Damage detection in composite structures under uncertainties via random vibration
Loading...
Date
2025-03-05
Authors
Δημογιάννης, Γεώργιος
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Composite materials are increasingly used in various engineering applications, particularly in aerostructures, due to their exceptional mechanical properties and high strength-to-weight ratio. However, their sensitivity to uncertainties caused by varying Environmental and Operating Conditions (EOCs) poses significant challenges for effective damage detection. Vibration-based Structural Health Monitoring (SHM) in composites has become a field of significant interest mainly because vibration is naturally available for most structures. A major issue in vibration-based SHM stems from the fact that variations in the structural dynamics caused by the varying EOCs and other uncertainty factors can be so pronounced that they partially or even completely obscure the dynamic changes associated
with early-stage damage. Therefore, distinguishing the effects of EOCs and uncertainties from those induced by damage is crucial for effective vibration-based SHM. In the present study, the problem of random vibration-based robust damage detection for a composite aerostructure under temperature, manufacturing uncertainty and a distributed excitation profile is numerically addressed. A three-dimensional, four-node shell element of linear order withsix degrees of freedom per node is employed for the Finite Element Method (FEM). Temperature varies within the [−40,20]℃ range and is modelled to affect the material properties and damping ratio of the structure. Manufacturing uncertainty is modelled as a variation in the material properties involving the vertical and transverse directions.
Additionally, a distributed excitation profile that simulates the real in-flight conditions, based on the Dryden turbulence model, is considered. Delamination is implemented by stiffness degradation in the middle layers of the structure and is considered at three distinct locations: (i) near the boundary condition, (ii) near the middle, and (iii) near the tip of the structure. Three methods, all utilizing acceleration-response signals and based on Multiple Input Multiple Output (MIMO) Transmittance Function (TF) representations of the structural dynamics estimated via stochastic Vector AutoRegressive with eXogenous excitation (VARX) models, are employed for damage detection: (i) an U-PCA-VARX method, and two Multiple Model (MM)-based methods: (ii) an U-MM-VARX method, and (iii) a PCA-enhanced, U-PCA-MM-VARX, method. The study is extensive, based on hundreds of Monte Carlo simulations conducted across a wide range of temperatures. The inspection phase of the damage detection process is structured into two distinct sub-phases: one within the [0,20]℃ temperature range, corresponding to inspections conducted under controlled laboratory conditions, and the other within the [−40,0]℃ range, representing experiments conducted under a real operating regime. All methods achieve excellent damage detection results for both inspection phases considered.
Description
Keywords
Vibration-based SHM, Robust damage detection, Uncertainty, Multiple model method, Principal component analysis