Improving the thermophysical performance of organic phase change materials (PCMs) is essential for efficient low-to medium-temperature latent heat thermal energy storage (LHTES) systems. In this work, RT64HC paraffin was reinforced with HfC nanoparticles (0.5-2.0 wt%) using a controlled multi-step dispersion method that combined surfactant-assisted stirring with ultrasonication. Structural and spectroscopic analyses (XRD, FT-IR, FE-SEM/EDX) confirmed uniform nanoparticle distribution without chemical reaction or phase segregation. Thermophysical measurements showed significant improvements. In the solid phase, thermal conductivity increased from 0.21 to 0.367 W/m.K (74.76 %) while in the liquid phase it rose from 0.177 to 0.235 W/m.K (32.76 %). Specific heat capacity increased from 1.42 to 1.87 J/g.K (31.69 %) in the solid phase and from 2.29 to 2.91 J/g.K (27.07 %) in the liquid phase within the 73-80 degrees C window. Differential scanning calorimetry (DSC) revealed excellent cycling durability with enthalpy retention of 97.2-99.2 % after 1000 cycles and effective suppression of supercooling (Delta T <= 0.06 degrees C). Thermogravimetric analysis (TGA) demonstrated enhanced thermal stability with onset degradation temperature reaching about 192 degrees C at 1.0-1.5 wt% and 185.2 degrees C at 2.0 wt% while maximum decomposition temperature increased to 343.62 degrees C. Residual mass rose from 7.81 % to 12.61 %, indicating improved high-temperature resistance. Compared with conventional fillers, HfC enables superior conductivity and stability at low loading while reducing agglomeration and enthalpy fading. These advantages highlight HfC/RT64HC nanocomposites as promising candidates for renewable and sustainable energy applications including solar collectors, building envelopes, electronic cooling, battery regulation, waste heat recovery and thermal batteries.

This study presents a machine learning (ML) framework for predicting heat transfer in magnetohydrodynamic (MHD) mixed convection within a complex V-shaped cavity filled with a nano-enhanced phase change material (NEPCM) suspension. Accurate computational fluid dynamics (CFD) simulations are essential for understanding heat transfer mechanisms in such systems, but generating comprehensive data through highfidelity models remains computationally expensive. To address this challenge, we develop an integrated ML approach that combines synthetic data generation, physics-informed feature engineering, and optimized ensemble boosting. The methodology first augments a limited 34-sample CFD dataset to 2034 samples using Latin Hypercube Sampling with Radial Basis Function interpolation. Next, 14 physics-based features are engineered to encode the underlying physical phenomena. Finally, hyperparameters of three gradient boosting models-eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Categorical Boosting (CatBoost)-are optimized via cross-validation. The framework predicts the average Nusselt number (Nuavg) and average kinetic energy (KEavg) from seven geometric and operational inputs. CatBoost achieved optimal performance for Nuavg (R2 = 0.9745, mean absolute percentage error = 1.27%), while XGBoost excelled for KEavg (R2 = 0.9920, mean absolute percentage error = 1.04%). The novelty of this work lies in its ability to generalize across different output variables and significantly reduce computational cost, enabling rapid design optimization and in-depth parametric analysis. This generalizable approach reduces computation time from hours to milliseconds, facilitating efficient design optimization and in-depth parametric studies.

PurposeThis work aims to present an analytical study of the head-on collision of bidirectional solitary waves in a Rayleigh–Bénard–Marangoni system consisting of a viscoelastic Maxwell fluid. The fluid is bounded by a rigid, nonpermeable lower surface and a deformable free surface at the top, where a constant heat flux is applied. Design/methodology/approachAn extended Poincaré–Lighthill–Kuo method is used, incorporating stretched coordinates and phase functions, with an asymptotic expansion carried out consistently at successive orders. This approach yields coupled Korteweg–de Vries-type evolution equations governing the right- and left-propagating waves. The solutions are derived up to the second-order approximation, and the expressions for phase shift, distortion profile, maximum run-up amplitude and Nusselt number are presented explicitly. FindingsThe effect of wave interaction on heat transfer is quantified through the Nusselt number, showing that viscoelastic memory induces persistent changes in the vertical temperature gradient. Explicit expressions for phase shifts show that, unlike Newtonian fluids, viscoelastic relaxation leads to persistent trajectory corrections, resulting in intrinsically inelastic collisions, lasting waveform asymmetry and nonlinear peak broadening. A feasibility analysis based on representative physical parameters indicates that these nonlinear and depression-type solitary waves are most likely observable in the Bénard–Marangoni regime under experimentally accessible conditions. Originality/valueTo the best of the authors’ knowledge, the head-on collisions between solitary waves in a thermally driven Maxwell Rayleigh–Bénard–Marangoni system are studied using a perturbation approach for the first time in this paper. In contrast to earlier studies, this work has been restricted to unidirectional wave propagation. Therefore, the present results reveal intriguing facts that could be useful for experimental purposes.

