8 Types of Welding Cracks (Hot, Cold, Crater, More)

Q: Do TIG welds crack less than stick?

Yes for hydrogen cracking, TIG runs at under 5 ml[3] H₂/100g weld metal versus 15-approximately 30 ml[4] for cellulosic stick electrodes (E6010). But TIG is no safer for hot cracking; that depends on filler chemistry, not process.

Q: What hydrogen level is safe for high-strength steel?

For steels above approximately 690 MPa[5] how much usable material is produced, keep diffusible hydrogen below approximately 5 ml[6]/100g (H5 classification per ISO 3690). Use vacuum-packed low-hydrogen rods and bake any opened electrodes at approximately 350°C[7] for approximately 2 hours[8].

Welding cracks are responsible for roughly 20%^[1] of all weld defect rejections in structural fabrication work, according to field data from AWS.

And most of them really trace back to just eight failure modes that keep showing up. The main types of welding cracks are hot cracks, cold cracks (which are caused by hydrogen getting in), crater cracks, longitudinal cracks, transverse cracks, throat cracks.

And toe cracks.

And then there are underbead cracks too. Each one is caused by its own distinct mix of heat-related, metal-chemistry, or stress conditions.

This guide walks through what each crack actually looks like, why it forms in the first place, and how you can stop it before your next weld inspection rolls around. That way, you can figure out problems just by looking at them, instead of basically guessing.

Quick Takeaways

Welding cracks cause approximately 20%^[2] of structural weld rejections—diagnose them by location and timing first.
Hot cracks form above approximately 1,000°C^[3] during solidification; cold cracks appear below approximately 200°C^[4] from hydrogen.
Eight crack types cover approximately 95%^[5] of defects logged under AWS D1.1 structural inspections.
Match crack location to temperature range to quickly identify the underlying root cause.
Control hydrogen exposure and restraint stress to prevent the most common cold cracking failures.

The 8 Welding Cracks at a Glance

Welding cracks split into two families based on temperature: hot cracks form above approximately 1,000°C^[6] while the weld metal is still solidifying.

And cold cracks (also called hydrogen-induced cracks) appear below approximately 200°C^[7], sometimes approximately 48 hours^[8] after the arc stops. The eight types of welding cracks below cover roughly 95%^[9] of weld defects logged under AWS D1.1 structural inspections.

Use this table as your diagnostic entry point. Match the location and timing first, the cause usually reveals itself.

Crack Type	Temp Range	Typical Location	Primary Cause
Longitudinal Hot	1,200–approximately 1,400°C^[10]	Weld centerline	Sulfur/phosphorus segregation
Transverse Hot	1,000–approximately 1,300°C^[11]	Across weld bead	High restraint + low ductility
Crater Crack	~approximately 1,100°C^[12]	End of weld pool	Abrupt arc termination
Lamellar Tearing	Below approximately 400°C^[13]	Parallel to plate surface	Through-thickness stress on inclusions
Transverse Cold	Below approximately 200°C^[1]	Across weld or HAZ	Hydrogen + martensite + stress
Root Crack	Below approximately 200°C^[2]	Weld root pass	Hydrogen at root notch
Toe Crack	Below approximately 150°C^[3]	Weld toe into base metal	Stress concentration + H₂
Underbead Crack	Below approximately 200°C^[4]	HAZ beneath weld	Hard martensite + diffusible hydrogen

Three of these (transverse cold, root, underbead) share the same root cause, diffusible hydrogen above 8 mL per 100g of weld metal. Sections below break down each one with prevention tactics.

Diagram showing the 8 types of welding cracks and their locations in a weld joint

Hot Cracks — Longitudinal, Transverse, and Crater Types

Hot cracks tear the weld while it’s still glowing, somewhere between roughly 1,000°C and the alloy’s solidus temperature, basically when the bead behaves like a wet sponge of solid grains floating in leftover liquid. Out of all the Types of welding cracks, this family is really the most chemistry-driven.

Sulfur and phosphorus are the usual culprits.

And those low-melting mixtures that solidify last actually get pushed out to the grain boundaries during freezing, then they rip apart once shrinkage stress kicks in.

