A hyper reduced order model for the transient thermo-mechanical behavior analysis of high temperature blades

This paper presents the development of a hyper-reduced-order model combining proper orthogonal decomposition and the discrete empirical interpolation method. The developed model solves the problem of directly measuring and quickly predicting the instantaneous thermal state of the high-temperature rotor blades of thermal machinery. This work begins by establishing a snapshot matrix and obtaining a reduced orthogonal basis using a high-fidelity, full-order finite element model. Interpolation nodes and an interpolation matrix are then determined by adopting discrete empirical interpolation, enabling the rapid construction of a nonlinear system matrix and load vector in the thermodynamic coupling equation. Finally, temperature and strain fields are reconstructed. Numerical examples show that compared with the commercial software full-order finite element model, the calculation speed of the hyper-reduced-order rapid calculation model is 24.63 times and 21.17 times higher than that of the commercial software full-order finite element model, respectively. The traditional eigen-orthogonal decomposition reduced-order method offers limited increases in speed when dealing with nonlinear systems, and the overall calculation accuracy of the model in this paper is within the acceptable range.