基于POD和代理模型的气固流化床流场预测  

Flow Field Prediction of Gas-Solid Fluidized Bed Based on POD and Surrogate Model

作　　者：冯佳磊 赵晓辉 屠楠 方嘉宾 胡瑞河 许一博 FENG Jialei;ZHAO Xiaohui;TU Nan;FANG Jiabin;HU Ruihe;XU Yibo(School of Mechanical and Electrical Engineering,Xi′an Polytechnic University,Xi′an 710600,China;Northwest Electric Power Design Institute Co.,Ltd.,China Power Engineering Consulting Group,Xi′an 710075,China;School of Chemical Engineering and Technology,Xi′an Jiaotong University,Xi′an 712000,China)

机构地区：[1]西安工程大学机电工程学院,陕西西安710600 [2]中国电力工程顾问集团西北电力设计院有限公司,陕西西安710075 [3]西安交通大学化学工程与技术学院,陕西西安712000

出　　处：《气体物理》2025年第1期45-56,共12页Physics of Gases

基　　金：国家自然科学基金(22378321);中国能源建设股份有限公司重大科技项目(CEEC2021-ZDYF-02)。

摘　　要：针对CFD数值模拟气固流化床中两相流动复杂、计算成本高昂的问题,建立了一种基于本征正交分解(proper orthogonal decomposition,POD)和代理模型的非反应性气固流化床流体动力学降阶模型,通过POD-RBF方法和POD-Kriging方法分别对流化床的气固两相流动过程进行了预测,将不同气速下的降阶模型解与全阶模型解进行了比较,并探究了气体速度对床层运动的影响。结果表明,在用于生成POD的设计参数范围内,降阶模型与全阶模型结果之间的平均相对误差(mean relative error,MRE)不超过5%,两种方法对设计工况的预测精度均高于非设计工况。对于流化床内压力、固含率、气体速度的分布,POD-RBF方法的预测精度均高于POD-Kriging方法,其中对压力场的预测精度最高,对速度场的预测精度最低。降阶模型在保持较高的重建和预测精度的同时,将计算效率提高了近400倍。Aiming at the problems of complex two-phase flow and high computational cost in CFD numerical simulation of gas-solid fluidized beds,a reduced-order model for investigating the fluid dynamics in a non-reactive gas-solid fluidized bed was established based on the proper orthogonal decomposition(POD)and the surrogate model.The gas-solid two-phase flow process in the fluidized bed was predicted by adopting the POD-RBF method and the POD-Kriging method,respectively.Besides,the reduced-order model solutions with different gas velocities were compared with the full-order model solutions,and the effect of gas velocity on the bed motion was also studied.The results show that the mean relative error(MRE)between the reduced-order model and the full-order model does not exceed 5%in the range of design parameters used to generate the POD,and the prediction accuracy of the two methods is higher for the design conditions than for the non-design conditions.For the distributions of pressure,solids content,and gas velocity in the fluidized bed,the prediction accuracy of the POD-RBF method is higher than that of the POD-Kriging method.The prediction for the pressure field has the highest accuracy and the velocity field has the lowest one in the fluidized bed by employing these two reduced-order methods.Moreover,the reduced-order model improves the computational efficiency by nearly 400 times while maintaining high reconstruction and prediction accuracy.

关 键 词：气固流化床 两相流 数值模拟 模型降阶 本征正交分解 

分 类 号：O359[理学—流体力学] V211[理学—力学]