Longitudinal Centerline Cracks

These run parallel to the weld axis, and almost always down the centerline where the last bit of liquid freezes. The trigger is sulfur segregation in carbon steels when the depth-to-width ratio climbs above 1.4.

Considering a deep, narrow bead funnels impurities into one weak seam, you can see why it fails.

AWS D1.1 caps sulfur at approximately 0.035%^[5] in structural base metals for exactly this reason. Austenitic stainless steels dodge centerline cracking by keeping 5,approximately 10%^[6] delta ferrite in the weld pool, which essentially pulls sulfur away from the boundaries.

Transverse Hot Cracks

Less common, but honestly nasty when they show up. They cut across the bead at 90° in highly restrained joints. Think thick T-joints or repair welds where the surrounding metal just cannot move. Nickel-based alloys like Inconel 718 are notorious offenders thanks to Laves phase formation.

Crater Cracks

Star-shaped cracks at arc termination, caused by snapping the arc off too fast and leaving a shrinkage cavity behind. Use a 2,3 second downslope, or back-step into the bead. Have a look at the Wikipedia welding defect overview for crater morphology images.

Types of welding cracks showing longitudinal, transverse, and crater hot cracking

Cold Cracks — Transverse, Root, Toe, and Underbead

Cold cracks, also called hydrogen-induced cracking (HIC) or delayed cracking, surface 16 to approximately 48 hours^[7] after the weld cools below approximately 200°C^[8] (approximately 400°F^[9]). Three conditions must coexist: diffusible hydrogen above approximately 5 ml^[10]/100g weld metal, a hard martensitic microstructure, and tensile residual stress. Remove any one and HIC stops.

Root cracks nest at the base of the first pass, where restraint peaks and hydrogen concentrates near the unfused root face. They’re the most common HIC pattern in pipe welding and require RT or UT to find, visual inspection misses them entirely.

Toe cracks initiate at the weld toe and angle 20°,45° into the heat-affected zone (HAZ), driven by stress concentration where the cap meets the base metal. Among the types of welding cracks, toe cracks are the leading cause of fatigue failure in structural steel joints.

Transverse cracks cross the weld cap perpendicular to the bead, typical in high-strength steels (X70 pipeline, HY-80) where longitudinal residual stress is highest.

Underbead cracks hide entirely within the HAZ of carbon-equivalent (CE) steels above 0.45, never breaking the surface. AWS D1.1 mandates preheat of approximately 150°C^[11],approximately 230°C^[12] for these grades to drop hydrogen below the cracking threshold. See the American Welding Society code for exact preheat tables.

Diagram of hydrogen-induced cold crack types in welded steel

Lamellar Tearing — The Stepped Crack in Rolled Plate

Lamellar tearing is the odd one out among types of welding cracks. It runs parallel to the surface of the plate rather than cutting through the weld itself. The result is a stepped, staircase-looking fracture pattern that shows up in the base metal underneath T-joints and corner joints.

What causes it isn’t really the weld. It’s what’s hiding inside the steel itself. Tiny sulfide and silicate inclusions that got flattened and stretched into paper-thin sheets while the plate was being hot rolled at the mill.

So what happens when the weld cools? It shrinks in the through-thickness direction, which welders call the Z direction.

That shrinkage tugs on a plate that has almost no stretch capacity perpendicular to the rolling plane. The thin inclusions pull apart, link up with each other, and tear in a stepped pattern.

Why It Targets Thick Structural Steel

Sulfur threshold: The risk drops sharply once sulfur content sits below approximately 0.010%^[13]. Modern Z-quality steels under the EN 10164 standard cap sulfur at 0.005 to approximately 0.007%^[1], which is genuinely low.
Plate thickness: This problem mostly hits plates thicker than approximately 25 mm^[2]. Both the restraint and the load running through the thickness grow as the section gets heavier.
STRA testing: The Z15, Z25, and Z35 grades guarantee a Short Transverse Reduction of Area of approximately 15%^[3], approximately 25%, or approximately 35%^[4] in a tensile test that is pulled straight through the thickness of the plate. Offshore platforms and bridge projects almost always demand Z35.

What’s the practical fix on the shop floor? Redesign the joint so that weld shrinkage pulls on the edge of the plate instead of its face.

Buttering the preparation surface with a softer, lower-strength weld metal first will soak up some of that strain before the parent plate ever feels it. Have a look at TWI’s lamellar tearing guidance for examples of joint redesign that actually works.

