Asset maintenance of thick section fibre-reinforced composite structures

Abstract

Fibre-reinforced polymer (FRP) composites are popular in engineering, because they generally have higher specific strength and stiffness compared to other materials (e.g., some metals, polymers and ceramics). The use of FRPs is most often associated with thin, lightweight structures and components (e.g., aerospace, sporting goods, luxury and high-performance vehicles). As such, the vast majority of FRP research is targeted towards thin laminate structures and components. However, there exist thick-section FRP structures—such as Hunt-class mine countermeasures vessels (MCMVs)—that were designed decades ago (e.g., vessels resembling MCMVs appear in c. 1970s publications). Some of these early designs—or derivatives thereof—entered production, and have been in regular service for many years. Due to the comparatively sparse literature available, these structures now constitute an asset maintenance dilemma. For these reasons, this thesis focuses on the asset maintenance of thick-section FRP structures from multiple perspectives.

Firstly, the efficacy of two Non-destructive Testing (NDT) techniques to detect delamination flaws in thick section composites was evaluated. In the first technique, a Full Matrix Capture (FMC) Total Focusing Method (TFM) ultrasound was used. A range of delaminations were generated during manufacturing of six glass FRP blocks, which ranged in total specimen thickness from 20 to 120 mm in 20 mm increments, whilst delamination thickness, delamination in-plane dimensions, and delamination depth location were selected as test variables. For this inspection setup, research indicates that 3 mm thick delaminations are identifiable when embedded at depths up to 74 mm, reducing to 36 mm for surface-to-surface contact delaminations. Regardless of FRP specimen thickness, delaminations were observed when embedded at depths up to 74 % of FRP thickness, beyond which signal decay, noise and mechanically-benign acoustic features limited the success of industrially-representative inspection methods. An inverse relationship was observed between the FRP thickness and the ability to find delaminations. Furthermore, data shows that when delaminations are inserted at greater depths in the FRPs, the accuracy of measuring their depths using this setup increases.

Thereafter, in an innovative study, the efficacy of Ground Penetrating Radar (GPR) to detect delaminations in these same FRP composite specimens was evaluated, using a 2000 MHz microwave antenna. Therein, FRP specimen thickness, delamination depth location, antenna orientation and delamination cavity dryness were selected as variables. For perpendicular antenna orientation, research indicates that 3 mm thick dry delaminations are identifiable—after background removal and other post-processes—up to a depth of 87 mm in the 100 mm plate, and 107 mm in the 120 mm plate. These corresponded to the deepest-set delaminations in the test matrix, and furthermore, were deeper set (and in thicker FRP) than could be detected using advanced ultrasonic techniques. Since this is, apparently, the first successful inspection of thick FRP structures using GPR, it is speculated that additional research could unlock the true potential of GPR inspection of FRPs, with respect to the FRP specimen thickness and the characteristics of the damage/defect targets to be identified.

When the detection of damage in a structure (e.g., via NDT) is considered, it is logical to subsequently question what the consequences of that damage are, with respect to the original mechanical properties in the undamaged state. During NDT campaigns, the effect of found damage on the mechanical properties can be assessed, if the significance of similar damage has been parametrically quantified. This would empower asset owners/operators with data to determine appropriate next-steps, when these damages are identified during in-service NDT campaigns. To achieve this, a partial-factorial investigation, exploring the effect of delamination flaws on the mechanical properties of glass FRPs in flexure, is presented. Test variables included: the FRP thickness (expressed in number of plies); delamination through-thickness location; and delamination in-plane area (that was determined parametrically). The mean flexural strength and stiffness of each pre-flawed case was normalised using the corresponding reference case (that which had the same thickness but no deliberately inserted delaminations prior to testing). Potential differences in specimen manufacturing quality were accounted for using the respective measured densities. Curves having exponential form were fitted using the mean, normalised specific strength data, such that the relationships could be established between each of the test variables and the normalised specific flexural strength. In this way, the effect of delaminations of various sizes, positioned at various locations within laminates of various thickness, was quantified. Surface plots were created that connect all of the test variables to the normalised specific flexural strength, so that the consequence of an arbitrary delamination in another laminate can be inferred. These results are intended to be used in tandem with in-field NDT techniques, to empower owners and operators to determine design-appropriate limits for permissible damage in FRP structures. Furthermore, it is hoped that this research will inform their asset maintenance and repair procedures in a data-driven manner, such that the occurrence of would-be unnecessary and environmentally harmful repair processes, is minimised.

Some theories that relate stresses to strains (in FRP structures) are predicated on assumptions that are difficult to prove mathematically. It is common in the literature for researchers to develop axiomatic Equivalent Single Layer (ESL) theories that describe the behaviour of FRP plates and beams, often intended for use on "thick" FRPs. However, there remains a shortage of literature that defines the properties that a laminate must have for it to qualify as "thick", and therefore, the suitability of the ESL theories described in the literature, is unknown. In the penultimate chapter of this thesis, the definition of the descriptor "thick", with respect to FRP beams, was challenged. The research methodology was based on observing the effect of laminate thickness (expressed in number of plies) on the relationships recorded between through-thickness position and the Digital Imaging Correlation (DIC)-measured in-plane strains. The experimental design and the test matrix were selected on the supposition that the descriptor "thick" (and thereby, the suitability of any ESL theory axioms), could be directly linked to the number of plies used to construct a given laminate. Therefore, changes to the quantity of plies in a given beam would be most likely to result in changes to the characteristics of the resulting through-thickness strain relationships (i.e., strain profiles). The through-thickness strain profiles that were measured using DIC, and their corresponding beam thicknesses appeared to be mostly uncorrelated, with linear strain profiles observed regularly in both the thinnest and thickest specimens, whilst strain profiles that could be described as nonlinear were observed in the middle-thickness beams. Although the data presented does not show any easily extracted trends between the number of plies in a laminate and the nature and distribution of through-thickness measured strain profile, the research remains valuable as a foundation upon which further experiments can be conducted.