Fig. 1. Test rigs and testpiece geometry: (a) Detailed shape of the press-fitted part. (b) Test rig for Minden type tests. (c) Test rig for Vitry type tests.

words, a rapid increase in crack propagation rate is observed when the propagating cracks originated from different sources are merged with one another to form a major   crack.

For the reasons given determination of the limiting size for non- propagating cracks and selection of a proper multiaxial fatigue cri- terion well-describing the multiaxiality of the applied load case is essential. In the present study, two different multiaxial criteria, Dang Van [14] and Liu–Mahadevan [15], were applied to the stress distribution along press-fit contact obtained by finite element (FE) analysis.These two criteria, with a suitable modification for Dang Van [16] have presented remarkable results for the investigation of microcrack advance under influence the rolling contact fatigue [17].The obtained equivalent stress for each criterion was checked against the fatigue limit of the material expressed as a function of defect size through the El Haddad correction [18].The prediction capability for the proposed approach has been evaluated by    com-

Table 1

Axle geometry at  press-fit.

paring full-scale axle test results, performed as a part of Euraxles project with the data produced by the analytical approach [19]. Axles with two different axle geometries were tested in test benches until a fatigue life of 107 cycles was achieved. Tested axles are dismantled and prepared for macro and micro   examination.

In macro examination, magnetic particle inspection (MPI) of the failed and run-out axles was done in order to detect indications of macro cracking under press-fit seat region. Inspected orientations of major cracks are measured and expressed in distance with respect to the contact edge. Obtained experimental results are compared with the critical site estimations made by the proposed analytical method.

Micro examination was done on the samples sectioned from the cracked regions identified by macro examination. Sectioned sam- ples were examined under SEM in order to investigate size,orienta- tion and distribution of non-propagating cracks as well as the defect size from which a crack starts to propagate. Crack size mea- surements taken from the SEM images were used in characteriza- tion of critical size for non-propagating cracks. Obtained results are compared with the critical size estimations for the    corresponding

F4-Minden 165 147 1.12 75 15 190

the limiting conditions for fretting fatigue of railway axles has been

  discussed.

Table 2

Full scale fretting fatigue test results on railway   axles.

Test rig type Test # rnom  [MPa] Test results Failure location – main crack

F1-Vitry 1 153 Failed (N ¼ 1:69 × 106 cycles) Wheel  seat  at  about  20 mm from  the T-transition

F1-Vitry 2 141 Run-out (N ¼ 107 cycles)

F1-Vitry 3 153 Failed (N ¼ 1:47 × 106 cycles)