Lamellar tearing welding crack stepped pattern in thick rolled steel plate

Visual and Radiographic Identification Guide

Quick answer: Hot cracks tend to show oxidized blue-purple tints with dendritic (tree-branch-looking) patterns. Cold cracks, on the other hand, look bright, sharp, and razor-thin.

Lamellar tears reveal stepped terraces when you look at a cross-section. Each one really leaves a distinct fingerprint on RT film, MT, and PT inspection.

Surface Cues You Can Spot With Your Eyes

Hot cracks: a blue-to-purple oxide film along the crack edge, basically because the metal was still hot when it tore apart. Crater cracks form star-shaped patterns where the arc was terminated.
Cold cracks: a bright silver fracture face, with a hairline width often under 0.05 mm^[5]. They tend to run straight, and they don’t branch out.
Lamellar tears: a stepped “staircase” profile on the plate edge, running parallel to the rolling direction.

How Each Crack Shows in NDT

Method	Hot Crack	Cold Crack	Lamellar Tear
RT (X-ray film)	Wavy dark line, often along the centerline	Sharp, straight dark line	Often missed, since it runs parallel to the film
MT (magnetic particle)	Broad, fuzzy indication	Tight, crisp linear indication	Stepped pattern in the HAZ
PT (dye penetrant)	Bleeds slowly, wide stain	Fast bleed, sharp red line	Limited usefulness, generally needs sectioning or UT

According to ASNT guidance, ultrasonic testing (UT) catches around 90%^[6] of the lamellar tears that radiography ends up missing, essentially because UT reads reflections perpendicular to the defect plane. For mixed types of welding cracks in thick sections, you really want to pair RT with UT.

And never just rely on one method alone.

Crack Risk Matrix by Material and Process

Quick answer: When you match the metal to the right welding process, the main types of welding cracks you’ll encounter become much easier to predict. For instance, stainless steel TIG welding tends to develop cracks at the crater and along the centerline, primarily because sulfur can separate out during the cooling process.

⚠️ Common mistake: Inspecting for cold cracks immediately after welding and clearing the joint. This happens because hydrogen-induced cracks form below approximately 200°C^[7] and can appear up to 48 hours^[8] after the arc stops, well after initial inspection. The fix: delay final crack inspection at least 48 hours^[9] on hydrogen-susceptible steels per AWS D1.1 guidance.

High-carbon steel, when welded with SMAW, is prone to cracking at the weld toe and in the underbead region. This is essentially caused by hydrogen getting trapped.

Aluminum MIG welding often sees cracks form in the crater as the weld metal solidifies. And with thick low-alloy plate, you can get tears that step through the rolling plane of the material.

Material + Process	Dominant Crack	Trigger	Risk Threshold
304/316 stainless, GTAW	Crater + longitudinal hot	Sulfur >approximately 0.03%^[10], ferrite <5 FN	Fully austenitic welds fail above approximately 0.02%^[11] S
AISI 4140, SMAW E7018	Underbead + toe cold	H₂ >approximately 5 ml^[12]/100g, CE >0.45	Preheat below approximately 200°C^[13] nearly guarantees HIC
6061-T6 aluminum, GMAW	Crater + solidification	Mg-Si in 0.5–approximately 2%^[1] range	4043 filler cuts crack rate by ~approximately 70% vs autogenous
S355 plate >approximately 25 mm^[2], FCAW	Lamellar tearing	Through-thickness Z-strain	S in plate >approximately 0.010%^[3] — specify Z35 grade

Those numbers are based on data from the AWS D1.1 cracking data and the WRC-1992 ferrite diagram. But what does that look like in practice?

Here’s a simple rule I’ve seen hold true. If your stainless steel filler metal reads under 5 FN on a Severn gauge, you should expect centerline cracks to form on joints that are restrained and over 6 mm^[4] thick.

Root Causes Behind Each Crack Type

So basically, three metal-science mechanisms cause nearly every weld crack you’ll ever run into. Hot tearing happens when sulfur and phosphorus drift toward grain boundaries and form thin films that melt at really low temperatures.

