Assessment of Delaminations in Composite Laminates using Vibration Monitoring

Abstract

Fibre-reinforced polymer composite laminates are now commonly used in many structural applications, especially in the aerospace industry, where margins of safety are kept low in order to minimise weight. Timely detection and assessment of delaminations in composite laminates is therefore critical, as they can cause loss of structural integrity affecting the safe operation of the aircraft. Traditional NDI methods are time consuming and expensive since they require aircraft to be grounded for inspections. The current trend therefore is towards implementation of structural health monitoring systems which can monitor the structure in situ without down time. Vibration monitoring is one method which has the potential for such application. Among different methods of vibration monitoring, frequency monitoring is the simplest to implement, requiring only single point measurement, and is accurate and reliable. However, while reductions in natural frequencies clearly indicate the presence of damage, the assessment of the location and severity of the damage from measured shifts in frequencies requires the solution of the inverse problem. There have been few comprehensive studies in the past on the efficacy of delamination assessment from measured frequencies solving the inverse problem, especially for laminated plates.

In this thesis, three different methods of solving the inverse problem to estimate delamination parameters from measured frequency shifts are investigated: the graphical method, artificial neural network and surrogate assisted optimisation, all applied initially to composite beams with delaminations at different interfaces. The graphical technique has been extended to the three variable problem of assessing delamination size, axial location and interface location for the first time. The artificial neural network and surrogate assisted optimisation are also applied to the five-variable problem of delaminations in plates (two dimensions and three coordinates for the location). The inverse algorithms techniques are first validated through numerical simulation and their effectiveness and accuracy in predicting delamination parameters are assessed with experimental results from vibration tests conducted on fibre reinforced composite beams and plates with embedded delaminations. The robustness of the inverse algorithms is studied by adding different levels of random artificial noise to the numerical data and their impact on prediction accuracy is determined.