Analysis of thermophysical and transport properties of nanofluids using machine learning algorithms
Accurate calculations of thermophysical properties (TPP) and fluid transport properties are crucial for solving and enhancing a myriad of heat and mass transfer issues, particularly in the field of engineering and associated disciplines. This is even more significant in the case of advanced fluids,...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Elsevier
2024-02-01
|
Series: | International Journal of Thermofluids |
Subjects: | |
Online Access: | http://www.sciencedirect.com/science/article/pii/S2666202724000089 |
_version_ | 1797306904193531904 |
---|---|
author | O.M. Amoo A. Ajiboye M.O. Oyewola |
author_facet | O.M. Amoo A. Ajiboye M.O. Oyewola |
author_sort | O.M. Amoo |
collection | DOAJ |
description | Accurate calculations of thermophysical properties (TPP) and fluid transport properties are crucial for solving and enhancing a myriad of heat and mass transfer issues, particularly in the field of engineering and associated disciplines. This is even more significant in the case of advanced fluids, such as nanofluids, which play a pivotal role in the progression of new technologies. The significance of TPP in heat transfer studies is undeniable, yet there is a notable scarcity of reliable TPP values for nanofluids that can be uniformly applied to numerous problems and complex industrial systems. This shortage poses a unique challenge for nanofluids. The current study introduces NanoFluid Explorer (NanoFEx), a comprehensive MATLAB-based database specifically designed to catalog the TPP of nanofluids. The database is divided into two main parts. The first part employs Gaussian regression, supported by experimental data derived from existing literature, using kernel functions commonly used for this type of analysis. The second part features a neural network, designed for efficiency, that is trained on the aforementioned database and supplemented by an additional database formed from the Gaussian-regression output. This setup allows NanoFEx to recapture TPP values for nanofluids that may be absent in the existing literature. The preliminary findings demonstrate that NanoFEx could prove to be an invaluable resource for both theoretical and experimental researchers studying heat and mass transfer in nanofluids. Moreover, NanoFEx confirms thermodynamically consistent values with any discrepancies attributed to finite-precision computations. Furthermore, NanoFEx achieved over 93% accuracy in predicting key properties. The novelty and relevance of this study stem from the comprehensiveness of the experimental results database. In summary, this research provides an extensive and, to the best of the authors' knowledge, the most reliable dataset currently available for TPP of nanofluids. This significant contribution aims to unify nanofluid TPP research and optimize experimental processes while setting a crucial standard for future nanofluid applications. |
first_indexed | 2024-03-08T00:48:35Z |
format | Article |
id | doaj.art-fd1d51313c034f2f83b0ac4d4493bb4c |
institution | Directory Open Access Journal |
issn | 2666-2027 |
language | English |
last_indexed | 2024-03-08T00:48:35Z |
publishDate | 2024-02-01 |
publisher | Elsevier |
record_format | Article |
series | International Journal of Thermofluids |
spelling | doaj.art-fd1d51313c034f2f83b0ac4d4493bb4c2024-02-15T05:25:37ZengElsevierInternational Journal of Thermofluids2666-20272024-02-0121100566Analysis of thermophysical and transport properties of nanofluids using machine learning algorithmsO.M. Amoo0A. Ajiboye1M.O. Oyewola2Department of Mechanical Engineering, University of Ibadan, Ibadan, Oyo State, 200284 Nigeria; Corresponding author.Statistical and Optimization Methods, Pratt & Whitney, East Hartford, Connecticut, USADepartment of Mechanical Engineering, University of Ibadan, Ibadan, Oyo State, 200284 Nigeria; School of Mechanical Engineering, Fiji National University, Fiji; Department of Mechanical Engineering, University of Alaska Fairbanks, Fairbanks, USAAccurate calculations of thermophysical properties (TPP) and fluid transport properties are crucial for solving and enhancing a myriad of heat and mass transfer issues, particularly in the field of engineering and associated disciplines. This is even more significant in the case of advanced fluids, such as nanofluids, which play a pivotal role in the progression of new technologies. The significance of TPP in heat transfer studies is undeniable, yet there is a notable scarcity of reliable TPP values for nanofluids that can be uniformly applied to numerous problems and complex industrial systems. This shortage poses a unique challenge for nanofluids. The current study introduces NanoFluid Explorer (NanoFEx), a comprehensive MATLAB-based database specifically designed to catalog the TPP of nanofluids. The database is divided into two main parts. The first part employs Gaussian regression, supported by experimental data derived from existing literature, using kernel functions commonly used for this type of analysis. The second part features a neural network, designed for efficiency, that is trained on the aforementioned database and supplemented by an additional database formed from the Gaussian-regression output. This setup allows NanoFEx to recapture TPP values for nanofluids that may be absent in the existing literature. The preliminary findings demonstrate that NanoFEx could prove to be an invaluable resource for both theoretical and experimental researchers studying heat and mass transfer in nanofluids. Moreover, NanoFEx confirms thermodynamically consistent values with any discrepancies attributed to finite-precision computations. Furthermore, NanoFEx achieved over 93% accuracy in predicting key properties. The novelty and relevance of this study stem from the comprehensiveness of the experimental results database. In summary, this research provides an extensive and, to the best of the authors' knowledge, the most reliable dataset currently available for TPP of nanofluids. This significant contribution aims to unify nanofluid TPP research and optimize experimental processes while setting a crucial standard for future nanofluid applications.http://www.sciencedirect.com/science/article/pii/S2666202724000089Energy systemFluid-heat-mass transferGauss kernelRegressionNanofluidNeural network |
spellingShingle | O.M. Amoo A. Ajiboye M.O. Oyewola Analysis of thermophysical and transport properties of nanofluids using machine learning algorithms International Journal of Thermofluids Energy system Fluid-heat-mass transfer Gauss kernel Regression Nanofluid Neural network |
title | Analysis of thermophysical and transport properties of nanofluids using machine learning algorithms |
title_full | Analysis of thermophysical and transport properties of nanofluids using machine learning algorithms |
title_fullStr | Analysis of thermophysical and transport properties of nanofluids using machine learning algorithms |
title_full_unstemmed | Analysis of thermophysical and transport properties of nanofluids using machine learning algorithms |
title_short | Analysis of thermophysical and transport properties of nanofluids using machine learning algorithms |
title_sort | analysis of thermophysical and transport properties of nanofluids using machine learning algorithms |
topic | Energy system Fluid-heat-mass transfer Gauss kernel Regression Nanofluid Neural network |
url | http://www.sciencedirect.com/science/article/pii/S2666202724000089 |
work_keys_str_mv | AT omamoo analysisofthermophysicalandtransportpropertiesofnanofluidsusingmachinelearningalgorithms AT aajiboye analysisofthermophysicalandtransportpropertiesofnanofluidsusingmachinelearningalgorithms AT mooyewola analysisofthermophysicalandtransportpropertiesofnanofluidsusingmachinelearningalgorithms |