Lesson 9.7 — Visual Weld Inspection: Spotting Porosity, Undercut, and Lack of Penetration

§01 — Learning Objectives

Learning Objectives

By the end of this lesson, the technician will be able to:

Objective 1 (Cognitive/Understanding): Identify the principal visual weld discontinuities — porosity, undercut, overlap/cold lap, lack of fusion/penetration (where visible), cracks, craters, slag inclusions, spatter, and profile defects — and explain the mechanism and structural significance of each.
Objective 2 (Cognitive/Understanding): Distinguish discontinuity from defect — a finding becomes a rejection only against the governing code's acceptance criteria — and state where the technician's visual screening ends and the certified inspector's (CWI) authority begins.
Objective 3 (Field/Practical): Execute a disciplined visual inspection — lighting, cleaning, fillet gauges, magnification, systematic coverage — and produce a findings package (measurements, photos, locations) that feeds the formal inspection chain and 9.8's NDT.

§02 — Field Scenario

Field Scenario

The Crack That Hid Beside the Weld: A vibrating screen's frame weld is reported "cracked" by an operator. The technician arrives expecting to find a split bead. The bead itself is fine — uniform, well-profiled, no visible flaw down its centerline. Case closed as an operator false alarm?

Not with the flashlight held right. Sweeping the light at a low angle across the joint (raking light — the inspector's oldest trick, because surface-breaking flaws cast shadows that flat lighting erases), the technician picks up a hairline shadow running not in the weld but alongside it — in the narrow band of base metal at the weld toe. A wipe with solvent, a 10× loupe, and the line resolves into a true crack: a toe crack in the heat-affected zone (9.2/9.5's HAZ — where the metallurgy is roughest and the geometry concentrates stress), about three inches long, fine enough to vanish under paint and flat light.

Two lessons travel together here. First, technique: cracks at weld toes are the classic fatigue location on vibrating equipment — inspection looks at the toes and the HAZ band, not just the bead, with raking light on cleaned metal. Second, governance: a crack is the one discontinuity with no acceptance threshold — cracks of any size are rejectable in every code — and on a structural frame this finding doesn't get ground out and rewelded by whoever found it. It gets documented (photos with scale, location by reference, length marked), the equipment dispositioned with operations (a cracked structural member on vibrating equipment is a now-problem), and routed into the 9.10 chain — engineering evaluation, qualified repair under procedure, and likely NDT (9.8) to confirm the crack's true extent before and after repair, since the visible hairline is frequently the short version of the subsurface story.

Visual inspection is the cheapest, fastest, first-line NDT method — and its product is honest findings, properly routed, not field verdicts.

§03 — Concept Overview

Concept Overview

Discontinuity vs. Defect — the Word That Carries the Authority

A discontinuity is any interruption in the weld's ideal structure — pore, undercut, profile irregularity. A defect is a discontinuity that exceeds the acceptance criteria of the governing code (AWS D1.1 for structural, ASME for pressure work) — sized limits, density limits, location rules. The distinction is the lane marker: technicians find and measure discontinuities; acceptance/rejection of code work belongs to the certified inspection authority (CWI or equivalent) working to the applicable criteria. The exception that proves the rule: cracks are defects everywhere, at any size — no code accepts them.

