Design optimization of solar collectors with hybrid nanofluids: An integrated ansys and machine learning study

Document Type

Journal Article

Publication Title

Solar Energy Materials and Solar Cells

Volume

271

Publisher

Elsevier

School

School of Engineering

RAS ID

65628

Funders

Universiti Teknologi Malaysia / Civil and Environmental Engineering Department, King Fahd University of Petroleum & Minerals, Saudi Arabia

Comments

Alawi, O. A., Kamar, H. M., Abdelrazek, A. H., Mallah, A. R., Mohammed, H. A., Homod, R. Z., & Yaseen, Z. M. (2024). Design optimization of solar collectors with hybrid nanofluids: An integrated ansys and machine learning study. Solar Energy Materials and Solar Cells, 271, article 112822. https://doi.org/10.1016/j.solmat.2024.112822

Abstract

The current study discussed the integration between two computational approaches to evaluate the hydrothermal properties, such as pressure drop ( P), energy efficiency (ηeng), and absorbed energy (Qabs) of solar collectors using hybrid nanofluids. The physical problem was solved through a 3D model using Ansys 2021R1. After that, the three outputs were predicted through MATLAB R2022b using Optimize a Boosted Regression Ensemble by implementing Least-Squares Boosting (LSBoost) with Bayesian optimization (BO). Various working fluids were tested: distilled water (DW), Graphene oxide (GO-DW) nanofluids, and hybrid graphene oxide/silicon dioxide (GO/SiO2-DW) nanofluids in the mixing ratio (50:50), in the concentration range 0.01-1 vol%. The simulations covered a range of Reynolds numbers (300 ≤ Re ≤ 1500) and inlet temperatures (30 °C, 40 °C, 50 °C, and 60 °C). The results indicated that, the maximum variation compared to the first and second validations for mass flow rates and different nanofluid concentrations was between 3.96-4.71% and 4.32–6.45%. At an inlet temperature of 30 °C, GO-DW-1% exhibited the highest P, eng, and Qabs, while GO-SiO2-DW-1% closely followed with slightly lower energy values. The maximum errors between computational fluid dynamics and machine learning predictions for ( P), ( eng), and (Qabs) ranged from 2.40% to 6.32%, depending on the type of nanofluids (single or hybrid) and the volume concentration. To conclude, the current approach showed a promising result in the photothermal devices.

DOI

10.1016/j.solmat.2024.112822

Access Rights

subscription content

Link to Full Text

Share

 
COinS
 

Link to publisher version (DOI)

10.1016/j.solmat.2024.112822