Assessing composite joints on ships
Lightweighting ship joints can help facilitate ship fuel reductions. Geir Ólafsson, Rachael Tighe, Stephen Boyd, Richard Trumper and Janice M Dulieu-Barton discuss a method for testing composite joints in maritime applications.
Efforts to reduce the weight of ships have increased, often motivated by financial and environmental concerns as reductions in weight can lead to a reduction in fuel use. However, there are other advantages to be gained from lightweighting exercises during ship design, which are becoming important for a number of maritime organisations. For example, weight reduction of superstructures placed high up on a vessel reduces the centre of gravity and improves stability, allowing for sleeker hull forms. This in turn leads to additional financial and environmental benefits as fuel consumption is further reduced. Traditional maritime structural design is heavily reliant on steel.
But significant reductions in weight can be achieved by substituting steel for fibre-reinforced polymer composite materials which offer greater specific strength and stiffness. In some applications, such as naval ships, composites also allow for the integration of multiple functions, for example structural performance with thermal and radar signature reduction. Yet, attaching composite superstructures to a vessel usually necessitates some form of composite-to-steel joint. Currently mechanical fasteners are used, but this adds to the part count and weight, and requires drilling of the composite laminate which can initiate damage. As such, ship designers are seeking alternative connection methods. Adhesive bonding overcomes many limitations associated with fasteners, but has only been implemented on a small number of vessels, such as the La Fayette class frigates of the French Navy.
A key concern is that it is difficult to prove the integrity of bonded joints, and particularly hybrid joints, which facilitate the connection between the metallic hull and composite superstructure. Joint defects or damage can occur during manufacturing or in-service, significantly reducing joint strength and stiffness.
Defect identification is hindered by complex geometry and combination of differing materials within the joint. On top of this, vessel operators face a potentially greater challenge once defects are identified since at present, there is no accepted framework to assess how defects affect residual joint strength and service life. The project, therefore, aims to address the two key aspects for assessment of bonded joint integrity – defect identification and defect criticality. Initially, methods useful for defect identification and characterisation in typical maritime joints were developed.
To this end, existing non-destructive testing (NDT) techniques have been reviewed, with the aim of extending their applicability to joints in marine structures. A partnership between BAE Systems, UK, and the University of Southampton, UK, has resulted in an integrated sensor technology being developed and a patent application submitted, capable of indicating defect size and location. In parallel, the work aims to develop a framework for the assessment of defect criticality for proving structural integrity of joints.
The method integrates a detailed numerical model with data-rich validation testing based on full field imaging to achieve high fidelity models. The modelling space can then be used to incorporate inspection data relating to defect size and location, and to assess likely damage propagation, in addition to residual joint strength and service life.
Armed with such tools, vessel operators and certification authorities would be in a position to make informed decisions for remedial actions, planned maintenance and life extension.
For several decades, partly motivated by the increased use of composite materials, there has been a greater focus on NDT of composites. Initially, traditional NDT techniques such as ultrasound and radiography were used, many of which were originally developed to inspect metallic specimens.
While these techniques are still commonly used to this day, no particular one is capable of detecting damage in typical maritime joints, due to geometrical, material and logistical restraints.
In other industries in recent decades, thermographic NDT techniques have become increasing popular, particularly pulse thermography (PT), which has become an established NDT technique for aerospace applications. PT inspections use an infrared detector to provide full field measurement by monitoring the thermal decay of the surface of a component after pulse heating. Defects are identified by differences in thermal properties between defective and non-defective regions of components since most defects cause a local reduction in thermal diffusivity.
However, for the inspection of composite materials, the probing depth is limited by low thermal diffusivity. In such activities, noise and inherent setup error or hardware limitations can obscure information describing the presence of a defect. The current work, therefore, considered strategies to minimise error in thermographic inspections focusing on data post-processing. This included a technique to characterise spatial error that can then be subtracted from raw data. It also helps with the layering of multiple existing techniques such as thermal signal reconstruction (TSR) and pulse phase thermography (PPT) to optimise defect identification in thick composite components typical of maritime designs.
Specimens were manufactured with simulated defects of a known size and location by placing shaped Teflon inserts within a laminate. For comparison, the difference in mean temperature between defective and non-defective regions is normalised by standard deviation in the non-defective region. Using raw data alone, defects were identified at a depth of up to 0.6mm, however using the developed processing algorithm, they were identified at 1.8mm depth.
