Zonal Reduced-Order Modeling of Unsteady Flow Field

Abstract

The utilization of real-world data in cyberspace is becoming attractive in various fields due to the massive growth of sensing and networking technologies. It is expected to utilize such a data-rich environment to improve engineering simulations in computer-aided engineering (CAE). Data assimilation is one of methodologies to statistically integrate a numerical model and measurement data, and it is expected to be a key technology to take advantage of measured data in CAE. However, the additional cost of data assimilation is not always affordable in CAE simulations. In this study, we consider the cost reduction of numerical flow simulation with the help of a reduced-order model, which encodes a flow field into a low-dimensional representation. Since the prediction accuracy of existing ROMs are limited in complex flow fields, we investigate here a zonal hybrid approach of a full-order model and a reduced-order model.