An improved partial similitude method for dynamic characteristic of rotor systems based on Levenberg-Marquardt method

Similitude theory is widely used to predict the structural responses of the full-size prototype through scaled models and is categorized into complete and partial similitude according to the way of scaling parameters. Partial similitude is more flexible in practical design engineering, but its accuracy still needs to be improved. This study presents an improved partial similitude approach, which can be used to establish the scaling laws and understand the scaling behaviors of rotor systems. The aim is to overcome the insufficient accuracy of the existing methods in predicting the dynamic characteristic of rotor systems. The power in scaling laws is represented by a function, which is related to the scaling factor of the prototype to be predicted. In this work, the power function is developed by incorporating the Levenberg-Marquardt method and sensitivity analysis. The accuracy and effectiveness of the proposed partial similitude method are validated through numerical and experimental case studies, where the distortion in geometrical dimension is considered. The dynamic characteristic of prototypes is predicted through the scaled model. Furthermore, the accuracy of the proposed method is evaluated by using the existing methods. Both the numerical and experimental case studies indicate that the proposed method achieves higher accuracy than the existing methods in predicting the dynamic characteristic of rotor systems.

Automated summary

Similitude theory is widely used in predicting the structural responses of full-size prototypes through scaled models. Partial similitude, more flexible in practical design, needs improvement in accuracy. An improved partial similitude approach is presented in this study to establish scaling laws for rotor systems and to overcome the insufficient accuracy of existing methods in predicting dynamic characteristics. The accuracy and effectiveness of this approach are validated through numerical and experimental case studies, showing higher accuracy than existing methods.