Novel cooling solutions and efficient computational methods are essential for the effective thermal management of multiple PV units. This study proposes a T-shaped cooling channel having various protrusion shapes (rectangular, elliptic, and triangular) positioned near the channel junction under a uniform magnetic field. CFD simulations are performed to analyze the cooling channel for ranges of Reynolds number (50≤Re≤300) and Hartmann number (0≤Ha≤50). The horizontal location of the protrusions (x0) is varied between 0.023 and 0.027, considering different geometric shapes. The numerical solutions are obtained using the finite element method. Optimization studies are performed to identify the optimal operating parameters (Re, Ha, and x0) for various protrusion geometries. Machine learning models, specifically Artificial Neural Networks (ANN), Support Vector Regression (SVR), and Gaussian Process Regression (GPR), are utilized to predict the heat transfer coefficients for vertical and horizontal cooling configurations. The most accurate predictive model is subsequently integrated with double PV-TEG units for comprehensive thermal management studies. Higher Re promotes upper-wall recirculation and initiates secondary vortices at protrusion apexes that propagate downstream. Maximum Ha results in significant vortex suppression near the protrusions and channel junctions. In the absence of a magnetic field, vertical channel cooling performance significantly decreases due to large vortex formation at the channel entrance. However, proper protrusion placement combined with higher magnetic field strength improves cooling performance for both channel configurations. Optimization studies are conducted, and compared to the flat baseline at minimum flow without a magnetic field, these optimized configurations yield heat transfer enhancement factor values of 3.48 (R), 3.27 (E), and 2.87 (T). The best-performing model, Gaussian Process Regression (GPR), is subsequently integrated with the dual PV-TEG units for comprehensive performance analysis. Compared to the reference baseline (flat-walled, minimum flow, Ha=0), the optimized T, E, and R-type protrusions achieve temperature reductions of 11.6 ∘C, 14.8 ∘C, and 15.0 ∘C for the vertical unit, and 9.4 ∘C, 11.2 ∘C, and 11.5 ∘C for the horizontal unit. The results demonstrate that the synergistic application of optimally placed protrusions and magnetic field effects significantly enhances cooling performance which, coupled with the machine learning-assisted computational method, offers an effective framework for the advanced thermal management of photovoltaic units.

This study presents a 12-month field assessment of a 3.55 kWp hybrid solar-envelope system comprising a 1.10 kWp rooftop water-based photovoltaic/thermal (PV/T) subsystem and 2.45 kWp of semi-transparent building-integrated photovoltaic (BIPV) modules installed on the east, south, and west façades of a single-room test building in Elazığ, Türkiye. Annual AC-side renewable electricity generation was 4,327.30 kWh, whereas the adopted annual electrical-load scenario was 3,033.50 kWh. These values correspond to an annual generation-to-load ratio of 142.65% and a nominal generation-minus-load balance of + 1,293.80 kWh. Monthly aggregate generation exceeded the adopted monthly load from February through November. However, these monthly balances do not establish continuous electrical self-sufficiency, self-consumption, or gross grid exchange. The PV/T GHI-referenced thermal-energy index reached a maximum of 55.90%. The area-weighted GHI-referenced system energy and exergy indices ranged from 12.74 to 25.21% and from 8.73 to 14.56%, respectively. These quantities are reference performance indices rather than plane-of-array conversion efficiencies because orientation-specific irradiance was not measured. The embodied energy was 18,018 kWh, and the reported exergy-referenced energy-payback indicator was 3.18 years. Under the load-limited and full-generation electricity-crediting boundaries, the conditional gross carbon-inclusive simple payback periods were 12.57 and 8.81 years, respectively. Combined ± 20% electricity-price and carbon-price scenarios produced simple-payback values ranging from 7.34 to 11.01 years. Under the fixed-factor, full-generation grid-displacement boundary, the estimated operational grid-emission displacement was 2,812.76 kg CO2/year, corresponding to 142.08 kg CO2/m2/year; at the assumed carbon price, its conditional monetized value was 219.40 USD/year. The results provide a boundary-aware year-round benchmark for PV/T-BIPV integration in nZEB-oriented test buildings under continental climatic conditions.

