Fracture mechanics of bi-material interfaces for composite pipe repair

Abstract

Composite repairs evolved as popular choice for rehabilitation of oil and gas pipelines from corrosive material loss. However, corrosion can develop into local through-wall defects. In this case, fluid pressure acts directly on repairs, forms blisters and applies stresses on repair-pipe interface bonds. A practical model is needed to evaluate and design composite repairs against interface failure. This study investigated fracture mechanics aspects of failure through crack propagation along the repair-pipe interface. Thick fibre-reinforced plates were examined as repairs for circular ‘sharp-edged’ through-holes in stiff metal substrates. Blister formation and propagation onset were analysed. Energy release rates were investigated as measure of interface failure. Two types of blister tests were conducted; using fluid pressure and shaft-loading with different punch heads. A novel method of determining energy release rates in pressure blister tests was developed. Digital image correlation was used to track blister volumes, which are directly related to energy release rates. Existing and newly derived analytical solutions for each test method were compared with measurements and simulations using the virtual crack closure technique. Energy release rates were found to be influenced differently by repair, defect and shaft geometries. Contrary, critical loads could be plotted as function of defect size to repair thickness ratio. To reduce geometry dependence, ‘volumetric’ energy release rates were introduced by adjusting for defect and repair geometries. Similar to load curves, these are functions of defect size to repair thickness ratios, potentially simplifying fracture criteria. Shaft-loading could not be recommended as general fluid pressure replacement, because of differences between the test methods. Design criteria against debonding by ISO/TS 24817 [1] and ASME PCC-2/4 [2] based on energy release rates were reviewed against results presented. An improvement to the formulation in the standards was suggested. An alternative process for qualification and dimensioning was proposed based on empirical formulations for volumetric energy release rates and critical loads as design criteria.