Genetic-Algorithm-Optimized 1D Beam Rotordynamics Model Using a 3D Solid High-Fidelity Model

Abstract

Abstract. Accurately predicting critical speeds and shaft dynamic responses to imbalance is essential in rotordynamic analysis. Comprehensive simulations must be performed prior to manufacturing gas turbines or jet engines to avoid costly post-production design modifications. While 3D solid finite element models provide high accuracy, they demand substantial computational resources for both preprocessing and post-processing. Consequently, industries continue to rely on 1D beam models due to their efficiency, despite their inherent limitations in terms of accuracy. This article presents a novel approach for optimizing 1D rotor models using a genetic algorithm (GA). Correction parameters for the mass, transverse moment of inertia, and Young's modulus are introduced into the 1D beam model and optimized using a GA with the objective of minimizing the discrepancies in amplitude, phase angle, and gravity-induced displacement compared with a 3D solid model. Transient analyses are conducted for three cases: (1) a conventional 1D beam model, (2) a 3D solid model, and (3) a GA-optimized 1D beam model. The results demonstrate that the GA-optimized 1D beam model closely replicates the behavior of the 3D solid model, whereas the conventional 1D beam model exhibits significant deviations, particularly near critical speeds. Additionally, the results reveal that variations in disk size substantially affect critical speeds, leading to changes in imbalance amplitude and overall dynamic response.