High temperature fatigue of a polycrystalline nickel base superalloy

Abstract

A high temperature fatigue crack growth study on an experimental disc alloy of approximate composition Ni-14.75Cr-14-19Co-4.75Mo-3Al-3.75Ti-1.75Ta-0.7Hf-0.06Zr-0.02C-0.0175B wt- has been undertaken. Comparison of constant load fatigue crack growth tests conducted in air and a vacuum at 725C and in air at room temperature indicate that an oxidising environment has a major influence on crack growth rates over a wide range of applied stress intensity range. In particular it contributes to enhanced embrittlement of grain boundary regions, promoting an early transition to intergranular failure along with a concomitant increase in growth rate. Constant stress intensity factor range tests at high R ratio showed that a decrease in frequency at 725C in both air and a vacuum caused an increase in the crack growth rate per cycle due to time dependent crack growth. This was most significant in air at 725C rather than in vacuo, although in both instances low frequency tests were accompanied by wholly intergranular crack growth. At lower R ratios the influence of time dependent processes is less pronounced, especially in a vacuum. For the purposes of prediction a straightforward linear summation model using sustained load crack growth results combined with high frequency growth rates was found to model reasonably the influence of frequency on crack growth rates for the air tests at 725C. This can be rationalised by the observation that in air intergranular failure proceeds by linking of prior cracked or embrittled grains some distance ahead of the crack tip. At lower values of stress intensity range a slight under prediction of growth rates was evident, but improvements could be made through the use of triangular waveform data. For the vacuum tests, the linear summation model was consistently found to over predict growth rates due to the transitory nature of the sustained load crack growth rates under repeated loading and was not generally deemed suitable. Under these conditions damage occurs close to the crack tip and direct interaction between time dependent and time independent mechanisms will occur. This is not taken into account by a summation approach and more accurate modelling of damage formation in the varying strain fields ahead of the crack tip is required to predict this.