Smart tracking of the influence of alumina nanoparticles on the thermal coefficient of nanosuspensions: application of LS-SVM methodology

The thermal conductivity of working fluids is among the most important thermophysical property in all heat transfer equipment. Accurate estimation of the nano-fluids thermal conductivity is a prerequisite for designing and optimizing the associated heat-based equipment. Therefore, the present study simulates the thermal conduction coefficients of water-alumina nano-suspensions using the least-squares support vector machines (LS-SVM). The best structure of this paradigm is determined using a combination of trial-and-error and statistical analyses. After that, it is validated by both available empirical correlations and intelligent models available in the open literature. Our LS-SVM paradigm predicted 282 experimental data samples available in fifteen references with the absolute average relative deviation (AARD) of 1.24%, mean squared errors (MSE) of 0.0007, root mean squared errors (RMSE) of 0.026, and regression coefficient (R-2) of 0.9586. The leverage technique justifies that minor parts of experimental data are outliers (similar to 6.03%) and have an insignificant negative effect on the derived LS-SVM generalization. The designed simulator shows that temperature and alumina concentration positively affect the nano-fluids thermal conductivity, and alumina size reduces the thermal behavior of water-alumina nano-suspensions.

Automated summary

The study utilizes LS-SVM to simulate the thermal conduction coefficients of water-alumina nano-suspensions, achieving accurate prediction of experimental data. The designed simulator indicates that temperature and alumina concentration positively affect thermal conductivity.