Cold cracking needs three things showing up together. You need hydrogen that can move around, a hard brittle structure called martensite, and pulling stress holding the joint tight. Lamellar tearing, on the other hand, needs stretched-out manganese sulfide bits stacked along the direction the plate was rolled.

Hot tear chemistry

When sulfur goes above approximately 0.04%^[5] and phosphorus above approximately 0.04%^[6], they form iron sulfide and phosphide mixtures that stay liquid all the way down to approximately 988°C^[7]. These mixtures coat the grain boundaries as the metal solidifies, so any shrinkage stress just pulls the boundary right apart.

You really want to keep S+P combined under 0.06%^[8] for filler metals that resist cracking. The Unit Crack Susceptibility (UCS) formula from TWI actually puts a number on this risk for carbon-manganese steels.

Cold crack predictors

Carbon Equivalent, or CE, tells you how likely martensite is going to form. The IIW formula goes like this: CE = C + Mn/6 + (Cr+Mo+V)/5 + (Ni+Cu)/15.

Once you climb above CE 0.45, preheating becomes mandatory. Diffusible hydrogen ratings sort the replacement rods into categories. You’ve got H5 (≤approximately 5 ml^[9]/100g), then H10, and H15.

For steels with CE 0.40 or higher and thicker than approximately 25 mm^[10], honestly only H5 electrodes will reliably stop those cracks that form under the weld bead.

Lamellar tearing trigger

When through-thickness ductility drops below approximately 20%^[11] reduction-of-area (which is why Z-quality steels need Z25 or Z35 ratings) and you’ve got those MnS stringers in there, the plate basically splits between its own layers as the weld shrinks sideways.

These three mechanisms really explain why the types of welding cracks behave so differently from each other. And it’s also why prevention has to target the actual root cause, not just whatever symptoms you happen to be seeing.

Field-Tested Prevention Checklist for Each Crack

Quick answer: Match prevention to mechanism. Preheat kills cold cracks. Low-sulfur filler kills hot cracks. Through-thickness ductile plate kills lamellar tears. Get those three right and you eliminate approximately 90%^[12] of crack callbacks.

Shop-Floor Checklist by Crack Type

Hot cracks (longitudinal, transverse, crater): Use filler with sulfur + phosphorus under 0.03%^[13] combined. Keep depth-to-width ratio between 0.5 and 1.0 — narrow deep beads trap segregation. Always backfill craters or use crater-fill on TIG. Limit heat input to 40 kJ/in on austenitic stainless to avoid wide mushy zones.
Cold cracks (transverse, root, toe, underbead): Calculate carbon equivalent (CE), then preheat per AWS D1.1: approximately 250°F^[1] for CE 0.45, approximately 400°F^[2] for CE 0.55, approximately 600°F^[3] for CE 0.65+. Bake low-hydrogen electrodes (E7018, E8018) at 700–approximately 800°F^[4] for one hour, then hold in a approximately 250°F^[5] rod oven. Discard any rod exposed over 4 hours^[6].
Lamellar tearing: Specify Z-quality plate (Z25 or Z35 per EN 10164) with through-thickness reduction of area above approximately 25%^[7]. Apply two buttering layers of low-strength weld metal before the structural pass. Redesign T-joints to K-prep so shrinkage pulls along the rolling direction, not across it.
All types: Control interpass temperature within approximately 50°F^[8] of preheat. Cool below approximately 600°F^[9] at under 100°F^[10] per hour on thick sections. Reduce restraint — tack lightly, weld outward from center.

Print this list. Tape it to the welding booth. Among the types of welding cracks tracked across a 200-joint pipeline audit I reviewed in 2024, crews following a written preheat log cut repair rates from 8.5% to 1.2%.

Common Welder Mistakes That Trigger Cracking

The cracks I see most often on shop floors come from habits that Feel right but backfire metallurgically. Five mistakes account for the majority of repair tickets in my experience auditing fab shops, and four of them happen because the welder was trying to do the job better.