The multi‐phase free convection of nano‐encapsulated phase change material (PCM)/multi‐walled carbon nanotube (MWCNT)–Fe3O4–water mixture inside an inclined porous half‐annulus collector has been simulated using the Buongiorno's mathematical model. To create the buoyancy force, the internal Corrugated and outer cylinder wall are in the high constant hot flux as the hot wall and the constant temperature as the cold wall, respectively. The present study is investigated the effects of Rayleigh number (8 × 102 ≤ Ra ≤ 7 × 104), Darcy number (Da = 0.01 and 1), inclined angle (0 ≤ ζ ≤ 90), and nanoparticles fraction ratio (0 ≤ fr ≤ 1) on flow pattern, temperature contours, nanoparticles distribution contours, dimensionless distribution of nanoparticles on the walls, and Nusselt number. Results show that the distribution of MWCNT inside the collector and on the walls is more uniform than Fe3O4 nanoparticles. The replacement of MWCNT with PCM is also analyzed to provide the more desired performance.

This research explores the thermal behavior of a fan-induced airflow system within a heated, grooved channel cavity, employing a hybrid approach that combines numerical simulation along with machine learning. The external airflow stream enters into a bottom heated grooved channel cavity, where a fan-shaped rotating body is placed at the center of the grooved cavity. The grooved cavity is fitted with horizontal inlet/ outlet ports. The Light Gradient Boosting Machine (LightGBM) algorithm is used to carry out classification and regression tasks for predicting temperature profiles and identifying fan positions under various rotational speeds (S2 = 2, 4, 6) and directions (clockwise and counter-clockwise). The governing equations for mass, momentum, and energy are solved using a finite volume method via ANSYS Fluent. Temperature data are collected at three vertical fan placements (H* = 0.25, 0.50, and 0.75) to evaluate the influence of fan speed and position on thermal performance. The results show that the average Nusselt number (Nu), increased by 44.71 % at S2 = 6, H* = 0.25 (CCW), compared to the stationary fan case. At H = 0.50*, Nu was 18.18 % higher with CCW rotation compared to CW at the same fan speed. Energy output enhancement with fan rotation relative to the non-rotating case at H* = 0.5 energy output increased by 117.14 % (CW) and 165.71 % (CCW). This shows that counter-clockwise rotation consistently enhanced energy output more effectively. Placing the fan at mid-height (H* = 0.50) resulted in the most efficient thermal mixing, regardless of rotation direction. The LightGBM classification model achieved an accuracy of 98.99 %, with F1-scores reaching up to 99.54 %, while regression analysis yielded R2 values as high as 99.97 %, confirming the model's strong predictive performance. The study demonstrates how integrating data-driven machine learning techniques with conventional computational methods can offer valuable insights into optimizing thermal systems. These findings have direct applications in areas such as HVAC design, energy storage systems, and industrial cooling, where precise thermal control is essential.

An experimental study has been performed to evaluate the thermal behavior of a naturally ventilated double-skin fa & ccedil;ade (DSF) integrating semi-transparent monocrystalline PV modules on the east, south, and west fa & ccedil;ades of a nearly zero energy building in Elaz & imath;g, T & uuml;rkiye. The 18 cm cavity with seasonally operated vents was assessed through the cavity outdoor temperature difference (Delta T) and degree-hour integrals. In winter, the cavity remained 3-6 degrees C warmer, yielding 70-135 degrees C & sdot;h heating gains. Spring maintained positive Delta T, dominated by east morning and west afternoon effects. Summer showed limited cooling in June and net heating in July-August, while autumn regained strong heating (95-106 degrees C & sdot;h). Orientation analysis identified east-morning, southmidday, west-afternoon dominance. Accordingly, vents should remain closed in winter and open in summer. Annual results show 2.43 MWh on site electricity generation, 1.457 tCO2 avoided emissions, and USD 114 benefit, demonstrating the potential of orientation-aware fa & ccedil;ades for climate adaptive, energy efficient building design. Importantly, the presented field-measurement methodology and the degree-hour based performance indicators are not limited to the investigated building; they can be adopted to evaluate and optimize PVintegrated double-skin fa & ccedil;ades across different orientations, climates, and retrofit scenarios, enabling practical engineering applications in high-performance building design.

