A welded joint is not one material. It is at least three. Practice usually specifies a filler metal at least as strong as the plate it joins, so the weld deposit is often the last part of the joint to give way. What fails instead is frequently the narrow band of base metal alongside it — never molten, but heated, cooled and metallurgically rewritten by an arc passing an inch away. That band is the heat-affected zone.
Three metals in one joint
The fusion zone solidified from a molten pool and has a cast structure, with grains that grew inward from the cold walls of the joint. The base metal beyond the joint is unchanged, still carrying whatever microstructure it was supplied with. Between them sits the heat-affected zone, which was never liquid but was driven through a rapid heating and cooling cycle that a metallurgist would recognise as an uncontrolled heat treatment.
The zone is narrow — often a matter of millimetres — and it is not uniform across its width. Nearest the fusion line the metal reached temperatures high enough to coarsen the grain; further out it was merely tempered or partly transformed. Each sub-zone behaves differently under load, and a crack tends to find the least favourable one.
Why the weld metal is often the strongest part
Filler metals for structural and pressure work are commonly selected to overmatch the strength of the base metal, so that yielding occurs in the plate rather than in the joint. Pointing at the weld as the weak link is therefore often wrong. Where the deposit is sound and the filler correctly matched, the least favourable metal in the assembly is usually on either side of it.
What the arc does to metal it never melted
Rapid cooling in a hardenable steel can transform the coarse-grained region next to the fusion line into a hard, low-toughness microstructure. The same thermal cycle can do the opposite in a quenched-and-tempered steel, over-tempering a band of parent plate and leaving a softened zone that yields before anything around it.
Both outcomes follow from heat input and cooling rate rather than from manual skill, which is why a weld dispute so often moves from the person holding the electrode to the procedure that told them how much heat to use.
The hardness traverse
A line of indentations run across the joint — base metal, heat-affected zone, weld metal, and out the other side — is the most economical measurement in weld metallurgy. It converts an invisible thermal history into numbers, and the shape of the profile is diagnostic. A sharp peak at the fusion boundary indicates fast cooling and a hardened zone; a trough indicates local softening.
Hardness is also the practical screen for susceptibility to hydrogen-assisted cracking, which is why several service standards place hardness limits on welded joints — NACE MR0175/ISO 15156 for sour service being the most cited. A measured hardness outside the limit for the service is a finding in its own right, independent of whether anything has cracked.
The macroetch section
Cutting through the joint, polishing the face and etching it reveals in one image what no external inspection can: the fusion boundary, the extent and shape of the heat-affected zone, the number and sequence of passes, the depth of penetration, and the position of any crack relative to all of them.
Macroetch examination is itself a code requirement for certain qualification tests, so the technique is familiar to fabricators and inspectors rather than devised for litigation.
Cracks that never touch the weld metal
Hydrogen-assisted cold cracking commonly sits in the hardened heat-affected zone beneath the final bead rather than in the deposit, and it can appear hours or days after the arc was extinguished. Lamellar tearing sits further out still, in the parent plate, following bands of non-metallic inclusions where a restrained joint pulled on the plate through its thickness.
Reheat cracking, in susceptible low-alloy steels, appears in the coarse-grained zone during post-weld heat treatment or high-temperature service. What these share is a location: a visual inspection of the weld face will not find any of them.
Heat input, preheat and interpass control
Cooling rate through the transformation range is governed by heat input, section thickness, and the temperature of the surrounding steel when the arc arrives. Preheat and interpass requirements exist to slow that cooling, keeping the zone softer and giving diffusible hydrogen time to escape. AWS D1.1 sets preheat requirements by material group and thickness for structural steel, and the site records are among the first documents an examiner asks for.
What the codes require of the zone nobody sees
Procedure qualification under ASME Section IX tests the joint as a whole, and where the construction code calls for notch-toughness testing, specimens are taken with the notch in the heat-affected zone as well as in the weld metal — an explicit acknowledgement that the two behave differently. AWS D1.1 for structural steel and AWS D1.5 for bridge work impose their own qualification and preheat regimes. None guarantees a particular microstructure in production; they establish that the procedure was capable of producing an acceptable one under test.
What has to survive to be examined
An investigation of this kind needs the joint intact with a generous margin of parent plate on both sides, because the zone of interest extends beyond the visible weld. Grinding, gouging out a suspect area, or repair welding removes precisely the microstructure in question, and a repair pass imposes a fresh thermal cycle over the record of the original. Where a section must be cut for transport, the cut should be well away from the joint and add no heat to it.
This article is general technical orientation, not a failure analysis, an engineering opinion, or advice on any specific matter. Determining the cause of a particular incident requires hands-on examination by a credentialed expert.