Detection of fastener preload loss in a hybrid composite-to-metal bolted joint

Abstract

In this work we present two novel nonlinear time series analysis techniques-nonlinear cross prediction error and chaotic amplification of attractor distortion-for detecting preload loss in a hybrid metal-to-composite bolted joint. The two techniques involve imposing a chaotic steady-state on the structure and analyzing features from the resulting reconstructed attractors. The joint preload is controlled from a "fully tight" 10200 Ibf preload condition to complete failure (no preload), and numerous vibration tests are performed at discrete increments in that range. The two nonlinear techniques are compared to a linear autoregressive model fit for detection capability.

Year