Experimental investigation and develop ANNs by introducing the suitable architectures and training algorithms supported by sensitivity analysis: Measure thermal conductivity and viscosity for liquid paraffin based nanofluid containing Al2O3 nanoparticles

The objective of this experimental investigation is to assess the variations of thermal conductivity and viscosity of liquid paraffin-Al2O3 nanofluid containing oleic acid surfactant against temperature, nanoparticle mass concentration and surfactant concentration. The experiments are performed in the temperature range of 20-50 degrees C, nanoparticle mass concentration range of 1-5%, and surfactant/nanoparticle mass ratio of 1:3, 2:3 and 3:3. The results showed that the nanofluid behaves as a shear thinning fluid. Besides, it was found that boosting the nanoparticle concentration causes an increase in the thermal conductivity and viscosity, while augmenting the temperature results in a decrease in the viscosity and an increase in the thermal conductivity. Moreover, it was observed that the viscosity increases with surfactant concentration, while the thermal conductivity initially rises with surfactant concentration and then reduces. Furthermore, the Artificial Neural Network (ANN) was implemented to model the thermal conductivity and viscosity of the nanofluid using experimental data. The findings depicted that the thermal conductivity model predicts the outputs with RMS, RMSE, MAE and R-2 values of 0.0381, 0.0018, 0.0015 and 0.982, respectively. Meanwhile, these values for the viscosity model were respectively 0.0662, 0.0179, 0.0044 and 0.96. (C) 2018 Elsevier B.V. All rights reserved.