Intelligent prediction of thermodynamic performance in MHD Oldroyd B trihybrid nanofluids using artificial neural networks Authors: Mamoon Aamir, Chemseddine Maatki, Aqsa Zafar Abbasi, Rajab Alsayegh, Karim Kriaa, Walid Hassen, Lioua Kolsi Published: 2026 ABSTRACT This research aims to describe the flow characteristics and entropy creation of an Oldroyd-B tri hybrid nanofluid under conditions of MHD with heat transfer through a hybrid numerical—machine learning framework. The nonlinear boundary layer governing equations for momentum and heat transport are transformed to a commonly used ordinary differential equations (ODE) form via similarity transformations and solved using MATLAB's bvp4c solver. A mesh independence study has verified numerical. The main innovation of this study is combining ANN modelling with the nonlinear numerical calculation of Oldroyd-B tri hybrid nanofluid flow to create an accurate predictive surrogate modelling tool for complex thermofluid systems. Data collected from the bvp4c solver was then used to train a feed forward ANN, using Levenberg–Marquardt backpropagation algorithm. The trained ANN was able to accurately predict velocity, temperature and entropy creation, with regression accuracies above 0.999 and mean square error values less than \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-6}$$\end{document}, indicating a high degree of predictive power. Parametric analysis indicated that increasing the magnetic parameter caused a large reduction in velocity field (approximately 18–25%) as a result of the Lorentz force acting in against the flow. Also, the generation of heat increased the temperature profile by approximately 20%, therefore increasing Entropy Generation within the thermal boundary layer. In addition, tri-hybrid nanoparticles have better thermal conductivity and heat transfer performance than nanofluids made from conventional materials because of their ability to improve the thermal performance of fluids. A predictive framework for ANN-based modelling of nonlinear fluid transport has been developed that reduces the computational cost of obtaining accurate numerical solutions compared with traditional methods. This new framework has the potential to allow engineers and scientists to model advanced nanofluids with greater accuracy than previous approaches, thereby providing valuable information for thermal management, high-performance cooling systems, and energy conversion technologies. Introduction Effective management of thermal energy is a central function in numerous industrial and engineering applications, such as high-performance cooling systems, chemical reactors, microelectronic devices, heat exchangers, biomedical procedures, and radiators. Among the numerous solutions that have been devised for enhancing thermal performance, the use of nanofluids, base fluids containing nanoscale particles, has become of significant interest as a result of their improved thermal conductivity and better heat transport properties. The enhanced thermophysical characteristics of nanofluids are directly attributed to the high thermal conductivity and surface area to volume ratio of nanoparticles, resulting in enhanced convective heat transfer rates compared with traditional fluids. Due to this advantage, nanofluids have been widely researched in various thermal systems including drug delivery, chemical processing, thermal control, and energy conversion technologies. There is a considerable amount of literature that has investigated the impact of nanoparticles on heat transfer enhancement. Ellahi et al. [1] considered the role of nanoparticles on magnetic field under thermal transport, whereas Zeeshan et al. [2] maximized nanoparticle-included fluid performance in magnetically affected environments. Zeeshan et al. [3] also investigated the effect of hybrid nanoparticles on combined convective heat transfer in Casson fluids and Ellahi et al. [4] showed improved heat transfer by using nanoparticle suspensions. Tlili [5] examined flow and thermal properties in eccentric annuli for internal heat generation using Al2O3 and Ni nanoparticles in an ethylene glycol base fluid. Algehyne et al. [6] conducted numerical simulations of hybrid nanofluid flow along a vertical permeable sheet, and Hassan et al. [7] investigated MHD flow of hybrid nanofluids over a bidirectional stretching sheet with Ag and MoS2 nanoparticles using various base fluids (EG, EO, and water), and they concluded that Ag–MoS2/water was the most efficient in terms of heat transfer. Arif et al. [8] utilized the finite element method (FEM) for investigating thermal enhancement in Williamson fluids with mono (CuO–blood) and hybrid (CuO–Al2O3–blood) suspensions, and Ali et al. [9] found that hybrid nanoparticles in a porous medium enhanced shear stress but decreased the heat transfer rate. Further research [10–13] has uniformly proved the better thermal behavior of nanoparticle suspensions in different fluid systems. The internal heat generation and Joule (Ohmic) heating further affect thermal transport in electrically conducting fluids, especially in the presence of a magnetic field. Ohmic dissipation and volumetric heat generation can have a major effect on temperature distribution, influence boundary layer formation, and lower the efficiency of the system if not appropriately addressed. A. Few studies have been performed to investigate the effects of these phenomena: Shamshuddin et al. [14] investigated steady nanofluid flow along a vertical surface with internal heating and Ohmic dissipation, and Ajibade et al. [15] presented analytical solutions for viscous MHD flows with internal heat generation. Artificial neural network framework Artificial neural networks (ANNs) have emerged as powerful computational tools for solving highly nonlinear differentiable equations associated with complex transport processes in thermal-fluid systems. Since ANNs represent nonlinear mappings, have the ability to learn dynamic behaviors, and are able to capture complex parameter dependencies, they are often used in modeling multi-physics problems wherein traditional numerical methods struggle with convergence or stability challenges. Recently, ANNs have successfully been implemented to model fluid dynamics and heat transfer processes including non-Newtonian rheology, magnetohydrodynamic (MHD) effects, and coupled thermal–electromagnetic fields [16, 17]. Neural network models using optimization methods, including Levenberg–Marquardt and scaled conjugate gradient methods, have shown high accuracy for modeling flow characteristics, entropy generation, and heat transfer rates of nanofluids. Gupta and Wakif [18] are examining heat and mass transfer in the flow of nanofluid between two discs using neural networks. An innovative feature of this research is the use of the Levenberg–Marquardt artificial neural network (ANN) to analyse the training, testing, and validation procedures. The constructed ANN model's validity was shown by its accuracy throughout the training, validation, and testing stages. Gupta [19] examined the mass and heat transport of a ternary hybrid nanofluid between two parallel plates with an angled magnetic field using an artificial neural network. The effects of several physical factors on temperature, concentration, and velocity are discussed and shown graphically. This work builds off this modeling framework to examine the thermohydrodynamics of an Oldroyd-B tri-hybrid nanofluid under the influences of Joule heating, heat generation, and magnetic fields. The integrated intelligent solver and non-Newtonian transport model enables improvements for predictive capability and computational efficiency, enhances understanding of entropy generation, and develops high-performance thermal devices. In spite of the huge advancements in nanofluid and non-Newtonian fluid research, available literature has mostly focused on mono- and bi-nanoparticle suspensions. Much fewer efforts have gone towards tri-hybrid nanofluids, particularly under magnetohydrodynamic (MHD) conditions in conjunction with Joule heating and internal heat generation—factors that considerably influence entropy production and thermal performance. Additionally, the recent development of artificial neural networks (ANNs) has provided new prospects for more efficient solution of highly nonlinear differential equations governing transport phenomena with enhanced accuracy and computational efficiency. The merging of ANN-based solvers with multi-physics flow models offers a strong platform for the precise prediction of velocity, temperature, and entropy fields, thus facilitating the design of next-generation heat transfer systems. Driven by these shortcomings, this paper examines the thermohydrodynamic behavior attributes of an Oldroyd-B viscoelastic fluid with tri-hybrid nanoparticles (Al2O3,TiO2, and SiO2) in an ethylene glycol base fluid utilizing intelligence computing. The example of Oldroyd-B viscoelastic fluid is given in Fig. 1. The model takes into consideration the effects of combined Joule heating, internal heat generation, as well as an externally imposed magnetic field. The governing equations were derived based on boundary-layer approximations, solved numerically, and supplemented with an ANN-based prediction framework.Fig. 1Application of Oldroyd-B fluids Application of Oldroyd-B fluids There has been significant research into using artificial intelligence (AI), machine learning (ML) and other computational techniques to examine multiphase systems involving complex nanofluid or magneto hydrodynamic (MHD) behavior over the last several years. Machine learning frameworks such as Tensor Flow have been used for the analysis of thermal and entropy generation from nanofluids [20]; hybrid models such as response surface methodology (RSM) & artificial neural networks (ANN) combined with sensitivity analysis have been used for improving heat transfer and energy efficiency in parabolic trough solar collectors and aerospace systems [21, 22]; and stochastic AI methods have been utilized to propel MHD fluids with peristaltic motion and Reiner-Philippoff fluids [23], develop ANN-based models of two-dimensional magnetized nanofluid flow [24] and optimize the Thermophysical properties of radially magnetized hybrid nanofluids [25]. AI driven methods have also been applied to electromagnetically driven Casson hybrid nanofluid flow [26], bio convective nanofluid flows with gyrotactic microorganisms [27] and the thermodynamic effects of aggregated nanoparticles [28]; additionally, sensitivity analyses were performed on hypersonic glide vehicle propulsion systems subject to magnetic fields and nanofluid effects [29] and advanced cooling techniques for aircraft propulsion systems using Carreau trihybrid nanofluids [30]. A range of complementary numerical and optimization methods have been developed; for example, there