358 Mattia Pujatti et al. / Procedia Engineering 74 (2014) 356 – 359

Connecting rods were designed on basis of a stress analysis carried out via a nonlinear finite element model. Although the model confirmed that the most stressed area is that where cracks propagated, the maximum ideal stress is about 135 MPa, much below the fatigue limit of the material. It was therefore concluded that the failures could not be attributed to an incorrect fatigue design, which would have otherwise affected much a higher number of parts.

A common cause of fretting is too low a radial pressure. Radial pressure is a fundamental parameter as it must be high enough to prevent sliding of the bearing shell into the housing, which originates adhesion spots that are the first stage of fretting damage, and low enough not to exceed the strength of the materials under contact. A simple way to evaluate it is via the bearing crush height, defined by ISO 6524. Using classical Hertz theory, it can be in fact shown that the radial pressure Prad  and the crush height Ic  are linked by the following relation:

where Ks and Ap are respectively the stiffness of the bearing and its projected area when fit into the housing. It was found that the design radial pressure was close to the lower limit of the prescribed range. It was also noticed that two of the three cracked connecting rods were reconditioned during scheduled maintenance operations and the housing oversized by machining to remove fretting. As a consequence, crush height was further reduced and fretting was instead aggravated. It was thus concluded that bearing radial pressure was not sufficient to hinder micro-slips at the origin of fretting damage, from which cracks were then allowed to initiate and grow under oscillating stresses.

Some simple calculations were performed to give a theoretical support to these findings. The portion of fatigue

life  spent  for  crack  propagation  was  estimated  via  the  classical  Paris    law

assumption of a stress ratio equal to zero. The fitting parameters C and m were obtained by undisclosed internal tests, while the stress amplitude was taken from the original finite element model. In order to integrate the law, the initial and final crack lengths have to be specified. The longest crack found without failure is about 0.42 mm, which was therefore chosen as the initial crack length. The final length was fixed at 110 mm on basis of the remaining cross-section of the failed parts. Under these assumptions, the number of cycles spent for crack propagation came out to be about 105, which is fully negligible compared to the expected lifetime of the component (above 108 cycles). More sophisticated fatigue crack growth models are not expected to change  significantly the outcome,  with  the crack propagation life always resulting in few percentage points of the total one.

An attempt to explain the threshold crack length was made using fracture mechanics concepts. The threshold stress oth required for a crack of length a to propagate can be put in relation with the fatigue limit of a smooth specimen o0  via the El Haddad-Topper-Smith equation [2]:

where the shape factor a2 was applied to account for the geometry of a real component [3]. The material constant a0 is defined by the relationship:

Equation (2) provides a simple way to assess quantitatively the fatigue strength reduction due to the presence of a crack, but does not consider the stress raising effect due to fretting. On basis of a simplified stress analysis, the stress concentration factor was estimated to be around 2. According to internal test results, the fatigue threshold 6Kth and the pulsating fatigue limit o0 are respectively 7.7 MPaKm and 384 MPa, while a equals 0.704 for a 45°-inclined edge crack [4].  By introducing these  data  into  Equation (2)  a  crack depth of 0.327  mm was  obtained,  in fairly     good