While the improvements are significant and applicable to many other applications, composite marine structures are often significantly thicker than this maximum value.
Therefore, a new inspection methodology was sought. A recent concept, for which patent protection is being pursued, was to use an embedded sacrificial sensor placed at known damage initiation sites. It has been shown that this electrical sensor can be electrically interrogated to provide both the location and size of a defect within a bonded joint. An electrical path to the sensor is required, so that the sensor can be interrogated using a simple electrical connector and a handheld meter. Hence, continuous structural health monitoring is possible, providing a real-time indication of the condition of the joint.
Once a defect is identified and the extent of damage is assessed, it becomes important to decide if and when remedial action is required. To avoid undesired asset downtime, it is financially beneficial if repairs can be made during planned maintenance windows. Conversely, unsafe structures must be repaired at the earliest opportunity regardless of planned activities, potentially involving operational restrictions imposed as vessels return to port.
The potential loss of an operational strategic asset, safety concerns and the reputational damage of failed structures, make understanding the significance of identified defects an important concern for vessel operators. In addition, certification authorities are reluctant to accept designs including adhesive bonding due to the need to prove joint integrity. And so work aims to prove joint integrity by encoding information from NDT inspections into numerical models to predict the residual strength, stiffness and life of a bonded joint.
The key enabler to this strategy is using data-rich experimental testing to obtain a high-fidelity numerical model. For the purposes of developing this validation methodology, a simplified single lap joint (SLJ) is considered as its mechanical behaviour is well documented in the literature.
To validate numerical models, digital image correlation (DIC) is used to obtain high resolution full field strain information across a field of view. DIC involves applying a random pattern to a component surface, typically by spraying black paint speckles onto a white painted background. The component is then imaged at different loads, and software is used to track how speckles move between load steps. This then provides a displacement field, which can be differentiated to give the strain field.
The advantage of this technique is that thousands of such measurements can be made at a time, and combined to form a strain field over a field of view, as opposed to isolated and often unrelated point measurements obtained using specialist gauges for such information. Due to the rich data obtained, it is possible to make meaningful pointwise comparisons to numerical models across of a full field of data. This allows for validation of the strain distribution, adding confidence that the physics and boundary conditions have been accurately captured.
An example of this data was displayed in a GFRP-to-steel SLJ, where the full field x and y displacements at the adherend ends of a single lap joint, subjected to 0.5kN of tensile load, were measured. These were used to calculate the transverse normal and shear strains. If the joint experienced loads in excess of their design load, it is these strains at the adherend ends which would eventually cause failure of SLJs and more complex joints. By obtaining a full field image of the strains that lead to failure, the method provides a new means of obtain high resolution, high-fidelity data for the purposes of model validation and a tool for improving joint design.
A comparison of this data was made with a numerical model, where the DIC shear strain data has been subtracted pointwise from the numerical shear strain data. This involves a processing procedure to spatially align both fields, and subsampling of the numerical data such that number of points in each field is equal. The fields can then be compared using various metrics, such as absolute difference or percentage difference.
Considering absolute difference between DIC and numerical strains, the results showed a good correlation between simulation and experimental data, except in the adhesive bond line. However, the adhesive was modelled using bulk properties available from a supplier.
The testing used to obtain these properties takes no account of the constraining effects of joint on the bond in the single lap joints. This work has developed a methodology to compare numerical models to full field experimental results, and validated that the mechanics of joints can be captured accurately in a numerical model.
Current work is focused on validating models that include damage propagation prediction, which forms the final step in the proof of concept planned for this project. With a validation tool available, the next step is to apply this technique to a realistic maritime joint, using typical materials and geometry.
Larger scale testing which captures realistic boundary conditions and structural effects would then finalise the work where a numerical model is capable of assessing the criticality of defects and damage identified from inspections.
Proving joint integrity is crucial if the potential benefits of composite superstructures is to be fully realised. This outlines a novel method of improving the probing depth of an existing NDE inspection enabling new applications for pulse thermography. In addition, a novel sensor has been developed capable of identifying and localising damage in bonded joints.
A methodology is also presented which can incorporate information from such devices into numerical models capable of predicting mechanical behaviour of such joints. By developing a framework for both bonded joint evaluation, and defect criticality assessment, the use of bonding in this manner in the maritime industry is facilitated. In turn, the use of composite superstructures for maritime applications is enabled, allowing for significant reductions in vessel operational costs.