Weld and Manufacturing Defect Analysis

Welding standards , AWS D1.1 for structural steel, ASME Boiler and Pressure Vessel Code Section IX for pressure welds, ISO 5817 for classification , define an acceptance hierarchy. Not all discontinuities are defects: a discontinuity becomes a defect when it exceeds the acceptance limit for its service category. Forensic analysis must establish whether the discontinuity found at a failure origin exceeded the applicable standard at the time of fabrication, and whether the inspection system that passed the weld was adequate to detect it.

Hydrogen cold cracking (HCC) in welds requires three simultaneous conditions: absorbed hydrogen (from moisture in the electrode coating, shielding gas contaminants, or base metal surface), a susceptible microstructure (martensite or lower bainite in the HAZ), and sufficient residual tensile stress. Remove any one and HCC does not occur. The same three-condition model applies to stress corrosion cracking, but each condition has a different source.

The metallurgical reason HAZ martensite is susceptible is its low hydrogen diffusivity combined with high hardness and reduced plasticity. Hydrogen absorbed at welding temperatures is partly trapped in the HAZ as it cools. Below about 150 degrees C, diffusion slows dramatically and the hydrogen is now essentially locked in. Under residual stress, it migrates to stress concentrations at HAZ grain boundaries , the same physics as service-environment HE, but the hydrogen source is the welding process itself.

• Prevention: preheat , raising the base metal temperature before and during welding slows the cooling rate, reduces the martensite fraction, and promotes hydrogen diffusion out of the HAZ before the temperature drops below the diffusion-active range. AWS D1.1 and BS EN 1011-2 provide preheating tables as a function of carbon equivalent and heat input.
• Prevention: low-hydrogen electrodes , E7018 and similar H4-grade (maximum 4 ml/100g diffusible hydrogen) or H2-grade electrodes reduce the hydrogen input substantially compared to cellulosic E6010 electrodes. Electrodes must be stored dry and used within the manufacturer's exposure time after opening.
• Detection: time-delayed inspection , because HCC can develop days after welding, final NDT on susceptible joints (high-strength steel, heavy sections, highly restrained joints) should be performed not immediately after cooling but after a 48-to-72-hour holding period to allow crack development.

Castings and forgings each have characteristic defect populations that reflect their solidification or deformation history. In a failure investigation, confirming that a discontinuity is manufacturing-origin (rather than service-induced) is critical because it changes the liability framework: was the component never fit for purpose, or was it degraded during service?

• Shrinkage porosity (casting): dendritic voids that form in the last liquid metal to solidify, typically at the thermal centre of the casting. Identifiable by their irregular, branching morphology on a fracture surface or cross-section, contrasted with smooth spherical gas porosity.
• Inclusions: non-metallic particles (oxide, sulfide, silicate) introduced by slag, mould material, or deoxidation products. Energy-dispersive X-ray in the SEM gives composition directly. Stringer inclusions aligned with the forging direction are the precursor to laminar tearing in heavy plate.
• Seams and laps (forging): surface defects created when oxidised metal folds over during hot-rolling or forging without fusing. They run parallel to the workpiece surface, are typically filled with oxide scale, and can be mistaken for cracks on visual examination.
• Cold shuts (casting): planar discontinuities where two streams of liquid metal met but did not fuse due to premature solidification. They are smooth, planar, and oxidised, distinct from hot tears which are rough and branching.

Metallographic section preparation is essential for manufacturing defect investigation. A cross-section perpendicular to the suspected defect plane, polished to 1-micron finish and etched to reveal the microstructure, will show whether the discontinuity is bounded by a heat-affected zone (service-induced fatigue or overload crack), oxide layers (cold shut, seam, lap), or an irregular void morphology (shrinkage). This context-in-the-metal cannot be replicated by fractographic examination of the fracture surface alone.

Non-destructive testing plays two roles in failure investigation: it was supposed to have detected the defect before service (the original quality assurance role), and it is now used post-failure to characterise remaining material and to understand what the original inspection would and would not have detected.

RAF Nimrod XV230 crashed in Afghanistan on 2 September 2006 after a fire in the No. 7 fuel tank bay killed all fourteen crew. The subsequent independent review by Charles Haddon-Cave QC, published in 2009, is one of the most detailed public forensic investigations into an airworthiness failure in British aviation history.

The Haddon-Cave Review found that the cross-feed duct in the No. 7 fuel tank bay, which carried hot pressurised air from the engines, had been inadequately assessed through a series of modifications and safety reviews dating back to the 1990s. The ducting joints included brazed couplings of uncertain quality; the hot-air system created conditions suitable for igniting leaking fuel. The investigation identified not a single dramatically defective weld as the precipitating cause, but a systemic failure of the safety case process: responsibilities for assessing the fuel-heat source interaction had been lost across organisational changes, and the duct was never subjected to the risk assessment that its proximity to fuel systems demanded.

For forensic engineering, the Nimrod case teaches three things. First, that weld and joint quality cannot be separated from the system-level safety case: a technically adequate weld in the wrong location, at the wrong temperature, without the right inspection regime, can be as lethal as a defective one. Second, that modifications to safety-critical systems must include fresh hazard analysis even when each individual modification appears minor. Third, that documentation of the safety basis , what was inspected, what was accepted, and why , is itself a forensic artefact: the absence of that documentation was a central finding.