Multiaxial High-Cycle Fatigue Criterion In Mechanical Components Subjected To Impact Load.

Abstract

In several industries, the required design lifetime of many components often exceeds 108
cycles. This requirement is applicable to aircraft (gas turbine disks 1010 cycles), automobiles (car engine
108 cycles), and railways (high speed train 109 cycles). Although a large amount of fatigue data has been
published in the form of S-N (where S is stress and N cycles numbers) curves, the data in the literature
has been usually limited to fatigue lives up to 107 cycles. Using traditional fatigue criterions, a nearhyperbolic
relationship between stress and fatigue life is assumed. Experimental results in steels show
that the fatigue fracture can occur beyond 107 cycles. This means that in very high cycles number the
endurance limit has not asymptotic behavior and the concept of infinite fatigue life is not correct. For this
reason, to assert the expected life time of steel components it is necessary to carry out very prolonged
tests. FEM (Finite Element Method)simulation is a good way to solve this problem in short times.
In this paper, we present results from numerical models analyzing mechanical components subjected
to high number of impact cycles using commercial software. Two formulations are applied to solve the
problem: Crossland, Dang Van criterions.
As the loads on the system appear from the impact of flexible elements, contact algorithms were used.
With methods based on Lagrange multipliers, contact conditions are infinitely rigid and induce numerical
perturbation. To avoid this problem relaxed contact conditions were used by adding a penalty function.
In the first trials, the time integration algorithm used for solving this structural dynamics problem
was Hilber-Hughes-Taylor (HHT) but it showed poor high-frequency dissipation. Finally, the integration
method used to solve the dynamic problem was the generalized- method, because it achieves high
frequency dissipation while minimizing unwanted low-frequency dissipation.