Designing a solid–fluid interface layer and artificial neural network in a nanofluid flow due to rotating rough and porous disk

Journal of Thermal Analysis and Calorimetry, Год журнала: 2023, Номер 149(2), С. 867 - 878

Опубликована: Ноя. 21, 2023

Язык: Английский

---

Predicting the thermal distribution in a convective wavy fin using a novel training physics-informed neural network method

Scientific Reports, Год журнала: 2024, Номер 14(1)

Опубликована: Март 25, 2024

Abstract Fins are widely used in many industrial applications, including heat exchangers. They benefit from a relatively economical design cost, lightweight, and quite miniature. Thus, this study investigates the influence of wavy fin structure subjected to convective effects with internal generation. The thermal distribution, considered steady condition one dimension, is described by unique implementation physics-informed neural network (PINN) as part machine-learning intelligent strategies for analyzing transfer fin. This novel research explores use PINNs examine effect nonlinearity temperature equation boundary conditions altering hyperparameters architecture. non-linear ordinary differential (ODE) involved reduced into dimensionless form utilizing non-dimensional variables simplify problem. Furthermore, Runge–Kutta Fehlberg’s fourth–fifth order (RKF-45) approach implemented evaluate simplified equations numerically. To predict fin's properties, an advanced model created without using traditional data-driven approach, ability solve ODEs explicitly incorporating mean squared error-based loss function. obtained results divulge that increase conductivity variable upsurges distribution. In contrast, decrease profile caused due augmentation convective-conductive values.

Язык: Английский

---

Levenberg-Marquardt technique analysis of thermal and concentration storage in cone-disk apparatus with neural network-enhancement

Thermal Science and Engineering Progress, Год журнала: 2024, Номер 50, С. 102529 - 102529

Опубликована: Март 15, 2024

Язык: Английский

---

Design of Bayesian stochastic networks for numerical treatment of Williamson fluid stretching flow model with mixed convected heat generation

Numerical Heat Transfer Part B Fundamentals, Год журнала: 2024, Номер unknown, С. 1 - 24

Опубликована: Март 18, 2024

In the presented article, a stochastic network paradigm through Bayesian Regularization backpropagation neural (BRB-NN) is designed to interpret dynamics of Williamson fluid stretching flow model with mixed convected heat generation (WF-SFM). The governing nonlinear PDEs system WF-SFM reduced set ODEs by incorporating appropriate transformations. reference datasets for anticipated BRB-NN approach are created Adams solver numerical solutions varying material variable We, buoyancy factor λ, temperature characteristic parameter ε, thermal relaxation γ, source δ, and Prandtl numbers Pr. knacks artificial intelligence (AI) based procedure then employed on generated dataset WF-SFM. bias training, biased testing, validation conducted compute approximate results sundry scenarios, outputs in good agreement data that validate worthy performance proposed which further justified absolute error, mean squared error histogram illustrations regression measures. viable terms square (MSE) achieved at levels ranging from E−11 E−13, consistently all scenarios accuracy justification effectively proven low level MSE, optimal metric index as well distribution instances histograms negligible magnitude.

Язык: Английский

---

Deep neural network-based prediction and computational fluid dynamics analysis of convective heat transfer in dusty fluid flow over heated surface

Physics of Fluids, Год журнала: 2025, Номер 37(2)

Опубликована: Фев. 1, 2025

Due to the necessity of creating customized nanomaterial designs, advanced coating processes have been developed on surface an isothermal sphere. To achieve accuracy in these processes, it is necessary understand thermodynamic behavior, rheology materials, and chemical reactions involved. The novelty present research analyzing heat transfer dusty fluid flow over a heated cross section cylinder. We develop mathematical model including factors for both phases demonstrate how affect field. Thermal radiation propagation simulated using Rosseland's flux diffusion model. Equations describing transport magnetic body force are expressed Cartesian coordinate system. A set boundary conditions has established momentum, thermal energy, concentration particles formulate conservation equations Using nonsimilar transform, nonlinear partial differential (PDEs) transformed into dimensionless PDEs. PDSolve technique, PDEs solved. significant factor engineering calculating velocity distributions, mass, transfers, which were results this research. calculation carried out MAPLE 2024, computational software tool. deep neural network program was designed, emphasizes machine learning predicting nature physical phenomena applications. As part validation process proposed research, some statistical metrics taken assess degree error between true values anticipated values. Based presented, presented approach most efficient method predict quantities These therefore recommended development industrial device setups.

