14.3 Evaluating Weldments

Official ASNT UT general topic 3 is Evaluation of Weldments. Topic 2 asked whether you could read a product form with a 0° probe. Topic 3 asks whether you can put a shear-wave beam through a weld volume, know which leg you are on, tell root geometry from a root crack, and name a reflector as planar or volumetric before you reach for an acceptance table. The next chapter will finish bonded structures, detection tricks, and general evaluation. This section stays inside the weld: 45° / 60° / 70° coverage, scan patterns, ID/OD geometry, root versus cap versus HAZ, lack of fusion versus crack versus slag versus porosity, skip positioning, both sides and both surfaces, and the honest line between technique articles and construction-code acceptance.

AWS D1.1 (Structural Welding Code — Steel), ASME Section V, Article 4 (ultrasonic examination of welds), and Article 5 (materials — the base metal beside the weld, the cladding, the plate you already studied) are the technique documents the industry actually opens. They are not ASNT exam books. Attribute them. Acceptance for an ASME job is in the construction code (Section VIII, Section III, B31.3, and so on) or the owner's specification. Acceptance for an AWS structural job is in D1.1's own acceptance clauses (statically versus cyclically loaded welds, indication classes). Section V does not reject a weld. A Level II who cites "Article 4 reject" has named the wrong book.

Volume coverage with 45°, 60°, and 70°

Most weld examinations in this program are angle-beam shear after refraction in steel. The three shop angles exist because no single refracted angle lights every fusion face and every thickness equally.

• 70° produces a long metal path and a long skip. It is kind to thinner walls and to shallow, near-surface planes (weld toe, under-bead, a shallow bevel). On thick product the path is so long that attenuation and plotting error grow.
• 60° is the middle worker: a usable compromise of skip length, residual energy, and incidence on many single-V faces.
• 45° produces a short skip and more residual energy. It is kind to thicker product, to vertical fusion faces (double-V lands, narrow-groove sidewalls that stand near 90°), and to first-leg looks at a deep root when 70° would skip past the volume.

The procedure's scan plan — not a habit — names the angles for that thickness and groove. A common teaching pattern is 70° on thinner carbon-steel butt welds, 70° plus 45° as thickness grows, and 45°/60°/70° when the groove has both shallow and steep faces. Do not invent a universal "always 70°" rule. Do not invent an unpublished ASNT thickness cutoff. If an item gives a thickness and a groove, pick the angle that hits the fusion face near normal and that places the required leg on the screen.

Coverage is volume, not a single centerline pass:

• The weld metal, both fusion faces, the root land, the HAZ, and the adjacent base metal the procedure includes.
• First leg (half V) and second leg (full V) as required so that ID-connected and OD-connected reflectors both have a chance to reflect back to the probe.
• Enough approach distance from the weld toe that the beam enters the volume at the correct skip. A probe jammed against the cap cannot see the root on the first half-skip.

If the cap reinforcement blocks the approach, the procedure will say whether you grind the cap, use a higher angle, examine from the other surface, or record a limitation. Improvising a new angle because the cap is in the way, without a procedure change, is not Level II authority.

Skip positioning: where the beam is when the peak appears

Skip distance is the surface distance from exit point to the next point the centerline of the beam strikes the same surface after bouncing off the opposite surface:

Skip ≈ 2 × T × tan(θ)
Half-skip ≈ T × tan(θ)

where T is the local thickness and θ is the actual refracted angle you measured on the IIW or DSC block, not the number molded into the wedge.

Useful tan values, for orientation only:

What that means on a plate butt weld examined from the OD / accessible face:

• First leg (before the half-skip node): the beam is heading toward the far surface. On pipe or a vessel that far surface is the ID. Root, incomplete penetration, ID undercut, and ID-connected cracks live here.
• After the bounce, second leg (toward the full skip): the beam is heading back toward the near surface. Cap, OD toes, OD undercut, and OD-connected cracks live here.
• 1.5 skip and beyond: used when access forces a longer approach or when the procedure wants a second look at the root after another bounce. Plotting error grows with every bounce. Do not live there if a first-leg path exists.

Depth of a reflector, once metal path (MP) and θ are known:

Depth ≈ MP × cos(θ) on the first leg, then folded back through the thickness on later legs. Surface distance from the exit point is MP × sin(θ). Digital instruments compute this if the calibrated angle and thickness are honest. A worn wedge with yesterday's angle in the menu plots a root indication in the cap.

ID versus OD is therefore a plot, not a vibe. If the metal path and the probe stand-off put the peak at the ID node, treat it as an ID-family reflector (root geometry or root discontinuity) until a second angle or the other surface contradicts you. If it plots in the cap envelope, treat it as cap-family. A peak that plots inside the weld volume, off both surfaces, is a volume candidate (slag, buried LOF, porosity cluster) — still confirmed with a second angle when the procedure asks.

Scan patterns: raster, oscillating, and the forgotten longitudinal pass

A calibrated probe that never visits the volume is not an examination.

