Trivariate Spectral Modeling of Space Shuttle Flight Data

Abstract

Acceleration time histories recorded during liftoff in six Space Shuttle flights are analyzed digitally. Accelerations are measured in the three axial directions of the shuttle using accelerometers calibrated for the 0–50 Hz range. The event analyzed involves the interval beginning at solid rocket booster ignition and lasting for 2.5 sec. During this interval the data are assumed to be realizations of a stationary trivariate random process. Power spectral density models are formulated using a two-stage auto/cross-correlation matching (ACM) technique. In this technique, the spectra of the vector process representing the measured data are modeled using an autoregressive (AR) filter. From this filter, a reduced-order autoregressive moving average (ARMA) model is formulated. Further, when considering a univariate random process, additional order reduction is achieved by using a novel modal-energy-based, system order-reduction technique applied to the initial AR model. The numerical results of this paper should be viewed in context with the usefulness for Space Shuttle related studies of information on the degree of correlation of excitations along different geometrical axes; also, the derivation of simplified univariate power spectra for design purposes is useful.