Estimation of flutter boundary from random responses due to turbulence at subcritical speeds

Abstract

The present paper describes a new efficient technique for estimating the flutter or divergence boundary from responses of a model to wind tunnel turbulence. The boundary can be predicted without estimating or measuring the dampings and frequencies of the aeroelastic modes. The sampled time response is modeled by the mixed autoregressive moving average process. The orders and coefficients of both autoregressive and moving average parts of the process are determined with the aid of Akaike's estimation procedure. The stability boundary is estimated by using Jury's stability determinants which are expressed in terms of the autoregressive coefficients alone. The modal frequencies and dampings are also evaluated from the coefficients. The technique has been applied with success to signals from a cantilever wing model tested in a low supersonic flow. Comparison between the actual and estimated flutter boundaries shows that an accurate estimation can be made from data obtained in a narrow range of the dynamic pressure which is sufficiently below the boundary.