Title: Soft computational modelling and regression analysis for thermal properties of nanofluids

Authors: R. Kavitha; P.C. Mukesh Kumar

Addresses: Department of Computer Science and Engineering, Parisutham Institute of Technology and Science, Thanjavur, 613-005, Tamil Nadu, India ' Department of Mechanical Engineering, University College of Engineering, Dindigul, 624-622, Tamil Nadu, India

Abstract: An emerging feature of nanofluids is its thermo physical properties, which leads to develop an enormous application in various fields especially in automotive sector as an engine coolant. Experimental and mathematical models were developed to predict the thermal properties of nanofluids but there is no accordance between them. For accurate prediction various machine - learning algorithms were used. In this paper, thermal conductivity ratio of CNT/H₂O were predicted using Gaussian process regression method with different covariance functions and optimised using hyper parameters. The predicted results were compared with the experimental values both possess a good agreement between them. The root mean square error (RMSE) value of squared exponential covariance function with hyper parameters is 0.014926 and regression coefficient value (R²) for overall data is 0.98. The outcome of the proposed model will reduce the experimental test runs and used for accurate prediction.

Keywords: nanofluids; soft computation; Gaussian process regression; GPR; covariance function; thermal conductivity.

DOI: 10.1504/IJRAPIDM.2019.100504

International Journal of Rapid Manufacturing, 2019 Vol.8 No.3, pp.243 - 258

Received: 18 May 2018
Accepted: 25 Jul 2018
Published online: 29 Jun 2019 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article