The Visual Catalog — Mechanism and Meaning

Porosity: gas bubbles frozen into solidifying metal — surface pinholes, clusters, or wormhole tracks. Mechanism: contamination (moisture, oil, paint, rust on the joint), lost shielding (wind on GMAW — 9.5's warning; damp low-hydrogen rods), or arc trouble. Significance: scattered light porosity is often acceptable within limits; clustered/aligned porosity, and any porosity on pressure boundaries, is serious — and surface porosity hints at more underneath (9.8's territory). Root cause reads: the joint or the shielding was dirty/compromised — fix the process, not just the weld.
Undercut: a groove melted into the base metal at the weld toe and left unfilled — the parent metal is now thinner beside the weld, with a built-in stress riser exactly where the HAZ lives. Mechanism: excessive current/travel speed, electrode angle. Measured (depth gauges/profile gauges) against code limits (commonly on the order of 1/32" depending on member and code — the criteria document governs). On cyclically loaded (vibrating, rotating-adjacent) members, undercut is a fatigue-crack starter — the Field Scenario's toe crack often begins in one.
Overlap / cold lap: weld metal that rolled over onto the base metal without fusing to it — looks like generous coverage, functions as a sharp unfused crevice (a crack-shaped notch by geometry). Visual tell: a rounded lip at the toe; a pick or the profile gauge catches the unfused edge. Mechanism: too-cold parameters, too-slow travel.
Lack of fusion / lack of penetration (LOF/LOP): the weld didn't fuse to the base metal (sidewall fusion) or didn't reach the joint root (penetration). Mostly invisible from outside — the deceptive GMAW failure of 9.5, and the reason CJP welds (9.6) carry NDT requirements. Visual inspection catches only its edges: visible root gaps on open-root joints, suspicious bead profiles, the unfused line at an accessible root or back side. The honest visual verdict is often "cannot verify — interior method required" (9.8).
Cracks: the zero-tolerance category. Types by location/timing: toe cracks and underbead/HAZ cracks (often hydrogen-related — damp rods, no preheat on hardenable material: 9.2/9.5's physics), crater cracks (star-shaped, in an unfilled crater at a weld's stop), centerline cracks (hot cracking down the bead's middle), transverse cracks. All rejectable; all documented and routed; visible length presumed to understate true extent until NDT says otherwise.
Crater / unfilled crater: the depression at an abruptly ended weld — thin, often crack-hosting (crater cracks). Codes require craters filled.
Slag inclusions: flux slag trapped in or between passes (SMAW/FCAW — 9.5) — visible when surface-breaking as glassy/dark inclusions; interpass cleaning failures are the mechanism.
Spatter, arc strikes: spatter is mostly cosmetic (but masks the surface — clean before inspecting); arc strikes (stray arc touches outside the joint) are not cosmetic — each is a tiny uncontrolled heat treatment (a hard, sometimes cracked spot on hardenable material — 9.2) and codes require their removal/inspection. An arc strike on a Q&T or code component is a finding, not a smudge.
Profile defects: excessive convexity/reinforcement, insufficient throat/leg (the fillet gauge's verdict vs. 9.6's called size), excessive concavity (undersized throat in disguise), and warpage/distortion of the assembly (a fit-up and heat-input story — 9.9).

§04 — How the Principle Works

How the Principle Works: The Inspection Method

Know the requirement first (9.6's decode): the symbol's type/size/length/contour, the general notes, and the governing code named by the job — inspection is comparison against requirement, not aesthetic judgment.
Clean and light: slag chipped, spatter and soot removed (wire brush — stainless-dedicated on stainless, 9.1), surface dry; raking light swept at low angles from multiple directions; 5–10× magnification for suspect areas. Most misses are lighting-and-cleanliness failures, not eyesight failures.
Cover systematically: the full weld length (both sides where accessible), the toes and HAZ band (the Field Scenario's territory), starts/stops and craters (defect-dense locations), and the root side where accessible.
Measure, don't adjective: fillet weld gauges for leg size and throat against the called size (a set of standard gauges, used square to the weld); undercut gauges for depth; tape for length/pitch of intermittent patterns (9.6); straightedge for member distortion. "Looks small" becomes "measures 3/16 against a 1/4 callout."
Document like evidence (10.8 discipline): photos with scale references and locations (member, joint, distance from a datum — 10.4 grids/10.6 welds where applicable), measurements recorded, the requirement cited alongside each finding.
Route by classification: non-code shop work → rework per the supervisor/fabricator with the findings package; code/structural/pressure work → the certified inspection authority and the 9.10 chain — the technician's screening feeds formal inspection, never replaces it; cracks and structural findings additionally get an immediate operational disposition (can this equipment run?) with operations/engineering, because paperwork has no schedule authority over a propagating crack.
Know visual's limits out loud: interior soundness (LOF/LOP, subsurface porosity/inclusions, crack depth) is invisible — the honest report says "visual acceptable / interior not verified," and CJP or cyclic-service findings escalate to 9.8's methods by requirement or by judgment.

§05 — Reference Matrix

Visual Discontinuity Reference Matrix

Discontinuity	Visual Signature	Mechanism	Structural Meaning	Disposition Note
Porosity	Pinholes, clusters, wormholes	Contamination, lost shielding, damp rods	Section loss; pressure-tightness risk	Limits per code; surface hints at subsurface
Undercut	Groove at toe, unfilled	Hot/fast parameters	Stress riser + thinned parent at HAZ	Measured vs. code limit; fatigue-critical members strict
Overlap/cold lap	Rolled, rounded toe lip; unfused edge	Cold/slow welding	Crack-shaped unfused notch	Rejectable; pick/profile gauge confirms
LOF/LOP	Mostly invisible; root gaps where accessible	Poor technique/parameters/prep	Joint not actually joined	"Cannot verify visually" → NDT (9.8)
Cracks (all types)	Hairlines — toes, craters, centerline; raking light	Hydrogen, restraint, HAZ hardening, hot cracking	Zero tolerance, every code	Document, secure equipment, route 9.10/9.8 — never solo-repair
Crater (unfilled)	End-of-weld depression, often star-cracked	Abrupt arc stop	Thin section + crack host	Fill required; check for crater cracks
Slag inclusion	Glassy/dark surface inclusions	Interpass cleaning failure	Section loss; more likely between passes	Surface = suspicion of interior
Arc strike	Stray arc spot off-joint	Careless electrode/work lead	Hard spot/micro-crack (9.2)	Finding on code/treated material — removal per code
Spatter	Adhered droplets	Parameters	Cosmetic, masks surface	Clean before inspecting
Profile (size/convexity/concavity)	Gauge disagreement with 9.6 callout	Technique	Under-strength or stress-raising shape	Measured vs. symbol + code

