Development of a Precision Vibration Analysis Framework for Structural Safety Prediction of Gas Turbine Blades

Abstract

Blades in the gas turbine engine are subjected to resonant excitation which causes high cycle fatigue accumulation, and eventually may lead to failure of the blades. To avoid this, the structural response associated with the resonant condition should be predicted, and it is indispensable to accurately predict the dynamic characteristics of the blade at its preliminary design process. In this thesis, an advanced vibration analysis framework including the capability to predict the crucial physical phenomena in gas turbine blades, i.e., geometric nonlinearity, high-speed rotational and thermal effects, is developed. Three-dimensional co-rotational (CR) solid element is employed for the geometric nonlinearity. On the other hand, a large amount of discretized elements may be required for more accurate analysis of a complex blade configuration, and this causes significant increase in computational cost. To overcome such problem, reduced order modeling based on the proper orthogonal decomposition (POD-ROM) analysis is also developed. The numerical examination is carried out aimed on the first-stage turbine blade of 75MW gas turbine engine under various operating conditions, i.e., high-speed rotation and high temperature. The present analyses are validated by comparing with the results obtained by the commercial software, ANSYS. As a result, it is found that the present analyses show good correlation by comparison the natural frequencies and mode shapes. And, by using the present POD-ROM, significant improvement in computational cost is accomplished when compared with the full order model (FOM) and ANSYS analysis. Also, the snapshot collection time for the initial POD-ROM analysis is significantly improved by parallel computation base on the domain decomposition.