Development and integration of full-field imaging techniques for assessment of composite structures

Abstract

A new versatile approach is proposed to assess the mechanical performance of composite structures. The focus of the PhD is on the development and integration of Thermoelastic Stress Analysis (TSA) and Digital Image Correlation (DIC) that are both surface based imaging techniques. The overarching goal is to create a Non-Destructive Evaluation (NDE) system for assessing the mechanical response of large composite structures using both TSA and DIC, and to use these techniques to detect defects and damage within these structures.One of the greatest challenges with inspecting large composite structures is the need for a multicamera setup to offer full coverage of the structure. For this to be financially viable, lower cost cameras are required. Part of this study investigates the feasibility of replacing expensive infra-red (IR) cameras based on photon detectors with low-cost microbolometers to conduct TSA, and utilising standard high-resolution white light cameras instead of expensive high-speed cameras to facilitate DIC on structures under cyclic loading.In TSA, the dynamic behaviour of cyclic loading has led to the use of the photon detector based IR cameras that have an instantaneous response over a controllable integration time. In contrast, microbolometer based IR cameras are much cheaper, however, their performance is intrinsic to the sensor material and is governed by a fixed thermal time constant that drives the response time. To assess the suitability of different microbolometer cameras, alongside the different image processing algorithms for TSA, a means of simulating the camera response has been devised. The response from two microbolometer cameras is simulated: a standard microbolometer (<£10k) and a very low cost (<£500) microbolometer based thermal core. The thermoelastic response is simulated for the case of a Brazilian disc specimen, which provides a complex 2D stress field, and in-plane elasticity has a known analytical solution. The results from camera simulations are compared with the experimental data, and the limitations of the use of microbolometer based cameras for TSA are highlighted. Additionally, three image processing algorithms for TSA, i.e. least-squares fitting, lock-in and FastFourier Transform (FFT), are compared. The study showed that the least-squares fitting is the most appropriate as the lock-in and FFT approaches are affected by spectral leakage. Most importantly, the results obtained with the thermal core are promising, strongly indicating that multiple camera TSA is a realistic possibility with an outlook to having cameras permanently installed on structures. However, at present, it is clear that the thermal core can only be used in a qualitative sense, as it has a very low response alongside significant attenuation caused by the low-pass filtering effect.An in-depth study of the variables affecting the microbolometer based TSA is performed by means of simulations and experiments. To define the microbolometer performance, a Simulink model is developed to study the influence of the fixed time constant, inbuilt ‘noise reduction’ filters, signal noise, and different waveform inputs. The study shows that for transient temperature signals, such as those necessary for TSA, microbolometers behave as low-pass filters, hence, the thermoelastic response is attenuated as the loading frequency increases beyond a cut-off frequency. A parametric study considering different test specimen materials, noise reduction features, camera frame rates, loading frequencies and loading amplitudes has been performed experimentally to confirm the findings of the model. This allowed a new calibration approach based on the loading frequency to be devised, which provides faster post-processing. The procedure is developed and validated using a CFRP sample with an internal defect by comparing the thermoelastic response obtained with the photon detector and microbolometer. Additionally, Brazilian disc experimental data collected with the noise reduction on and off enables the validation of the calibration technique with the analytical solution. Previous work on orthotropic composite laminates has indicated that the thermoelastic response might be driven by the temperature change in the resin-rich surface layer or the fibre reinforced substrate layers. It is clear that the response is dependent on the loading frequency used in the experiments and that heat transfer plays an important role. The ‘source’ of the thermoelastic response is studied in detail, in well-characterised glass and carbon reinforced composite laminates with different stacking sequences. Classical Laminate Theory (CLT) is used to determine the theoretical thermoelastic response from the resin-rich layer, the surface ply and homogenised through the entire laminate. Experiments are conducted where the thermoelastic response is obtained alongside strain measurements obtained simultaneously using DIC and strain gauges. To conduct the DIC during a cyclic test a new approach based on the least-squares algorithm developed for TSA is applied and named as ‘Least-Squares DIC’ (LSDIC). The measurement of the strain fields enables an independent ‘measure’ of the thermoelastic response to be derived without any influence of heat transfer. The approach enables the source of the thermoelastic response to be established categorically and highlights how differences in fibre volume fraction, surface resin layer and paint thickness must be accounted for if any form of quantitative study is to take place using TSA.Combining the stress metric from TSA and the strain measurement from LSDIC enables the extraction of a parameter related to the overall structural stiffness. To investigate a full-field ‘damage parameter’ based on integrating the thermoelastic response and the LSDIC strains, an experimental campaign was conducted on GFRP coupons with different stacking sequences. Specimens were loaded at different stress levels to create damage, and inspected with TSA and LSDIC at lower load levels, as well as with X-ray Computed Tomography (CT) scans between loadings to verify the presence of any damage. The approach suggested that there is degradation of the matrix coefficient of thermal expansion due to microcracking. Finally, microbolometer-based TSA and LSDIC are applied simultaneously in a structural composite component (T-joint of a wind turbine blade spar). The integration of TSA and LSDIC was facilitated by implementing an interpolation approach so that both data sets are constructed on the same image plane at the same spatial resolution to provide a demonstration of how the integrated approach could be applied at a structural scale.

More information

Identifiers

ORCID for Irene Jimenez Fortunato: orcid.org/0000-0002-7050-3303

Catalogue record

Export record