2 mm Stainless Tanks: Bowed Panels, Black Roots and Failed Leak Tests
A workshop in Galway is making twelve rectangular tanks from 2 mm 304 stainless sheet, each 800 × 600 × 900 mm, TIG welded with 1.6 mm 316L filler. The jig clamps only the base of each tank. The customer specification requires a pickled and passivated internal finish and a 0.5 bar hydrostatic leak test. After the first three tanks, the 800 × 900 mm side panels are bowing outward by 6–9 mm and the outside corner roots are coming out grey-black with a sugared appearance. Three of the first six tanks have failed the leak test on the corner seams. Delivery is due in five days, and the shop has no rolls or press capable of straightening the panels. The welder holds a valid IS EN ISO 9606-1 certificate for stainless TIG, and the WPS in the folder was originally written for 3 mm material at 90 A in a single pass. The foreman wants the batch back on track without scrapping the six part-built tanks.
1. Explain why the side panels are bowing 6–9 mm and set out the changes to technique, sequence and tooling you would apply to tanks 4–12. Assume the panels cannot be straightened after welding.
Read the worked answer
Cause. The bowing is shrinkage-driven, not a material fault. A single pass at 90 A on 2 mm sheet is far too much heat input: the weld pool is roughly three times the cross-section the joint needs, so contraction on cooling pulls the free edge of each panel inward and the panel bows outward to compensate. Clamping only the base lets the top edge move freely, and wide tack spacing lets the gap open as welding progresses, which makes the welder slow down and put in more heat.
Corrective actions.
- Rewrite the WPS for 2 mm: about 45–60 A, 1.6 mm filler, fast travel, stringer bead, no weave. Consider pulsed TIG to lower average current.
- Clamp or strongback the panels to a rigid jig on both faces, not just the base, and leave them clamped until the part is cool to touch.
- Tack every 50–75 mm and check the gap before welding.
- Use copper or aluminium chill bars clamped either side of the joint to draw heat out and hold the panel flat.
- Weld with a balanced sequence: backstep in 100–150 mm steps, alternate opposite corners, and complete verticals and horizontals alternately so shrinkage on one side is opposed by the other.
- Add a shrinkage allowance of about 1.5–2 mm per seam to the flat pattern so the finished tank meets the 800 mm dimension.
Do not plan to fix it afterwards. Flame straightening is not an option on 304 - local heating in the 450–850 °C range risks sensitisation and heat tint - and planishing work-hardens the sheet and can crack welds. Prevention is the only economical route, so the six part-built tanks should be re-sequenced rather than corrected.
2. The corner roots are grey-black and three tanks leaked. Explain what has happened metallurgically, what the specification demands, and how you would repair the failed tanks without making the problem worse.
Read the worked answer
What the colour means. Grey-black root colour and sugaring indicate the back of the weld was not shielded: oxygen and nitrogen from air reacted with the hot metal, oxidising chromium and forming nitrides and complex oxides. The passive chromic oxide layer is destroyed in that zone and the corrosion resistance is permanently reduced, even though the weld may look sound from the outside. That is also why the leaks concentrate on the seams: the contaminated root is porous and often has incomplete fusion at the edges.
What the specification requires. Full argon purging of the enclosed side before welding, with the oxygen level at the root held below about 100 ppm (many tank specifications demand 50 ppm or better). Purge dams or water-soluble tape should be set close to the joint to keep the purge volume small, and the purge must be maintained until the weld is below roughly 200 °C. After welding, heat tint is removed by pickling paste or a pickling machine, then the surface is passivated - this is an explicit requirement here, not cosmetic.
Repair method.
- Drain and dry the tank, then grind out the defective root back to bright, uncontaminated metal, extending 25 mm beyond each end of the defect.
- Dye-penetrant check the excavation to confirm the defect is gone; do not simply re-melt over the top, as this buries the contaminated layer and the leak returns.
- Re-weld under a qualified repair WPS with a proper purge, using dedicated stainless grinding discs - carbon steel discs leave free iron that rusts and is a common avoidable defect.
- Re-pickle the repaired area, re-test at 0.5 bar, and record the repair against the tank serial number.
3. Before restarting production you must confirm the paperwork is adequate. What would you check regarding welder qualification, WPS coverage and consumables, and what would you reject?
Read the worked answer
Welder qualification. The certificate must cover the process (141 TIG), the material group (stainless, ISO 9606-1 group 8), the joint type and the thickness range. A welder qualified on 3 mm plate may or may not cover 2 mm - check the stated range on the certificate rather than assuming, and check the certificate's validity date and that it was issued or re-validated by an examiner with the correct 6-monthly confirmation of continued employment.
WPS coverage. The existing WPS was qualified on 3 mm at 90 A; a 2 mm procedure with different current, gas and pass sequence is a different set of essential variables. It needs either a new WPS supported by a pWPS and, where required, a weld procedure test to IS EN ISO 15614-1, or at minimum a documented procedure revision with the supporting qualification. Welding to an unqualified WPS on a pressure-tested item is not acceptable, so this would be rejected.
Consumables and equipment. Check filler certificates and that 316L is actually what the specification calls for; check that argon is welding grade with a current cert, that the purge gas supply and flow meter are working, that the machine's calibration and pulse settings are documented, and that the pickling paste is in date and applied per the manufacturer's instructions with the correct PPE and containment.
Reject any attempt to run the batch on the 3 mm WPS on the grounds of the deadline; the leak-test failures are evidence that a shortcut has already been taken.