Modeling and performance optimization of non-Newtonian hybrid nanofluid solar HVAC systems with magnetic effects using neural networks

The current study develops a computational framework to examine the integration of Sutterby magneto-hybrid nanofluid (SMHNF) boundary layer behavior into solar-powered heating, ventilation, and air conditioning (HVAC) systems to enhance thermal performance and energy efficiency through artificial intelligence techniques. A hybrid nanofluid consisting of Cu (copper) and SiO₂ (silicon dioxide) nanoparticles dissolved in propylene glycol (PG) is used to represent the system under the influence of different physical effects, including activation energy, chemical reactions, and solar thermal radiation. To improve physical realism, a three-dimensional model is built that includes Smoluchowski thermal slipping and velocity slip characteristics. The governing partial differential equations are modified utilizing transformation analysis, which is derived from conservation laws, into nonlinear ordinary differential equations. The implementation of a linked numerical-ANN technique allows for the prediction and validation of system behavior under a variety of scenarios. The BVP4C solver in MATLAB offers a reference dataset for training a backpropagation neural network using the Levenberg-Marquardt algorithm. Strong consistency between artificial neural network (ANN) estimations and numerical findings is demonstrated by function fitness analysis with errors of approximately 10⁻⁰⁴, error histograms, mean squared error evaluation with range 10⁻¹⁰ to 10⁻⁰⁸, gradient of values 10⁻⁰⁹, and best regression analysis. The research results reveal that solar radiation and magnetic field strength considerably improve solar-HVAC systems' heat transfer performance. Physically, the results show that increases in slippery speed and rotation parameters reduce the velocity profile, whereas increases in thermal and radiative parameters raise the temperature field and enhance heat transfer. For the design and engineering aspects and optimization of comprehensive thermal management mechanisms in renewable energy applications, the suggested modeling and predicting methodology offers valuable insight.