Delayed Cracking on A514 Lifting Lugs
An OEM's lifting-lug contract requires 2 in. fillet welds on 1 in. A514 Grade B plate, welded with E11018-M per a qualified WPS calling for 200 F minimum preheat, 250-350 F interpass, and a 48-hour delay before inspection. In January the shop floor sits at 46 F overnight and the plate is stored outside under a tarp. On Tuesday the rod oven at station 4 was down for repair, so the crew ran E7018 from a portable quiver that had been standing open since Monday. On Wednesday morning, three of eight lugs show toe cracks on the base-metal side, roughly 1/4 in. long. The lugs are already fit to the frame, and the customer's witness point is Thursday. The welders say the cracks are "just surface" and want to gouge and re-run. The shop has two rosebud torches and a contact pyrometer that reads about 15 F low against the calibrated shop unit.
1. What is the likely mechanism, and what evidence would confirm it before you commit to a repair plan?
Read the worked answer
Hydrogen-assisted delayed cracking, and every condition for it is present.
Three factors combine:
- A514 Grade B is a quench-and-tempered low-alloy steel with a carbon equivalent high enough to be hydrogen-crack sensitive, and it depends on hydrogen-controlled consumables and preheat.
- Preheat was effectively absent. At 46 F plate against a 200 F minimum, the weld cools fast and hydrogen has no time to diffuse out.
- The E7018 in an open quiver is the decisive clue. Low-hydrogen electrodes are only low-hydrogen while dry; a quiver standing open in a shop in January is an oven that isn't. E7018 also undermatches a 100 ksi base metal, so the joint is being asked to carry load with weld metal that was never specified for it.
Timing and location fit: toe cracks on the base-metal side appearing the morning after welding are the classic delayed-hydrogen signature. A hot crack would have shown immediately, usually at the crater or centerline.
Evidence to gather: crack location and orientation relative to the weld, time since welding, electrode lot and quiver history, measured preheat and interpass from a calibrated instrument (not the one reading 15 F low), and MT or PT on every lug. If the customer requires it, hardness or hydrogen checks on a sample.
2. What is your containment, including the five lugs that show no cracks?
Read the worked answer
Quarantine all eight and stop producing more.
- Cracks under a lifting lug are a device-failure mode, not a cosmetic one. MT or PT all eight lugs, 100% of the weld plus at least 1 in. of adjacent base metal. "Just surface" is a verbal guess; the test is the answer.
- Do not gouge and re-run a cracked lug in place. Visible surface cracks usually indicate a deeper path, and re-welding into hydrogen-charged, hardened base metal reproduces the problem at a larger scale. Remove cracked lugs, grind out to sound metal, verify with MT, then re-weld.
- Restore the consumable controls: E11018-M held at 250-300 F, issued in a portable holding oven, returned and logged at shift end, and anything that sat open reconditioned or scrapped. A log with a name on it is what changes behavior.
- Restore preheat: 200 F minimum, verified with the calibrated reader, interpass capped at 350 F, and controlled cooling. Two rosebuds is thin coverage for eight lugs - raise blankets as a purchasing issue before the next run.
- Honor the 48-hour hold before inspection. Compressing it to make Thursday's witness point is precisely how a crack that has not opened yet ships to a customer.
3. Build the recovery plan and decide how much to tell the customer and when.
Read the worked answer
Sequence, cost, then the customer conversation.
- Sequence: MT/PT all eight, excavate the cracked ones, re-weld under the corrected conditions, 48-hour hold, final MT/PT, dimensional check. Realistically three to four days once cure time is respected, which puts Thursday's witness point at risk. Tell the customer before they travel, not when they arrive.
- Cost: repair labor on eight lugs, electrode reconditioning, possibly a metallurgical evaluation at roughly $1,500-3,000 if the OEM asks. That is cheaper by orders of magnitude than a field failure on a lifting device.
- Customer: report the escape yourself, with root cause and containment, and invite their inspector into the hold point. Lifting-lug work is safety-critical and self-reported containment is what preserves the contract when something goes wrong.
- Upstream fixes: no welding on A514 below preheat, measured and logged per lug; store the plate indoors or preheat before fit-up; add the rod ovens to the PM schedule; and confirm with engineering that the WPS was written for this base metal and heat-input range before the next batch is released.