Prediction of Ground Noise Footprint using Reduced Order Model

Abstract

A helicopter is superior in terms of mission flexibility and versatility due to its unique ability to perform vertical takeoff and landing. But its use is limited by the distinctive noise caused by Blade Vortex Interaction (BVI) noise. To solve this problem, it is necessary to accurately and efficiently predict the noise generated by helicopter rotor. In this study, a Reduced-order model (ROM) is constructed to efficiently predict the noise hemisphere generated by rotors, and its effectiveness and accuracy are demonstrated by applying it to an AH-1G rotor. To analyze the aerodynamic noise at various locations and express it in multiple forms of noise metrics, information on the aerodynamic load distribution and blade passage information are required, ROM is constructed to model them. Using high-fidelity URANS calculations for five variables consisting of operational conditions and control inputs, the load distribution in the normal, chordwise, and spanwise directions are modeled. To verify accuracy, the predicted load distribution by the ROM is compared to the one calculated by CFD, resulting in an l2 error of approximately 3% and an accuracy of less than 2% for the aerodynamic coefficient. Additionally, the location and intensity of blade-vortex interaction noise that occurs during descent are accurately predicted. Using the constructed ROM, control inputs that satisfy the trim state for operational conditions can be calculated, and tonal noise in the noise hemisphere located under the rotor is calculated using the FW-H formula based on the load distribution and blade passage information. The average OASPL of the noise hemisphere is predicted with an error of less than 0.47 dB, and the BVI noise is predicted with an average error of 0.49 dB. Finally, a framework for predicting ground noise footprints for various paths in a time-efficient manner is presented using the constructed ROM. The noise impact on the ground from three paths according to the descent method is evaluated.