Artificial Intelligent Model to Enhance Thermal Conductivity of TiO₂-Al₂O₃/Water-Ethylene Glycol-Based Hybrid Nanofluid for Automotive Radiator

Abstract

Vehicular cooling system is one of the priorities for the automobile industry, aiming to achieve sustainability and energy efficiency. Currently, coolants are being utilized in cooling systems that exhibit super heat transfer capabilities. Hybrid nanofluids as a coolant is offer enhanced heat transfer rate and improved efficiency and eco-friendliness of vehicle engine cooling systems. This study aims to analyze the conductivity of a hybrid nanofluid consisting of distilled water and ethylene glycol (in a ratio of 40 and 60) with Al2O3 and TiO2 particles to evaluate its suitability as a coolant for vehicle engines using intelligent techniques. The volume concentration and the temperature varied from 0.02%-0.1% and $30~^{\circ }$ C- $80~^{\circ }$ C, respectively. The experimental findings led us to develop an artificial neural network (ANN) model. This model consists of a layer containing 9 neurons designed to estimate thermal conductivity. ANN model was constructed using input parameters such as volume concentration and temperature, with the output being the conductivity. Furthermore, apart from utilizing the ANN, we employed techniques like support vector machine (SVM) and curve fitting (CF) approaches to analyze the experimental data. This allowed us to calculate values such as the correlation coefficient (R) and mean square error (MSE). The increase in thermal conductivity reached a maximum of 40.86% when the temperature was $80~^{\circ }$ C, and the volume concentration was 0.1%. The results obtained indicate that the suggested ANN model aligns closely with the experimental data. Based on the assessment of the highest R-value and lowest MSE, this analysis demonstrates performance, with an R-value of 0.9998 and an MSE of $3.87415\times 10^{-06}$ . The training and testing phases exhibit remarkable performances with values of $4.86256\times 10^{-07}$ and $2.540599\times 10^{-06}$ , respectively. Moreover, when comparing the SVM and CF approaches, it was found that ANN modelling provided a level of accuracy in predicting the enhancement of conductivity in the hybrid nanofluid. These results demonstrate that the ANN can accurately predict thermal conductivity.

Keywords

• Thermal conductivity
• hybrid nanofluids
• artificial neural network
• support vector machine
• curve fitting