Failure of Rubber Components under Fatigue

Abstract

Rubber components under cyclic loading conditions often are considered to have failed as a result of the stiffness changing to an amount that makes the part no longer useful. This thesis considers three distinctive but related aspect of the fatigue failure exhibited by rubber components. The first considers the reduction in stiffness that can result from a phenomenon known as cyclic stress relaxation. The second considers fatigue crack growth encountered resulting in potentially catastrophic failure. The final issue relates to the complex topography of the resulting fatigue fracture surfaces. Previous work has shown that the amount of relaxation observed from cycle to cycle is significantly greater than that expected from static relaxation tests alone. In this thesis the reduction in the stress attained on the second and successive loading cycles as compared with the stress attained on the first cycle in a stress strain cyclic test of fixed strain amplitude has been measured for elastomer test pieces and engineering components. Adopting the approach of Davies et al. (1996) the peak force, under cyclic testing to a specific maximum displacement, plotted against the number of cycles on logarithmic scales produces a straight line graph, whose slope correlates to the rate of cyclic stress relaxation per decade. Plotting the rate of stress relaxation per decade against the maximum average strain energy density attained in the cycle reduces the data measured in different deformation modes for both simple test pieces and components to a single curve. This approach allows the cyclic stress relaxation in a real component under any deformation to be predicted from simple laboratory tests (Asare et al., 2009). Earlier work (Busfield et al., 2005) has shown that a fracture mechanics approach can predict fatigue failure in rubber or elastomer components using a finite element analysis technique that calculates the strain energy release rate for cracks introduced into bonded rubber components. This thesis extends this previous work to examine real fatigue measurements made at both room temperature and 70±1ºC in both tension and shear using cylindrical rubber to metal bonded components. Dynamic testing of these components generated fatigue failures not only in the bulk of the component but also at the rubber to metal bond interface. The fatigue crack growth characteristics were measured independently using a pure shear test piece. Using this independent crack growth data and an accurate estimate for the initial flaw size allowed 3 the fatigue life to be calculated. The fracture mechanics approach predicted the crack growth rates accurately at both room temperature and 70±1ºC (Asare et al., 2011). Fatigue crack growth often results in rough fatigue crack surfaces. The rough fatigue crack surface is, in part, thought to result from anisotropy being developed at the front of a crack tip. This anisotropy in strength whereby the material is less strong in the direction that the material is stretched might allow the fatigue crack to grow in an unanticipated direction. It might also allow the crack front to split. Therefore the final part of this thesis examines how, once split, the strain energy release rate associated with growth of each split fatigue crack develops as the cracks extend in a pure shear crack growth test specimen. The aim being to understand how the extent of out of plane crack growth that results might allow a better understanding of the generation of particular crack tip roughness profiles. Using a method of extending one split crack at a time, whilst keeping a second split crack at a constant length, it has been possible to evaluate the initial strain energy release rates of split cracks of different configurations in a pure shear specimen. It was observed that, for a split crack in a pure shear specimen, the initial strain energy release rate available for crack growth depends on the precise location of the split crack. It is also clear that the tearing energy is shared evenly when the crack tip is split into two paths of equal length, but as one crack accelerates ahead it quickly increases in tearing energy and leaves the slower crack behind. It is thought that this phenomenon is responsible for a lot of the roughness observed on the resulting fracture surfaces.