Vibration Signatures of Fatigue: Indicator Comparison and Prognostic Assessment in Resonance Fatigue Testing
A. Mestre Pinheirinho Simões Costa (TU Delft - Aerospace Engineering)
J.A. Pascoe – Mentor (TU Delft - Aerospace Engineering)
V. Yaghoubi Nasrabadi – Mentor (TU Delft - Aerospace Engineering)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Resonance vibration testing accumulates fatigue cycles far faster than conventional servo-hydraulic methods by driving a specimen at its own natural frequency, and the same response that applies the load also carries information about the damage accumulating within it. However, the literature offers no systematic comparison of competing vibration-based damage indicators measured simultaneously on the same metallic specimens, and no characterisation of how far ahead of failure such indicators support a useful remaining-life estimate.
This thesis addresses both gaps. Eight notched Al~6061-T6 beam specimens were driven to failure at a fixed excitation frequency and interrogated between load blocks with a scanning laser Doppler vibrometer, yielding 33 features per block. A five-layer pipeline ranked those features by sensitivity, combined them into six competing damage indices, and estimated remaining useful life by trend extrapolation and by a population Gaussian Process. Two further specimens were interrupted at the point of detection for optical
fractography.
Measurement reliability, rather than indicator choice, governed what the campaign resolved. Only two modes could be tracked stably, mode-shape correlation between two undamaged blocks of the same specimen fell as low as 0.785, and mode labels did not consistently correspond to the same physical mode. Modal parameters changed abruptly at the transition to full-amplitude loading rather than progressively, which physical inspection attributed to the clamping and load-transfer arrangements. Indicators requiring no modal identification, principally the harmonic distortion index and amplitude RMS, proved more dependable than frequency, damping, or coherence, and no damage index reached a resolved population-level prognostic horizon.
These outcomes form a traceable account of where vibration-based fatigue prognosis succeeds, where it does not, and which failures are attributable to method rather than principle. It delivers a ranked indicator comparison, the first prognostic horizon characterisation for this application on metals, quantified design requirements for resonance fatigue fixturing, and three reusable open software tools.