微胶囊浆体相变换热多尺度建模及其耦合处理

Multi-scale modeling of heat transfer during phase change of microencapsulated slurry and the coupling strategy between different scales

为了准确预测微胶囊浆体的相变流动过程,构建了关联型多尺度耦合模型。通过双向耦合信息传递方式完成了微胶囊和载流体间的热交换过程,实现了微观尺度的胶囊相变换热模型与宏观尺度的浆体流动换热模型之间的关联耦合。针对尺度间的耦合处理,提出了3种相变数据的提炼简化与信息传递方案。结果表明,多元回归分析方案适于各参数关联关系明显的工况,具有较好的预测能力;RBF神经网络方案呈现出整体准确而个别偏离的特点,因而适用于模型具备容错能力的情形,且通过增加训练样本可提高其预测性能;分析解算方案在粒径小于3mm时可获得理想结果,最大误差仅为2.91%。

A multi-scale model was proposed based on correlative algorithm for accurate prediction of flow and heat transfer during phase change of microencapsulated phase change material slurry(MPCMS).Interchange of heat transfer information between microcapsule and carrier fluid was achieved by two-way data-based coupling strategy,and flow and heat transfer model in macro-scale was correlated with a phase-change heat transfer model in micro-scale.Furthermore,three strategies were used to refine and reduce massive phase change data in micro-scale,which further simplified the process of transferring thermal information between macro-scale and micro-scale models.The results show that the multi-regression method is useful for dataset regression and prediction.RBF network has a great overall prediction performance based on known phase change information,while deviation of few data also exists.Therefore,the RBF network can be used only for models permitting mistakes.However,increase of training can improve its prediction performance.The maximum relative error is approximately 2.91%when the microcapsule size is less than 3 mm.