is research into porous cavity and porous block entropy generation [31] and machine learning for both dust and bio-magnetic hybrid nano-fluids [32–34]. There have also been studies of semi-analytical rheological methods applied to biomedical flows [35] along with examples of magnetic dipole and radiative effects in permeable materials [36]. All of the above studies contribute to the development of a multi-layer neural network model for mixed convection MHD Prandtl nano-fluids containing bio-logical microorganisms [37]. These studies demonstrate that the combination of advanced numerical models with machine learning models and thermofluid dynamics provides a powerful framework for improved prediction accuracy and optimization and energy-efficient design of thermal systems. Tri-hybrid nanofluids have been a focus of much research in recent years because adding three types of nanoparticles to a base fluid will increase Thermophysical properties more than adding just one type or combining two types of nanoparticle together (hybrid). In fact, because of their combined effect, multiple nanoparticles have been shown to increase thermal conductivity, heat transport capabilities and thermal stability compared to either mono or hybrid nanofluids. Thus, tri-hybrid nanofluids are becoming more common to find in modern thermal systems and energy engineering applications. In this paper, we use the tri-hybrid Thermophysical model developed in the paper referenced [38] as the basis for our study and demonstrate the authors’ derivation and validation on the governing equations for tri-nanofluids. This reference study looks at how to physically model tri-hybrid nanofluids whereas our study uses a different method to study the properties of such complex nanofluid systems by testing whether we could use machine learning technologies to accurately predict their thermodynamic behavior. To do this we used an artificial neural network trained by Levenberg–Marquardt backpropagation algorithm to approximate the numerical solutions of the velocity, temperature and entropy generation profiles of the Oldroyd-B tri-hybrid nanofluid model. Research questions How do the magnetic interaction parameter (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M$$\end{document}) and heat generation/absorption parameter (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda $$\end{document}) influence velocity, and temperature in tri-hybrid Oldroyd-B nanofluids?What are the opposing impacts of relaxation Deborah number \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_1$$\end{document} and retardation Deborah number \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document} on the structure of viscoelastic flow and heat transfer mechanism?How does the simultaneous variation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M, \lambda , \beta_1,$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document} control the Nusselt number, skin friction, and entropy generation rate?Can an ANN model be able to predict accurately the nonlinear behavior of magnetic, thermal, and viscoelastic parameters in tri-hybrid non-Newtonian fluids? How do the magnetic interaction parameter (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M$$\end{document}) and heat generation/absorption parameter (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda $$\end{document}) influence velocity, and temperature in tri-hybrid Oldroyd-B nanofluids? How do the magnetic interaction parameter (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M$$\end{document}) and heat generation/absorption parameter (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda $$\end{document}) influence velocity, and temperature in tri-hybrid Oldroyd-B nanofluids? What are the opposing impacts of relaxation Deborah number \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_1$$\end{document} and retardation Deborah number \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document} on the structure of viscoelastic flow and heat transfer mechanism? What are the opposing impacts of relaxation Deborah number \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_1$$\end{document} and retardation Deborah number \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document} on the structure of viscoelastic flow and heat transfer mechanism? How does the simultaneous variation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M, \lambda , \beta_1,$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document} control the Nusselt number, skin friction, and entropy generation rate? How does the simultaneous variation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M, \lambda , \beta_1,$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document} control the Nusselt number, skin friction, and entropy generation rate? Can an ANN model be able to predict accurately the nonlinear behavior of magnetic, thermal, and viscoelastic parameters in tri-hybrid non-Newtonian fluids? Can an ANN model be able to predict accurately the nonlinear behavior of magnetic, thermal, and viscoelastic parameters in tri-hybrid non-Newtonian fluids? Novelty and research contribution With the increase in demand for high-performance heat management systems in aerospace, microelectronic cooling, and renewable energy technologies, intelligent predictive models with high accuracy are required to capture the nonlinear thermohydrodynamic behavior of viscoelastic nanofluids. It is essential to understand the interaction among \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M, \lambda , {\beta}_{1},$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document} to reduce entropy generation and enhance energy transfer in sophisticated magnetic and thermal scenarios. The main contributions of this work are:Intelligent modeling innovation: Presentation of a leading ANN-based hybrid solver on the Levenberg–Marquardt algorithm, with ultra-low MSE (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-11}-{10}^{-9}$$\end{document}) and stable convergence for nonlinear viscoelastic MHD systems.Unique parameter focus: Exhaustive examination of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M, \lambda , {\beta}_{1},$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document}, showing their individual and collective effects velocity, and temperature profiles — \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{1}$$\end{document} reducing and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document} augmenting velocity and temperature.Tri-hybrid nanofluid innovation: Development of an Oldroyd-B \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Al_2O_3-TiO_2-SiO_2/EG$$\end{document} nanofluid model with MHD and internal heat generation for the very first time. Intelligent modeling innovation: Presentation of a leading ANN-based hybrid solver on the Levenberg–Marquardt algorithm, with ultra-low MSE (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-11}-{10}^{-9}$$\end{document}) and stable convergence for nonlinear viscoelastic MHD systems. Intelligent modeling innovation: Presentation of a leading ANN-based hybrid solver on the Levenberg–Marquardt algorithm, with ultra-low MSE (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-11}-{10}^{-9}$$\end{document}) and stable convergence for nonlinear viscoelastic MHD systems. Unique parameter focus: Exhaustive examination of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M, \lambda , {\beta}_{1},$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document}, showing their individual and collective effects velocity, and temperature profiles — \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{1}$$\end{document} reducing and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document} augmenting velocity and temperature. Unique parameter focus: Exhaustive examination of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M, \lambda , {\beta}_{1},$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document}, showing their individual and collective effects velocity, and temperature profiles — \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{1}$$\end{document} reducing and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document} augmenting velocity and temperature. Tri-hybrid nanofluid innovation: Development of an Oldroyd-B \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Al_2O_3-TiO_2-SiO_2/EG$$\end{document} nanofluid model with MHD and internal heat generation for the very first time. Tri-hybrid nanofluid innovation: Development of an Oldroyd-B \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Al_2O_3-TiO_2-SiO_2/EG$$\end{document} nanofluid model with MHD and internal heat generation for the very first time. Problem formulating In this study, we investigate the entropy generation characteristics of an Oldroyd-B viscoelastic fluid, where the base fluid is ethylene glycol \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$({C}_{2}{H}_{6}{O}_{2}),$$\end{document} mixed with various nanoparticles and flowing over a non-uniformly moving surface. The base fluid is assumed to have internal volumetric heat generation, and when subjected to an external magnetic field, it experiences Ohmic (Joule) dissipation. Three types of nanoparticles, alumina (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Al_2O_3$$\end{document}), titania (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$TiO_2$$\end{document}), and silica (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$SiO_2$$\end{document}) are uniformly dispersed into the base fluid to form a homogeneous tri-hybrid nanofluid. The wall temperature is considered non-uniform and varies quadratically with the horizontal coordinate, while the ambient temperature is maintained constant. The surrounding fluid far from the wall is assumed to be quiescent. A schematic representation of the physical configuration is illustrated in Fig. 2.Fig. 2Schematic diagram [38] Schematic diagram [38] To make the mathematical description simpler, the governing partial differential equations (PDEs) are formulated under classical boundary-layer approximations, collapsing them to a lower-dimensional set of equations [38].1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{\partial u}{\partial x}+\frac{\partial v}{\partial y}=0,$$\end{document}2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&u\frac{\partial u}{\partial x}+v\frac{\partial u}{\partial y}+{\gamma}_{1}\left[{u}^{2}\frac{{\partial }^{2}u}{\partial {x}^{2}}+{v}^{2}\frac{{\partial }^{2}u}{\partial {y}^{2}}+2uv\frac{{\partial }^{2}u}{\partial x\partial y}\right]={v}_{thnf}\frac{{\partial }^{2}u}{\partial {y}^{2}}\\ &+{v}_{thnf} {\gamma}_{2}\left[u\frac{{\partial }^{3}u}{\partial x{y}^{2}}+v\frac{{\partial }^{3u}}{\partial {y}^{3}}-\frac{\partial u}{\partial x}\frac{{\partial }^{2}u}{\partial {y}^{2}}-\frac{\partial u}{\partial y}\frac{{\partial }^{2}v}{\partial {y}^{2}}-\frac{{\sigma}_{thnf}}{{\rho}_{thnf}} {B}_{0}^{2}\left(u+{\gamma}_{1}v\frac{\partial u}{\partial y}\right)\right] ,\end{aligned}$$\end{document}3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$u\frac{\partial T}{\partial x}+v\frac{\partial T}{\partial y}={\left(\frac{1}{\rho {C}_{p}}\right)}_{thnf}[{k}_{thnf}\frac{{\partial }^{2 }T}{\partial {y}^{2}}+{\mu}_{thnf}\frac{{\gamma}_{2}}{{\gamma}_{1}}{\left(\frac{\partial u}{\partial y}\right)}^{2}+{\sigma}_{thnf}{B}_{0}^{2}{u}^{2}+{Q}_{0}(T-{T}_{b})]$$\end{document} The mathematical model that governs the analysis at hand centers on the conservation of mass, momentum, and energy for an incompressible, steady, two-dimensional Oldroyd-B viscoelastic nanofluid composed of ethylene glycol (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${C}_{2}{H}_{6}{O}_{2}$$\end{document}) as its base fluid with the homogeneous suspension of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Al_2O_3, TiO_2,$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$SiO_2$$\end{document} nanoparticles forming a tri-hybrid nanofluid with increased thermophysical properties. The flow is considered within the confines of a non-uniformly moving surface that experienced a transverse magnetic field that gives rise to Lorentz forces and Joule (Ohmic) heating and an internal volumetric heat generation would likewise affect the thermal effects. The governing equations are derived under the assumptions of thin boundary layers to govern the effects of the flow and heat transfer phenomena. The continuity equation (Eq. 1) ensures conservation of mass through the condition of a divergence-free velocity field. The momentum equation (Eq. 2) includes the inertia of the fluid, viscous diffusion and viscoelastic effects related to relaxation and retardation times as well as electromagnetic body forces produced by the application of the magnetic field, thereby accounting for the nonlinear viscous response along with the viscoelastic stress and magnetohydrodynamic forces on the boundary layer. In the momentum equation for the magnetohydrodynamic (MHD) model, the magnetic interaction term indicates how the Lorentz force acts on the fluid due to its conductivity and an external magnetic field applied to it. In this paper, the assumption that the magnetic Reynolds number is small enough that can be neglible to the induced magnetic field from the fluid and only consider the applied magnetic field [38]. The energy equation (Eq. 3) governs the thermal transport process, incorporating the effects of heat conduction, viscous dissipation, Joule heating, and internal heat generation. Following boundary conditions are implemented for numerical solutions of modelled problems [38]:4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&u={U}_{w}={u}_{0}x , \,\,v={V}_{w} ,\,\, T={T}_{w}={T}_{b}+{c}_{0}{x}^{2}\,\,at\,\, y=0\\ &u\to 0,\,\, T\to {T}_{b},\frac{\partial u}{\partial y}\to 0\,\, at \,\,y\to \infty .\end{aligned}$$\end{document} The relaxation time, retardation time, and heat production or absorption coefficient are denoted by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\gamma}_{1}, {\gamma}_{2},$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${Q}_{0}$$\end{document}, respectively. The transformations allow PDEs to be converted into ODEs [38].5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta =y\frac{\sqrt{{u}_{0}}}{{v}_{bf}} ,\,\, u={u}_{0}x{f}^{\prime}\left(\eta \right), \,\,v= -\sqrt{{u}_{0}{v}_{bf}f\left(\eta \right)}, \theta \left(\eta \right)=\left(T-\frac{{T}_{b}}{{T}_{w}-{T}_{b}}\right)$$\end{document} Hence, extensive calculations lead to [38]6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{{A}_{1}}{{A}_{0}}\left[{f}^{{\prime \prime \prime}}+{\beta}_{2}\left({f}^{{\prime \prime}2}-f{f}^{iv}\right)\right]+\left(1+{M}^{2}{\beta}_{1}\right)f{f}^{{\prime \prime}}-{f}^{{\prime}2}-{\beta}_{1}\left({f}^{2}{f}^{{\prime \prime \prime}}-2f{f}^{\prime}f{f}^{{\prime \prime}}\right)-\frac{{A}_{2}}{{A}_{0}} {M}^{2}{f}^{\prime}=0$$\end{document}7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{{A}_{4}}{{A}_{3}} {\theta }^{{\prime \prime}}+\mathrm{Pr}\left[\frac{{A}_{1}}{{A}_{3}}\frac{{\beta}_{2}}{{\beta}_{1}} Ec{f}^{{\prime \prime}2}+f{\theta }^{\prime}+\frac{{A}_{2}}{{A}_{3}} Ec{M}^{2}{f}^{{\prime}2}-2\theta {f}^{\prime}+\frac{\lambda }{{A}_{3}} \theta \right]=0$$\end{document} With the following streamlined BCs [38]8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f\left(\eta \right)=S=-\frac{{V}_{w}}{\sqrt{{u}_{0}{v}_{bf}}} , {f}^{\prime}\left(\eta \right)=1, \theta \left(\eta \right)=1, at \eta =0$$\end{document}9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${f}^{\prime}\left(\eta \right)=0, {f}^{{\prime \prime}}\left(\eta \right)=0, \theta \left(\eta \right)=0 as \eta \to \infty $$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${A}_{0}=\frac{{\rho}_{thnf}}{{\rho}_{bf}} , {A}_{1}=\frac{{\mu}_{thnf}}{{\mu}_{bf}} , {A}_{2}=\frac{{\sigma}_{thnf}}{{\sigma}_{bf}}, {A}_{3}=\frac{{\left(\rho {C}_{p}\right)}_{thnf}}{{\left(\rho {C}_{p}\right)}_{bf}}$$\end{document} and\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${A}_{4}=\frac{{k}_{thnf}}{{f}_{bf}}$$\end{document}. Here, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varphi $$\end{document} represents the volume fraction of nanoparticles. also the subscript \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$thnf$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$bf$$\end{document} are represented for nanomaterials and base fluid. The Deborah quantities of relaxation and retardation time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{1}={u}_{0}{\gamma}_{1}$$\end{document} and\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}={u}_{0}{\gamma}_{2}$$\end{document}, the Eckert number \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{c}=\frac{{v}_{0}^{2}}{{c}_{0}{\left({C}_{p}\right)}_{bf}}$$\end{document}, the Prandtl number \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Pr=\frac{{v}_{bf}}{{\alpha}_{bf}}$$\end{document} and the heat generation /absorption parameter \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda =\frac{{Q}_{0}}{{\left(\rho {C}_{p}\right)}_{bf}{u}_{0}}$$\end{document} the suction parameter \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S= -\frac{{v}_{w}}{\sqrt{{u}_{0}{v}_{bf}}}$$\end{document} and\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${M}^{2}= \frac{{\sigma}_{bf}{B}_{0}^{2}}{{u}_{0}{\rho}_{bf}}$$\end{document}. For the Thermophysical properties of tri-nanoparticles and their numeric value refers to [38], presented in Tables 1 and 2.Table 1Thermophysical Properties of base fluid and nanofluidsThermophysical PropertyFormulaThermal Conductivity\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{nf}={K}_{bf}\left[\frac{{K}_{3}+2{K}_{bf}-2{\phi}_{3}\left({K}_{bf}-{K}_{3}\right)}{{K}_{3}+2{K}_{bf}+{\phi}_{3}\left({K}_{bf}-{K}_{3}\right)}\right]$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{hnf}={K}_{nf}\left[\frac{{K}_{2}+2{K}_{nf}-2{\phi}_{2}\left({K}_{nf}-{K}_{2}\right)}{{K}_{2}+2{K}_{nf}+{\phi}_{2}\left({K}_{nf}-{K}_{2}\right)}\right]$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{thnf}={K}_{hnf}\left[\frac{{K}_{1}+2{K}_{hnf}-2{\phi}_{1}({K}_{hnf}-{K}_{1})}{{K}_{1}+2{K}_{hnf}+{\phi}_{1}({K}_{hf}-{K}_{1})}\right]$$\end{document}Viscosity\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mu}_{thnf}=\frac{{\mu}_{bf}}{{\left(1-{\phi}_{3}\right)}^{2.5}{\left(1-{\phi}_{2}\right)}^{2.5}{\left(1-{\phi}_{1}\right)}^{2.5}}$$\end{document}Electric Conductivity\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\sigma}_{nf}={\sigma}_{bf}\left[\frac{{\sigma}_{3}\left(1+2{\phi}_{3}\right)+2{\sigma}_{bf}\left(1-{\phi}_{3}\right)}{{\sigma}_{3}\left(1-{\phi}_{3}\right)+{\sigma}_{bf}\left(2+{\phi}_{3}\right)}\right]$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\sigma}_{hnf}={\sigma}_{nf}\left[\frac{{\sigma}_{2}\left(1+2{\phi}_{2}\right)+2{\sigma}_{nf}\left(1-{\phi}_{2}\right)}{{\sigma}_{2}\left(1-{\phi}_{2}\right)+{\sigma}_{nf}\left(2+{\phi}_{2}\right)}\right]$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\sigma}_{thnf}={\sigma}_{hnf}\left[\frac{{\sigma}_{1}\left(1+2{\phi}_{1}\right)+2{\sigma}_{hnf}(1-{\phi}_{1})}{{\sigma}_{1}\left(1-{\phi}_{1}\right)+{\sigma}_{hnf}(2+{\phi}_{1})}\right]$$\end{document}Heat Capacitance\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\left(\rho {C}_{p}\right)}_{thnf}=\left(1-{\phi}_{3}\right)\left[\left(1-{\phi}_{2}\right)\left(\left(1-{\phi}_{1}\right){\left(\rho {C}_{p}\right)}_{bf}+{\phi}_{1}{\left(\rho {C}_{p}\right)}_{1}\right)+{\phi}_{2}{\left(\rho {C}_{p}\right)}_{2}\right]+{\phi}_{3}{\left(\rho {C}_{p}\right)}_{3}$$\end{document}Density\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rho}_{thnf}=\left(1-{\phi}_{3}\right)\left[\left(1-{\phi}_{2}\right)\left(\left(1-{\phi}_{1}\right){\rho}_{bf}+{\phi}_{1}{\rho}_{1}\right)+{\phi}_{2}{\rho}_{2}\right]+{\phi}_{3}{\rho}_{3}$$\end{document}Table 2Thermal propertiesPhysical property\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${C}_{2}{H}_{6}{O}_{2}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$A{l}_{2}{O}_{3}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Ti{O}_{2}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Si{O}_{2}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${C}_{p} / {\mathrm{Jkg}}^{-1}{\mathrm{K}}^{-1}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2415$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$765$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$686.2$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$745$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$k / {\mathrm{Wm}}^{-1}{\mathrm{K}}^{-1}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$0.253$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$32.9$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$8.953$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1.4013$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho\,\, {\mathrm{kgm}}^{-3}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1113.5$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$6310$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$4250$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2270$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sigma / \mathrm{S}\times {\mathrm{m}}^{-1}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$4.3\times {10}^{-5}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$5.96\times {10}^{7}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2.4\times {10}^{6}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$3.5\times {10}^{6}$$\end{document} Table 1Thermophysical Properties of base fluid and nanofluidsThermophysical PropertyFormulaThermal