【摘要】 由于超临界流体在拟临界区复杂的物性变化,其管内流动的对流换热模式与亚临界流体有很大不同。除实验研究和理论分析外,许多学者建立了用于预测换热系数的超临界流体换热关联式,但现有的关联式均不能在宽广的参数范围上取得满意的预测效果。因此,在建立了超临界水竖直上升流宽范围换热实验数据库的基础上,运用机器学习方法分析了影响换热的典型无量纲数与努塞尔数的关系,建立客观评价指标,选出适当的关联形式并建立新的换热关联式。换热关系的非线性分析结果表明,其非线性特征在密度比上的表现最为显著。结合非线性特征的趋势和客观评价指标的对比,在换热关联式中引入对数型的密度比项,由此提出宽范围的超临界水竖直上升流换热关联式,其在数据库上的预测表现明显优于现有关联式。更多还原

【Abstract】 Due to the complex physical property changes of supercritical fluid in the pseudo-critical region,the convective heat transfer mode of the tube flow is very different from that of subcritical fluid.In addition to experimental research and theoretical analysis,many scholars have established supercritical fluid heat transfer correlation for the prediction of heat transfer coefficient,but none of the existing correlation can achieve satisfactory prediction results over a wide range of parameters.Therefore,based on the established heat transfer experiment database of the vertical upflow of supercritical water,the machine learning methods are used to analyze the relationship between the typical dimensionless numbers that have impact on heat transfer and the Nusselt number.According to objective evaluation indicators,appropriate correlated form is selected and a new heat transfer correlation is established.The results of non-linear analysis of heat transfer relationship show that this nonlinearity is most pronounced in the density ratio.Combining the trend of the non-linear characteristics and the evaluation results of objective indicators,the logarithmic density ratio term is introduced into the heat transfer correlation,thereby a wide-range heat transfer correlation of the vertical upflow of supercritical water is established and its prediction performance on the database is much better than existing correlations.更多还原