Scholarlink

<p>This work implies to improve the flat plate solar collector (FPSC) performance accompanied with this mono (Al<sub>2</sub>O<sub>3</sub> and MgO) and hybrid nanofluids (Al<sub>2</sub>O<sub>3</sub> + MgO) 1:1 under varying volume proportions (0.2, 0.4, 0.6 and 0.8%) with various mass rates (0.016, 0.033 and 0.05&#xa0;kg&#xa0;s<sup>−1</sup>) in comparison with Water + Ethylene Glycol (65:35) as reference fluid. Energy Dispersive X-Ray Analysis (EDAX) and scanning electron microscopy (SEM) investigation was utilized to investigate the elemental composition and surface morphology of the nanofluids. Outcome denoted that working fluid’s thermal conductivity increased by 81% when hybrid nanofluid (HNFs) was mixed with the reference fluid. This improvement was 72% for Al<sub>2</sub>O<sub>3</sub> and 55.8% for MgO nanofluids individually, compared to reference fluid. Experimental findings showed that the output temperature, heat transfer rate, collector efficiency and pressure drop were 41.8&#xa0;°C, 753.9&#xa0;Wm<sup>−2</sup>, 71% and 12,065.4&#xa0;Nm<sup>−2</sup> respectively achieved with input parameters of 0.6vol.% HNF at 0.05&#xa0;kgs<sup>−1</sup> mass rate. This study was initially designed using the Design of Experiments (DoE), employing a central composite design within response surface methodology (RSM).The model was evaluated for probability of 95% confidence level and the statistical equation was developed for each response such as outlet temperature, heat transfer rate, collector efficiency and pressure drop. The predicted equation was evaluated against the experimental data, and the difference was determined to be lesser than 5%.Then, the predicted equation trained in artificial neural networks (ANN) and the predicted results were compared with RSM for good agreements with experimental data. The inter connection among input and output factors is displayed using a feedforward multi-layer perceptron network. The findings indicate that the performance characteristics can be properly modeled by the ANN with regression coefficients (R<sup>2</sup>) ranging from 0.9064 to 1. The accuracy of ANN prediction was better than RSM.</p> Graphical abstract <p></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Experimental investigation and numerical predictions on the performance of FPSC utilizing mono and hybrid nanofluids using ANN and RSM

  • 期刊论文

  • 发表日期：

    2025年7月22日

查看全文
  • G. Saravanan,
  • S. Murugapoopathi,
  • K. Muninathan

摘要

This work implies to improve the flat plate solar collector (FPSC) performance accompanied with this mono (Al2O3 and MgO) and hybrid nanofluids (Al2O3 + MgO) 1:1 under varying volume proportions (0.2, 0.4, 0.6 and 0.8%) with various mass rates (0.016, 0.033 and 0.05 kg s−1) in comparison with Water + Ethylene Glycol (65:35) as reference fluid. Energy Dispersive X-Ray Analysis (EDAX) and scanning electron microscopy (SEM) investigation was utilized to investigate the elemental composition and surface morphology of the nanofluids. Outcome denoted that working fluid’s thermal conductivity increased by 81% when hybrid nanofluid (HNFs) was mixed with the reference fluid. This improvement was 72% for Al2O3 and 55.8% for MgO nanofluids individually, compared to reference fluid. Experimental findings showed that the output temperature, heat transfer rate, collector efficiency and pressure drop were 41.8 °C, 753.9 Wm−2, 71% and 12,065.4 Nm−2 respectively achieved with input parameters of 0.6vol.% HNF at 0.05 kgs−1 mass rate. This study was initially designed using the Design of Experiments (DoE), employing a central composite design within response surface methodology (RSM).The model was evaluated for probability of 95% confidence level and the statistical equation was developed for each response such as outlet temperature, heat transfer rate, collector efficiency and pressure drop. The predicted equation was evaluated against the experimental data, and the difference was determined to be lesser than 5%.Then, the predicted equation trained in artificial neural networks (ANN) and the predicted results were compared with RSM for good agreements with experimental data. The inter connection among input and output factors is displayed using a feedforward multi-layer perceptron network. The findings indicate that the performance characteristics can be properly modeled by the ANN with regression coefficients (R2) ranging from 0.9064 to 1. The accuracy of ANN prediction was better than RSM.

Graphical abstract