Fatigue cracking at a stiffener-to-flange weld on an ageing gantry
A 12-year-old conveyor gantry in S275 steel carries a reversing load of about 2×10^6 cycles per year. Cracks 40–80 mm long have been found at the toe of a 6 mm continuous fillet weld that attaches a transverse stiffener to the tension flange, at three separate locations along the span. The original design drawings specify a 6 mm fillet and state that the detail is “non-load-carrying, category 56”. Site measurements give a measured stress range at the detail of about 60 MPa. The stiffener is welded on both sides, the weld toe angle is steep, and there is no toe treatment. The owner wants to gouge out the cracks, reweld the same detail, and return to service within a 48-hour shutdown. An alternative proposal is to remove the stiffeners entirely, but the webs show signs of breathing under load.
1. Was the design category correctly applied? What is the actual failure mechanism?
Read the worked answer
The declared category 56 should be challenged. A continuous fillet attaching a transverse stiffener to the tension flange is a welded detail at the tension fibre; even when the stiffener is nominally non-load-carrying, the weld termination creates a notch and the local stress concentration raises the effective stress range. Such details are normally assessed at category 50 or lower in EN 1993-1-9 Table 8.1, and at 36* where the stiffener is genuinely load-carrying; the exact row depends on the end detail and the attachment. The arithmetic also matters: the fatigue limit at 2×10^6 cycles equals the detail category, so a 60 MPa stress range already exceeds category 56 at the design life, and the gantry has accumulated roughly 24×10^6 cycles. On its own stated assumptions the detail was marginal. The crack pattern - three separate locations, all at the toe on the tension flange, growing from the toe - is classic toe-initiated fatigue, not a material defect or a single bad weld. The steep toe angle and lack of treatment explain the initiation sites. The as-built detail is therefore worse than the drawing assumption, and reweldng the same geometry will restore the same notch.
2. Evaluate the two repair options and specify a repair procedure.
Read the worked answer
The 48-hour reweld is the cheapest and the least durable: it removes the crack but not the cause, and a repair weld often has a harder, rougher toe than the original, so the next crack may appear sooner. Removing the stiffeners addresses the notch but must be checked for web breathing and plate buckling; if the web relies on the stiffener, removing it trades a fatigue problem for a stability problem. The better repair improves the detail. Excavate each crack to sound metal by gouging or grinding, confirm by MT or PT that no crack remains, re-weld with a low-hydrogen consumable, then treat the toe - grind to a smooth radius, TIG dress or HFMI - and, where the design allows, cut the stiffener end back to a taper or radius so the end weld is soft. That raises the effective detail category and removes the sharp notch. Check the other stiffeners on the span: three cracks mean the detail is systematic, not local. If the owner insists on the 48-hour window, phase the work: make the repairs safe and treat the toes now, and plan the stiffener redesign for the next planned shutdown.
3. What post-repair and ongoing inspection would you specify?
Read the worked answer
After repair, MT or PT every repaired toe, and use UT if there is any doubt about crack depth or if the excavation was deep. Record the repair locations and the toe treatment so the next inspector knows what was done. Then inspect earlier than the normal cycle: a first in-service check at about three months, with MT at the same locations and a close visual for toe cracks; if clean, move to 12 months and review the remaining fatigue life using the improved detail category. The inspection method must match the defect: dye penetrant or magnetic particle at the toe, not a distant visual from the walkway. Also confirm by calculation that the repaired detail can carry the measured stress range for the intended remaining life, and if it cannot, plan the redesign rather than relying on inspection. The inspection interval should be justified by that calculation, not by habit. Finally, make sure the owner understands that passing one inspection is not a statement that the detail is now adequate.