Semi-Empirical Correlation of Gas Cooling Heat Transfer of Supercritical Carbon Dioxide in Microchannels

Abstract

This paper provides a comprehensive review of existing correlations for supercritical heat transfer of CO 2 in microchannels, as well as a comparison of these correlations with experimentally measured data. Based on the experimental data, a new semi-empirical correlation is developed to predict the gas cooling heat transfer coefficient of supercritical CO 2 in microchannels, within an error of 15% for most (91%) of the presented experimental data that were obtained in an 11-port microchannel tube with an internal diameter of 0.79 mm and with a pressure range of 8 to 10 MPa and mass flux range of 300 to 1200 kg/m 2 s.

Experimental study of in-tube cooling heat transfer and pressure drop characteristics of R134a at supercritical pressures

“…Thus, the effects of the property variations along the entire length of the test section must be considered, especially the specific heat, c p , which most significantly affects the heat transfer at supercritical pressures [29]. The ratio of the densities evaluated at the wall temperature to that evaluated at the bulk temperature is also important since this describes the effect of the density gradient in the radial direction.…”

“…Horizontal tube Re f > 2 Â 10 4 ID = 7.73 mm Dang and Hihara [16] 2004 CO 2 Experimental Horizontal tube 3 Â 10 4 < Re f < 1.5 Â 10 5 ID = 1-6 mm Son and Park [24] 2005 CO 2 Experimental Horizontal tube 2.2 Â 10 4 < Re f < 1.5 Â 10 5 ID = 7.75 mm Kuang et al [29] 2008 CO 2 Experimental Microchannels -ID = 0.5-2 mm the present study. The R134a inlet pressure was measured using a pressure transducer (Model EJA430A).…”

Neural network prediction of cooling heat transfer characteristics of supercritical R1234ze(E) in horizontal tube

“…或 Gnielinski [2] 方程)通过考虑径向密度变化 [3,4] 、径向比热变化 [5,6] 、浮升力效应 [7,8] 、流动加 速效应 [9,10] 等已建立起了大量的适用于不同工质、不同工况范围的超临界传热努塞尔数关联 式。然而，虽然经验关联式的预测结果与实际数据间的平均绝对偏差已经很小，但是其在拟 临界温度(T pc )附近对于传热系数的预测值与实际数据之间仍然存在着相当大的差异。因此， 寻求超临界传热预测的新方法是必要的。 近年来，随着计算机算法的迅猛发展，人工神经网络(ANN)在数据预测方面的应用受 到了特别关注，并已经成功地应用于众多领域 [11][12][13][14] 。当前，已有部分学者探究了 ANN 在预测 超临界流体管内对流传热系数方面的准确性和有效性。Ye 等人 [15] 将 ANN 模型预测结果与 Jackson、Hwan Yeol Kim、Bringer 等人建立的经验关联式进行了对比，指出 ANN 模型具有优 异的学习能力和令人满意的泛化性能，尤其是在浮升力和流动加速可以忽略的情况下。Zhu 等人 [16] 建立了具有两层隐藏层的反向传播神经网络(BPNN) ，并将其与 Bishop、Jackson、 Morky 和 Yu 的经验关联式进行对比。 无论是对正常传热、 传热强化还是传热恶化情况， BPNN 均能够更加准确、快速地预测超临界 CO 2 的传热系数。Rajendra Prasad 等人 [17] 基于 CFD 模拟 数据集训练了具有四层隐藏层、每层 15 个神经元的 BPNN 模型，其预测结果的绝对平均偏差 仅为 3.49 %。除了上述针对于超临界 CO 2 的 ANN 传热预测研究外，也有部分学者 [18,19] 图 1 传热数据集的全部数据点分布 Fig. 1 Distributions of all data points for the heat transfer dataset 2.2 传热经验关联式建立 参照文献 [22] ， 基于以上传热数据集的 1657 组数据点， 采用分段线性方式， 以 Dittus Boelter 致网络结构的复杂和训练时间的延长。已有学者 [15,19,20,23] 证明，具有一层隐藏层的 BPNN 足以 获得令人满意的预测精度。因此，本工作均采用一层隐藏层网络结构，如图 2 所示。 图 2 具有一层隐藏层的 BPNN 结构 Fig.…”

Computational investigations on heat transfer of turbulent supercritical CO2 in large tubes using RANS modelling

“…Two commonly considered Nusselt formulations were suggested for constant-property fluids: Dittus-Bolter equation [19] and Gnielinski correlation [20]. Experimental data [22,33,110,194,207,208] indicate consistent failure of these correlations in predicting turbulent sCO2 heat transfer, particularly near . The errors are mainly due to predominant property variations in the radial direction caused by the radial temperature gradient.…”

“…The errors are mainly due to predominant property variations in the radial direction caused by the radial temperature gradient. Hence, it was suggested to modify the existing correlations by evaluating the fluid properties at the film temperature (the average value of bulk and wall temperature) [22] or by introducing correction factors (usually the ratio of specific heat and density) to represent wall-to-bulk property variations [24,25,33,35,37,39,43,194,207]. These modifications improved the prediction accuracy under certain conditions.…”