Page 106 - Failure Analysis Case Studies II

91

HOT BLADE
Fig. 9. Weld formation in final closure of plastic panels to form large diameter hoop. The panels are bent round to meet
a hot knife, the knife is withdrawn, and the two ends pressed together to form the final joint. One problem is that the
panels are still in a state of strain, which is not relieved. Stress relaxation will occur with time, but residual stresses may
contribute to early failures if the design itself is faulty.

present, the spark would travel through the gap and the defect be discovered. The method was
relatively crude, and could not detect pinholes which did not connect the two opposite surfaces of
the weld, for example.
The large circumferential hoops are created first, in the following way. Flat sheets previously
cut to the appropriate length with the table saw are welded up until the last joint is ready to be
made. The ca 8.55 m length (in the case of the failed tank) is then bent round by hand to form a
complete circle, and the ends thermally welded by the machine to make the hoop (Fig. 9). The
hoop is therefore under a continuous bending stress, with a large tensile component in the outer
surface, and a compressive component on the inner. This is discussed further, especially the effect
on the gross stress to which the welds are exposed as a result, in Part I1 of this investigation.
Similar hoops made in the same way are then stacked and welded to create the tower. The
operator thought that the horizontal welds thus made were somewhat weaker than the butt welds
made initially, but certainly much stronger than hand made welds.

7. Materials analysis programme

Material from both panels and welds was also tested for its chemical integrity. The tests employed
included infra-red spectroscopy (using a Fourier Transform or FTIR spectrometer) and melting
behaviour (using a Differential Scanning Calorimeter or DSC). The objective in these tests was to
see if there were any perceptible differences in chemical behaviour between ‘good’ and ‘poor’ weld
material, between weld and panel material and between independent samples of polypropylene
and material taken from the tank.
FTIR spectroscopy showed that the tank material was an ethylene-propylene copolymer. Little
variation in chemical composition could be detected from several weld sample and normal tank
material. Moreover, there was no evidence of oxidation, especially in the welds. The results of
various DSC experiments conducted on the various tank materials are shown in Table 2.
The thermograms showed that both sets of samples had very similar endotherms, with melting