Competition of ANN and RSM techniques in predicting the behavior of the CuO-liquid paraffin

In this article, the estimation of CuO-liquid paraffin nanofluid viscosity was assessed using response surface method (RSM) and artificial neural network (ANN) methods. Since CuO-liquid paraffin nanofluid is Newtonian, two parameters of temperature and mass fraction were introduced in ANN and RSM techniques at 25-100 degrees C, 0.25-6wt.%. Both methods map the three-dimensional input space to one-dimensional space (viscosity). A response surface cubic model was approved by applying ANOVA and calculations showed an R-2 value of 0.923 and a maximum margin of deviation of 10.482%. Efforts revealed that ANN with five neurons takes precedence over others. The R-2 and maximum deviation margin were 0.994, and 3.266%, respectively. Finally, the comaprison of ANN and RSM methods indicated that the ANN method is more accurate than the RSM for conducting the nanofluid viscosity. The accuracy of ANN was such that for 50% of points, MOD was less than 1%. For MOD in the range of 0-2%, 90% of points can be predicted with an error of less than 2%. This figure for RSM was only 37%.

Automated summary

In this article, the estimation of CuO-liquid paraffin nanofluid viscosity was assessed using response surface method (RSM) and artificial neural network (ANN) methods. The results showed that the ANN method is more accurate than the RSM method.