Experimental Analysis of a Novel Piezoelectric Flow Injection Valve for Flutter Boundary Mapping in Turbomachinery

Flutter remains a challenge for turbomachinery design and operation. Current testing methods often rely on a binary “flutter” / “no flutter” determination to identify the flutter boundary. Measuring actual aeromechanical damping provides a more accurate determination of the flutter boundary and reduces risk of entering flutter. Moreover, it provides data validation for engine simulations. It is possible to measure aeromechanical damping by using actuators to excite modes of interest during engine testing and measuring their vibratory response. However, the limited bandwidth and authority of current actuators restrict how broadly this approach can be applied. Our collaborative team recently developed a flow-injection valve using piezoelectric actuation to expand testing. This paper presents the development of a novel Non-contact Oscillatory Valve Actuator, NOVA, and characterizes its performance in benchtop testing. Primary performance metrics include mass flow rate variation through the valve (authority) and frequency of operation (bandwidth). Collected data indicates some frequency dependence of authority metrics, notably a reduction in performance across a broad range from 700-800 Hz with a notable increase in performance (even relative to the nominal baseline) by 900-1000 Hz. Performance generally scaled linearly with voltage amplitude. Although deviations from anticipated linear relationships and losses of expected authority under certain operating conditions are not ideal, it is possible to use internal valve measurements to assess performance. Most importantly, the frequency dependence is repeatable across various operating pressures and mean mass flow rates. These measurements provide feedback that enables adjustment of operating parameters during on-engine testing, such as increasing fluid supply pressure, to meet performance requirements. Ongoing analysis and valve development is investigating potential sources of frequency dependence and determining how to improve performance linearity.