Application of neural network on heat transfer enhancement of magnetohydrodynamic nanofluid

The applications of neural networks (NNs) on engineering problems have been increased for obtaining high precision results. In this study, a new type of NN known as the group method of data handling (GMDH) is applied to obtain a formulation of a heat transfer rate. The numerical method of control volume-based finite element method (CVFEM) is applied as a robust and reliable numerical approach for simulation of magnetohydrodynamic (MHD) flow of a nanofluid inside an inclined enclosure with a sinusoidal wall. A water-based nanofluid with Cu nanoparticles is used as main fluid in our model. Maxwell-Garnetts (MG) and Brinkman models are applied to calculate effective thermal conductivity and viscosity of nanofluid, respectively. This study tries to find that GMDH-type NN is a reliable technique for calculation of MHD nanofluid convective based on specified variables. Our findings clearly demonstrate that GMDH-type NN is more reliable than the CVFEM approach and this technique could efficiently identify the patterns in data and precisely estimate a performance. Comprehensive parametric studies are done to disclose the impact of significant factors such as electromagnetic force, buoyancy, nanoparticle volume fraction, and direction of enclosure on heat transfer rates. According to obtained results, heat transfer rate rises with the growth of buoyancy effects, the concentration of nanoparticles, and slope of domain while it reduces when Hartmann number is increased.