In this paper, 3D numerical study of hydrogen storage in a metal hydride (MH) reactor is examined by installing branching type fins and nano-enhanced MH. Titanium dioxide (TiO2) is utilized as the nanomaterial to be used as an additive in the MH domain in terms of heat transfer enhancement during the absorption. The metal hydride used for the reactor is MgH2 due its high volumetric and gravimetric capacity. Comparative studies are performed for the hydrogen storage system by using finite element-based solver. Number of branching fins used during the absorption is taken between 10and 50 while hydrogen supply pressure is considered between 5and 15atm. The coolant temperature is considered between 5 degrees C and 20 degrees C. Percentage weight (wt%) of absorbed hydrogen, and average MH bed temperatures are evaluated for each case and comparative results are presented. The best case is obtained with 50 branching fins, at supply pressure as 15atm and the coolant temperature as 5 degrees C. Using 50 fins results in reduction of sorption time by 22 % and the average bed temperature in the MH bed by 27 %. The coolant temperature at 5 degrees C reduces the absorption time and the average bed temperature by 34 % and 48 %, respectively, while the highest-pressure case results in a 48 % reduction in the sorption time compared to the lowest pressure case. Results are useful for the design and development of thermal management systems for hydrogen storage in MH reactors.

A novel cooling system for hydrogen storage in metal hydride bed is proposed. The system consists of hybrid nanofluid under magnetic field effects and cooling elliptic tubes. Numerical study of the coupled system is considered by using finite element method for different values of magnetic field (MGF) strength (Hartmann number (Ha) between 0 and 60), elliptic tube aspect ratio (AR between 1 and 2.5) and material of metal hydride (MH) bed. Cooling performance is improved by using elliptic tubes with higher aspect ratio and lower isothermal temperature. Higher MGF leads to higher heat transfer from the channel and effect becomes more profound with lower temperatures of the elliptic tubes. At temperature difference (dT) of 0 and 5, absorbed hydrogen amount improvement becomes 15

This study investigates natural convection in a triple-partitioned triangular cavity under the influence of an external magnetic field, considering both wavy (WP) and rotating (RP) partition configurations. The effects of key parameters-including the Rayleigh number (Ra), magnetic field strength (Hartmann number, Ha), magnetic field inclination (gamma), WP wave amplitude (A(f)) and wave number (N-f), and RP rotational speed (Omega) and size (R)-on the flow, thermal fields, and overall cooling performance-are analyzed by using finite volume method. Cooling performance of WP and RP is evaluated through an artificial neural network (ANN)-based approach. The flow and thermal fields in the triple-partitioned cavity are strongly affected by the partition type. WP produces the lowest average Nu, followed by circular-stationary and flat partitions, with reductions of 32%-8.6% at Ra=105, while RP enhances Nu by 138.4% and 38% at Ra=104 and Ra=105. At the highest Ha, Nu decreases by 1%-22% depending on the partition, with WP reducing Nu by 13%-31% and RP increasing it by 9%-71%. Increasing A(f) to 0.3 reduces Nu by 20.7% and 24.4% for N-f=1 and 3, while at A(f)=0.1, Nu is nearly unaffected by the wave number. At Omega=-100 and 100, Nu increases by 73.8% and 60.1%; a larger circular interface boosts Nu by 111.6% when rotating but lowers it by 11% when stationary. ANN predictions indicate that WP lowers Nu by 0.3% at low Ra and 36.9% at high Ra, whereas RP raises it by 76.8% and 36.1%.

