The plate tears, not the weld. Lamellar tearing is a through-thickness (z-direction) failure of rolled steel plate, just beside the weld — a material and design problem, not a welder's technique problem. Welders cannot weld their way out of it.
Lamellar tearing is a cracking mechanism in the base metal of rolled steel plate, just outside the weld, caused by through-thickness (z-direction) tensile strain acting on elongated, flattened non-metallic inclusions — mainly manganese sulfide (MnS) — that were aligned with the rolling plane during plate manufacture. The cracks run parallel to the plate surface, in steps, in the base metal adjacent to the weld. The weld metal itself is usually sound.
It is almost always found on thick plate at T-joints, corner joints, and welded connections where shrinkage pulls in the plate's thickness direction — shipbuilding, heavy equipment, offshore structures. It matters because it is notoriously hard to repair: the tears run deep along inclusion planes, so the fix is design and material selection, not re-welding.
Stepwise cracking, z-direction cracking, through-thickness tearing, underbead tearing.
The signature is location and orientation: cracks in the base metal, running parallel to the surface, never inside the weld.
Three things, in this order: plate with flattened inclusions, a joint that pulls through the thickness, and enough shrinkage to pull it. Technique is a minor player; material and design are the whole game.
The material is the starting point. Plate is made by rolling: a hot slab is squeezed into plate, and the non-metallic inclusions it contains — sulfides, oxides, silicates — are flattened and elongated into thin, disc-like pancakes aligned with the rolling plane. The steel is strong in the rolling directions but weak through the thickness, where the pancakes act as millions of microscopic delamination planes. Through-thickness ductility is measured as the % reduction of area on a z-direction test piece — and this number, not the tensile strength, decides whether a joint tears.
The joint delivers the strain. At a T-joint, the weld metal pulls the web and the flange together as it shrinks. If the joint is restrained — the flange cannot move — the shrinkage must be accommodated by strain in the plate, and at the weld toe that strain is directed straight through the plate thickness: z-direction strain.
The plate tears. Where the z-strain meets the inclusion pancakes near the weld, the plate tears along the inclusion planes. The tears start at the stress raisers where inclusions cluster, then grow and link with short vertical steps — producing the characteristic staircase, or lamellar, crack. The tearing sits in the base metal just off the weld, in the zone where shrinkage strain is highest, and the weld metal itself stays sound.
Flattened MnS pancakes aligned with the rolling plane are the tear planes. Sulfur content sets the count; rolling practice sets the shape.
Shrinkage at a restrained T-joint pulls through the plate thickness — the one direction rolled plate is weakest.
Tears run along inclusion planes and link with short vertical steps — the stepwise crack that gives lamellar tearing its name.
The controlling variables are therefore material and design: inclusion content sets how much strain the plate can take in z, joint design sets how much z-strain reaches the plate, and weld size and sequence set how much strain is generated. Change the design or the steel and the problem disappears; change the welder's technique and it mostly stays.
Lamellar tearing is prevented on the drawing board and in the material spec — welding practice only buys margin.
Specify Z-quality steel for the joint — Z25 or Z35 guarantees 25% or 35% minimum reduction of area through the thickness. Low-sulfur steel (S ≤ 0.005%) is the modern fix. If the design loads the plate through its thickness, change the design.
Avoid T-joints that pull in z. Prefer full-penetration butt details where possible, or re-design so shrinkage is accommodated in the weld direction instead of through the plate thickness.
Butter the plate: deposit a ductile weld layer on the flange face before laying the T — the buttering takes the strain instead of the plate. Keep the weld as small as the design allows and balance the sequence so shrinkage is even.
Thinking a bigger weld makes it stronger. On a restrained T-joint in tearing-prone steel, a bigger weld generates more z-strain and more tearing. And don't ignore the material spec — if the drawing says Z35, it was specified because the joint pulls through the thickness.
