Experimental analysis of mode I crack propagation in adhesively bonded joints by optical backscatter reflectometry and comparison with digital image correlation

Highlights

• Optical Backscatter Reflectometry was used to monitor DCB adhesive joints.

• OBR allows for the onset of damage in the bondline to be identified.

• Indications by OBR are in agreement with Digital Image Correlation;

• Process zone size could be estimated combining DIC and visual inspection.

Abstract

The relationship between the response of backface strain distributed sensing by Optical Backscatter Reflectometry and the damage in the adhesive was studied using double cantilever beam specimens. Digital Image Correlation and visual inspection provided information about the crack-tip position and the extension of the cohesive process zone. The comparison between the features of the backface strain pattern and the position and size of the cohesive zone led to the conclusion that, for the studied adhesive, backface distributed sensing allows for the onset of plastic deformations ahead of the crack-tip to be identified, opening new perspectives for fatigue crack monitoring and in-service measurements.

Introduction

The use of adhesively bonded joints in aeronautical and automotive fields has grown significantly in recent years due to their high strength to weight ratio, design flexibility in multi-materials joints and more uniform stress distributions concerning mechanical joints. However, adhesive joints present a lack of long-term reliability because of environmental and mechanical degradations [1], [2], [3], [4], [5]. Consequently, to increase the in-service reliability and safety of adhesively bonded joints in primary structures, either a damage tolerant design [6] or a Structural Health Monitoring (SHM) approach [7], [8] are, nowadays, being developed and implemented.

The damage tolerance philosophy assumes that any given structure contains defects able to propagate during its operational life. Therefore, in order to ensure in-service safety, this method requires the optimised definition of the periodicity (“inspection interval”) of service interruptions for maintenance and the application of Non-Destructive Testing (NDT). To this aim, life predictions based on fracture mechanics concepts and the capability, i.e., the Probability of Detection, of the adopted NDT techniques must be available [6].

On the other hand, SHM is based on real-time, or on-demand, monitoring of the structural integrity of a component or structure during service, also called “diagnostics”, and it is a key for the real-time estimation during service, also called “prognostics”, of the “remaining useful life”. To this aim, a reliable evaluation of the monitored damage (at least, its size and location), the permanent instrumentation of the monitored structure, and the application of Big Data analytics methods to the acquired data are needed [7], [8], [9].

Despite the above-mentioned conceptual and methodological differences between a damage tolerant design and SHM, both approaches rely on an accurate assessment of damage size and its location, particularly in the case of adhesively bonded joints. From this perspective, it is worth remarking that the in-service structural integrity of adhesively bonded joints can be affected by damage (i.e. crack) initiation and propagation. That can be a consequence of either the presence of manufacturing defects (such as voids, kissing bonds and inclusions among many others) or in-service loading conditions (such as fatigue or impacts) or both simultaneously, by the geometry of the joint and by the material of the substrates [1].

Composite bonded joints can fail in the substrate (i.e., fibre-tear breaking, delamination, and stock-break failure), in the substrate/adhesive interface (adhesive failure), or the adhesive layer (cohesive failure). On the other hand, metallic adhesively bonded joints mainly undergo adhesive and/or cohesive failures [10]. This work aims at testing the capability of a distributed backface strain sensing method to monitor the crack propagation in the bondline. Therefore, metallic substrates were chosen to exclude all the above-mentioned damage mechanisms typical of composite joints, which might have hampered the interpretation of results.

The backface strain measurement is widely reported in the literature [5], [11], [14]. The traditional version of this method uses a discrete array of punctual strain sensors such as, for example, conventional strain gauges fixed on the external surface of one or both substrates of the adhesive joint. Once a crack in the adhesive propagates, the stiffness of the joint and its strain response change. Therefore, variations in the backface strain pattern of the adhesive joint are assessed by the sensors, and the crack-tip position can be inferred. The main advantage of backface strain measurement consists of the possibility of being used as an in-service damage monitoring method both in static and dynamic loading conditions and can be applied regardless of the type of substrate material. On the other hand, discrete sensing provides valuable information just in the regions of the joint close to and around the applied punctual sensors [15], [16].

