Optimization, modeling, and prediction of relative viscosity and relative thermal conductivity of AlN nano-powders suspended in EG

Abstract

Optimization of relative thermal conductivity (TC) and relative viscosity of mixtures of aluminum nitrides nanoparticles suspended in ethylene glycol nanofluid (NF) was investigated. The effect of temperature and volume fraction (VF) of nanoparticles on relative TC and relative viscosity of this NF was assessed in a double-tube heat exchanger. The response surface and multilayer perceptron methods were used to predict the relative TC and relative viscosities of the NFs. The response surface method yielded relative TC of the NF with a regression coefficient of R2 = 0.9882 and relative viscosity of the NF with a regression coefficient of R2 = 0.9932, respectively, indicating the accuracy of the models. The values of R = 0.9982, R2 = 0.9959, mean squared error = 1.05E−06 and mean absolute error = 6.54E−04 were obtained from the outputs of the designed neural network. The optimized results of non-dominated sorting genetic algorithm illustrated that high temperature has a key role compared to the VF of nanoparticle for the selection of the optimal points, in a double-tube heat exchanger. Moreover, optimization results with response surface method confirmed that 2.4790% VF of nanoparticles with 1.0605 relative TC and 1.0235 relative viscosity along with 0.9081 desirability function at 50 °C were the optimal conditions for using this NF.