Repairing lamellar tearing is a design decision, not a shop reflex. The default "grind it out and re-weld" fails more often than it works.
| Step | Action | Why |
|---|---|---|
| 1 | Confirm the diagnosis — UT from the accessible side; map the tear's extent and depth. | Tearing looks like other cracks; the wrong diagnosis gets the wrong repair. |
| 2 | Stop welding on the joint immediately. | Every additional pass adds shrinkage and grows the tear. |
| 3 | Attempt excavation only if the tear is shallow and localized — grind to sound metal, verify by UT/MT, re-weld with buttering and a low-restraint sequence. | Tears run deep along inclusion planes; excavation often misses them. |
| 4 | If the tear is deep or extensive, stop the excavation — redesign or replace the plate. | Repairs over inclusion planes re-tear; there is no "burn it out and forget it" fix. |
| 5 | Document and involve engineering — the repair must be designed, not improvised. | This is a design/material failure, not a welder's re-work. |
Lamellar tearing is the weld defect where "grind it out and re-weld" routinely fails — the new weld puts the same z-strain on the same inclusion planes and the tear re-opens, deeper. Major tears usually mean plate replacement. Confirm the repair scope against the governing code and engineering before any excavation.
Tearing has no process signature — it has a joint signature. These are the industries and processes where it shows up most.
The classic report location: fillet T-joints in thick flange plate, shipbuilding and structural. The process is incidental — the restrained T-joint and the plate's z-ductility are the story.
Heavy equipment frames — high-deposition welds deliver the shrinkage strain quickly. Same fix as everywhere: joint design, Z-quality steel, buttering.
Large box sections and crane beams — single-sided, high-restraint welds maximize z-strain on the flange. A classic location for tearing in thick plate.
Less common — thinner sections and lower restraint — but a restrained T-joint in dirty plate tears in any process. Offshore and marine fabrication see it across the whole arc range.
When tearing appears, the parameters to question are the ones on the drawing and the mill certificate — not just the weld settings.
| Parameter | Check | Typical Fix |
|---|---|---|
| Joint design | Load direction vs plate thickness — is the joint pulling in z? | Re-design to avoid z-strain; full-penetration butt details over T's |
| Steel Z-grade | Z25 / Z35 rating per EN 10164; % RA in the thickness direction | Specify Z-quality or low-sulfur steel for the joint |
| Sulfur content | Chemistry certificate, S % | Low S (≤ 0.005%) dramatically raises through-thickness ductility |
| Weld size / throat | As designed vs as welded — oversize welds over-strain the plate | Weld to size; every extra pass is extra z-strain |
| Pass sequence | Balanced vs one-side-first | Balance shrinkage across the joint; butter the flange before the T |
| Preheat | Reduces strain-rate sensitivity, never cures tearing | Hold per WPS, but never rely on preheat alone |
The Material Thickness guide covers the restraint side of the equation, and the Filler Metal Selector helps choose a ductile buttering filler where the design allows it.
The one consumable decision that matters is the buttering filler — everything else is conventional.
Buttering needs a ductile, crack-resistant filler — see the consumables library for selection across processes and base metals. Positioners help balance the sequence and keep the joint at workable angles. The real "equipment" for this defect is the material spec and the drawing — check those first.
Tearing hides parallel to the surface — it is easy to miss precisely because it is oriented the wrong way for most inspection habits.
Acceptance depends on the code and project: AWS D1.1, ISO 5817, API 1104 and project specifications each set their own limits for this discontinuity. Check the governing document before judging a weld acceptable — there is no universal pass/fail rule.
The questions welders actually ask about lamellar tearing — answered straight.
Almost never. It is a material and design failure: the steel's through-thickness ductility was insufficient for the joint's through-thickness strain. The welder can reduce the risk with welding sequence, buttering, and weld size, but cannot weld his way out of a bad joint design on dirty plate.
Z-quality steel — Z25 or Z35, which guarantees 25% or 35% minimum reduction of area in the through-thickness direction — and low-sulfur steel (sulfur at or below about 0.005%). Modern clean steel with few inclusions is far less prone, because the inclusions are the tear planes.
No. Preheat reduces the strain rate sensitivity a little and helps some, but it does not cure tearing — the inclusion planes and the z-direction strain are still there. That is why the engineering solution is Z-quality steel and joint design, not preheat.
Very difficult. The tears run deep along inclusion planes, excavation often misses them, and re-welding puts the same z-direction strain on the same planes. Shallow, localized tears can sometimes be excavated and re-welded with buttering; deep or extensive tearing usually means redesign or plate replacement.
Ultrasonic testing from the accessible side is the standard method — the indications run parallel to the plate surface, often in steps, just off the weld. Radiography can see it where access allows. Magnetic particle or penetrant only work if the tear breaks the surface, which is often late in the story.