Statistical and artificial based optimization on thermo-physical properties of an oil based hybrid nanofluid using NSGA-II and RSM

Optimization of thermal conductivity (TC) and viscosity of Al2O3-MWCNT/thermal oil hybrid nanofluid with NSGA-II and RSM was investigated. Effect of temperature and volume fraction (VF) on heat conduction and viscosity of the nanofluid was studied. Modeling of nanofluid properties of heat conduction and viscosity were done with RSM and MLP methods. Nanofluid TC and viscosity models of RSM have the regression coefficient of R 2=0.9959 and R 2=0.9989, respectively and adjusted regression coefficient of these models are Radj2=0.9947 and Radj2=0.9984, respectively. Based on these values, it can be concluded that this model is suitable for nanofluid TC and viscosity prediction. In MLP modeling the best topology and structure selected between more than 100 investigated topologies and optimal number of neurons determined for each hidden layer. Obtained results from MLP modeling, maximum residual values of nanofluid TC are +0.009, and -0.006 and maximum residual values of nanofluid viscosity are +0.016 and -0.02. Considering result of MLP, it may be concluded that the designed model is highly capable of predicting heat conduction and viscosity of the nanofluid. In optimization with NSGA-II, optimum viscosity and heat conduction were reported in maximum operating temperature. Furthermore, in RSM optimization, the optimum condition using this nanofluid was achieved in 49.99 °C and VF of 1.49% with TC of 0.1820 (W/mK), viscosity of 0.1174 (Pa.sec) and total desirability function of 0.9725. Desirability is a criterion for evaluation of optimization process accuracy. Experimental results revealed that temperature enhancement has a positive effect on both heat conductivity and viscosity of the nanofluid to reach the best nanofluid efficiency. Also, it was concluded that temperature and VF have direct effects on heat conductivity.