§06 — Field Application Checklist

Field Application Checklist: Visual Weld Inspection Pass

Boundaries: inspection of in-service equipment under the machine's safe condition (isolation where guards open or movement is possible — Module 1); heights/access per 1.8; no grinding "to see better" on code work without inspection-authority direction (grinding alters evidence and dimensions).
Assemble the requirement set (10.1/9.6): drawing + revision, symbols decoded, notes, governing code identified, prior inspection records from the CMMS (10.8) for trend context.
Prepare the surface and the light: clean per Section 4 (material-correct brushes), raking light + magnification staged; mark the weld into zones for systematic coverage on long joints.
Inspect in the high-yield order: toes/HAZ both sides → starts/stops/craters → bead body → root side where accessible → adjacent base metal for arc strikes and distortion.
Gauge everything the symbol dimensions: fillet legs/throat, undercut depth, intermittent length/pitch, contour vs. the 9.6 callout.
Photograph findings with scale and location; mark crack extents on the metal per site practice (paint pen beyond visible ends — NDT will test past the marks).
Classify and route (Section 4, step 6): code/structural/pressure → CWI/9.10 chain with operational disposition for cracks; non-code → documented rework path. The findings package travels complete: requirement, measurement, photo, location.
Record the inspection (10.8) — including clean results: "inspected per D1.1 visual criteria, no recordable findings" is baseline data the next inspection trends against, and the entry that proves the screening happened.

§07 — Knowledge Check

Knowledge Check

During a PM, a technician visually inspects the welds on a hopper's support frame (structural, AWS D1.1 work per the drawing; 1/4" fillets called out). Findings: (a) one weld gauges 3/16" leg over its middle 4 inches but 1/4" elsewhere; (b) scattered fine surface porosity, roughly 1 pore per inch, on another weld; (c) a 2-inch hairline at a weld toe that catches the pick and shows under raking light, on a member adjacent to the vibratory feeder mount; (d) heavy spatter throughout. The technician's proposed disposition: grind and reweld the undersized section himself (he's handy with a MIG), log the porosity as acceptable since it's light, wire-brush the hairline to "see if it's just a scratch" and monitor it next PM, and note the spatter. What is the correct correction to this plan?

A) The plan inverts the priority order and crosses two authority lines. The hairline (c) is the emergency, not the monitoring item: a toe crack on a structural member beside a vibration source is the zero-tolerance, fatigue-propagating finding — it gets documented (photo, scale, marked extents), an immediate operational disposition with supervision/engineering (can the hopper run?), and routing into the 9.10 chain with likely NDT (9.8) to establish true extent; 'brush it and watch it' converts a known crack into a scheduled failure, and aggressive brushing can smear metal over the evidence. The undersized weld (a) is a real, measured nonconformance — but on D1.1 structural work its repair is not a handy-with-MIG job: rework on code welds runs under qualified welder + WPS + reinspection (9.5/9.10), so the finding routes with the package rather than being freelanced. The porosity (b) and spatter (d) are the judgment items — but judged against D1.1's acceptance criteria by the inspection authority for code work, with the technician's contribution being honest measurement (density, size, location), not the verdict; logging 'acceptable' is lane-crossing even when the guess is probably right. Correct order: secure/disposition the crack today, package all findings to the CWI/engineering chain, and let rework proceed under procedure. B) All four findings should simply be ground smooth and rewelded in one session to restore the frame to drawing condition efficiently. C) The plan is sound — it prioritizes the measurable defect (size) for immediate repair and applies judgment to the minor findings. D) The only error is the porosity call — any visible porosity on structural work is rejectable, so that weld also needs rework.

§08 — Source List

Source List

AWS D1.1 — Structural Welding Code — Steel, visual inspection acceptance criteria (Clause on workmanship/inspection)
AWS B1.11 — Guide for the Visual Examination of Welds
CWI program references (AWS QC1) — certified inspector authority and scope context
ASME Section V / VIII references — pressure-work examination context (with 9.4/9.8/9.10)
Lessons 9.2, 9.5, 9.6, 9.8, 9.10, 10.8 cross-references — HAZ metallurgy, process defects, symbol requirements, NDT escalation, repair limits, and findings documentation