The ability to approximate the nanofluid properties such as viscosity, thermal conductivity, and specific heat capacity will greatly assist in the modeling and design of nanofluidic systems. The purpose of this study was to present an adaptive neuro-fuzzy inference system (ANFIS) model for estimating the viscosity of Water/Glycerin nanofluid-containing Cu nanoparticles. The model inputs consist of two variables of temperature and volume concentration of nanofluids which have a great influence on the nanofluid viscosity. The experimental data were divided into two categories: training (three-quarters) and testing (a quarter of the data). The grid partition and subtractive clustering approaches were employed to determine the ANFIS configuration. The mean value of the relative error of 5.18% and the root mean square error of 0.0794 were obtained by comparing the target and model output values for the testing data. Proper matching of ANFIS prediction results with the test data set indicates the validity of the model. In addition, an empirical correlation was developed based on the form presented in the literature. The constants of the equation were determined by the genetic algorithm (GA) searching technique. The comparison of the prediction accuracy of the two models showed the complete superiority of the ANFIS.
