Short versus long cracks fatigue crack growth rate, a novel experimental approach

Abstract : Some aircraft components, such as turbine disks, are subjected to very strict certification stages. Other than the fact that the material has a non-linear behaviour (Nickel base superalloy), the complex loading it experiences makes the definition of a fatigue cycle difficult. Moreover, during operation or maintenance those components are exposed to the creation of surface anomalies such as dents or scratches. Besides creating an initial residual stress state that is multiaxial, they create micro-cracks that lead into a significant effect of T-stress. This study aims to highlight the effect of T-stress during a mode I fatigue crack propagation and to compare with the results on effective short cracks. The crack propagation under a loading in terms of stress intensity factor depends on its initial length. In fact, the crack growth rate of a short crack is greater than the one for the long crack [1,2]. Previous work done by Brugier on the propagation of short cracks showed that the effect of T-stresses cannot be neglected unlike long cracks where its effect is reasonably neglected. Thus, assuming that the difference between short and long cracks stems from T-stress (mode I), the goal is to see what is its effect on the fatigue propagation. Bi-axial tests have been performed, at the LMT on the multiaxial machine ASTREE, on cruciform specimens containing a 20mm notch, in order to propagate a long crack as if it were a micro-crack. This can be possible by changing the biaxiality state, i.e. creating a (KI, T) loading specific to a micro-crack, so that the crack tip “sees” the same loading as a micro-crack tip. Based on a 2D numerical model of the specimen, adequate loadings taking in account the crack length are calculated in order to get the right T-stress. Then, 40k cycles sequences divided into two 20k cycles blocks with different T-stress conditions have been applied during the tests. The results about T=0 MPa to T=-280 MPa tests show a major potential increase when the negative T-stress is applied. Thanks to the calibration law, this means that the crack length increases during a few thousands of cycles before getting back to a stabilized state. The effect of T-stress is along 90 mm and during about 5000 cycles. The discontinuity between the two blocks is due to phenomena out of interest. SEM observations are currently held in order to determine the crack growth mechanisms. Those results will be compared to the one with real short cracks, the challenge here is that it is difficult to get a mechanically short crack, meaning a large crack front but a small length. The comparison of the results obtained on short and long cracks will allow us to validate the assumption that the same fatigue crack growth is obtained if the loading (KI,T) is applied. If confirmed, it will be possible to measure short crack growth using tests on long cracks.
Complete list of metadatas
Contributor : Yoann Guilhem <>
Submitted on : Wednesday, October 24, 2018 - 4:45:45 PM
Last modification on : Friday, May 24, 2019 - 5:27:05 PM


  • HAL Id : hal-01903881, version 1


Besnik Sadriji, Bumedijen Raka, François Brugier, Yoann Guilhem, Didier Soria, et al.. Short versus long cracks fatigue crack growth rate, a novel experimental approach. 12th International Fatigue Congress (Fatigue 2018), May 2018, Poitiers, France. ⟨hal-01903881⟩



Record views