Prediction and modeling of MWCNT/Carbon (60/40)/SAE 10 W 40/SAE 85 W 90(50/50) nanofluid viscosity using artificial neural network (ANN) and self-organizing map (SOM)

• Published in: Journal of thermal analysis and calorimetry Vol. 134; no. 3; pp. 2275 - 2286
• Main Authors: Maddah, Heydar, Aghayari, Reza, Ahmadi, Mohammad Hossein, Rahimzadeh, Mohammad, Ghasemi, Nahid
• Format: Journal Article
• Language: English
• Published: Cham Springer International Publishing 01.12.2018; Springer Nature B.V
• Subjects: Analytical Chemistry; Artificial neural networks; Automotive engineering; Carbon; Chemistry; Chemistry and Materials Science; Correlation coefficients; Inorganic Chemistry; Mathematical models; Measurement Science and Instrumentation; Multi wall carbon nanotubes; Nanofluids; Nanoparticles; Neural networks; Neurons; Physical Chemistry; Polymer Sciences; Predictions; Self organizing maps; Standard deviation; Viscosity; Viscosity; Nanofluid; Self-organizing map; Artificial neural networks
• ISSN: 1388-6150; 1588-2926
• DOI: 10.1007/s10973-018-7827-1

The present study investigated and predicted the relative viscosity of multiwall carbon nanotube/carbon (60/40)/SAE 10 W 40/(Society of Automotive Engineers) SAE 85 W 90(50/50) at different temperatures and the different volumetric fraction by applying artificial neural networks based on experimental data. Several samples of nanofluid were provided by adding nanoparticles in 0%, 0.1%, 0.3%, 0.5%, 0.8% and 1% volumetric concentrations. Dynamic viscosity of the nanofluid was measured in the temperature range of 25–50 °C. Initially, a self-organizing 6 × 6 hexagonal network was used. A total of 36 neurons were chosen. The winner neuron was neuron 25, having assigned the most data to itself. Then 25 neurons were used for the neural network, which had a very good performance. Temperature and concentration were considered as input variables, while the relative viscosity was the output parameter of the neural network. Mean-square error, correlation coefficient and standard deviation were utilized in order to assess the results. Based on the obtained results, the best model was double-layer perceptron neural network with 25 neurons. The mean square error, correlation coefficient and standard deviation were equal to 2.0193e−008, 1 and 0.00021082, respectively. Therefore, the model is able to predict relative viscosity with appropriate accuracy.