Fracture Surface No. 1
Fatigue initiating from the toe of a transverse
non-load-carrying fillet weld
it iS. virtually impossible to design a complex
welded structure without fillet welds, Oue to their
shape, and the presence of small defects (2) at the
wold foe (Figure 1) these constitute severe stress
raisers, either across the plate where the stress is
transverse to the weld, or at the weld end for welds
parellel to the direction of stressing. Consequently,
the fatigue strengths of many welded structures are
dictated by the unavoidable presence of fillet welds
‘The fracture surface is typical of that commonly
observed in the case of transverse non-load-carrying
fillet welds. Figure 2 shows the SN curve from the
series of which this spacimen was part, and the actual
test result for the specimen is indicated by an arrow in
the diagram,
Points to note on the fracture surface (Figure 3)
tare that the fatigue failure has initiated at the toes of
the welds on the two opposite surfaces of the load.
carrying plate and that it has initiated at a number of
points along these weld toes (a, b, c,d and p, a).
‘These separately initiated fatigue cracks on each side
have been joined up by steps in the fracture surface
more or less perpendicular to its plane, Once the
Separately initiated cracks on each side had joined up.
the fatigue cracks on the two opposite sides of the
‘Stress concentration
‘at weld end
Fig 1, Non-load-carrying tet welds.
specimen grew in an approximately semi-elliptical
fashion until they, in turn, eventually joined up in @
similar fashion by shear lips (Figure 4). The final
ductile fracture (X) represents a relatively small pro
portion of the overall area of the fracture surface
Scanning electron microscopy of the flatter features,
of the fracture surface would show the individual
fatigue striations associated with each cycle of
loading. The shiny regions at each end of the load-
carrying section were caused by the two broken
halves hammering together immediately after failure
(ie, before the testing machine had stopped)
‘Other features to note are the surface ripples of the
weld beads, which indicate that each weld was made
in two passes, each starting from one of the ends of
the non-load-carrying attachment and being deposi
ted towards the centre, where the smooth circular
‘stop position of the second weld bead can be seen (F)
Both surface ripples and start and stop positions can
provide easy initiation sites for fatigue failure if
suitably oriented with respect to the stress system
Note particularly the undercut at G which would have
provided @ particularly good initiation site for fatigue
in this specimen, if it had occurred on the other side of
the weld bead.
Longitudinal
weld
‘Transverse weld
Stress concentration
at weld toe220
lower limit stress =0
180
§ *
140
g eee wee .
5 NN
a
100
ote 408 TOOT TE too D3 es to?
Endurance, cycles
Fig. 2. S-WW curve for transverse non-foad-carrying fet welds.
Fig. 3 Fatigue tracture initiated at 9 transverse non-tord-carrying fillet weld.
w -— ~~ - Te
Initiation
“i — ore
Intermediate
)
Advanced
developmentTacture Surface No. 2
Fatigue initiating at the edge of a transverse
butt weld
A perfect butt weld whose surface is flush with the
plate surface should have very little stress concen-
trating effect, but in practice, because excess metal is
deposited, the fatigue strength of transverse butt
welds is ctiticelly dependent on the weld profile (3)
Strengths at 2 million cycles can vary between 75 and
480 N/mm* and although the profile can be improved
by grinding, commercially this is very expensive.
The sample in the collection is of a butt weld
(Figure 5), deposited using an automatic welding
process, in this case submerged-arc welding. An S—N
curve for this type of joint is shown in Figure 6, and
again the result for the actual specimen is indicated.
‘Automatic welding produces a very uniform weld toe
profile and it is of interest to observe on the fracture
surface (Figure 5) that fatigue crack initiation has
taken place over almost the entire length of one side
of the specimen (A~A’). Also, because of the uniform
nature of the weld toe, the crack has initiated almost
entirely on one plane and such stepping as occurs
(B, C) between separately initiated cracks is far less
marked than in the case of the fillet weld described
above which was made using the manual metal-arc
process. The applied stress range in this specimen
was less than half the yield stress of the plate material
80 that the fatigue crack could propagate almost
entirely through the section before failure occurred
with the formation of the very small shear lips (D)
Other features of interest on the fracture surface are
the opening up of inclusions (E, F) in the rolling
plane of the plate material at the extremities of the
fatigue crack, just before final fracture. This de
lamination was caused by the extensive strain
‘occurring at the crack tip as final failure started, and is,
ot confined to fatigue fractures.
Fig. 5 Fatigue fracture initiated atthe edge of a transverse butt weld.
300
260:
220
180
stress, N/mm2
140
Upper 1
Stress concentration
at weld toe
100 |_
3x 104
oa 105
lower limit stress =0
, 108 107
Endurance, cycles
(9.6 S-N curve for transverse butt welds