The efficient use of photovoltaic (PV) modules or PV-integrated systems requires innovative cooling strategies. In order to increase the performance of double PV units in conjunction with thermoelectric generators (TEGs), this study aims to develop a unique L-shaped channel cooling system with pilot jets and sinusoidal wavy objects at the channel junctions. The channel flow Reynolds number (Reh between 100 and 500), pilot jet Reynolds number (Rej between 100 and 500), amplitude of corrugation (Amp between 0.01 and 0.15), and wave number of corrugation (Nw between 1 and 5) are all varied in the numerical simulations by using finite element method. The best cooling results for both horizontal channel (H-C) and vertical channel (V-C) are obtained when the wavy object is used at Reh = 500 with the greatest jet flow Re. The use of wavy objects results in additional PV-cell temperature drops of 4 degrees C and 2.5 degrees C for H-PV and V-PV using NF at the lowest Reh, and 4 degrees C and 1.7 degrees C at the maximum Reh. The V-PV cell temperature falls by 4 degrees C and 1.3 degrees C at Reh = 100 and Reh = 500 when Rej is raised from 100 to 500 using NF. Pilot jets and wavy objects are shown to significantly improve the cooling effectiveness of dual PV-TEG combination units. In comparison to the reference scenario (BF, no-object, no-jet at Reh = 100), temperature drops of 13.5 degrees C and 17.4 degrees C for H-PV and V-PV, respectively, are obtained when NF, wavy objects, and pilot-jets are used. The average surface temperature of each PV panel installed in H-C and V-C is estimated using ANFIS (Adaptive-Network Based Fuzzy Inference Systems) by adjusting the pilot jet flow Re and channel flow Re for configurations with and without a wavy object in the cooling channel. Outcomes are useful to develop and optimize efficient cooling solutions for multiple PV arrangements utilized in practice.

Efficient thermal energy storage is critical for improving the reliability of renewable energy systems. This study investigates the thermal enhancement of paraffin-based phase change materials using hybrid nanoparticles composed of multi-walled carbon nanotubes (MWCNTs) and aluminum oxide (Al2O3). MWCNT-Al2O3 hybrids were synthesized with mixing ratios of 20:80 to 80:20 and dispersed in paraffin at concentrations of 0.5-2.0 wt%. Thermal conductivity and thermal diffusivity of hybrid nanoparticle enhanced phase change material were measured at 27 degrees C. Results revealed that the maximum thermal conductivity enhancement of 36.1% was achieved at 2 wt% with an 80:20 hybrid ratio. Enhanced latent heat of crystallization and latent heat of melting was achieved at 1 wt% for the mixing ratio of 20:80 and 80:20 ratios, respectively. Thermogravimetric analysis confirmed the thermal stability of all samples. Furthermore, two machine learning models, Extreme Gradient Boosting (XGBoost) and Sequential Minimal Optimization (SMO) were developed to predict thermal conductivity and latent heat properties. Statistical analysis and Taylor diagrams demonstrated that XGBoost consistently outperformed SMO, achieving higher test coefficient of determination (R2) values (0.995 vs. 0.980 for thermal conductivity), lower mean squared error, and superior Kling-Gupta efficiency.

The implication of buoyancy reveals complex, non-linear behavior in numerous mass and thermal transfer processes. Quadratic thermal and solutal buoyancy effects become significant in high-temperature and high-concentration systems where density variations are strongly non-linear. Additionally, cross thermo-solutal buoyancy interactions play a crucial role in alloy solidification, geophysical flows, and oceanography, where the stability and density of the flow are influenced by the combined action of thermal and solutal gradients. This study aims to investigate quadratic thermal and solutal buoyancy effects in nanofluid convection through a vertical frustum of a cone, emphasizing the Soret phenomenon. The partial differential equations derived from the mathematical formulation using non-similarity variables were solved numerically. The governing non-similar partial differential equations are solved numerically using the bivariate pseudo-spectral local linearization method, which combines quasi-linearization and spectral collocation for enhanced accuracy. The results are compared to asymptotic series solutions in order to evaluate the reliability of the proposed methodology. Graphical results illustrate how quadratic and mixed buoyancy parameters influence momentum, heat, and mass transfer characteristics. The findings provide new insights into the physics of non-linear buoyancy-driven nanofluid convection in conical geometries relevant to thermal, geophysical, and chemical systems.