Conductivity\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{nf}={K}_{bf}\left[\frac{{K}_{3}+2{K}_{bf}-2{\phi}_{3}\left({K}_{bf}-{K}_{3}\right)}{{K}_{3}+2{K}_{bf}+{\phi}_{3}\left({K}_{bf}-{K}_{3}\right)}\right]$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{hnf}={K}_{nf}\left[\frac{{K}_{2}+2{K}_{nf}-2{\phi}_{2}\left({K}_{nf}-{K}_{2}\right)}{{K}_{2}+2{K}_{nf}+{\phi}_{2}\left({K}_{nf}-{K}_{2}\right)}\right]$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{thnf}={K}_{hnf}\left[\frac{{K}_{1}+2{K}_{hnf}-2{\phi}_{1}({K}_{hnf}-{K}_{1})}{{K}_{1}+2{K}_{hnf}+{\phi}_{1}({K}_{hf}-{K}_{1})}\right]$$\end{document}Viscosity\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mu}_{thnf}=\frac{{\mu}_{bf}}{{\left(1-{\phi}_{3}\right)}^{2.5}{\left(1-{\phi}_{2}\right)}^{2.5}{\left(1-{\phi}_{1}\right)}^{2.5}}$$\end{document}Electric Conductivity\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\sigma}_{nf}={\sigma}_{bf}\left[\frac{{\sigma}_{3}\left(1+2{\phi}_{3}\right)+2{\sigma}_{bf}\left(1-{\phi}_{3}\right)}{{\sigma}_{3}\left(1-{\phi}_{3}\right)+{\sigma}_{bf}\left(2+{\phi}_{3}\right)}\right]$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\sigma}_{hnf}={\sigma}_{nf}\left[\frac{{\sigma}_{2}\left(1+2{\phi}_{2}\right)+2{\sigma}_{nf}\left(1-{\phi}_{2}\right)}{{\sigma}_{2}\left(1-{\phi}_{2}\right)+{\sigma}_{nf}\left(2+{\phi}_{2}\right)}\right]$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\sigma}_{thnf}={\sigma}_{hnf}\left[\frac{{\sigma}_{1}\left(1+2{\phi}_{1}\right)+2{\sigma}_{hnf}(1-{\phi}_{1})}{{\sigma}_{1}\left(1-{\phi}_{1}\right)+{\sigma}_{hnf}(2+{\phi}_{1})}\right]$$\end{document}Heat Capacitance\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\left(\rho {C}_{p}\right)}_{thnf}=\left(1-{\phi}_{3}\right)\left[\left(1-{\phi}_{2}\right)\left(\left(1-{\phi}_{1}\right){\left(\rho {C}_{p}\right)}_{bf}+{\phi}_{1}{\left(\rho {C}_{p}\right)}_{1}\right)+{\phi}_{2}{\left(\rho {C}_{p}\right)}_{2}\right]+{\phi}_{3}{\left(\rho {C}_{p}\right)}_{3}$$\end{document}Density\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rho}_{thnf}=\left(1-{\phi}_{3}\right)\left[\left(1-{\phi}_{2}\right)\left(\left(1-{\phi}_{1}\right){\rho}_{bf}+{\phi}_{1}{\rho}_{1}\right)+{\phi}_{2}{\rho}_{2}\right]+{\phi}_{3}{\rho}_{3}$$\end{document} Thermophysical Properties of base fluid and nanofluids Table 2Thermal propertiesPhysical property\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${C}_{2}{H}_{6}{O}_{2}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$A{l}_{2}{O}_{3}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Ti{O}_{2}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Si{O}_{2}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${C}_{p} / {\mathrm{Jkg}}^{-1}{\mathrm{K}}^{-1}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2415$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$765$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$686.2$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$745$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$k / {\mathrm{Wm}}^{-1}{\mathrm{K}}^{-1}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$0.253$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$32.9$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$8.953$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1.4013$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho\,\, {\mathrm{kgm}}^{-3}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1113.5$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$6310$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$4250$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2270$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sigma / \mathrm{S}\times {\mathrm{m}}^{-1}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$4.3\times {10}^{-5}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$5.96\times {10}^{7}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2.4\times {10}^{6}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$3.5\times {10}^{6}$$\end{document} Thermal properties The physical and thermal behavior of the system is further constrained by the boundary conditions defined in Eq. 4, which prescribe the velocity distribution, suction or injection velocity, and surface temperature along the non-uniformly moving boundary. These conditions ensure that the velocity and temperature fields adhere to the physical characteristics of the flow near the surface, while asymptotically satisfying free-stream conditions far from the wall. To enable analytical tractability and numerical solvability, a similarity transformation is incorporated in Eq. 5 that simplifies the governing partial differential equations into a set of coupled, nonlinear ordinary differential equations defined in terms of dimensionless similarity variables. This transformation encapsulates the dominant dynamics of the viscoelastic nanofluid system, emphasizing the role of pertinent physical parameters while largely demystifying the computational intricacy. The new momentum and energy equations, provided in Eqs. 6 and 7, model the coupled transport processes under the influence of magnetohydrodynamic, viscoelastic stress, and thermal enhancement caused by nanoparticles. Equation 6, for instance, models the momentum transport within the boundary layer in terms of inertial forces, viscous shear, elastic stresses involving relaxation and retardation times, and Lorentz force due to the applied magnetic field. Equation 7 controls the thermal field, including conductive and convective heat transfer, viscous dissipation expressed in terms of the Eckert number, Joule heating, and internal heat generation or absorption expressed by the parameter λ. The Prandtl number and magnetic interaction parameter also become important factors in controlling the temperature field and entropy generation. The corresponding dimensionless boundary conditions given by Eqs. 8 and 9, determine the near-wall and far-field velocity and temperature profiles, guaranteeing the mathematical well-posedness of the problem as well as the physical correctness of the solutions. Methodology The transformed equations governing the boundary layer (the transforming process creates the similarity conditions) are a set of nonlinear ordinary differential equations subject to two sets of boundary conditions for each equation. Due to the high degree of nonlinearity and coupling of the two systems of equations (for velocity and temperature), analytical solutions are typically not practical to implement; thus a numerical method was employed to solve for the solution profiles. In this case, the set of equations was solved using a boundary measure problem (symmetrical approximation) solution technique and this data was then used to create an artificial neural network (ANN) model to predict the behaviour of the thermodynamic properties of the tri-hybrid nanofluid system. Numerical solution using BVP4C solver The MATLAB bvp4c solver uses a collocation method based on the three-stage Lobatto IIIA formula for solving nonlinear boundary value problems. The bvp4c solver is often used to compute solutions to nonlinear two-point boundary value problems related to fluid mechanics and heat transfer. The transformed system of ordinary differential equations can be expressed as:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{dY}{d\eta }=F(\eta ,Y)$$\end{document}and has boundary conditions represented as:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$B\left(Y\left(0\right),Y\left(\infty \right)\right)=0$$\end{document}where\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Y=\left[\begin{array}{c}\begin{array}{c}f\\ f^{\prime}\\ f^{\prime \prime }\end{array}\\ \begin{array}{c}\begin{array}{c}f^{\prime \prime \prime}\\ \theta \end{array}\\ \theta ^{\prime}\end{array}\end{array}\right]$$\end{document}is a vector of dependent variables. The system is first converted into a set of first-order equations to yield:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{array}{c}\begin{array}{c}{y}_{1}=f\\ {y}_{2}=f^{\prime}\\ {y}_{3}=f^{\prime \prime}\end{array}\\ \begin{array}{c}\begin{array}{c}{y}_{4}=f^{\prime \prime \prime}\\ {y}_{5}=\theta \end{array}\\ {y}_{6}=\theta^ {\prime}\end{array}\end{array}$$\end{document} Thus,\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{array}{c}\begin{array}{c}{y}_{1}^{\prime}={y}_{2}\\ \begin{array}{c}{y}_{2}^{\prime}={y}_{3}\\ {y}_{3}^{\prime}={y}_{4}\end{array}\\ {y}_{4}^{\prime}={F}_{1}\left({y}_{1},{y}_{2},{y}_{3},{y}_{4},{y}_{5},{y}_{6}\right)\end{array}\\ \begin{array}{c}{y}_{5}^{\prime}={y}_{6}\\ {y}_{6}^{\prime}={F}_{2}\left({y}_{1},{y}_{2},{y}_{3},{y}_{4},{y}_{5},{y}_{6}\right)\end{array}\end{array}$$\end{document}the nonlinear functions derived from the momentum and energy equations are represented by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${F}_{1}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${F}_{2}$$\end{document}. The overall iterative process utilized to arrive at the final solution will continue until residual errors are reduced to acceptable tolerance levels using collocation points. For purposes of this study, convergence tolerance will be set at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-9}$$\end{document} to provide sufficient numerical stability and accuracy for derived solution profiles. Grid independent study The use of a mesh independence study on the numerical accuracy and reliability of the solutions obtained to solve the nonlinear boundary value problem via the bvp4c solver had to vary the number of mesh points in the computational domain. To check the convergence behaviour of the numerical solution, the maximum residual values produced throughout the iterative process were recorded. Three separate mesh distributions were analyzed for this study. The solver will automatically change the mesh size during the iterative process until it is within tolerance. Tolerance of convergence for the solver was set to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-9}$$\end{document}. A summary of all the different mesh sizes’ numerical performance can be found in Table 3.Table 3Mesh independence study for numerical solutionMesh pointsMaximum residualODE function callsBoundary condition calls237\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$9.064\times {10}^{-10}$$\end{document}11,384112297\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$7.249\times {10}^{-10}$$\end{document}793490298\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$7.265\times {10}^{-10}$$\end{document}736489 Table 3Mesh independence study for numerical solutionMesh pointsMaximum residualODE function callsBoundary condition calls237\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$9.064\times {10}^{-10}$$\end{document}11,384112297\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$7.249\times {10}^{-10}$$\end{document}793490298\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$7.265\times {10}^{-10}$$\end{document}736489 Mesh independence study for numerical solution Based on the results presented in Table 