NONDESTRUCTIVE TESTING OF LAMINATED COMPOSITES

Abstract

An iterative method is presented for the estimation of the size and depth of unbends and voids in laminated composites bonded by relatively flexible adhesives from holographic and shearographic fringes recorded using the double-exposure technique with an incremental vacuum pressure applied between the exposures. The method is based on the theory of plates on elastic foundation. It does not require the current flexibility of the adhesive to be pre-determined. The surface over the defective region of the laminate is modelled as a fully-clamped thin plate loaded by a pressure of magnitude equal to the applied vacuum uniformly distributed over the actual defect area and supported on an elastic medium of the Winkler type. The model is valid for laminates with adhesive layers of varying degrees of flexibility. The iterative method is evaluated using (i) laminated plates and strips with thick and relatively flexible adhesive layers, such as aluminium laminates bonded by a foam adhesive, and (ii) laminates with thin and inflexible adhesive layers, such as multi-layered glassfibre-reinforced plastic (GRP) plates. The experimental results show that the size and depth of the artificially created unbond in the GRP plate are accurately assessed. The results also show that the sizes of the unbonds and voids in the foam adhesive-bonded laminated plates are estimated reasonably accurately. The location of the defects in these laminates can only be deduced qualitatively from an examination of both sides of the laminates. The sizes of the large unbends in the foam adhesive-bonded laminated strips are grossly under-estimated. In addition, the results show that defects of the void-type, simulated by removing a section of the foam adhesive prior to bonding, can be distinguished from defects of the inclusion-type, simulated by the mylar air pocket. Overall, the findings demonstrate that the iterative method is applicable to laminates bonded by adhesives of varying degrees of flexibility where the defect to be assessed is small compared to the laminate and located away from the edges of the laminate. The assessment of the unbonds and voids in the foam adhesive-bonded laminated plates and strips by a calibration method is also explored for the purpose of comparison. The method is based on a model of an elastically-clamped plate and a clamping coefficient pre-determined from a range of known void sizes. The experimental results show that the sizes of the unbonds are estimated reasonably accurately and that their locations can be deduced qualitatively from an examination of both sides of the laminates. The results also show that the sizes of the internal void are under-estimated. The results suggest that this method cannot distinguish between a large internal void and a small adherend-to-adherend void and that only the size and location of the unbond can be determined with certainty. The need to pre-determine the clamping coefficient and its subsequent change with time caused by degradation of the adhesive is also expected to limit the practical use of this method. The findings of this thesis have been recently published (Shang et al. , 1994 a; 1995).