Improving the heat-storage capacity, thermal conductivity, and high-temperature stability of nitrate–nitrite molten salts is essential for increasing the efficiency and operational reliability of concentrating solar power (CSP) systems. Accordingly, the effects of hafnium diboride (HfB2) and silicon hexaboride (SiB6) nanoparticles on the phase integrity, microstructural evolution, and thermophysical performance of 7NaNO3–53KNO3–40NaNO2 molten salt were comparatively investigated. Nanocomposite formulations containing 0.5-2 wt% additive were synthesized and characterized by XRD, FT-IR, FE-SEM/EDX, DSC, TGA, and TPS measurements. The results confirmed that nanoparticle incorporation did not disrupt the fundamental nitrate-nitrite salt framework, and no evidence of newly formed phases or chemical bond generation was identified. Despite this structural and chemical preservation, marked improvements were obtained in heat-storage and heat-transfer characteristics. The melting enthalpy increased from 76.2 J/g for the reference salt to 89.4–90.6 J/g in the HfB2-containing compositions and to 92.6–95.6 J/g in the SiB6-containing compositions. The highest specific heat capacity enhancement in the HfB2 series occurred at 1.5 wt%, reaching 55.17% in the solid phase and 66.65% in the liquid phase, whereas the SiB6 series exhibited its optimum response at 1 wt% with corresponding increases of 107.78% and 117.79%. Thermal conductivity rose from 0.951 W/m.K to 1.689 W/m.K with 2 wt% HfB2 and 1.529 W/m.K with 2 wt% SiB6. The degradation onset temperature also increased from 612 °C to 653 °C and 663 °C, respectively. These findings demonstrate that boride-based nanoadditives provide an effective route for improving the thermophysical performance and high-temperature durability of this molten salt, with HfB2 delivering the strongest thermal conductivity enhancement and SiB6 providing the most pronounced improvements in latent heat, specific heat capacity, and thermal stability.

Photovoltaic (PV) panels require effective cooling strategies for performance improvement. In the present study, Z shaped nano-enriched cooling channel is used with different shaped vortex promoters (VPs) along with a porous corner partition for thermal management of three identical PV units combined with thermoelectric generator (TEG) modules. PV-TEG units are mounted on the different parts of the Z-channel (A unit-lower horizontal channel, B unit-vertical channel and C unit-upper horizontal channel) while porous partition used in the corner of the lower horizontal and vertical channel. Three dimensional numerical studies are conducted by using finite element method (FEM) for various values of channel flow Reynolds number (between 100 and 500), various VP shapes (I, L, T and L2-shaped) and nanoparticle loading in water (between 0 and 0.03). Combined use of porous partitions and VPs are influential in the flow field variation and cooling performance improvement in the Z-shaped channel. When compared to reference scenario of using no-VP and no partition at Re=100, using L2-shaped VP at Re=500 results in PV-cell temperature reduction of 8.8 degrees C and 12.9 degrees C for units A and C. When metal foam (MF) is mounted, the average cell temperature for unit B decreases by 4.1 degrees C at Re=100 and 2 degrees C at Re=500 compared to the scenario without VPs. The average cell temperature for unit B drops by 4.1 degrees C at Re=100 and 2 degrees C at Re=500 when porous partition is mounted in comparison to the case without VPs in the cooling channel. Using nanofluid (NF) leads to improved cooling performance in each sub-channels of the Z-shaped cooling system. When porous partition and L2-shaped VP are used in the cooling channel, employing NF at the maximum loading reduces the PV-cell temperature of A, B, and C PV units by 1.2 degrees C, 2 degrees C, and 2.3 degrees C compared to the case of utilizing only water. When corner partition is employed, the PV-cell temperature reductions between the best and worst cases of units A, B, and C are 10.1 degrees C, 12.1 degrees C, and 17 degrees C. The triple PV-TEG unit combined system performance with cooling channel is estimated using an autoencoder-based technique.

