Experimental evaluation and artificial neural network modeling of thermal conductivity of water based nanofluid containing magnetic copper nanoparticles
Experimental evaluation and artificial neural network modeling of thermal conductivity of water based nanofluid containing magnetic copper nanoparticles
复制标题
DOI：
10.1016/j.physa.2019.124127
复制
发表时间：
2020-08-01
影响因子：
3.3
通讯作者：
Kumar, Ravinder
中科院分区：
文献类型：
--
作者：
Ghazvini, Mahyar;Maddah, Heydar;Kumar, Ravinder
纳米流体的导热系数对流体的传热能力起着至关重要的作用。将纳米颗粒添加到基础油中可以提高导热比。在本研究中，利用实验数据，应用人工神经网络方法对改性纳米铜/水纳米流体的导热比进行了建模。利用KD2Pro测量了纳米流体在不同流体温度、纳米粒子浓度和直径下的导热系数。此外，为了模拟纳米流体的导热系数与温度、固体体积分数和直径的关系，通过应用人工神经网络并考虑实验数据，提出了一个关联式。根据统计精度分析，该任务的最佳建模结构是具有6个隐含神经元的两层前馈神经网络模型。该模型预测实验数据的平均绝对百分比误差(MAPE)为1.09%，均方误差(MSE)为2.5×10(-4)，决定系数(R-2)为0.99。基于比较结果，神经网络模型能够较好地预测纳米流体导热系数的提高。(C)2020爱思唯尔B.V.保留所有权利。
Thermal conductivity of nanofluids performs as a crucial role in heat transfer capacity of fluids. Nanoparticles' addition to a base fluid results in enhancing thermal conductivity ratio. In this investigation, the thermal conductivity ratio of modified copper nanoparticles/water nanofluid is modeled by applying artificial neural network approaches utilizing experimental data. The nanofluids' thermal conductivity at various fluid temperatures, nanoparticle concentration and diameter was measured experimentally using KD2Pro. Additionally, in order to model the nanofluid's thermal conductivity with respect to temperature, solid volume fraction, and diameter, a correlation is proposed by applying artificial neural networks and considering the experimental data. According to statistical accuracy analysis, the best structure to model this task is a two-layer feedforward ANN model with 6 hidden neurons. This model predicted the experimental data with Mean absolute percentage error (MAPE) of 1.09%, mean square errors (MSE) of 2.5 x 10(-4), and coefficient of determination (R-2) of 0.99. Based on the comparison results, the ANN model is able to predict the enhancement in the thermal conductivity of nanofluids favorably. (C) 2020 Elsevier B.V. All rights reserved.