Page 195 - Failure Analysis Case Studies II

180

boxes. The ideal samples to examine would be those cut across suspected weld lines, since then the
defects would be subjected to the highest stress when the samples were pulled in tension.
Four samples were cut, two from the new tank in a region showing serious flow lines, and two
from the failed tank showing similar flow lines, but well away from the failure crack. One of each
pair was cut parallel to the flow lines, the other at right angles to the flow lines. The samples were
about 7 cm long, and possessed a ca 1 cm narrow central region about 5 cm long, so giving a
dumbell shape. The samples were cut very carefully using a small hacksaw along the pre-marked
shape, and polished by hand with a series of finer emery papers to remove any edge imperfections
which could cause premature fracture.
All the specimens were tested at room temperature (ca 25°C) and ambient humidity (ca 50%),
using a constant strain rate of 0.1 mm s-’. Three of the four samples broke centrally, the fourth
(No. 2) on the edge of the shoulder. All the samples broke in a brittle fashion, i.e. the two parts of
the fracture surfaces could be fitted back together. The results were as follows:

Sample No. 1 (new, lateral) og = 84 MN m-’
Sample No. 2 (new, parallel) oIj = 8 1 MN m-2

Sample No. 3 (old, parallel) os = 80 MN m-2
Sample No. 4 (old, lateral) crI, = 55 MN m-2

The elongation to break was very similar for the first three samples, at approximately 9%, and
10% for the final sample.
It may be noted that the best result of 84 MN m-* calculated for sample No. 1 fell somewhat
below the ideal value given in the data sheet for this polymer (140 MN m-2). Sample No. 3
appeared to have failed from a surface tool impression mark, present as a sharp corner across part
of the dumbell, for example. Inspection of the fracture surfaces, however, showed them to be
reasonably free of internal voids and cold slugs of the kind found in the fracture surface (Fig. 12).
All the fracture surfaces showed a central ‘spine’ or cusp indicative of skin-core control of fracture,
quite unlike that of the critical fracture. The calculated elongations to break are rather greater
than the value of 6% quoted on the same sheet, both strength and elongation to break being given
for samples conditioned to ambient temperature and humidity essentially identical to those used
here. Comparison of the tensile strengths showed that the material from the new tank is superior
to that from the failed tank. Although there were visible flow lines in all the samples, failure did
not seem to be related to them in any clear, unambiguous way. In general, the tests reinforced
earlier impressions when cutting the tanks for analysis of relatively stiff but brittle mechanical
behaviour. The addition of 30% chopped glass fibre improves the stiffness of nylon, but at the
expense of strength.

5. Discussion

A reasonably clear picture of the failure emerged as a result of detailed examination of the
critical fracture surface, and comparison of the properties of new and unused tanks. The critical
crack probably resulted from the coalescence of several smaller cracks below the fan buttress.