Welding defects and inspection

Why it matters

Most fatigue and brittle failures of welded structures start at a weld defect — a toe undercut, a lack-of-fusion line, a hydrogen crack. Knowing what causes each defect lets you fix the procedure; knowing which inspection method finds it lets you prove a pressure vessel, pipeline or bridge girder is fit for service before it goes into use.

Key ideas

Main fusion-weld defects, causes and remedies.

• Porosity — gas pores (H₂, N₂, CO) trapped as the pool freezes. Causes: moisture, oil, rust or paint on the joint; damp electrodes or flux; lost gas shield (draught, wrong flow); arc too long. Remedies: clean and dry, bake electrodes, correct gas flow and arc length. Usually spherical, rounded indications on a radiograph.
• Slag inclusions — non-metallic slag trapped in the weld, typically between passes in SMAW, FCAW and SAW. Causes: poor inter-pass cleaning, too low current, wrong electrode angle, undercut grooves that hold slag. Remedy: clean each pass, adequate current and access.
• Lack (incomplete) fusion — weld metal not fused to base metal or to the previous bead. Causes: low heat input, high travel speed, poor torch angle, oxide films, cold lap in short-circuit GMAW. Planar and dangerous.
• Incomplete penetration — the root not fused through the thickness. Causes: root gap too small, root face too large, low current, high speed. Remedy: correct joint preparation, back-gouging and back-welding.
• Undercut — a groove melted into the base metal at the weld toe and left unfilled. Causes: excessive current or arc length, too high travel speed, wrong electrode angle. It is a sharp notch, very harmful in fatigue.
• Overlap (cold lap) — weld metal flowing over the base metal without fusing at the toe. Causes: low current, slow travel, large electrode.
• Cracks — the most serious defect.
  • Hot (solidification) cracks form along the weld centreline during freezing; promoted by S and P, high dilution and deep narrow beads (depth-to-width ratio above about 1).
  • Cold (hydrogen-induced) cracks form in the HAZ hours after welding; need hydrogen, a hard microstructure and tensile stress. Remedies: low-hydrogen consumables, preheat, controlled heat input, delayed inspection.
  • Lamellar tearing in rolled plate loaded through its thickness, caused by elongated inclusions.
  • Crater cracks at the end of a bead if the arc is broken abruptly.
• Burn-through, spatter, excessive reinforcement, misalignment, arc strikes — mostly workmanship faults.
• Distortion and residual stress — weld metal shrinks as it cools while restrained by cold base metal; tensile residual stress near the yield strength remains along the weld, balanced by compression farther away. Controlled by balanced welding sequence, presetting, fixtures, minimum weld size and post-weld stress relief.

Planar defects (cracks, lack of fusion, incomplete penetration) are far more dangerous than volumetric ones (porosity, rounded slag) because their sharp edges concentrate stress.

Inspection methods.

• Visual inspection (VT) — always first; finds undercut, overlap, surface cracks, profile and size errors, using gauges.
• Liquid (dye) penetrant testing (PT) — penetrant drawn into surface-breaking cracks by capillarity, developer draws it out. Any non-porous material; surface-breaking defects only.
• Magnetic particle testing (MT) — the part is magnetised; a crack disturbs the flux and attracts iron particles. Ferromagnetic materials only; surface and slightly sub-surface defects; most sensitive to cracks lying perpendicular to the magnetic field.
• Radiographic testing (RT) — X- or gamma rays pass through the weld onto film or a digital detector; less-dense defects show darker. Good for volumetric defects (porosity, slag); planar cracks show only when aligned with the beam. Gives a permanent record; needs radiation safety and access to both sides.
• Ultrasonic testing (UT) — a probe sends 1–10 MHz pulses; reflections from defects and the back wall are timed to give depth. Best for planar defects, one-side access, thick sections, instant result; needs a skilled operator and a couplant. Angle-beam probes inspect butt welds without removing the reinforcement.
• Eddy-current testing — surface cracks in conductive materials, tubes.
• Destructive tests — tensile, guided bend (face, root, side), Charpy impact, hardness traverse and macro-etch, used to qualify procedures and welders.

Formulas

d = v · t / 2 Straight-beam UT: d = depth of reflector (mm), v = sound velocity in the material (mm/µs; about 5.9 for longitudinal waves in steel — take from your data book), t = round-trip time (µs). Divide by 2 because the pulse goes there and back.

depth = s · cos θ and surface distance = s · sin θ Angle-beam UT: s = sound path length (mm), θ = refracted beam angle measured from the normal to the surface.

U_g = F · d / D Radiographic geometric unsharpness: F = effective source (focal spot) size (mm), d = object-to-film distance (mm), D = source-to-object distance (mm). Keep U_g small by a small source, film close to the weld and a long source distance.

H = η · V · I / v Heat input (J/mm) — too low gives lack of fusion and fast cooling; too high gives burn-through, distortion and coarse grains.

Worked examples

Example 1 — straight-beam UT (standard). A steel plate is tested with a longitudinal-wave probe (v = 5.9 mm/µs). A defect echo appears at 6.8 µs and the back-wall echo at 8.5 µs.

1. Defect depth d = v·t/2 = 5.9 × 6.8 / 2 = 20.1 mm.
2. Plate thickness = 5.9 × 8.5 / 2 = 25.1 mm.
3. The defect lies about 5 mm above the back wall — likely in the root region; check for lack of root fusion.

Example 2 — angle-beam UT and radiograph unsharpness (GATE level). (a) A 60° angle probe gives an echo at a sound path of 50 mm. (b) A radiograph uses a 3 mm focal spot at 600 mm from the weld, with the film 25 mm from the side of the weld facing the source.

1. Depth = s·cos θ = 50 × cos 60° = 50 × 0.5 = 25.0 mm.
2. Surface distance from the probe index = s·sin θ = 50 × 0.8660 = 43.3 mm.
3. U_g = F·d/D = 3 × 25 / 600 = 0.125 mm. Doubling the source distance to 1200 mm halves it to 0.0625 mm.

Common mistakes

• Forgetting to halve the ultrasonic round-trip time.
• Measuring the angle-beam angle from the surface instead of from the normal (swapping sin and cos).
• Using MT on austenitic stainless steel or aluminium — they are not ferromagnetic.
• Expecting PT to find sub-surface defects.
• Assuming RT reliably finds tight cracks at any orientation; it misses cracks not aligned with the beam, where UT is better.
• Confusing undercut (groove in base metal) with overlap (unfused excess metal) or incomplete penetration (unfused root).
• Inspecting high-strength steel welds immediately; hydrogen cracks can appear up to 48 hours later.

For GATE PI

Expect MCQs matching defects to causes and remedies, matching NDT methods to the defects and materials they suit, planar versus volumetric defects, and hot versus cold cracking. Numericals cover UT depth from time of flight, angle-beam geometry, radiographic unsharpness and heat-input reasoning about lack of fusion or burn-through.

Quick check

1. Which NDT method suits surface cracks in an aluminium weld?
2. A UT echo returns after 10.2 µs in steel (5.9 mm/µs). Depth?
3. Name two causes of porosity.
4. Why is undercut dangerous in fatigue?
5. Which crack type forms along the weld centreline during solidification?

Answers: 1. Dye penetrant testing (or eddy current). 2. 5.9 × 10.2/2 = 30.1 mm. 3. Moisture or contamination on the joint, damp electrodes or flux, lost gas shield, long arc (any two). 4. It is a sharp notch at the weld toe that concentrates stress where fatigue cracks start. 5. Hot (solidification) cracking.