采用 AK-MCS 的燃气轮机轮盘疲劳寿命预测及可靠性评估

To ensure the performance and reliability of gas turbine discs in practical work, comprehensive consideration is given to various uncertainties associated with the discs during the service of gas turbines. Reasonable quantification is achieved through probability representation, employing the active learning Kriging surrogate model and Monte Carlo simulation (AK-MCS). A disc life prediction agent model is established, considering the hybrid uncertainty of the gas turbine disc. The low-cycle fatigue life of the disc is reasonably evaluated. This is followed by the development of a comprehensive disc low-cycle fatigue life reliability analysis flow as well as the completion of the disc low-cycle fatigue life reliability evaluation and sensitivity analysis. The research results indicate that the maximum equivalent stress value of the disc is 773.518 MPa; the maximum strain value is 0.004 73; the local danger point is located at the corner below the third tooth surface of the wheel groove; and the predicted life is 1.177 × 10⁴ weeks. Given a design life of 8×10³ weeks, the failure probability of the disc can reach 0.043 1, with a reliability value of 0.956 9. The sensitivity analysis results show that the main parameters affecting the fatigue life include the contact pressure of the wheel groove tooth surface and the disc speed. The research results demonstrate the application advantages of AK-MCS in the disc reliability analysis, providing methods and reference data for the design of gas turbine discs in engineering.