Forced response system identification of gas turbine fan flutter

Abstract

Flutter in aero-engine fans continues to present major safety concerns and cost millions of dollars in maintenance from high-cycle fatigue. New technologies to improve safety and reduce cost require knowledge of damping levels, but determining the in-situ flutter damping of an installed fan remains challenging. This thesis introduces a first-of-its kind forced response system identification approach to measure fan flutter damping in a full gas turbine aero-engine. A reduced order modeling framework that captures the effect of actuator dynamics, flutter and stall dynamics, and sensor dynamics on forced response was developed for rigorous design of the experiment. A statistical analysis demonstrates that the experiment is capable of measuring flutter damping within 15% for most flutter modes throughout the entire operating range of the Pratt & Whitney 615 turbofan engine.

The least damped, and therefore most limiting, flutter modes yield the least error with flutter margin estimated within 1.5 points or 14%. A key enabling technology for this experiment is a high power zero-net mass flux actuator capable of exciting the flutter modes. Experimental characterization of the actuator dynamics at this thus far un-explored scale demonstrates the limitations of current reduced order models in predicting actuator performance. This thesis establishes the need for experimental characterization of such actuators for future forced response experiments. Virtual forced response system identification experiments demonstrate robustness of the approach to noise sources, with damping measurements succeeding even with noise levels above those typical in engine testing. Fans with different performance characteristics were simulated demonstrating that flutter damping can be measured with similar levels of error as compared with the PW615 engine.

Guidelines on system identification, data processing, and experimental setup are developed for future forced response experiments. Finally, detailed design demonstrates the experiment can be conducted safely and with no impact on the engine steady state performance. The thesis contributions are (1) a new approach to measuring fan flutter damping in aero-engines that enables the development of flutter mitigation technologies and advanced prognostics, and (2) a forced response reduced order modeling framework that provides new capability for experimental design and coupled engine system dynamic modeling that can identify detrimental engine conditions.