3, it is evident that the numerical solution quickly converges to an accurate representation of the solution with an increase in mesh density. The maximum residual value dropped significantly, became stable after approximately 297 points of mesh size. The last two mesh configurations were nearly identical indicating mesh independence has been achieved by the numerical solution. In the following simulations, a mesh size of approximately 297 grid points will be used to obtain numerical accuracy and computational efficiency. The large degree of residuals confirm the stability and reliability of the numerical solution. Artificial neural networks model ANN is used to perform further analyses of the thermodynamics of the system to develop a predictive model. ANN can be used to approximate complex nonlinear representations of the input parameters and their corresponding system responses. The numerical dataset from bvp4c is then utilized in this study as input to the ANN model for the prediction of velocity and temperature distributions. The ANN model architecture includes a feed-forward multilayer architecture with:An input layerHidden layersAn output layer An input layer An input layer Hidden layers Hidden layers An output layer An output layer Where input vector is defined by:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$X=\left[\eta , M,Pr,{\beta}_{1},{\beta}_{2},\phi \right]$$\end{document} The network output is defined by:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Y=\left[{f}^{\prime}\left(\eta \right),\theta \left(\eta \right),Entropy\right]$$\end{document} Both outputs represent velocity and temperature distributions, respectively. ANN mathematical formulation Each neuron's output is calculated using\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${z}_{j}=\sum_{i=1}^{n}{w}_{ij}{x}_{i}+{b}_{j}$$\end{document} With \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${w}_{ij}=$$\end{document} connection weight, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${x}_{i}=$$\end{document} input variable and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${b}_{j}=$$\end{document} bias term. The activation function used by each of the hidden-layer neurons is the log-sigmoid function:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\phi \left(z\right)=\frac{1}{1+{e}^{-z}}$$\end{document}while the final output of the neural network (NN) is computed using\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${y}_{k}=\sum_{j=1}^{m}{v}_{jk}\phi ({z}_{j})+{c}_{k}$$\end{document}where, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${v}_{jk}=$$\end{document} Variables representing the weights between the hidden and output layers. Levenberg marquardt scheme In order to optimize the NN's weights, we use the Levenberg–Marquardt Backpropagation Training Algorithm. This algorithm reduces the mean squared error (MSE) between the actual (numerical) outputs and the predicted outputs. The objective function is represented by\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E=\frac{1}{N}\sum_{i=1}^{N}{\left({y}_{i}^{num}-{y}_{i}^{ANN}\right)}^{2}$$\end{document} While the NN weights are updated in accordance with\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${w}_{k+1}={w}_{k}-{\left({J}^{T}J+\mu I\right)}^{-1}{J}^{T}e$$\end{document}where, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$J =$$\end{document} Jacobian Matrix, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$e =$$\end{document} Error Vector, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu =$$\end{document} Damping Factor. The Levenberg–Marquardt algorithm combines the strengths of both the Gauss–Newton Method and Gradient Descent, resulting in fast convergence for nonlinear regression problems. Statistical interpretation Neural networks are capable of learning complicated functional relationships directly from data that can approximate complicated predictive models–often right from observational data–with a sensibility to accuracy and efficiency. In this study, we present the use of a feed-forward neural network trained with a specific optimization algorithm (Levenberg–Marquardt) to develop a predictive model of the thermo-hydrodynamic characteristics of tri-hybrid nanofluid flow over a non-uniformly stretching surface with magneto hydrodynamic (MHD) interaction and internal heat generation. This machine learning framework is designed to approximate the nonlinear behavior of the system with improved precision compared to conventional modeling techniques. The values of parameters for all scenarios are presented in Table 4 whereas the whole procedure for this work is illustrated in Fig. 3. The network’s predictive performance and reliability are quantitatively evaluated through key assessment metrics, including mean squared error (MSE), gradient, epochs, and computational efficiency, as summarized in Table 5.Table 4Illustrations of four scenarios for three different casesScenarioCase\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Ec$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Pr$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_1$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta _2$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda $$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Omega $$\end{document}111.01.00.56.20.10.10.11.021.01.00.56.20.40.10.11.031.01.00.56.20.70.10.11.0210.11.00.56.20.10.10.11.020.61.00.56.20.10.10.11.031.01.00.56.20.10.10.11.0311.01.00.56.20.10.10.11.021.01.00.56.20.10.50.11.031.01.00.56.20.10.90.11.0411.01.00.56.20.10.10.11.021.01.00.56.20.10.10.71.031.01.00.56.20.10.11.21.0Fig. 3Graphical PresentationTable 5Results of LMBFFNN for THNFOBMHD modelSCMSEPerformanceGradientEpochsTime (s)TrainingValidationTesting112.08E-102.28E-107.88E-102.08E-109.96E-8949724.13E-114.16E-114.20E-114.14E-119.98E-8596636.97E-99.23E-96.97E-96.98E-91.59E-610006214.38E-83.99E-85.03E-84.16E-81.20E-6209121.45E-93.91E-91.94E-91.45E-99.95E-8348435.15E-105.34E-106.01E-105.15E-101.26E-710007314.82E-105.21E-104.22E-104.83E-109.97E-8844823.63E-73.94E-76.59E-73.36E-79.94E-5268231.97E-92.76E-94.17E-91.97E-91.01E-610009414.43E-104.21E-105.26E-104.43E-109.96E-8778925.26E-109.97E-104.49E-95.26E-109.95E-8608436.68E-105.90E-104.75E-106.68E-101.00E-87025 Table 4Illustrations of four scenarios for three different casesScenarioCase\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Ec$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Pr$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_1$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta _2$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda $$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Omega $$\end{document}111.01.00.56.20.10.10.11.021.01.00.56.20.40.10.11.031.01.00.56.20.70.10.11.0210.11.00.56.20.10.10.11.020.61.00.56.20.10.10.11.031.01.00.56.20.10.10.11.0311.01.00.56.20.10.10.11.021.01.00.56.20.10.50.11.031.01.00.56.20.10.90.11.0411.01.00.56.20.10.10.11.021.01.00.56.20.10.10.71.031.01.00.56.20.10.11.21.0 Illustrations of four scenarios for three different cases Graphical Presentation Table 5Results of LMBFFNN for THNFOBMHD modelSCMSEPerformanceGradientEpochsTime (s)TrainingValidationTesting112.08E-102.28E-107.88E-102.08E-109.96E-8949724.13E-114.16E-114.20E-114.14E-119.98E-8596636.97E-99.23E-96.97E-96.98E-91.59E-610006214.38E-83.99E-85.03E-84.16E-81.20E-6209121.45E-93.91E-91.94E-91.45E-99.95E-8348435.15E-105.34E-106.01E-105.15E-101.26E-710007314.82E-105.21E-104.22E-104.83E-109.97E-8844823.63E-73.94E-76.59E-73.36E-79.94E-5268231.97E-92.76E-94.17E-91.97E-91.01E-610009414.43E-104.21E-105.26E-104.43E-109.96E-8778925.26E-109.97E-104.49E-95.26E-109.95E-8608436.68E-105.90E-104.75E-106.68E-101.00E-87025 Results of LMBFFNN for THNFOBMHD model The Fig. 4 shows the convergence performance of the LMBFFNN model by illustrating the descent of the gradient, the controller mu, and validation tests across the training process for four cases. Each figure (S-1 to S-4) illustrates the descent of the gradient for the network, concurrent adjustment of the learning parameter mu, and the validation performance for the network functions over the training epochs. The gradual decrease in the gradient with increasing epochs reflects the quick and steady convergence of the network. The damping factor mu remained fairly constant at a low value (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-8}-10^{-9})$$\end{document}, which reinforces stable optimization of the training process. The validation tests remained at zero, demonstrating no overfitting, which also ensures a level of confidence in generalization of the network. The S-4 scenario converged in less epochs (≈ 702), which indicates a computational gain, while all other scenarios reached the minimum gradient by the conclusions of the tests of the networks at around \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{-6}-10^{-7}$$\end{document} by epoch 1000. These plots confirm the robustness, stability, and learning power of the proposed LMBFFNN for modeling the nonlinear thermo-hydrodynamic behavior of tri-hybrid Oldroyd-B nanofluids, as mentioned above.Fig. 4Learning rate curve for case 3 for all scenarios of LMBFFNN for THNFOBMHD Learning rate curve for case 3 for all scenarios of LMBFFNN for THNFOBMHD In machine learning-based modeling of intricate fluid flows, the Mean Squared Error (MSE) acts as a critical measure of predictive precision and dependability of neural network structures. As a differentiable loss function, it facilitates effective optimization in the Levenberg–Marquardt feed-forward network, allowing the model to converge towards minimal error in predictions and maximizing convergence. In the current research work, the developed ANN attained remarkably low MSE values, with the errors on training as small as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2.08\times {10}^{-10}, 4.13\times {10}^{-11}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$4.82\times {10}^{-10}$$\end{document}, while the validation and test errors were of the same order of magnitude. These uniformly low errors illustrate the model's robust ability to simulate nonlinear thermohydrodynamic behavior with great accuracy. In addition, the network exhibited solid and fast convergence at optimality within a wide range of about 209 to 1000 epochs, exhibiting computational efficiency as well as stability. The convergence behaviors presented in Fig. 5 demonstrate the model's capability to identify the underlying physics of the Oldroyd-B tri-nanofluid system with great precision and less deviation from predicted outcomes. The low MSE values throughout all the training stages and the stable convergence pattern demonstrate the robustness of the Levenberg–Marquardt algorithm in handling sophisticated nonlinear problems, which further substantiates its applicability in predictive modeling of sophisticated thermal-fluid