Raster (index) scan. Move the probe parallel to the weld along a line at a fixed stand-off, then index toward or away from the weld by a fraction of the element size so the beam overlaps (procedures often want on the order of 10% overlap; follow the written number). Cover from a stand-off that puts the first-leg beam on the near fusion face out to a stand-off that puts the required far-leg beam through the far toe. Both sides of the centerline are separate rasters.

Oscillating (swivel) scan. While you traverse, rotate the probe a few degrees left and right of the primary aiming direction. Fusion faces are not perfectly aligned with the weld axis. A tight lack-of-fusion plane that is a few degrees off square can disappear on a rigid raster and light up as soon as the beam is normal to it. Oscillation is how you stop pretending the bevel is a textbook 30°.

Longitudinal (transverse-discontinuity) scan. Turn the probe so the beam looks along the weld, not across it. This pass is for transverse cracks and other reflectors whose faces are perpendicular to the weld axis. A perfect transverse raster with 70° across the weld can miss a transverse crack that a 45° or 60° look down the axis would have caught. If the procedure requires it — and many weld procedures do — it is not optional artwork.

Speed and overlap. Too fast a scan is a coverage failure, not a productivity win. The instrument's pulse-repetition rate and your eye have to see the peak. Automated scanners record this; a hand scan does not get a free pass.

ID and OD geometry you must plot before you name

Welds are full of legal reflectors that are not discontinuities.

ID / root geometry

• Hi-lo (mismatch): the two IDs are not flush. The resulting corner is a sharp, often one-sided geometric reflector at the root node.
• Suck-back / concave root: a geometric groove on the ID. It can look like incomplete penetration on the screen and must be separated by plot, by a second angle, and often by visual or remote visual of the root when access exists.
• Excess penetration / root bead: a protruding root that returns a fat geometric echo at the ID.
• Counterbore corner: on pipe, the land-to-counterbore corner is a textbook ID geometric reflector beside the weld, easy to assign to the weld if you do not measure stand-off.
• Backing bar or backing ring: a strong planar-ish echo at the ID that is on the drawing. Read the drawing.

OD / cap geometry

• Reinforcement (cap): the toes are corners. Second-leg echoes at the cap envelope are often geometry.
• Undercut: may be rejectable by the construction code, but it is still a surface groove. UT will see a toe reflector; VT often owns the call.
• Uneven grind or a remaining weld ripple after flush-grind: leftover corners.

Geometry tends to peak at a predicted skip and to move in a characteristic way as you change stand-off: the echo walks as a surface-following reflector. A buried slag line or a fusion-face plane stays in the volume as you rock and index. That motion, plus a second angle, is how a Level II stops calling every bright root echo incomplete penetration.

When the procedure requires ID and OD access, use it. A geometric ID corner that is confusing from the OD is often obvious from the ID. A cap toe that clutters the second leg disappears if you can scan first-leg from the OD after a flush grind, or first-leg from the ID toward the cap.

Root versus cap versus HAZ — different neighborhoods

Name the neighborhood before you name the species.

Root. Incomplete penetration, root lack of fusion, root crack, hollow bead, suck-back, leftover unconsumed insert. First-leg from the OD (or a dedicated ID scan) is the usual look. Root calls without a plotted ID node are guesses.

Cap and weld toes. Toe cracks, undercut, incomplete fusion at a cap pass, overlap. Second-leg from the OD, or first-leg from the opposite surface, or a high-angle look at the toe. Cap geometry is the noise floor of this neighborhood.

Weld volume. Interpass lack of fusion, trapped slag lines, porosity clusters, a crack that grew out of a slag line. These plot between the surfaces. They should not walk like a cap toe when you index.

HAZ and fusion line. Hydrogen cracks, reheat cracks, and some toe cracks sit in the HAZ just outside the fusion face. They can be mid-aligned with the fusion face (and look like LOF) or normal to the surface (and look like a toe crack). A small stand-off change that walks the peak out of the weld into the plate is a HAZ clue. Base-metal laminations that open at the weld prep are a topic 2 leftover that becomes a weld problem — still a lamination by origin.

If an item gives you only "bright echo, 70°, somewhere near the weld," demand (or assume the question already gave) metal path, stand-off, and thickness so you can say root, volume, or cap. Without those, the honest answer is "insufficient to locate."

Planar versus volumetric signatures

Characterization on the general exam is signature family, not a full sizing lab. The next chapter goes deeper. Here you need four weld species and one geometric control.

Lack of fusion (LOF). Planar. Oriented along a fusion face (bevel, interpass, or adjacent-bead wall). Strongest when the beam is near-normal to that face — which is why 45° and 70° are not interchangeable. Often one-sided: a bevel LOF that lights up from side A may be quiet from side B. Sharp, persistent as you traverse along the weld if the missed fusion is long. Does not behave like a sphere when you swivel.

Crack. Planar, often sharper than slag, sometimes branched. Addresses: root, toe, HAZ, crater (if a stop was left in). A vertical crack likes a 45° or 60° look more than a grazing 70°. A toe crack may want 70°. Cracks can mode-convert and throw extra peaks; do not invent a second discontinuity for every satellite blip until you have plotted them. Like LOF, a crack is directional. Unlike LOF, it is not obliged to sit on a bevel — it can cut across beads.

