Using artificial neural network and parametric regression to predict the effect of mechanical vibrations on heat transfer coefficient of a counter flow heat exchanger containing MWCNTs-water nanofluid

In this study, the effect of surface vibrations and temperature changes on the heat transfer coefficient of MWCNTs and water nanofluid was evaluated both experimentally and numerically utilizing different concentrations of MWCNTs and water nanofluid. The test results indicate that the application of mechanical vibrations increases the heat transfer coefficient. It has also been observed that the heat transfer coefficient increases with increase in the concentration of nanofluid and fluid temperature. To predict the heat transfer coefficient at any arbitrary combination of the independent variables within the studied ranges, both parametric and non-parametric methods are studied. For the parametric model, the heat transfer coefficient is obtained as a function that is of the first order with respect to flow rate, temperature, and of the second order with respect to nanofluid concentration and vibration amplitude. The robust least square algorithm is used to find the predictors of the parametric model. Also as a non-parametric study, a two-layer feed-forward artificial neural network is used to predict the heat transfer coefficient. The results show that the generalization of the both models is acceptable for non-trained datasets; nevertheless, the prediction error of the parametric model is higher than the non-parametric one, especially for high values of the vibration amplitudes.