18 TH INTERNATIONAL CONFERENCE ON COMPOSITE MATERIALS DAMAGE TOLERANCE ANALYSIS OF ADHESIVELY BONDED REPAIRS TO COMPOSITES STRUCTURES C. H. Wang 1 *, J. Y. Goh 1 , J. Ahamed 1 , A. Glynn 2 and S. Georgiadis 2 1 School of Aerospace, Mechanical and Manufacturing Engineering, RMIT University, Melbourne, Australia 2 Boeing Research & Technology Australia, Melbourne, Australia * Corresponding author (chun.wang@rmit.edu.au) A damage tolerance methodology is developed to predict the strength of bonded repairs containing manufacturing flaw or in-service damage. A series of experiments have been carried out using specimens with varying disbond lengths to investigate the effect of pre-existing flaws on the load-carrying capacity of scarf joint. The experiments show a considerable reduction in strength with an increase in disbond length, greater than the reduction in bond area. Computational models using the cohesive zone technique have been employed to predict the strength of scarf joint containing pre-existing flaws. The hybrid cohesive-interfacial model has been found to provide good correlation with the experimental tests, providing a promising technique for damage tolerance analysis of scarf repairs. Keywords : Composite repair, damage tolerance, cohesive zone model growth behaviour of pre-existent disbonds in bonded 1. Introduction repairs. Airworthiness certification of adhesively bonded scarf repairs remain a significant challenge due to This paper presents an experimental and numerical the lack of non-destructive inspection techniques analysis of the damage tolerance behaviour of that are capable of detecting weak or kissing bonds. adhesively bonded scarf joints. Additional Currently no structural credit is given to bonded experiments were carried out to characterise the repairs for safety-critical or primary aircraft fracture properties pertinent to disbonding and structures, which must be able to sustain design delamination damage, which are used to identify the loads without repairs. In other words, repairs are fracture parameters of cohesive zone models assumed to be fully ineffective when performing employed to predict the load-carrying capacity of structural integrity assessment. This approach limits scarf joints containing embedded flaws of varying the application of the bonded repair technology to sizes. largely secondary structures or primary structures with negligible damages. It is found that stresses vary in the adhesive along the scarf due to different ply angles [2]. It is also To overcome the aforementioned challenge, it is found that the mode of failure along the bondline necessary to develop a damage tolerance assessment varies with respect to the lay-up angle of the bonded methodology for bonded structures to demonstrate ply. This is due to a reduced load carrying capability compliance with airworthiness standards [1]. In this in the adhesive at the 45 o and 90 o plies. This leads to approach a disbond is assumed to exist along the failure in the plies. The cohesive zone model shows adhesive bond; the disbond size is set by the promises in predicting the strength reduction due to detection limit of approved non-destructive pre-existing flaws. It is found that a complex model inspection techniques. This approach has two major is required to capture the complex crack branching implications on the design and certification of behaviour and its stress fluctuations along the repairs. Firstly, repairs need to be designed to bondline [3]. sustain the design loads in the presence of disbonds. Secondly, a validated damage tolerance analysis methodology is required to predict the complex 1
DAMAGE TOLERANCE ANALYSIS OF ADHESIVELY BONDED REPAIRS TO COMPOSITES STRUCTURES 2. Experimental 3. Experimental Results The mode I and II fracture toughness of the The scarf joint specimens were observed to have composite laminate and the adhesive was measured fractured by a mixture of cohesive and interfacial using double cantilever beam (DCB) and end failure. Examples of scarf joints with a 12mm notched flexure (ENF) tests respectively. This is an embedded PTFE and without the embedded PTFE element level test to obtain parameters used for the are presented in Fig.2 and 3 respectively. It can also subsequent numerical analysis. Force-displacement be seen from the bands on the specimen that intralaminar failure occurred at the 90 o plies. This is curves were obtained from both sets of tests. The DCB tests were loaded up to a predetermined due to intralaminar failure from matrix cracking displacement and the ENF tests were loaded up to through the ply. failure. 2.1. Scarf joints with embedded flaws Composite laminates made of VTM 264 uni- directional carbon-fibre-reinforced polymer (CFRP) prepreg material were bevelled to form scarf angle of 5 degrees. Scarf joints were secondarily bonded using VTA260 film adhesive. Both the CFRP prepreg and the film adhesive are from the Advanced Composites Group. The test specimens were fabricated to resemble a scarf repair to a damaged structure. The scarf and repair panel have the same quasi-isotropic layup [45/0/-45/90] 2S . Two types of specimens were manufactured. The Fig.2. Failure mode for a scarf joint with a pre-existing flaw. first type was a scarf joint with an embedded flaw simulated by thin polytetrafluoroethylene (PTFE) strips of 3, 6, and 12 mm in length, referring to Fig.1. Four specimens of each flaw size were fabricated for testing, in addition to three specimens of the baseline scarf joints without flaws. Tests were conducted under quasi-static tension until failure. The loads and machine displacements were recorded. Fig.3. Failure mode for a scarf joint without a pre-existing flaw. Fig.1. Schematic of scarf joint containing pre-existent flaws 2
DAMAGE TOLERANCE ANALYSIS OF ADHESIVELY BONDED REPAIRS TO COMPOSITES STRUCTURES η ⎛ ⎞ G 4. Cohesive Zone Model ⎜ ⎟ + − = G ( G G ) II G ⎜ ⎟ IC IIC IC c (4) + G G ⎝ ⎠ 4.1. Model theory I II The constitutive equation for linear-elastic systems is 4.2. Model calibration σ = k δ (1) i i i The cohesive element parameters were derived using an inverse method from the experimental results. where σ , k and δ denote the stress, stiffness and Force-displacement curves were correlated between displacement [4]. Loading mode (I or II) is denoted experimental and numerical analyses for the DCB by subscript i . When the stresses reach a maximum and ENF tests. The parameters for the fracture σ , which is then followed u value, fracture occurs at i strength and the fracture energies were refined in by a linear softening behaviour to a displacement this process. An example of the calibration is seen in δ f G . expressed in terms of fracture toughness, Fig. 5. i c 2 G (2) The properties of the cohesive zone models, with the δ = f i , c i element size being 0.5mm, are listed in Table 1. It is σ u i clear that the adhesive have a lower strength but Eqs.1 and 2 describe the traction-separation higher fracture energies than the composites. behaviour as illustrated in Fig. 4. ENF Results for Film Adhesive In most cases, bonded joints are under mixed mode 1600 loading. Therefore, a damage initiation model based 1400 on the quadratic stress criterion is given as 1200 1000 (3) 2 2 Load (N) ⎛ σ ⎞ ⎛ ⎞ τ AVERAGE ⎜ ⎟ + ⎜ ⎟ = 800 1 ⎜ ⎟ ⎜ ⎟ FE Analysis σ u σ u ⎝ ⎠ ⎝ ⎠ 600 1 2 400 200 0 0 2 4 6 8 Displacement (mm) Fig.5. An example of cohesive model calibration on a DCB force-displacement curve Table 1 : Properties of cohesive zone models (modified) Adhesive model Composite model Property Tensile Shear Tensile Shear Strength 56 58 45 85 (MPa) Fracture Fig.4. Linear traction law for the cohesive zone model [4] Energy 1.13 7.75 0.462 1.60 This is followed by the Benzeggagh-Kenane (B-K) (kJ/m 2 ) fracture criterion for damage evolution given as 3
Recommend
More recommend
