Investigation of different transfer functions for modeling the dynamic viscosity of hybrid nanolubricant containing MWCNTs and MgO nanoparticles using the response surface methodology (RSM)

Abstract

In this investigation, to derive the model and find the most impact, response surface methodology (RSM) is used. In the RSM, range selected for each factor is very important. The range of each factor must be coded and placed in the range of − 1–1 so that the regression analysis can be performed well because the unit of the independent variables is not the same and even if the range is the same, the ranges are not the same. the viscosity of 10W40based hybrid nanofluid (HNF) has been modeled using the response surface methodology (RSM). The presented data for this modeling are so extensive that the ratio of maximum to a minimum was 20.8 and the need to use the transfer function is felt. Four transfer functions of Natural-Log, Log₁₀, Square-Root and Inverse-Sqrt are selected for review and the best function is introduced. Parameters R², Adjusted R², Predicted R², C.V (%) and the distance of the parameters from the best are considered as criteria for selecting the best transfer function. The results show that the Square Root transfer function has the best performance such that the values of R², Predicted-R², Adjusted-R², C.V and the distance of the parameters from the best are equal to 0.9999, 0.9998, 0.9998, 0.5517 and 0.17, respectively, which shows the high precision of this transfer function. The Inverse-Sqrt transfer function also has the weakest performance among the studied transfer functions.