Limit strains for severe accident conditions synthesis report of the EU-project LISSAC contract no. FIKS-CT1999-00012

Abstract

The local failure strains of essential reactor vessel components are investigated. The size influence of the components is of special interest. Typical severe accident conditions including elevated temperatures and dynamic loads are considered. The main part of work consists of test families with specimens under uniaxial and biaxial static and dynamic loads. Within one test family the specimen geometries and the load conditions are similar, the temperature is the same; but the size is varied up to reactor dimensions. Special attention is given to geometries with a hole or a notch causing non-uniform stress and strain distributions typical for reactor components. There are indications that for such non-uniform distributions size effects may be stronger than for uniform distributions. Thus size effects on the failure strains and failure processes are determined under realistic conditions. Several tests with nominal identical parameters are performed for small size specimens. In this way some information is obtained about the scatter. A reduced number of tests is carried out for medium size specimens and only a few tests are carried out for large size specimens to reduce the costs to an acceptable level. To manufacture all specimens sufficient material was available from the unused reactor pressure vessel Biblis C consisting of the material 22NiMoCr37. Thus variations of the mechanical material properties, which could impair the interpretation of the test results, are quite small. This has been confirmed by an adequate number of additional quality assurance tests. A key problem was the definition of failure and the determination of the local strains at failure for very different specimens under varying load conditions. Here appropriate methods had to be developed including the so-called vanishing gap method and the forging die method. They are based on post test geometrical measurements of the fracture surfaces and reconstructions of the related strain fields using finite element calculations, for instance. To deepen the understanding of structural degradation and fracture and to allow extrapolations, advanced computational methods including damage models have been developed and validated. The problems to be treated here are quite difficult. Micro-structural effects, for instance, play an important role. Therefore several approaches were tried in parallel. In some cases so-called non-local concepts, in other cases the description of stochastic properties at the grain size level are considered. The experimental results indicate that stresses versus dimensionless deformations are approximately size independent up to failure for specimens of similar geometry under similar load conditions. Also the maximum stress is approximately size independent, if failure occurs after the maximum stress is reached. Cracks are initiated, if the local equivalent strain - here expressed as a true or logarithmic strain, respectively - reaches a critical value, called the local failure strain. It turned out to be more than 50 % for large specimens approaching the dimensions of the reactor pressure vessel. The local failure strains are size dependent. They reach values around 150 % for small specimens with thicknesses or diameters of a few millimetres. The parameter describing the size effect is the radius of holes or notches located in critical specimen regions. It is very remarkable and it simplifies the applications that the shape of the specimen and the type of load plays a minor role, only. Some of the above findings about size effects can be understood by theoretical investigations considering the stochastic, micro structural character of the material. However quantitative predictions based on micro structural models are still beyond the current knowledge. As expected, the scatter of the results on structural failure is considerable. However statistical evaluations indicate that the failure strains will hardly fall below a lower threshold.