Design and analysis procedures which include dynamic crack propagation and crack arrest are currently under development in order to assure the safety of such structures as nuclear pressure vessels. This type of design and analysis procedure requires a knowledge of the relationship of stress intensity factor (K) as a function of crack speed (å). The minimum value of stress intensity factor (designated as KIm) on the K versus å relationship is of particular importance because crack arrest will occur if the stress intensity factor falls below this minimum value.
Direct determination of the K versus å relationship and of the value of KIm for pressure vessel grade steels is an extremely difficult and complex task. For this reason methods have been developed to approximate KIm values. One such method was developed by Battelle Columbus Laboratory (BCL) and another by Materials Research Laboratory in Chicago. One difficulty involved in an analysis of dynamic crack propagation and arrest is the interaction of the specimen and loading system, and its effect on crack behavior.
This paper describes a series of experiments conducted to determine the influence of loading system on crack jump length and stress intensity as a function of velocity. Rectangular double-cantilever models made from Homalite 100, a birefringent polymer, were tested with two different wedge loading systems. One system used steel loading pins and aluminum wedge. The second system used both plastic pins and plastic wedge. A Cranz-Schardin high speed camera was used to photograph isochromatic fringe patterns associated with propagating and arrested cracks. Dynamic stress intensity factors were determined from the isochromatic fringe loops using a three-parameter Westergaard stress function.
The high speed photographs also were used to determine crack length as a function of time. From this relationship the crack velocity was determined. The results were compared with the predictions of a one dimensional finite difference computer code written by BCL. The crack jump for the models loaded with the plastic pins and wedge was found to be about twice the crack jump for the models loaded by metal pins and wedge for the same initial pin displacement. The amount of energy stored in pin contact stress was computed and when this additional energy was added to the computer input it was found that the proper crack length as a function of time was predicted by the BCL code.