Page 457 - Failure Analysis Case Studies II

440

detected in zones far from the fracture surface. In addition, other cracks were detected under the
fracture surface with a defined orientation, that seems to support the existence of internal stresses
in these areas. Moreover, the microhardness of the material close to the fracture is very high,
reaching values up to 500 HV close to the outer diameter.
As regards the chromium-plated sleeve, this shows macroscopically at the OD close to the
fracture many grooves of different widths and depths. Microscopically these correspond to dam-
aged zones without the chromium-plated layer, with different microstructural characteristics and
with significant grain growth. These microstructural changes were also reflected in the fractographic
examination. In addition, the fractography indicated that the final fracture of the sleeve was
produced at the ID.
All these observations, related to the changes in the microstructure and in the microhardness,
in the shaft and in the sleeve close to the fracture, point to both having-been subjected to very high
temperatures, above the recrystallization temperature. It is well known that one of the more clear
indications that a metal has been overheated is the coarse grain fracture surface that results [2].
The characteristics detected in the shaft cracking, macroscopically and microscopically, suggest
that the cracking originated in quench cracks due to localized overheating. This process results
from stresses produced as a consequence of the volume increase accompanying the austenite to
martensite transformation. When a steel is quenched, untempered (hard and brittle) martensite is
formed at the outer surface first. As cooling continues, the austenite transformation progresses
towards the center of the piece, accompanied by a volumetric expansion. This results in internal
stresses that place the surface in tension and cracking occurs. Moreover, the resultant martensite
is very hard and brittle, making the material very susceptible to cracking [Z]. On the other hand,
in the shaft failure the appearance of internal cracks (another type of thermal crack), may have
contributed to the failure. Internal cracks occur when the overheating is too fast as a result of
unequal temperatures of the surface relative to the center of the mass [Z].
Taking into account that the operating temperature conditions cannot produce the overheating,
it may be deduced that the heat source has been friction between the shaft, or its associated
components, and the casing. This implies that, firstly, the failure of the graphite-nickel bearing was
produced. Then the chromium-plated sleeve started to rub with the pump casing as demonstrated by
the presence of copper and lead, coming from the original bronze bearing in contact with the
casing. This fact is also supported by the grooves observed in the outer surface of the sleeve. The
friction with the casing originated an overheating that was transferred through the sleeve to
the shaft, producing finally a microstructural change in the shaft material and its consequent
embrittlement.

6. Conclusions

The failure sequence has been as follows:

(1) The graphite-nickel bearing fails.
(2) The chromium-plated sleeve rubs against the casing and the heat generated by this friction is
transferred through the sleeve to the shaft.