Numerical Investigation of Nanofluid Mixed Convection in an Inclined Channel and Predicting Nusselt Number with Artificial Neural Networks

Hamid Teimouri, Navid Bozorgan
Artificial Neural Networks (ANNs) are used as a new approach in determination of Nusselt number of copper water nanofluid in an inclined channel with three heat sources. For training the ANNs, the simulation results are obtained by Finite Volume Method (FVM). The effects of independent parameters, including the Reynolds number, Rayleigh number, inclination angle, and the solid volume fraction of nanoparticles, on the streamlines, isotherm lines, and the average Nusselt number have been studied.
more » ... Artificial neural networks (ANN) used to find a relation involve independent parameters for estimating the Nusselt number. The back propagation-learning algorithm with the tangent sigmoid transfer function is used to sequence the ANN. Finally, analytical relations for the nanofluid mixed convection in a channel are derived from the available ANN. It is shown that the coefficient of multiple determination (R 2) between the FVM and ANN predicted values is equal to 0.99866, maximum relative error is less than 5.9128% and mean square error is 1.13×10-3. Results show that the obtained formulation is obviously within acceptable limits.