Язык: Английский

---

Evolutionary Computing for the Radiative–Convective Heat Transfer of a Wetted Wavy Fin Using a Genetic Algorithm-Based Neural Network

Biomimetics, Год журнала: 2023, Номер 8(8), С. 574 - 574

Опубликована: Дек. 1, 2023

Evolutionary algorithms are a large class of optimization techniques inspired by the ideas natural selection, and can be employed to address challenging problems. These iteratively evolve populations using crossover, which combines genetic information from two parent solutions, mutation, adds random changes. This iterative process tends produce effective solutions. Inspired this, current study presents results thermal variation on surface wetted wavy fin algorithm in context parameter estimation for artificial neural network models. The physical features convective radiative heat transfer during wet conditions also considered develop model. highly nonlinear governing ordinary differential equation proposed problem is transmuted into dimensionless equation. graphical outcomes aspects profile demonstrated specific non-dimensional variables. primary observation decrease temperature with rise parameters convective-conductive parameters. implemented offers powerful technique that effectively tune network, leading an enhanced predictive accuracy convergence numerically obtained solution.

Язык: Английский

---

Analysis of Heat Transfer Behavior of Porous Wavy Fin with Radiation and Convection by Using a Machine Learning Technique

Symmetry, Год журнала: 2023, Номер 15(8), С. 1601 - 1601

Опубликована: Авг. 18, 2023

The impact of convection and radiation on the thermal distribution wavy porous fin is examined in present study. A hybrid model that combines differential evolution (DE) algorithm with an artificial neural network (ANN) proposed for predicting heat transfer fin. equation representing variation reduced to its dimensionless arrangement numerically solved using Rung, e-Kutta-Fehlberg’s fourth-fifth order method (RKF-45). study demonstrates effectiveness this model, results indicate approach outperforms ANN parameters obtained through grid search (GS), showcasing superiority DE-ANN terms accuracy performance. This research highlights potential utilizing DE improved predictive modeling sector. originality it addresses problem by optimizing selection algorithm.

Язык: Английский

---

Energy consumption prediction in an office building by examining occupancy rates and weather parameters using the moving average method and artificial neural network

Heliyon, Год журнала: 2024, Номер 10(4), С. e25307 - e25307

Опубликована: Фев. 1, 2024

Occupancy rate refers to the level of usage and presence individuals within a building or specific space. This factor can have significant impact on energy consumption. When occupancy in is high, naturally, consumption also increases. correlation might be due increased use lighting, heating, cooling, higher numbers electrical electronic devices, similar factors associated with people building. One modern methods field involves empirically utilizing monitoring tools buildings analyzing relationship between such utilization through artificial neural network tools. In this research, camera sensitive entry exit was installed at entrance an office Tehran, Iran. By doing so, accurately monitored. next stage, by investigating building's consumption, amount predicted using statistical method (moving average). The results indicate errors 9.8 4.5 for respective methods, highlighting that yields most accurate outcomes. Moreover, study's findings suggest direct correlation: as rates increase, values rise.

Язык: Английский

---

Heat transfer analysis in magnetohydrodynamic nanofluid flow induced by a rotating rough disk with non-Fourier heat flux: aspects of modified Maxwell–Bruggeman and Krieger–Dougherty models

Nanoscale Advances, Год журнала: 2023, Номер 5(21), С. 5941 - 5951

Опубликована: Янв. 1, 2023

Non-Newtonian fluids have unique heat transfer properties compared to Newtonian fluids. The present study examines the flow of a Maxwell nanofluid across rotating rough disk under effect magnetic field. Furthermore, Cattaneo-Christov flux model is adopted explore transport features. In addition, comparison fluid without and with aggregation performed. Using similarity variables, governing partial differential equations are transformed into system ordinary equations, this then solved by employing Runge-Kutta Fehlberg fourth-fifth order method obtain numerical solution. Graphical depictions used examine notable effects various parameters on velocity thermal profiles. results reveal that an increase in value Deborah number decreases profile. An relaxation time parameter artificial neural network employed calculate rate surface drag force. R values for skin friction Nusselt were computed. demonstrate networks accurately predicted values.

Язык: Английский

---

A neural network approach to modeling magnetohydrodynamic stagnation point Ree-Eyring flow over a convectively heated stretched surface

International Journal of Modelling and Simulation, Год журнала: 2024, Номер unknown, С. 1 - 14

Опубликована: Апрель 7, 2024

This study investigates the application of artificial intelligence (AI) in fluid dynamics, mainly using a neural network trained by Levenberg – Marquardt method (NN-BLMM), to model magnetohydrodynamic (MHD) stagnation point Ree Eyring flow. We focus on this flow over convectively heated stretched surface integrating Cattaneo Christov heat model. The initial complex nonlinear partial differential equations (PDEs) are transformed into ordinary (ODEs) suitable similarity variables. A dataset was generated Lobatto IIIA numerical solver analyze effects various and thermal parameters. NN-BLMM then rigorously evaluated through training, testing, validation phases compared with reference data. ensures model's precision effectiveness. observe that an increase Powell parameter notably reduces fluid's shear resistance, implying decrease viscosity. Concurrently, transfer rate within medium increases internal generation parameter. These findings highlight robustness simulating emphasizing AI's potential provide deeper understanding non-Newtonian behaviors. research has important implications for industrial applications which precise control properties is crucial.

Язык: Английский

---