微小通道内过冷流动沸腾阻力特性实验及预测研究

Experimental and predictive study on pressure drop of subcooled flow boiling in a mini-channel

实验探究了微小圆管内(内径1 mm)过冷水流动沸腾的阻力特性,参数范围:热通量4.0~5.6 MW/m^(2),压力3.0~5.0 MPa,质量流速2000~4200 kg/(m^(2)‧s),进口热力学干度-0.50~-0.10。获取了质量流速、压力、热通量等参数对过冷沸腾阻力的影响,并重点关注其预测方法。将测试数据与典型阻力关联式对比,结果表明,由于高热流、微通道等特殊因素,导致大部分阻力关联式的预测精度不够理想。为更准确预测高热流过冷沸腾阻力,基于LeakyReLU函数,建立了遗传算法优化的极限学习机模型(GA-ELM),其预测精度优于传统关联式(平均绝对误差为2.0%),且泛化性良好。研究工作可为微小尺度流动换热系统设计优化提供支撑。

The pressure drop characteristics of subcooled water flow boiling in a mini-tube(1 mm)were experimentally investigated.The experimental parameters were as follows:heat flux 4.0—5.6 MW/m^(2),pressure 3.0—5.0 MPa,mass flow rate 2000—4200 kg/(m^(2)‧s),and inlet thermodynamic quality-0.50—-0.10.The effects of mass flow rate,pressure,heat flux and other parameters on the subcooling boiling resistance were obtained,and the prediction method was focused on.Comparison of the experimental data with typical pressure drop correlations indicates that the accuracy of the prediction for most pressure drop correlations is not sufficient due to special factors such as high heat flux and micro-channels.In order to predict the pressure drop of subcooled boiling of high heat flux more accurately,the extreme learning machine model optimized by genetic algorithm(GA-ELM)is established based on LeakyReLU function.The prediction accuracy of GA-ELM is better than the traditional correlations(the average absolute error is 2.0%),with well generalization ability.This study will support design optimization of micro/mini-scale flow heat transfer systems.