Page 160 - Failure Analysis Case Studies II

145
5. CONCLUSIONS

(1) Available evidence showed that the primary cause of the failure was the combination of the
high alloy filler material (E312) used to weld the high carbon steel rods (En9) and the subsequent
nitriding.
(2) The welds contained extremely large and elongated grains with hairline grain boundary cracks
and carbide precipitation both at grain boundaries and within the grains. The carbides probably
originated during nitriding and resulted in a hard, brittle weld.
(3) The piston rods failed by intergranular cracking through the centre of the weld. Once the cracks
were sufficiently long to result in bending of the rod away from the adapter, the cracks branched
and propagated along the length of the rod. Such cracking was facilitated by the brittle ferrite/
pearlite microstructure with the as rolled grain structure.
(4) Other contributory factors included the absence of preheating, differences in thermal expansion
coefficients of the weld and parent materials, the highly constrained nature of the weld joint,
the width of the weld preparation and the poor penetration of the weld.
(5) The “rosette welds” did not appear to have played any role in the prevention or retardation of
the failures.
(6) The blanking during carburising of those sections of the adapters that were to be welded was
not satisfactory and randomly varying degrees of carburising occurred in metal that was
subsequently welded. This increased the hardenability of the weld and the likelihood of cracking.

6. RECOMMENDATIONS
6.1. Weld preparation geomeiry
The weld preparation had to be changed for two reasons. Firstly, the width of the weld was
reduced and the depth of penetration increased, since the strength of a butt weld is only determined
by its depth, while excessive width can result in excessive dilution, high heat input and residual
stress. Secondly, the weld bead was designed such that the gap between the En9 tube and the En32
adapter is not tangential to the weld interface nor parallel to the weld centreline (Fig. 8).
This was achieved by a double “J preparation” with an included angle of 30” and by grooving
the adapter by up to 0.7 mm.

6.2. Filler material
This investigation showed that the E312 filler material was not suitable for the welds under
consideration. A plain carbon filler material such as AWS A5.18 ER70 S-6 was recommended as a
replacement, since this would not be adversely affected by nitriding. Trials with the recommended
filler showed that the temperatures at which nitriding takes place resulted in a beneficial tempering/
stress relieving heat treatment. In addition, the alloying of this material due to the dilution by the
parent materials rendered the thermal expansion properties of the filler somewhere between those
of the En9 and the En32 steels. The residual stresses in the plain carbon weld metal should thus be
substantially lower than those in the highly alloyed stainless steel weld metal.
Because of significant differences in thermal expansion coefficients between high alloy austenitic
steels and low alloy ferritic steels, the use of austenitic filler metals for the welding of steels with low
weldability must be approached with extreme caution. This is especially true where the joint is
highly constrained.

6.3. Welding procedure
Preheating is essential when welding steels that have good hardenability, since this reduces
the susceptibility to embrittlement of the heat affected zone and formation of low temperature
transformation phases.
The recommended minimum preheat temperature was established at 244°C using a standard
formula and a carbon equivalent of 0.71 for the En9 piston rod. If significant carburisation of the
adapter had occurred, this temperature would have to be increased.