelerating the hydrogen dilution. In all the scenarios, fan blowing downwards always helps to accelerate the hydrogen dilution. This study offers a pragmatic approach to effectively mitigate the risk of hydrogen leakage during the refilling process.

Previous efforts regarding adsorption-based osmotic heat engines (ADOHE) are focused on two adsorption bed configurations and no literatures have been reported on ADOHEs with multi-bed configurations as well as principles for system configuration adjustment under different operating conditions. In this study, a four-bed adsorption-based regeneration is applied to the osmotic heat engine to upgrade the power generation and refrigeration performance. Cascade and parallel configurations distinguished by the thermal energy utilization models are presented. Performance indicators of heat recovery rate and exergy efficiency are employed for evaluating the system capability to utilize low-temperature waste heat. Results reveal that compared to the parallel configuration, cascade configuration exhibits lower work extracted, cooling capacity output and coefficient of performance (COP), while higher electrical efficiency, energy recovery rate and exergy efficiency. In addition, the cascade configuration effectively reduces the refrigerant outlet temperature fluctuations, thus improving the quality of cooling capacity. To demonstrate the practical application potential, configurationswitching control strategy optimization under real-time heat source with varying temperatures and flow rates is carried out via machine learning and genetic algorithms. The cascade configuration is preferred for higher temperature heat sources, while parallel configuration is preferred for lower temperature heat sources. The maximum electrical efficiency and COP of 0.27 % and 0.85 are obtained under the optimal configurationswitching control strategies. Compared to the parallel and cascade configurations, electrical efficiencies are improved by 5.95 % and 4.98 %, while the COPs are improved by 3.99 % and 4.84 %, respectively. This study may offer rational and potential guidance for designing and operating upgraded adsorption-based osmotic heat engines.

To screen drug targets of ovarian aging from a genetic perspective. Systematic analyses were conducted with cis-expression quantitative trait loci data of druggable genes extracted as instrument variables. Summary statistics were from large genome-wide association studies for age at menopause. The following colocalization analysis was utilized to examine whether identified genes and ovarian aging shared causal variants. Furthermore, clinical validation was conducted by comparing expression of identified genes in granulosa cells from women with normal or diminished ovarian reserve (DOR) who went through in vitro fertilization (IVF) and by evaluating correlation of targeted gene expression with ovarian function and IVF outcomes. Moreover, single-nuclear RNA (snRNA) seq and drug database were analyzed to find target cells within the ovary and potential drugs targeting identified genes. Systematic analyses identified five therapeutic targets of ovarian aging, including four protective factors (BRCA1, KLHL18, PNP, SRPK1) and one risk factor (PDIA3). The change in expression level of four protective factors has been verified in clinical validation. Particularly, both BRCA1 and SRPK1 have been downregulated among advanced-aged women with DOR and were positively correlated with anti-Müllerian hormone and antral follicle count. Specific target cells and potential small molecule targeted drugs of these genes were identified through snRNA analysis and searching in the drug database. By systematic genetic analyses combined with clinical validation, we identified five potential druggable genes for ovarian aging, providing theoretical basis and promising direction of therapeutic genetic targets for ovarian aging in the future.