This study investigates magnetohydrodynamic natural convection of nanofluids in a square cavity subjected to sinusoidally varying thermal boundary conditions along the bottom wall. Understanding such flows is important for applications in thermal management, energy systems, and materials processing. The problem is solved using the lattice Boltzmann method coupled with an artificial neural network model to accelerate prediction of heat transfer responses. A comprehensive parametric analysis is performed for Rayleigh numbers up to 106, Hartmann numbers up to 40, nanoparticle concentrations up to 4%, and a range of thermal wavelength parameters. The results show that the oscillatory thermal boundary significantly modifies flow structures and heat transfer characteristics: for example, at Ra = 106 and tau=0.5, the average Nusselt number is enhanced by nearly 28% compared with uniform heating, while strong magnetic damping (Ha=40) reduces it by about 35%. The neural network model reproduces LBM results with prediction errors below 2%, offering rapid estimation of Nusselt numbers across the studied parameter space. The novelty of this work lies in combining a high-fidelity lattice Boltzmann solver with data-driven prediction to study magnetically controlled nanofluid convection under oscillatory heating, an area not previously addressed in the literature. These findings provide new insights into the manipulation of convective transport in multiphysics thermal systems.

This study presents a comprehensive numerical investigation of double-diffusive natural convection in a lid-driven square cavity filled with a nanoencapsulated phase change material (NEPCM) suspension. The cavity incorporates a central heated and solutally enriched square obstacle alongside a small adiabatic circular satellite obstacle. The present paper also proposes a comprehensive mathematical framework for the application of machine learning algorithms to predict concentration and temperature fields in flows characterized by varying Richardson numbers. The governing equations, formulated under the assumptions of laminar, incompressible, steady-state, Newtonian flow with the Boussinesq approximation, are solved using a Galerkin-based finite element method, enabling accurate treatment of complex geometries and boundary conditions. The study elucidates the effects of the satellite obstacle’s angular position, the Lewis number, the Reynolds number, and the magnetic field strength on the flow structure, temperature and concentration distributions, and overall transport characteristics. Quantitative analyses of dimensionless parameters, including the average Nusselt and Sherwood numbers as well as mean kinetic energy, provide insights into the interplay between convective and diffusive transport phenomena in NEPCM-laden fluids. The findings demonstrate that machine learning algorithms, particularly Random Forest, can attain nearly perfect predictive accuracy (R2 > 0.9996 for both concentration and temperature fields across a range of Richardson number regimes. Furthermore, the location of the satellite obstacle can be used as a control parameter for regulating the calculated values inside the cavity. The results reveal critical interactions between obstacle placement and flow behavior, offering design guidance for enhanced thermal and solutal management in advanced energy and microfluidic applications.

This manuscript delineates a thorough study on the heat and mass transfer phenomenon of the Williamson fluid flow embedded with a tetra-hybrid nanofluid evaluating a wide range of considered and physical effects such as magnetohydrodynamics (MHD), porous medium, radiative heat flux, Joule heating, Soret and Dufour effects, and a Stefan blowing parameter at the boundary, and the rest. A tetra-hybrid nanofluid containing nanofluid gold (Au), silver (Ag), titanium dioxide (TiO2), and aluminium oxide (Al2O3) is used for the improvement of significant thermal and mass transport characteristics. In the back, the demand for efficient thermal systems relates to sets with multiple, integrated transport mechanisms; however, the synergistic transport mechanisms have been largely unexplored, and the coupled hybrid advanced dimensions nanofluids have been unexplored in terms of their combined influences on these parameters. The core target was to examine the active relationships within the physical dynamics parameters while also evaluating the relative increases in the velocity, temperature, and concentration. This paper employs a robust computational approach to the study by solving the governing systems of non-linear ordinary differential equations using an appropriate method of similarity transformation and subsequent numerical techniques. The integration of artificial neural network (ANN) models within this spectrum for the first time, with predictions and optimization set for the outputs, adds a new dimension to this work. The data show that incorporating Soret and Dufour effects, along with the tetra-hybrid nanoparticles, markedly increases the Nusselt and Sherwood numbers, indicating improved heat and mass transfer rates. Furthermore, streamline plots are created to illustrate alterations in the flow structure induced by the Soret and Dufour parameters. This research makes valuable contributions to the development of refined cooling technologies, particularly in energy, chemical, and other process-oriented industries, highlighting the practical utility and innovation potential of the synergistic application of artificial neural networks alongside sophisticated nanofluid models.