Slag. Intermediate. Elongated along the weld, often in interpass pockets or along a fusion wall that was not cleaned. Irregular, faceted, sometimes a "tail" as you walk off the peak. More forgiving of angle than a tight crack, less omnidirectional than a gas pore. A long slag line can mimic LOF on one angle and look volumetric on another. Use the second angle and the plot (volume versus face).

Porosity. Volumetric. Isolated pores or clusters. Rounded, usually lower amplitude for their 'size' than a plane of the same envelope, and similar from several aiming directions. A single pore is a blip that does not stretch far along the weld. A cluster is a noisy packet that still does not lock to one fusion-face angle. Do not call a 40 mm traveling planar peak "porosity" just because porosity is a friendlier word.

If the procedure uses DAC or TCG, you still characterize before you apply the recording threshold. A geometric ID corner 14 dB over DAC is a loud geometric ID corner. Amplitude does not convert it into slag.

Both sides and both surfaces when the procedure requires them

A single-V weld sitting on a backing bar, examined with one 70° probe from one flange face, is a partial look at the volume. Fusion face A, fusion face B, the root land, and the cap toes do not all see that beam at a useful incidence.

Full coverage language in weld procedures usually means some combination of:

• Both sides of the weld (from plate A and from plate B).
• Both surfaces (OD and ID, or top and bottom on plate).
• More than one angle on each of those approaches.

That can be up to four approach surfaces × two or three angles. The procedure will cut that matrix down when access, thickness, or a flush-ground cap makes a path redundant. When the procedure does require both sides and both surfaces, skipping the ID because "the first-leg root looked quiet" is a coverage failure. When access is only one side (a T-K-Y chord, a vessel with no ID entry, a lap), the procedure must state the limitation and often adds angles or a supplementary method. A Level II does not silently convert a two-side procedure into a one-side examination.

T-, K-, and Y-joints, nozzles, and fittings are the practical extreme: one surface, ugly geometry, and a scan plan that is a drawing, not a habit. Topic 3 items that mention a T-joint are often testing whether you know the backwall is gone (topic 2 geometry) and whether the weld volume still needs the angles and sides the plan named.

Technique references versus acceptance

Keep the books in their lanes.

ASME Section V, Article 4 tells you how to examine a weld ultrasonically: which block, how to build DAC, which scans, how to record. Article 5 tells you how to examine materials. Neither article is, by itself, the accept/reject table for a pressure vessel or a piping weld. The referencing construction code (Section VIII Division 1, Section III, B31.1, B31.3, NBIC repair context, and so on) or the owner's specification states what amplitude, length, and character are unacceptable. Specific-exam items will lean on this harder; the general exam still expects you not to say "failed Article 4."

AWS D1.1 is a construction code that also contains ultrasonic technique (IIW/DSC calibration, scanning patterns, the indication-rating arithmetic that uses reference level and indication level and attenuation factor) and acceptance by weld category and loading. On an AWS structural job, D1.1 really is the acceptance book — but it is acceptance because it is the construction code, not because it is a Section V analog.

The employer's written procedure sits under those documents and under the written practice. It freezes angles, holes, recording level, and whether both surfaces are required. The general exam will not ask you to recite an unpublished ASNT dB table. It will ask whether you know which document type answers "how do I scan" versus "does this weld pass."

Realistic exam scenarios

• 1 in plate, 70° only, probe jammed against an as-welded cap. You do not have first-leg root coverage. Move back to the half-skip stand-off or use the other surface / another angle as the plan requires.
• Peak at a metal path that plots exactly on the ID counterbore corner, one side only, gone from the other side. Geometry until a second angle or VT says otherwise — not automatic LOF.
• Long traveling peak that is 12 dB over DAC from side A on 45° aimed at the bevel, quiet from side B, quiet on 70°. Planar, fusion-face, one-sided: lack of fusion until proven otherwise.
• Short, rounded blips at several metal paths, similar amplitude from 45° and 70°, no travel along the weld. Porosity family.
• Procedure requires both surfaces; ID is accessible and you skipped it. Incomplete examination, even if the OD DAC shots were pretty.
• Item asks which book rejects the weld. Construction code or D1.1 acceptance clauses — not "ASME V Article 4" as a reject stamp, and not ASNT's general-exam outline.

What topic 3 is testing

Evaluating weldments is a scan plan plus a plot plus a signature family. 45°, 60°, and 70° exist to hit different faces and thicknesses. Skip math tells you ID versus OD. Raster, oscillation, and a longitudinal pass turn the plan into coverage. Both sides and both surfaces are required when the procedure says they are. Root, cap, and HAZ are neighborhoods. LOF and cracks are planar and directional; porosity is volumetric; slag sits in between; geometry lives on the surfaces at predicted skips. AWS D1.1 and ASME V Articles 4 and 5 tell you how to run the examination. The construction code tells you whether the weld stays. That is topic 3 as the general exam uses it.