As transistor densities surge and boost frequencies intensify transient CPU heat fluxes, passive thermal management with phase change materials (PCMs) offers a zero-power alternative. However, the low thermal conductivity of PCMs demands high-conductivity fins, and no existing study has employed topology optimization to design a multi-layer stacked PCM heat sink for long-duration cooling. The present study's novelty is a four-row PCM heat sink with aluminum fins generated by density-based topology optimization (SIMP, 20% solid fraction, Helmholtz filtering) under steady operating conditions. The optimized geometry is simulated via a transient enthalpy-porosity CFD model resolving conjugate heat transfer, laminar natural convection, and phase change, validated against experimental data (RMSE=2.44 K, NRMSE = 3.26%). Over eight hours of continuous CPU load, sequential melting occurs: the lowest row fully melts in 66 min, the second in 275 min, while the third reaches a liquid fraction of 0.54 and the top remains solid. The heat-sink base temperature stabilizes at 355.72 K (approximate to 82.6 degrees C) after 8 h, which lies near the upper limit of typical commercial CPU operating temperatures, with a difference of only 3.51 K, compared to 388.36 K for a conventional single-layer radial-fin heat sink after just 1 h-a 42.4 K reduction, reflecting the combined effect of the multi-layer configuration and the topologyoptimized fin geometry. Latent heat (2.6 kJ per row) dominates the lower blocks, while natural convection is suppressed to micrometre-per-second velocities.
The rapid advancement of low-power microelectronic technologies has stimulated growing interest in vibrationbased energy harvesting as a promising alternative to conventional batteries. While batteries are widely used at these scales, their replacement becomes highly challenging for systems deployed within inaccessible or remote locations. The major drawback of linear harvesters is their limited frequency bandwidth. This limitation, together with the wide frequency spectrum of ambient excitations, significantly diminishes the energy conversion efficiency of linear harvesters. To address this challenge and improve the performance of such systems in practical environments, it is essential to broaden their frequency bandwidth. Among the various approaches for broadening the frequency bandwidth, employing nonlinear techniques has proven to be highly effective in enhancing the efficiency of vibration energy harvesters. This study proves the theoretical modeling of frequency bandwidth analysis in a piezoelectric energy harvester for a series of experimental cases considering magnetic nonlinearity and impact-induced nonlinearity. Applying both approaches together increases the system efficiency by about 80 % compared to using either method separately. Furthermore, the influence of key system parameters-including magnetic strength, input acceleration amplitude, electrical resistance, and initial gap-on the system's dynamic behavior is investigated. Finally, an efficiency index is introduced to assess the harvester's efficiency relative to the conventional design.
Abstract Mechanical splices enable continuity of reinforcement, yet their behavior under corrosion remains a concern. We experimentally compare spliced (threaded couplers) and unspliced bars in corroded and noncorroded conditions across common diameters (16–22 mm). An accelerated corrosion protocol was applied, followed by tensile testing and metallographic examination. Under uncorroded conditions, yield strength differences between spliced and unspliced bars were small ( + 0.29 % – 0.95 % ). Under corrosion, losses in mechanical performance were more pronounced, and fracture location could shift relative to the coupler. An American Concrete Institute (ACI) 318 acceptance screen ( 1.25 F y ) indicated that only selected noncorroded spliced specimens (e.g., M18ST and M20ST) met the minimum requirement, while corroded spliced specimens generally did not. These results clarify when mechanical splices should be avoided in zones with high corrosion risk or ductility demand and motivate protective detailing in aggressive environments.
Gas hydrate formation is a complex physicochemical process governed by coupled thermodynamic and kinetic factors, making accurate prediction challenging using conventional models. In this study, a machine learning (ML)-based framework was developed to predict and analyze the formation kinetics of CO2 and CH4 hydrates using differential scanning calorimetry (DSC) data. A Random Forest regression model was trained on experimental datasets covering a range of pressures and additive concentrations, including sodium dodecyl sulfate (SDS), lauryl alcohol ethoxylate-2 (LAE2), and tetrahydrofuran (THF). The model demonstrated excellent predictive performance, achieving an R2 of 0.972 and low mean squared error across independent condition-level validation folds. Comparative analysis with design of experiments (DoE) results showed that ML predictions more closely matched experimental data, particularly for heat-flow ratio and relaxation time of the hydrate formation, with significantly lower deviations. The study further identified optimal operating conditions for hydrate formation, highlighting the strong influence of pressure and additive concentration. While DoE provided general trends, ML captured nonlinear interactions more effectively, enabling improved prediction accuracy. Overall, the data consistently show that increasing pressure enhances all aspects of hydrate formation, while the type and concentration of additive determine the efficiency, with LAE2 (for CO2) and THF (for CH4) providing the most favorable conditions.
Recommender systems have become a fundamental component of modern online platforms,significantly influencing user experiences by delivering personalized content. However, theincreasing complexity of contemporary recommendation models has raised critical concernsregarding their transparency and fairness. This survey provides a comprehensive and unifiedreview of explainable and fair recommender systems, emphasizing the need to move beyondaccuracy-centric design toward trustworthy and responsible recommendation paradigms. Wefirst introduce the core concepts of recommender systems, explainability, and fairness, anddiscuss their roles in building transparent and accountable systems. We then propose a multidimensional taxonomy that systematically categorizes existing approaches based on objectives,intervention stages, explanation types, and fairness perspectives. Furthermore, we present adetailed analysis of state-of-the-art methods, including explainable recommendation techniques,fairness-aware models, and emerging approaches that jointly address both aspects. In addition,we examine commonly used evaluation metrics and provide a comprehensive analysis of thetrade-offs among accuracy, explainability, and fairness, highlighting the challenges of multiobjective optimization in real-world systems. Finally, we identify key open challenges, includingdata bias, methodological limitations, evaluation inconsistencies, and deployment constraints,and outline promising directions for future research. This survey aims to provide researchers andpractitioners with a unified perspective and to support the development of recommender systemsthat are not only accurate, but also transparent, fair, and trustworthy
