竖直通道内流动沸腾气泡脱离行为的数据驱动分析

Data Driven Analysis for Departure Behavior Characteristics of Flow Boiling Bubbles in the Vertical Channel

为了研究过冷沸腾过程中气泡的脱离特性,基于统计学分析方法,利用公开发表的实验数据和经验关联式构建驱动分析数据库,运用有限混合模型和标准方差分解方法对竖直过热壁面上流动沸腾气泡脱离直径的影响因素进行数据驱动分析研究。将描述流动沸腾的量纲一参数组作为模型的输入变量,气泡脱离直径作为输出变量,采用Kullback-Leibler散度和χ2距离对于输入变量的敏感性程度顺序分别进行评估。初步的数据驱动分析结果表明:雷诺数和气液密度比对气泡脱离直径的影响最为显著,与传统方法取得的结果具有一致性,证明了数据驱动方法在分析物理问题上的可行性;在现有数据库的基础上,有限混合模型比标准方差分解方法更具一般性,可以直接基于数据库进行应用。

To reveal the bubble detachment behaviors during subcooled boiling process,a data driven statistical analysis method according to the finite mixing model(FMM)and the standard variance decomposition(Sobol model)is employed to investigate the influence factors of bubble departure diameter on superheated wall surface in vertical flow boiling.The data driven database is constructed and modified based on the experimental data and empirical correlations collected from literature.The dimensionless parameters describing flow boiling are chosen as the input variables of this model,and bubble departure diameter as the output variable.Kullback-Leibler divergence andχ2 distance are adopted for determining the importance degree order of input variables.The preliminary results of sensitivity analysis show that the effects of Reynolds number and vapour-liquid density ratio on bubble departure diameter are the most significant among these inputs.The results coincide well with those obtained by the traditional methods,which indicates the feasibility of the data-driven method in analyzing engineering problems.In addition,FMM is more general than Sobol model based on the present database,and can be applied directly based on these data.