Title: Artificial intelligence-based viscosity prediction of polyalphaolefin-boron nitride nanofluids

Authors: Omer A. Alawi; Haslinda Mohamed Kamar; Mustafa Mudhafar Shawkat; Mohammed M. Al-Ani; Hussein A. Mohammed; Raad Z. Homod; Mazlan A. Wahid

Addresses: Department of Thermofluids, School of Mechanical Engineering, Universiti Teknologi Malaysia, 81310 UTM Skudai, Johor Bahru, Malaysia ' Department of Thermofluids, School of Mechanical Engineering, Universiti Teknologi Malaysia, 81310 UTM Skudai, Johor Bahru, Malaysia ' Department of Petroleum Engineering, School of Chemical and Energy Engineering, Faculty of Engineering, Universiti Teknologi Malaysia, 81310, Johor Bahru, Malaysia ' Center for Artificial Intelligence Technology (CAIT), Faculty of Information Science and Technology, Universiti Kebangsaan Malaysia (UKM), Bangi, Selangor, Malaysia ' School of Engineering, Edith Cowan University, 270 Joondalup Drive, WA 6027, Australia ' Department of Oil and Gas Engineering, Basrah University for Oil and Gas, Aljamhoryah, Alzahraa District, Altijari Street, Basra, Iraq ' Department of Thermofluids, School of Mechanical Engineering, Universiti Teknologi Malaysia, 81310 UTM Skudai, Johor Bahru, Malaysia

Abstract: Predicting viscosity's nanofluids can benefit all domains, including energy, thermofluids, power systems, energy storage, materials, cooling, heating, and lubrication. The objective of this study to predict the dynamic viscosity of polyalphaolefin-hexagonal boron nitride (PAO/hBN) nanofluids using four main parameters: shear rate, shear stress, nanomaterials mass fraction, and temperature. Moreover, three hybrid ensemble learning models (Bayesian ridge-random forest, Bayesian ridge-MLP regressor and Bayesian ridge-AdaBoost regressor) were developed for the current task. The forward sequential feature selector (FSFS) created four input combinations (models). Model 4 showed the best prediction accuracy, followed by models 2, 3 and 1. The computational findings showed that ensemble learner 1 was slightly outperformed by ensemble learner 3. Meanwhile, among the predictive models, ensemble learner 2 consistently placed third. Besides, the research results demonstrated that creating predictive models based on all input parameters can produce a precise prediction matrix. Overall, the study recommended exciting conclusions on predicting a nanolubricant's viscosity for use in heat transfer applicants.

Keywords: nanofluids; viscosity; polyalphaolefin; PAO; machine learning; ensemble learning; boron nitride.

DOI: 10.1504/IJHM.2024.138261

International Journal of Hydromechatronics, 2024 Vol.7 No.2, pp.89 - 112

Received: 01 Aug 2023
Accepted: 03 Nov 2023
Published online: 30 Apr 2024 *