Assimilation of Disparate Data for Improving the Performance Prediction of Body-Force Model

Despite the extensive application of three-dimensional Reynolds-averaged Navier-Stokes equation (RANS) in axial compressor numerical simulations, body-force model (BFM) also plays its own role profiting from its low computation cost. However, the computation accuracy highly depends on the modeling of blade force, which usually involves several parameter constants. In this work, data assimilation based on Ensemble Kalman Filter (EnKF) was employed to optimize these model constants in BFM. Previous work associated with data assimilation mainly focuses on employing only one data source. Considering the various measurement quantities in engineering practice, disparate data were incorporated into the assimilation method to improve the prediction. The test case of a low-speed axial compressor was provided. Only one single data source, i.e., total pressure ratio, was first employed as the observation data in EnKF. And to reveal the superiority of the disparate data assimilation, total pressure ratio and isentropic efficiency were then incorporated to improve the performance prediction. The converged results reveal the robustness of disparate data assimilation based on EnKF. At last, the rationality of the optimized constants is verified further through the great agreement between the measurement and the prediction of BFM, with regard to the radial profile and the performance at another rotational speed.

Automated summary

This study employed Ensemble Kalman Filter (EnKF) based data assimilation to optimize the model constants in body-force model (BFM). The assimilation method incorporated disparate data sources to improve the prediction accuracy. The results showed the effectiveness of the disparate data assimilation and the rationality of the optimized constants.