Steel-made components of modern fuel injection systems are designed for pressure
amplitudes of ≥300 bar (gasoline engines) and 2200 bar (diesel engines),
respectively. In order to evaluate the risk of field failure, for example, for a
service life of 300 000 miles, Wöhler pulsation tests are conducted at very
high-pressure levels far beyond the service pressure. In a standard procedure,
the results of these high-cycle fatigue (HCF) tests with an ultimate number of
cycles of 5∙106 are extrapolated down to real-life load amplitudes,
assuming that there is a unique function for the dependency of failure
probability PA on pressure amplitude Δp, regardless of the different
failure mechanisms and crack initiation sites, like surface imperfections,
internal defects, etc. Probability functions of the Weibull or the log-logistic
type are usually fitted to the dependency of failure rate on pressure amplitude
taken from the HCF experiments, and it is assumed that this functional relation
can be extrapolated over five orders of magnitude from Pf = 5% down to 1 ppm.
This assumption, which can lead to a false estimation of fatigue limits, is
questionable. Although different approaches have been proposed which take the
multitude of fracture mechanisms into account, the standard procedure is still
very common, maybe due to the complex nature of the subject. In this article, a
simple refinement of the procedure for fatigue stress evaluation is suggested,
considering the important role of non-metallic inclusions and their size
distribution. This approach allows for a safer estimation of fatigue limits
based on mechanical data and metallographic investigations. An example for the
difference of the standard and the simplified new approach with data derived
from high-pressure pulsation tests and fractographic analyses on case-hardened
steel specimens is given.