An insight into the prediction of TiO.sub.2/water nanofluid viscosity through intelligence schemes

• Published in: Journal of thermal analysis and calorimetry Vol. 139; no. 3; pp. 2381 - 2394
• Main Authors: Ahmadi, Mohammad Hossein, Baghban, Alireza, Ghazvini, Mahyar, Hadipoor, Masoud, Ghasempour, Roghayeh, Nazemzadegan, Mohammad Reza
• Format: Journal Article
• Language: English
• Published: Springer 01.02.2020
• Subjects: Comparative analysis; Neural networks
• ISSN: 1388-6150; 1588-2926
• DOI: 10.1007/s10973-019-08636-4

Viscosity can be mentioned as one of the most crucial properties of nanofluids due to its ability to describe the fluid resistance to flow, and as the result it affects other phenomena. The effects of nanofluids' viscosity on different parameters can be enumerated as pressure drop, pumping power, feasibility of the nanofluid, and its convective heat transfer coefficient. In this investigation, the viscosity of TiO.sub.2/water nanofluid was compared and analyzed with experimental data. The primary goal of this investigation was to introduce a combination of experimental and modeling approaches to predict viscosity values using four different neural networks. Between MLP-ANN, ANFIS, LSSVM, and RBF-ANN methods, it was found that the LSSVM produced better results with the lowest deviation factor and reflected the most accurate responses between the proposed models. The regression diagram of experimental and estimated values shows an R.sup.2 coefficient of 0.995 and 0.993 for training and testing sections of the ANFIS model. These values for MLP-ANN, RBF-ANN, and LSSVM models were 0.998 and 0.999, 0.996 and 0.997, and 0.997 and 1.000 for their training and testing parts, respectively. Furthermore, the effect of different parameters was investigated using a sensitivity analysis which demonstrates that the average diameter can be considered as the most affecting parameter on the viscosity TiO.sub.2-water nanofluid with a relevancy factor of 0.992123.