Hydrogen-assisted HAZ cracking in a restrained HSLA node: which factor dominates?
A 25 mm thick HSLA plate (0.10 C, 1.40 Mn, 0.30 Si, 0.20 Cr, 0.20 Ni, 0.10 Mo, balance Fe) is welded into a highly restrained box node by shielded metal arc welding with 3.2 mm E7018 electrodes taken from an unheated site store; the packet was opened three days earlier. Ambient temperature is 3 °C, there is no preheat, interpass is 60 °C and heat input is 1.1 kJ/mm. Ultrasonic testing 48 h after completion finds toe cracks running parallel to the fusion line, 1–2 mm into the heat-affected zone. Field hardness is 430 HV10. The written procedure was qualified on 12 mm plate in summer with the same electrode classification. The joint is a complete-joint-penetration tee with a backing strip.
1. Calculate the IIW carbon equivalent and the Pcm for this steel. Which index governs your preheat decision here, and what does it tell you?
Read the worked answer
IIW carbon equivalent: CE = C + Mn/6 + (Cr + Mo + V)/5 + (Ni + Cu)/15 = 0.10 + 1.40/6 + (0.20 + 0.10)/5 + 0.20/15 = 0.10 + 0.233 + 0.060 + 0.013 = 0.41
Pcm: Pcm = C + Si/30 + (Mn + Cu + Cr)/20 + Ni/60 + Mo/15 + V/10 + 5B = 0.10 + 0.30/30 + (1.40 + 0.20 + 0.20)/20 + 0.20/60 + 0.10/15 = 0.10 + 0.010 + 0.090 + 0.003 + 0.007 = 0.21
Which governs: at 0.10% C the steel is in the low-carbon, Pcm-governed regime, because Pcm weights carbon much more heavily than manganese and is the index that tracks hardenability when carbon is low. Both should be reported - the 0.41 CE warns that the manganese still drives hardenability in the HAZ, and the 0.21 Pcm says the steel is not intrinsically crack-prone if it is welded with low hydrogen and adequate heat input.
What it means practically: published preheat guidance for CE 0.40–0.45 at 25 mm gives roughly 100 °C minimum preheat, more when restraint is high and ambient is near freezing. The steel does not demand an exotic procedure - it demands a competent one, which this job did not have.
2. Of the four factors required for cold cracking, which is dominant here? Justify from the evidence, and explain why the procedure qualified on 12 mm plate was not transferable.
Read the worked answer
Cold cracking needs hydrogen, a susceptible (hard) microstructure, tensile stress and time near ambient. All four are present, but the evidence points to hydrogen as the dominant factor, with the hard HAZ as the enabling condition.
Justification:
- Location and orientation - cracks parallel to the fusion line 1–2 mm into the HAZ are the classic hydrogen crack morphology, not solidification cracking (which would be centreline in the weld metal) or lamellar tearing (which would be step-like in the parent plate under the tee).
- Time - detection at 48 h, not at inspection immediately after welding, means a delayed cracking mechanism; hydrogen diffuses over hours.
- Hardness - 430 HV10 is far above the 350 HV10 threshold, so the HAZ is essentially untempered martensite. At 1.1 kJ/mm on 25 mm plate at 3 °C, the cooling time from 800 to 500 °C is only a few seconds, so martensite is inevitable.
- Hydrogen source - E7018 is a low-hydrogen electrode only when it is dry. Opened three days earlier and stored unheated in winter, it reabsorbs moisture and can deliver 10–20 mL/100 g diffusible hydrogen instead of the 4–5 mL/100 g it was classified for. That is a tenfold increase in the hydrogen that must diffuse out of a highly restrained joint.
Why the 12 mm qualification did not transfer:
- Section thickness. 25 mm conducts heat away far faster than 12 mm; the t8/5 at the same heat input is much shorter, the HAZ harder, and the restraint (which scales with thickness) much higher.
- Season. The summer qualification had a warm ambient and a high initial plate temperature, which acted as an unrecorded preheat; 3 °C ambient removes it.
- Restraint. A box node tee with a backing strip is far more restrained than a test plate; restraint raises the triaxial stress that hydrogen cracking needs. A procedure is only qualified for the thickness, heat input, joint configuration and hydrogen level it was tested at.
3. Write the revised welding procedure with numbers, and specify the acceptance test that would prove it works.
Read the worked answer
Revised procedure:
- Hydrogen control. Use E7018 from a heated holding oven at 110–120 °C, issued in heated quivers that hold at 70–100 °C, with a maximum four-hour exposure before return or discard. Re-dry at 300–350 °C for 1–2 h any electrode of doubtful storage. Target H4/H5 classification, checked by diffusible hydrogen testing on the consumable lot.
- Preheat. 100 °C minimum, raised to 125 °C for the tee node because restraint is high and ambient is 3 °C. Measure 75 mm either side of the joint with contact thermocouples or temperature crayons, and verify before the arc is struck. Maintain interpass between 150 °C and 250 °C - the upper bound protects the parent metal's toughness.
- Heat input. Raise to 1.6–2.0 kJ/mm by stringer beads with a weave limit of three electrode diameters, aiming for a t8/5 of 10–15 s. This tempers the HAZ as it forms and lowers peak hardness.
- Post-heat. If full post-weld heat treatment is not required by the design code, apply an immediate hydrogen bake-out at 200–250 °C for 2 h, then slow cool under insulation. If the code requires PWHT, 580–620 °C for one hour per 25 mm achieves both stress relief and hardness reduction.
Acceptance testing:
- Re-qualify on 25 mm production thickness with the actual joint geometry and restraint, not a flat plate, and run a Y-groove or controlled thermal severity test alongside the butt qualification.
- Hardness traverse across the HAZ, root and cap: maximum 350 HV10, with the 430 HV10 of the failed joint as the explicit baseline to beat.
- Hold the qualification welds 72 h before sectioning and UT, so that delayed cracking has time to appear; section and examine at least three cross-sections metallographically.
- On production, UT at 48 h and again at 7 days for the first five nodes, and verify electrode quiver temperatures daily as a record, since consumable storage was the real failure in this job.