systems.Fig. 5Performance curves for case 1 for all scenarios of LMBFFNN for THNFOBMHD Performance curves for case 1 for all scenarios of LMBFFNN for THNFOBMHD Figure 6 shows the error distribution histograms and fitness evaluation of the Levenberg–Marquardt feed-forward neural network proposed for the Oldroyd-B tri-nanofluid model. The histograms provide an objective measure of prediction capacity, with the horizontal axis representing the magnitude of the errors and the vertical axis representing the magnitude of errors' frequency. The results show that prediction errors are mostly concentrated near zero error indicating the network indeed predicted the unobserved nonlinear function of the physical system well. Most error values are within a range near between \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$-6.97\times {10}^{-9}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$5.26\times {10}^{-10}$$\end{document}. Peak occurrences are at approximately \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2.08\times {10}^{-10}, 4.13\times {10}^{-11},$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$4.82\times {10}^{-10}$$\end{document}. Error values are extremely small and reflect the network capacity to accurately map the input parameters to a target response. Additionally, with no major outliers and most errors in range of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-10}$$\end{document} to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-9},$$\end{document} the distribution of error indicates the Levenberg–Marquardt is robust and produces capacities of prediction.Fig. 6Erroneous Histogram Plots for case 1 for all scenarios of LMBFFNN for THNFOBMHD Erroneous Histogram Plots for case 1 for all scenarios of LMBFFNN for THNFOBMHD Figure 7 illustrates the fitness curves for the Levenberg–Marquardt feed-forward neural network that was introduced with its associated predictability and generalization. The plots demonstrate the times series with the predicted outputs and actual target values for training, validation and testing. The predicted outputs lie close by the observed outputs for each set respectively and show evidence that the neural network articulated the highly complex non-linear relationships that exist for the full system sufficiently to approximate across flow and thermal conditions. Even greater evidence in supporting this deduction, can be seen in the form of the subtype of the fitness plot displayed in the bottom plots in Fig. 7. The residual plot illustrates the predicted responses closely follow the actual observed and does not show evidence of error difference across the predicted responses, suggesting stability and resultant predictability of the models. The series of consecutive correlations between predicted outputs with target data is encouraging and is in agreement with broader reliability observations of the neural network and the models capacity to approximate complex non-linear thermohydrodynamic interactions across flow, temperaturaure and thermodynamic state of the system, while avoiding acute over- or under-fitting of the model to produce outputs for the complex systems.Fig. 7Time series response curves for case 1 for all scenarios of LMBFFNN for THNFOBMHD Time series response curves for case 1 for all scenarios of LMBFFNN for THNFOBMHD Figure 8 exhibits regression trends for the nonlinear THNFOBMHD model, which makes use of LMBFFNN. Regression plots are significant to assess the model's capability to predict and to visualize the connection between parameter. Figure 8 provides evidence that the data points fit nicely around the regression line, which indicates that there is an acceptable fit between expected and actual values. The framework's accuracy and durability are demonstrated by an almost perfect alignment of predicted and actual values as well as a high training R-squared.Fig. 8Regression curve (Training, Testing, Validation and combined all) for case 1 for all scenarios of LMBFFNN for THNFOBMHD Regression curve (Training, Testing, Validation and combined all) for case 1 for all scenarios of LMBFFNN for THNFOBMHD Figure 9a, b depicts the effect of the relaxation time parameter \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{1}$$\end{document} on the velocity and temperature profiles, and their absolute errors, based on the estimates from the Artificial Neural Network (ANN) model. In Fig. 9a, the corresponding velocity gradient profile \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f^{\prime}(\eta )$$\end{document}, which declines smoothly with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta $$\end{document}, indicates an accelerating fall because of high values of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{1}$$\end{document}, meaning a thinner momentum boundary layer forms. In Fig. 9b, the absolute error results for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f(\eta )$$\end{document} demonstrate that the ANN-Levenberg–Marquardt method produces reliable errors, remaining close to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-4}$$\end{document}. Figure 9c illustrates the temperature profile \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta (\eta )$$\end{document} reported for values of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{1}$$\end{document}, demonstrating that when \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{1}$$\end{document} is increased, so is the decay rate of the temperature, which in turn produced a thinner thermal boundary layer. Figure 9d presents the absolute error results for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta (\eta )$$\end{document}, showing that the absolute error remained well below \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-3}$$\end{document}, demonstrating that the ANN-based approach is stable, convergent, and reliable. The Fig. 9€ shows the impact of the magnetic parameter M on the temperature profile and the associated absolute error based on the Artificial Neural Network (ANN) model. The temperature profile \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta (\eta )$$\end{document} in Fig. 9e shows a decreasing monotonically for larger \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta $$\end{document}. Larger \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M$$\end{document} led to a higher temperature in the boundary layer, demonstrating the resistive effect (Lorentz force) of the magnetic field that leads to a thicker thermal boundary layer. The respective Fig. 9f of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta (\eta )$$\end{document} provides the absolute error of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta (\eta )$$\end{document}. The resulting absolute error being uniformly low, largely maintained below \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-3},$$\end{document} further confirms the numerical stability and accuracy of the ANN–Levenberg–Marquardt algorithm. In Fig. 9, the consequences of the retardation time parameter \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document} and the heat generation/absorption parameter \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda $$\end{document} on both the velocity and temperature distributions along with their respective absolute errors predicted using the ANN model are presented. In Fig. 9g, h the impact of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document} on the velocity profile denoted \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f^{\prime}(\eta )$$\end{document} is shown, with velocity decreasing with increasing η, the larger value of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document} produces a more contoured reduction in velocity and, therefore, greater retardation time suppresses the flow and creates a thinner momentum boundary layer. Further, the associated absolute error with respect to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f^{\prime}(\eta )$$\end{document} is below \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-4}$$\end{document} confirming the ANN–Levenberg–Marquardt scheme is accurate and precise. In Fig. 9i, j the change of temperature profile denoted θ(η) occurs with changes in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document}. As \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta}_{2}$$\end{document} increases, the temperature distribution is reduced indicating the heat transfer is declining because of increased fluid relaxation effects. The associated absolute error in obtaining error for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta (\eta )$$\end{document} was also minimal around \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-4}$$\end{document} too, confirming the stability and convergence of the ANN model. In the end, Fig. 9k, l demonstrate the effect of the heat generation/heat absorption parameter \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda $$\end{document} on the temperature profile. A positive \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda $$\end{document} (heat generation) leads to a higher temperature \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta (\eta )$$\end{document}, while a negative \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda $$\end{document} (heat absorption) leads to a lower temperature; hence, these two effects correspond to the thermal energy transfer occurring in the system. For all these results the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$6.2$$\end{document} value of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Pr$$\end{document} is utilized. The corresponding absolute error values were also relatively small, thus confirming the reliability and the validity of the ANN-based computational framework.Fig. 9(a) Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{1}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f^\prime(\eta)$$\end{document}, (b) AE for variation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{1}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f^\prime(\eta)$$\end{document}, (c) Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{1}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document}, (d) AE for variation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{1}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document}, (e) Effect of M on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document}, (f) AE for variation of M on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document}, (g) Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{2}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f^\prime(\eta)$$\end{document}, (h) AE for variation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{2}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f^\prime(\eta)$$\end{document}, (i) Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{2}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document}, (j) AE