基于实验数据同化的扩张管分离流动数值模拟

Numerical simulation of flow separation in diffusers using measurement data assimilation

为了提高RANS(Reynolds-averaged Navier-Stokes equations)模型对分离流动的预测精度,本文采用集合卡尔曼滤波数据同化方法,利用扩张管出口流向速度作为同化数据,对BSL RSM(baseline Reynolds stress model)湍流模型的模型常数进行了优化。默认模型常数的BSL RSM湍流模型预测的壁面压力系数和分离区位置与实验结果差距很大。通过模型常数优化,提高了BSL RSM湍流模型对分离流动的预测精度,数值模拟与实验的壁面压力系数曲线更接近,流动分离区的位置与实验结果基本一致。首次证明了基于数据同化的湍流模型常数优化方法应用于分离流动准确预测的可行性与有效性。

In order to improve the prediction accuracy of RANS(Reynolds-averaged Navier-Stokes equa-tions)model for separated flow,the ensemble Kalman filter data assimilation method was adopted to optimize the model constants of BSL RSM(baseline Reynolds stress model)turbulence model by using the stream ve-locity at the outlet of an expansion pipe as the assimilation data.The wall pressure coefficient and the loca-tion of the separation zone predicted by the BSL RSM turbulence model with the default model constant dif-fered greatly from the test results.Through optimization of model constants,the precision of BSL RSM turbu-lence model to predict separated flow was improved.The wall pressure coefficient curve of the numerical sim-ulation was closer to that of the experiment,and the position of the flow separation zone was basically consis-tent with that of the experiment.The feasibility and effectiveness of applying the turbulence model constant optimization method based on data assimilation to the accurate prediction of flow separation were proved for the first time.