For this reason, other kinds of sensors are being studied, developed, and proposed. Several types of research investigated the performance of optical fibres as strain sensors in the backface strain measuring method. They present greater accuracy, long-term stability, and reduced sizes, providing less invasiveness in the system when compared to strain gauges or piezoelectric transducers [17], [18], [19]. The main types of adopted fibre optic strain sensors are Fibre Bragg Gratings (FBG) and Chirped Fibre Bragg Gratings (CFBG). Both measure the strain over the length of a Bragg grating. FBGs have a constant modulation period of the refractive index, which enables measuring uniform strains over the sensor’s length like a strain gauge [18]. In contrast, the period of CFBG varies linearly, thus enabling short length (limited to the Bragg gratings size) distributed sensing. Distributed sensing over a longer length can be achieved by Optical Backscatter Reflectometry (OBR) [20].

The OBR technique uses swept wavelength interferometry (a narrow band signal launched into the optical fibre by a low-power laser) to measure the Rayleigh backscattering caused by the intrinsic defects present in the core of the optical fibre and producing stochastic local modifications in the refractive index profile along its whole length. These random disturbances produce a unique Rayleigh backscatter profile representing a fingerprint of the whole given fibre. In addition, external temperature and/or strain stimuli produce changes of the optical fibre refractive index and, consequently, spectral and the local temporal Rayleigh backscattered profiles [21], [22]. From these changes of the Rayleigh local patterns, and after suitable calibration, mechanical and thermal strains can be derived and measured. Since the reflectometer gathers a point-by-point response from the fibre, strain measurements are distributed. The spatial resolution of this method is of a virtual kind, i.e., a virtual gage length is set by the user depending on the needs and can vary from fractions of millimetres to kilometres.

Although many researchers [20], [23], [24], [25] have already studied backface strain measurement by optical fibres, distributed strain sensing by OBR has not been fully explored as a backface strain monitoring method yet. An application of OBR to monitor crack growth in a double cantilever beam (DCB) specimen can be found in [26] for the case of hybrid co-cured metal-CFRP specimens. In the present paper, a similar approach is applied to mode I, quasi-static crack propagation in an adhesively bonded DCB with metal adherents. The present work aims to interpret the relationship between the backface strain profile measured by OBR and the essential features of a crack developing in the bondline of adhesively bonded joints, which still presents open points. To achieve this goal, in the present research, the distributed backface strain measurements along optical fibres provided by OBR are discussed based on visual inspections and strain distributions measured in the bondline by Digital Image Correlations (DIC).

Considering the application of experimental stress/strain analysis approaches to bonded joints, numerous research projects used Digital Image Correlation (DIC) to measure crack size, crack tip location, and determine adhesives' cohesive properties [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37].

As highlighted by Gorman et al. [31] and Reiner et al. [32], DIC can be a very accurate method to determine the energy release rate and the traction separation law of the adhesive joints in Double Cantilever Beam (DCB) tests. In addition, this method can give important information about the strain profile of the specimens. However, DIC can hardly be applied as an in-service monitoring technique because of the bulky equipment (out of the lab, for parts with non-planar surfaces, a 3D system is required) and the strict calibration requirements and complex procedures needed to obtain accurate measurements.

In this work, DIC was not used to determine the fracture properties of the studied adhesive like in [30], [31], [32], but as a validation tool for OBR results. OBR appears to be a promising technique for estimating the crack front position and the size of the process zone in the bondline of adhesive joints, particularly under fatigue loading. The present work is focused on the use of the backface OBR methodology to determine the strain profile of cracked adhesively bonded joints and to evaluate the relationship between the OBR response and the actual features of the present damage. A well-defined experimental case of study (quasi-static DCB test, i.e., an effective representative of mode I loading) is chosen to act in a more controlled and known scenario. OBR results are then compared and discussed with respect to DIC and visual inspection ones.