for variation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{2}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document}, (k) Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document}, (l) AE for variation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document} (a) Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{1}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f^\prime(\eta)$$\end{document}, (b) AE for variation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{1}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f^\prime(\eta)$$\end{document}, (c) Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{1}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document}, (d) AE for variation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{1}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document}, (e) Effect of M on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document}, (f) AE for variation of M on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document}, (g) Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{2}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f^\prime(\eta)$$\end{document}, (h) AE for variation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{2}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f^\prime(\eta)$$\end{document}, (i) Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{2}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document}, (j) AE for variation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{2}$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document}, (k) Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document}, (l) AE for variation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda$$\end{document} on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta(\eta)$$\end{document} As illustrated in Fig. 10, the relationship between local entropy generation number \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${N}_{s}$$\end{document} and similarity variable \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta $$\end{document}, is shown for mono, hybrid and tri-hybrid nanofluids where magnetic parameter values of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M = \mathrm{0.1,0.6,1.1}$$\end{document} were used. As can be seen from Fig. 10, increasing values of η resulted in slower rates of entropy production. Due to the existence of a boundary layer and the number of interaction forces acting upon an object near to it, it is anticipated that entropy generation rates would be greatest near the indicated surface, i.e., \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta = 0$$\end{document}. The presence of a magnetic field, which produces Lorentz forces, dramatically increases the generation of entropy in all nanofluid configurations, because these forces impede or resist fluid movement (resulting in increased velocity gradients) and increase the effects of Joule heating, further enhancing viscosity irreversibility of the flowing fluid by increasing friction forces acting on the fluid. Thus, the net amount of entropy generated increases. Alternatively, the peak amount of entropy generated increased by approximately 20–30% over the range of M from 0.1 to 1.1 (with increasing values indicating that the imposition of a magnetic field would have substantial influences on thermodynamic irreversibility within this system). An additional important observation from the data presented in Fig. 10 is that tri-hybrid nanofluids produced higher levels of total entropy than hybrid or mono nanofluids, illustrated by the order of increasing levels of total entropy generation, i.e., tri- hybrid, hybrid, mono. The changes in the heat transport characteristics of fluid systems with the inclusion of multiple types of nanoparticles are due to the increased effective thermal conductivity and modified viscosity that the multiple types of nanoparticles provide. The presence of tri-hybrid nanoparticles helps enhance heat transfer from the working fluid; however, there are greater thermal gradients and higher viscous interactions, hence producing higher entropy near the boundary of the wall. In contrast, the tri-hybrid nanofluid generates a higher amount of local entropy, but the tri-hybrid nanofluids also have a superior heat transfer efficiency. Therefore, there is a tradeoff between improving heat transfer and producing thermodynamic irreversibility in the design of a practical thermal system. The data indicate that although tri-hybrid nanofluids may have improved thermal performance, the optimal strength of the magnetic field used should be found to minimize the amount of entropy produced and to increase the efficiency of the system.Fig. 10Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M$$\end{document} on entropy generation for mono, hybrid and trihybrid nanofluid Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M$$\end{document} on entropy generation for mono, hybrid and trihybrid nanofluid The effect of the magnetic parameter \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M$$\end{document} on the thermal profile \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta (\eta )$$\end{document} of mono (NF), hybrid (HNF) and trihybrid (THNF) nanofluids is shown in Fig. 11. The temperature of all three nanofluids increases as M increases. A magnetic field produces a Lorentz force on currect-carrying fluids, known as an electromagnetic (EM) force, that opposes the fluid motion. The EM force reduces the velocity of the nanofluid and increases the viscous dissipation of energy and Joule heat generated in the boundary layer, resulting in thermal energy and therefore larger temperature distributions. The tri-hybrid has the highest temperature, followed by the hybrid and mono nanofluids. The tri-hybrid nanofluid has a higher temperature due to the presence of three species of nanoparticles in the base fluid, which increases the effective thermal conductivity of the nanofluid. The three different nanoparticles interact with each other synergistically to enhance the heat transfer processes in the boundary layer, leading to a greater rate of thermal diffusion than in the hybrid (or) mono-nanofluids. In addition, a significant increase in the thickness of the temperature boundary layer was observed when using a tri-hybrid nanofluid versus the other configurations. This suggests that the use of tri-hybrid nanoparticles could enhance their ability to hold and transfer thermal energy, as well as enhance heat transfer. Therefore, based on the findings of these experiments, we conclude that by incorporating multiple types of nanoparticles into one tri-hybrid configuration than just one type will provide a better level of thermal performance than either mono or hybrid configurations when subjected to a magnetic field.Fig. 11Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M$$\end{document} on temperature profile for mono, hybrid and trihybrid nanofluid Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M$$\end{document} on temperature profile for mono, hybrid and trihybrid nanofluid Comparative analysis To evaluate how effective, the Levenberg–Marquardt algorithm is at predicting, we compared its performance against several other ANN training algorithms: Bayesian Regularization & Scaled Conjugate Gradient is presented in Table 6.Table 6Comparison of levenberg marquardt with other schemes for S1C1AlgorithmMSEAERLevenberg Marquardt3.97E-109.27E-070.999Bayesian Regularization9.25E-107.12E-080.998Scaled Conjugate7.13E-084.91E-050.996 Table 6Comparison of levenberg marquardt with other schemes for S1C1AlgorithmMSEAERLevenberg Marquardt3.97E-109.27E-070.999Bayesian Regularization9.25E-107.12E-080.998Scaled Conjugate7.13E-084.91E-050.996 Comparison of levenberg marquardt with other schemes for S1C1 The results indicated that the Levenberg − Marquardt algorithm produces both lower prediction error and the highest correlation coefficient, thus demonstrating the highest degree of accuracy for thermodynamic predictions made in this study. As shown in the Table 7 provided below, the different prediction positions of the ANN model and BVP4C Model are analysed using absolute error analysis for four different similarity points: S1C1 (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta =0, 1, 2, \& 3$$\end{document}), which indicates the distance of both models from the surface. As outlined in the table below, the prediction errors for the three variables; the velocity function (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f(\eta )$$\end{document}), the first and second derivative for the velocity function, and the temperature profile (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta (\eta )$$\end{document}) and their first derivative have been recorded.Table 7Error Analysis for S1C1 for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta =0, 1, 2, 3$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{\eta}}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{f}}({\boldsymbol{\eta}})$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{f}}\boldsymbol{^{\prime}}({\boldsymbol{\eta}})$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{f}}\boldsymbol{^{\prime}}\boldsymbol{^{\prime}}({\boldsymbol{\eta}})$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{f}}\boldsymbol{^{\prime}}\boldsymbol{^{\prime}}\boldsymbol{^{\prime}}({\boldsymbol{\eta}})$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{\theta}}({\boldsymbol{\eta}})$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{\theta}}\boldsymbol{^{\prime}}({\boldsymbol{\eta}})$$\end{document}01.13E-053.92E-065.12E-05-0.000310.0003120.0011241-3.90E-057.40E-05-9.76E-057.68E-050.0003070.0003192-9.03E-067.60E-05-0.000120.000127-2.84E-051.75E-0534.61E-050.000704-0.000120.007801-1.40E-050.00024 Table 7Error Analysis for S1C1 for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta =0, 1, 2, 3$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{\eta}}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{f}}({\boldsymbol{\eta}})$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{f}}\boldsymbol{^{\prime}}({\boldsymbol{\eta}})$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{f}}\boldsymbol{^{\prime}}\boldsymbol{^{\prime}}({\boldsymbol{\eta}})$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{f}}\boldsymbol{^{\prime}}\boldsymbol{^{\prime}}\boldsymbol{^{\prime}}({\boldsymbol{\eta}})$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{\theta}}({\boldsymbol{\eta}})$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{\theta}}\boldsymbol{^{\prime}}({\boldsymbol{\eta}})$$\end{document}01.13E-053.92E-065.12E-05-0.000310.0003120.0011241-3.90E-057.40E-05-9.76E-057.68E-050.0003070.0003192-9.03E-067.60E-05-0.000120.000127-2.84E-051.75E-0534.61E-050.000704-0.000120.007801-1.40E-050.00024 