Ammonia (NH3)/mathanol (CH3OH) co-firing is a promising clean combustion strategy, while hydrogen (H2) production through NH3 cracking and CH3OH reforming can further enhance fuel reactivity. Therefore, studying NH3/H2/CH3OH ternary flames can provide a basis for the individual and combined utilization of these three fuels. In this study, spherical expanding NH3/H2/CH3OH flames were investigated at 473 K and 0.5 MPa over different equivalence ratios (phi), H2 fractions, and CH3OH fractions. Flame-shape deformation, crack length, and cell number were quantified by image processing, and linear stability theory was used to evaluate critical conditions and perturbation growth rates. The results showed that CH3OH addition weakens thermal-diffusion instability (TDI) on the lean side governed by molecular diffusion, while the effect becomes stronger on the rich side. H2 addition shows the opposite trend. With H2 addition only, the critical flame radius increases monotonically with phi. With further CH3OH addition, critical flame radius decreases first and then increases with phi. The turning point shifts toward a higher phi than that of NH3/air flame. The combined addition of H2 and CH3OH does not simply produce additive effects, but changes the competition between TDI and hydrodynamic instability (HI). Specifically, H2 mainly enhances preferential-diffusion effects under lean-burn conditions, whereas high CH3OH addition strengthens the relative contribution of HI by reducing flame thickness and increasing the thermal expansion ratio. Adding H2 and 30% CH3OH intensify late-stage crack development and cellularization under lean-burn and stoichiometric conditions. At 60% CH3OH, stronger HI accelerates cell splitting and causes exponential growth of crack length. NH3/air flame is more prone to large-scale asymmetric deformation of flame profiles. With CH3OH and H2 addition, higher flame speed suppresses large-scale asymmetric deformation of flame profile. Meanwhile, it amplifies electrode-induced disturbances and makes mediumto-large-scale wrinkling more pronounced. Abbreviations: BI, Buoyancy instability; CVCC, Constant-volume combustion chamber; db4, Daubechies 4 wavelet; HI, Hydrodynamic instability; LC, Linear curvature model; LS, Linear stretch model; NQ, Nonlinear quasi-steady model; TDI, Thermal-diffusion instability
Microchannel flow boiling has emerged as a promising technology for thermal management of electronic devices due to its high heat transfer coefficient, low pumping power, and compact cooling architecture. This review presents a comprehensive overview of recent advances in microchannel flow boiling heat transfer and performance optimization strategies. The fundamental mechanisms governing microchannel flow boiling are first summarized with emphasis on bubble dynamics, microscale confinement effects, interfacial phase change, and thin film evaporation. Then the effects of channel geometry, surface modification, and working fluid modification on heat transfer performance are further analyzed by elucidating how these strategies regulate the underlying mechanisms of microchannel flow boiling. Furthermore, emerging optimization strategies are discussed in terms of hybrid active passive regulation, coordinated regulation of multiple boiling mechanisms, and artificial intelligence (AI) assisted prediction and optimization. Representative approaches include topology-optimized microchannel structures, adaptive shape memory alloy (SMA)-based regulation, jet microchannel coupled cooling, and AI-assisted prediction and optimization methods. The practical limitations of existing strategies are also discussed in terms of manufacturability, durability, operational stability, cost, packaging compatibility, and system level integration. Finally, the current challenges and future research directions in microchannel flow boiling are outlined. This review provides guidance for the design and optimization of next generation thermal management systems for high power electronic and energy systems.
Leakage in district heating networks (DHNs) can reduce building heat supply reliability, increase heat-carrier losses and pumping energy demand, and delay maintenance when leakage signatures are confused with routine operations. Pressure wave monitoring offers a rapid diagnostic route, but leakage-induced transients may overlap with valve adjustment, pump regulation, and emergency control actions. This study develops Sparse High-Order Mutual Information-based Multi-View Feature Selection (SHMI-MVFE), an interpretable framework for pressure wave leakage identification under operational transients. A laboratory DHN platform was used to collect 2000 pressure records at 2000 Hz under six conditions: leakage, normal operation, valve adjustment, pump-speed adjustment, emergency valve opening/closure, and emergency pump start/stop. After Hampel-based spike suppression and zero-phase band-pass filtering, 61 time-domain, frequency-domain, time-frequency, nonlinear, and complexity descriptors were extracted. SHMI-MVFE combines ElasticNet-based pre-screening, triplet-wise joint mutual-information scoring, adaptive weighting, bootstrap selection frequency, and Shapley additive explanations (SHAP) ranking consistency to identify a compact feature subset. Compared with representative feature selection baselines across multiple classifiers, SHMI-MVFE reached a stable performance plateau with fewer features. Using 12 selected features, SHMI-MVFE with Extreme Gradient Boosting achieved a weighted F1-score of approximately 0.968 and a six-class accuracy of approximately 97%. Error and SHAP interaction analyses show that residual misclassifications mainly arise from leakage–valve overlap, while selected feature interactions capture damping, persistence, and spectral-shift patterns. The framework supports interpretable leakage monitoring and false-alarm-aware maintenance decisions for reliable DHN operation.
This work proposes a hybrid computational framework that integrates the Virtual Element Method (VEM) with neural network techniques to efficiently solve three-dimensional hyperelastic problems. The framework combines graph neural networks (GNNs), VEM, and physics-informed neural networks (PINNs), and incorporates a transfer learning strategy to improve the model generalization ability. Owing to its ability to operate on arbitrary polyhedral meshes, VEM provides a flexible and accurate discretization for hyperelastic problems involving complex geometries. Building on this property, the proposed framework employs VEM to construct high-fidelity physics constraints, utilizes GNNs to encode the topology of polyhedral meshes, and adopts PINNs to solve the governing equations under these constraints. Furthermore, the introduced transfer learning strategy enables rapid adaptation of the trained model to new material parameters and loading conditions. Numerical experiments demonstrate that the proposed framework achieves accuracy comparable to that of traditional high-fidelity numerical simulations while improving computational efficiency by one to two orders of magnitude. In addition, the method exhibits strong robustness and generalization capability under variations in parameters and loading scenarios. These results indicate that the proposed approach provides a promising and efficient solution for the rapid analysis of multi-scenario and multi-parameter three-dimensional hyperelastic problems.
