Scholarlink

<p>In this study, an Artificial Neural Network (ANN) model was developed to predict the specific heat capacity of Multi-Walled Carbon Nanotube (MWCNT)-Al₂O₃/Transformer Oil hybrid nanofluid as a function of temperature (20–65&#xa0;°C) and nanoparticle volume fraction. (Experimental investigations were systematically conducted at three distinct volume fraction (0,025, 0,05, and 0,1%) to establish a comprehensive dataset. The collected experimental data (<i>n</i> = 939 total samples) were partitioned for model development and validation, with 70% (657 samples) allocated for training, and 15% each (141 samples) for testing and validation purposes. Approximate average relative error of 0,05% was compared to the experimental results. The model’s predictions showed an average deviation of −0,005% with the experimental data, proving its high accuracy. The developed ANN model achieved a high prediction accuracy with a correlation coefficient of approximately 0.9998 and a mean squared error of about 0.65. In addition, when compared with the Pak and Cho and Xuan and Roetzel models, it was observed that ANN produced more reliable results, especially at high temperatures and concentrations. This study has shown that ANN-based models are more effective than traditional methods in predicting the thermal properties of hybrid nanofluids. The findings provide significant potential for improving thermal management in power transmission systems and increasing transformer efficiency.</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 ANN-Based Prediction of Specific Heat Capacity of MWCNT–Al₂O₃/Transformer Oil Hybrid Nanofluids

  • 期刊论文

  • 发表日期：

    2026年7月22日

查看全文
  • Oğuzhan Yıldız

摘要

In this study, an Artificial Neural Network (ANN) model was developed to predict the specific heat capacity of Multi-Walled Carbon Nanotube (MWCNT)-Al₂O₃/Transformer Oil hybrid nanofluid as a function of temperature (20–65 °C) and nanoparticle volume fraction. (Experimental investigations were systematically conducted at three distinct volume fraction (0,025, 0,05, and 0,1%) to establish a comprehensive dataset. The collected experimental data (n = 939 total samples) were partitioned for model development and validation, with 70% (657 samples) allocated for training, and 15% each (141 samples) for testing and validation purposes. Approximate average relative error of 0,05% was compared to the experimental results. The model’s predictions showed an average deviation of −0,005% with the experimental data, proving its high accuracy. The developed ANN model achieved a high prediction accuracy with a correlation coefficient of approximately 0.9998 and a mean squared error of about 0.65. In addition, when compared with the Pak and Cho and Xuan and Roetzel models, it was observed that ANN produced more reliable results, especially at high temperatures and concentrations. This study has shown that ANN-based models are more effective than traditional methods in predicting the thermal properties of hybrid nanofluids. The findings provide significant potential for improving thermal management in power transmission systems and increasing transformer efficiency.