Error Analysis for S1C1 for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta =0, 1, 2, 3$$\end{document} The forecaster, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f(\eta )$$\end{document}, produced by the ANN model accurately produced predictions for the entire domain for the same cases as the BVP4C Model. The majority of prediction difference was between \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-5}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${10}^{-4}$$\end{document}. Therefore, this indicates the majority of predictions made by the ANN were accurate and confirmed to be accurate by comparisons to the BVP4C Model. When comparing predictions for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta =0$$\end{document} (where the boundary layer meets the wall (lower)) of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f(\eta )$$\end{document}—the measured value—and its first derivative, prediction error are very small; for example, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1.13e-5$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$3.92e-06$$\end{document}, respectively. Both values demonstrate that the ANN model has adequately modelled the standard boundary conditions and steep gradients of the boundary layer. When the similarity variable increases, the prediction error became consistently small, indicating that the ANN continues to provide a high degree of predictive accuracy over the entire domain. As shown in the Table 7, the prediction error from the second derivative is larger than for the first derivative; in the case of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta =3$$\end{document}, the difference is \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$7.80e-03$$\end{document}. This is because the higher-order derivatives are much more sensitive regarding small numerical fluctuations and approximations. Nevertheless, the deviations are still small enough that the predicted velocity/temperature profiles are assumed to be very accurate. In general, the errors are all still small enough that the ANN–Levenberg–Marquardt method will provide a very good representation of the numerical solutions provided by bvp4c, meaning that it can be considered a very reliable surrogate model in predicting nonlinear flow and heat transfer characteristics of the three hybrid nanofluid system. Sensitivity analysis of ANN architecture The ANN models will evaluate how much variation exists within the predicted output by changing the number of hidden layers, changing the design of the ann and different data sets used to train the model. The configuration used to analyze these models has 10 neurons in one hidden layer and is divided 80% for training and 10% for both validation and testing, and the Levenberg–Marquardt algorithm was used to evaluate this model. The results in Table 8 show that as the number of neurons increases (up to a maximum of approximately 10), the accuracy of prediction increases, and that the 10 neuron architecture represents the best possible architecture in terms of computational and prediction capacity.Table 8Sensitivity of ANN accuracy with respect to number of hidden neuronsHidden NeuronsTraining MSEValidation MSETesting MSER-value62.91E-063.12E-063.47E-060.999181.84E-062.03E-062.16E-060.999410 (baseline)9.72E-071.08E-061.15E-060.9998129.31E-071.05E-061.17E-060.9998149.28E-071.12E-061.21E-060.9997 Table 8Sensitivity of ANN accuracy with respect to number of hidden neuronsHidden NeuronsTraining MSEValidation MSETesting MSER-value62.91E-063.12E-063.47E-060.999181.84E-062.03E-062.16E-060.999410 (baseline)9.72E-071.08E-061.15E-060.9998129.31E-071.05E-061.17E-060.9998149.28E-071.12E-061.21E-060.9997 Sensitivity of ANN accuracy with respect to number of hidden neurons In comparing Table 9, the ANN model with a single hidden layer predicted equally well as the ANN models with multiple hidden layers for this problem situation; thus, additional network layers did not increase the accuracy of the prediction, indicating that the non-linear relationship between the governing parameters and the solution variables can be modeled using a moderate sized ANN structure.Table 9Comparison of different ANN architecturesArchitectureHidden LayersNeuronsTesting MSER-valueANN-11101.15E-060.9998ANN-228–61.43E-060.9996ANN-31141.21E-060.9997 Table 9Comparison of different ANN architecturesArchitectureHidden LayersNeuronsTesting MSER-valueANN-11101.15E-060.9998ANN-228–61.43E-060.9996ANN-31141.21E-060.9997 Comparison of different ANN architectures As demonstrated in Table 10, the ANN models produce relatively consistent predictions for each data set, but the 80% for training, 10% for validation and 10% for testing configuration produced the highest quality generalization as this provides an adequate amount of data for training and has independent data for validating and testing.Table 10Effect of dataset split on ANN accuracyTraining/Validation/Test SplitTesting MSER-value70/15/151.52E-060.999680/10/101.15E-060.999985/10/51.19E-060.9997 Table 10Effect of dataset split on ANN accuracyTraining/Validation/Test SplitTesting MSER-value70/15/151.52E-060.999680/10/101.15E-060.999985/10/51.19E-060.9997 Effect of dataset split on ANN accuracy Conclusion The present research examines the thermo-hydrodynamic properties of an Oldroyd-B viscoelastic tri-hybrid nanofluid (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$A{l}_{2}{O}_{3}-Ti{O}_{2}-Si{O}_{2}/$$\end{document} ethylene glycol) through the Levenberg–Marquardt-based Artificial Neural Network (ANN) method. The model includes the magnetic interaction, Joule heating, and internal heat generation/absorption phenomena to investigate the nonlinear behavior of viscoelastic transport. The boundary-layer approximation-derived governing equations were reduced to nonlinear ODEs via similarity transformations and solved numerically. The governing equations were solved numerically using MATLAB’s bvp4c solver and a mesh independence study was performed in order to confirm the accuracy and convergence of the numerical results. In addition, an artificial neural network (ANN) was constructed using the Levenberg–Marquardt backpropagation algorithm to act as a surrogate model of prediction for the numerical data. The results provide a number of important physical insights on how heat transfer and entropy production mechanisms work. First, the magnetic field creates a resistive Lorentz force, which greatly diminishes the velocity of the fluid and increases the thickness of the momentum boundary layer and therefore modifies how heat is transported throughout the system. Second, the heat generation parameter significantly increases the thickness of the thermal boundary layer and increases the production of entropy in the system. Third, the viscoelastic characteristics of the Oldroyd-B fluid change how momentum is distributed and change how thermal and mechanical energy are coupled together. Finally, the addition of tri-hybrid nanoparticles increases the thermal conductivity of the base fluid, which results in higher rates of heat transfer than is found with conventional nanofluids. Engineering perspective The findings from this research will benefit engineers in regards to providing essential design information regarding systems in which effective heat transfer and minimizing the amount of entropy produced are of importance. The enhanced thermal conductivity of the Oldroyd-B tri-hybrid nanofluid can be utilized to design a working fluid used in such applications like aerospace thermal control systems, turbine blade cooling, microelectronics heat sinks, polymer extrusion processes, and solar thermal collectors. The data obtained regarding the magnetic field strength and internal heat generation allow for effective regulation of flow resistance, temperature distribution throughout the system, and reducing the amount of thermodynamic irreversibility. The surrogate model created using an artificial neural network allows for thermal performance to be predicted quickly, without continuously performing numerical simulations, thus being very helpful for design optimization, performing parametric studies, and providing real-time control of advanced thermal systems. Limitations The current results hold significant value for thermal management systems, micro scale cooling devices, energy conversion systems, polymer processing flows, and advanced heat exchange technologies because they provide better thermal conductivity and optimized entropy generation which contributes greatly to the efficiency of the system. While the current findings do have value there are a number of limitations associated with the current study. Firstly, this study forms a steady-state, 2-dimensional analysis for boundary layer flow; however, other real-world physical phenomena such as thermal radiation, aggregation of nanoparticles, and unsteady flow are not addressed in this study. Additionally, this study developed an artificial neural network (ANN) model based on data generated from numerical simulations and not from experimental data which may limit validation against actual database. Lastly, this study did not include any evaluation of pumping power or flow stability requirements which are also very important in making the results relevant to engineers. Future work These results may be built upon in future work by considering the many physical phenomena from real life, like thermal radiation or chemical reactions. Likewise, there is potential for future work to build upon the present analysis by considering unsteady flow or three-dimensional fluid motion to better predict real-world events. Additionally, using new architectures in machine learning will assist in predicting complex nonlinear thermofluid systems; this includes deep neural networks, physics informed neural networks, and hybrid machine learning using optimization-based training methods. Future studies can extend the current model for radiative heat transfer, unsteady and three-dimensional flow cases, or hybrid nanoparticles with bio-inspired base fluids for biomedical heat management. Also, use ANN as a fast surrogate for optimization, but warn that extrapolation needs retraining or solver checks. The discussed LMBFFNN approach can be coupled with higher-order soft-computing methods like genetic algorithms, adaptive neuro-fuzzy inference systems (ANFIS), or deep neural networks to optimize parameters that control entropy reduction and heat transfer enhancement. In addition, experimental confirmation of tri-hybrid nanofluids under MHD and internal heating conditions can further validate the computational results of the present study. Also, this study can be extended for QNN [39, 40] for the simulation of Oldroyd-B tri-hybrid nanofluid. Also, ANN can be implemented to the flow studies in engineering application e.g., [41, 42, 43].