Specimen geometry effect on the accuracy of constitutive relations in a superplastic 5083 aluminum alloy

Current experimental methods are influenced by the end effects that cause non-uniform strain rates in the gauge section and material flow within the grips. A series of tension tests and finite element models confirm this for an Al-5083 alloy. Both the tests and the finite element simulations predict that the actual strain rate begins at about 60% of the desired strain rate and increases gradually with strain. Testing and models are also run to compare specimens with and without alignment holes in the grips. It is shown that alignment holes increase flow from the grips and thus introduce additional error in the tests. Further modelling is carried out to evaluate the improved accuracy of specimens with increased length-to-width ratios. In general, this study show that a specimen with 50% reduction in the standard gauge width and double the standard gauge length give strain rates within 10% of the desired value throughout the test.