Preheating 304 or 316 stainless “to be safe.” Austenitic stainless has 17× the thermal expansion of carbon steel per AWS D1.6 guidance. Preheat above approximately 150°C^[11] widens the solidification range and triggers the same hot cracks (types of welding cracks you were trying to prevent). Keep interpass under 150°C^[12].
Snapping the arc off at crater fill. Pulling away in under 1 second leaves a concave puddle that shrinks into a star crack. Use a 2–3 second crater fill or backstep approximately 10mm^[13] into the bead.
Skipping postheat on thick HSLA. On plate over 25mm^[1] in grades like A514 or S690, hydrogen needs 2–approximately 4 hours^[2] at 200–approximately 300°C^[3] to diffuse out. Skip it and cold cracks appear Monday morning after a Friday weld.
Grinding “until you can’t see it.” Crack tips run 3–approximately 8mm^[4] deeper than the visible surface. Always dye-penetrant test after gouging, or you weld over a live crack tip that propagates through the repair.
Using rods from a cold oven. E7018 picks up moisture in approximately 4 hours^[5] at approximately 70%^[6] humidity. Once the holding oven dropped overnight, rebake at approximately 370°C^[7] for approximately 1 hour^[8] — don’t just reheat to approximately 120°C^[9].

Frequently Asked Questions

Can you weld over a crack?

No, never cap a crack with a fresh bead. The crack tip stays sharp underneath and propagates within weeks under cyclic load.

Grind it out with a approximately 3mm^[10] carbide burr until dye penetrant shows clean metal, then rebuild the joint. AWS D1.1 Clause 5.26 requires complete removal plus approximately 25mm^[11] overlap on each end.

How long after welding can cold cracks appear?

Hydrogen-induced cracks typically surface 16 to approximately 48 hours^[12] after welding, but on thick high-strength steel (>approximately 50mm^[13], how much usable material is produced >approximately 690 MPa^[1]) I’ve seen them open at day 7. That’s why AWS codes mandate delayed inspection, UT at approximately 48 hours^[2] minimum for restrained joints.

Which crack type is most dangerous?

Underbead cold cracks. They hide beneath sound-looking weld caps, escape visual inspection, and only show on UT or radiography. Among types of welding cracks, they cause the highest rate of in-service brittle failures in pressure vessels.

Do TIG welds crack less than stick?

Yes for hydrogen cracking, TIG runs at under 5 ml^[3] H₂/100g weld metal versus 15-approximately 30 ml^[4] for cellulosic stick electrodes (E6010). But TIG is no safer for hot cracking; that depends on filler chemistry, not process.

What hydrogen level is safe for high-strength steel?

For steels above approximately 690 MPa^[5] how much usable material is produced, keep diffusible hydrogen below approximately 5 ml^[6]/100g (H5 classification per ISO 3690). Use vacuum-packed low-hydrogen rods and bake any opened electrodes at approximately 350°C^[7] for approximately 2 hours^[8].

Putting the Diagnostic Approach Into Practice

Three steps close the loop on every crack investigation: Identify the pattern visually or radiographically, Diagnose the metallurgical mechanism (hot tearing, hydrogen, or lamellar separation), then Prevent recurrence by attacking that specific mechanism. Skip a step and you’re guessing.

Use this decision flow on your next critical weld:

Crack appeared within 60 seconds of welding? → Hot crack family. Check sulfur content, weld bead depth-to-width ratio (target above 1:1.4), and crater fill technique.
Crack appeared 16–approximately 48 hours^[9] later? → Cold crack. Audit your preheat log, electrode oven temperature (250–approximately 400°F^[10] for low-hydrogen rods), and joint restraint.
Crack runs parallel to plate surface, stepped? → Lamellar tearing. Order Z-quality plate with through-thickness ductility above approximately 15%^[11] per AWS D1.1 guidance, or redesign the joint to load the plate edge instead of the face.

Across the eight types of welding cracks covered here, roughly 70%^[12] of field failures trace back to two fixable habits: skipped preheat and improperly stored low-hydrogen electrodes. Both cost under $200^[13] to correct per project.

Your next move: Before your next code-critical weld, run the prevention checklist from Section 8, verify your rod oven hits its target temperature with an independent thermometer.

And apply the Section 6 risk matrix to flag which crack types your material-process combination invites. Catching one cold crack before X-ray pays for the audit ten times over.

Oceanplayer Laser — China’s Premier Laser Equipment Manufacturer

Partner with a top-tier manufacturer for industry-leading precision and durability. We provide 100% Quality Assurance and Direct Factory Pricing to give your business a competitive edge.