6.3 MT Evaluation Techniques

Evaluation Techniques is official MT general topic 9 on the outline administered prior to 15 December 2026. After 15 December 2026 the same work sits inside the revised domain interpret, evaluate, and document, with codes and procedures as the authority for accept/reject. The teaching point does not change: particles are not a verdict. Particles are a pattern you classify, measure, and compare to a written standard.

ASTM E1444 and ASTM E709 describe how to produce and view indications. ASME Section V, Article 7 tells you how to perform the examination for ASME work. Neither document is, by itself, the accept/reject table for a weld or a forging. That table lives in the construction code, the customer specification, or the procedure that invokes them. A Level II who rejects a part because "it looks like a crack" without opening that table has not evaluated the indication.

Classify first: relevant, nonrelevant, or false

Use the same vocabulary as the rest of this guide and as the method standards.

• A relevant indication is caused by a discontinuity that must be evaluated against the acceptance standard. A sharp linear buildup at a carbon-steel weld toe is relevant until proven otherwise.
• A nonrelevant indication is caused by a known geometry or a permitted condition, not by a rejectable discontinuity. Thread roots, keyway corners, press-fit edges, abrupt section changes, the edge of a drilled hole, and the weld-toe radius can all leak flux. Magnetic writing (local residual patterns from contact with another magnetized object or from incomplete demagnetization) is also nonrelevant once you have identified it.
• A false indication is not caused by a leakage field from the part. Lint that fluoresces, drain lines of wet bath on a horizontal surface, dirt, scale flakes, over-dusted dry powder, a cracked UV filter that adds visible glare, or particles trapped in spatter are false.

Classification is a process, not a glance.

1. Confirm lighting, media, and magnetization were valid. A pattern seen under 50 lux of white light on visible powder is not ready to classify.
2. Wipe, blow, or rinse gently if the procedure allows a check for false media piles, then reapply if required.
3. Look at the geometry. Does the pattern follow a thread, a toe, or a section change for its entire length?
4. Change field direction or demagnetize and reinspect if magnetic writing is suspected. Writing often disappears or moves; a crack does not.
5. Only then call the pattern relevant, nonrelevant, or false.

A nonrelevant geometric collector can hide a real crack at the same location. If the procedure requires it, grind or machine the geometry (within remaining-thickness rules) and reinspect. Do not stamp "nonrelevant" on a toe simply because toes collect powder.

Linear versus rounded

After a pattern is relevant, describe its shape in the language the acceptance standard uses.

The 3:1 length-to-width idea is a common code convention, not an ASNT secret number. Some specifications use different wording ("crack-like," "elongated"). Use the definition in the governing standard. If the stem does not give a definition, treat a sharp, crack-like particle line as linear and a blob as rounded.

Linear is not automatically rejectable. Rounded is not automatically acceptable. A 1-inch linear seam on a bar may be rejectable; a 1/16-inch linear slag pop on a weld may be acceptable. The table decides.

Measure against the acceptance standard — not against instinct

Record length along the indication, location (weld, HAZ, base metal, distance from a datum), and orientation (parallel to the weld, transverse, radial). Some procedures also want width or a sketch. Measure the particle pattern with a scale or optical aid the procedure allows. Do not "split the difference" because you think the crack is shorter than the powder.

Then open the acceptance standard named in the procedure. Examples you will see in industry — not as ASNT answers — include ASME construction-code surface tables, AWS D1.1 visual/MT workmanship limits, and customer aerospace tables that reject any linear indication in a critical zone. The exam item will usually print the limit in the stem ("linear relevant indications longer than 1/8 in / 3 mm are rejectable"). Your job is to apply that sentence, not to invent a shop custom.

Worked logic the specific exam likes:

• Indication is relevant and linear, measured length 5 mm, limit 3 mm → reject (defect).
• Indication is relevant and rounded, diameter 2 mm, limit 3 mm isolated rounded → accept if spacing rules are also met.
• Indication follows a keyway corner, no crack-like extension, procedure calls geometry nonrelevant → accept as nonrelevant, and record that classification.
• Indication "looks ugly" but the standard has no rejectable condition → you do not reject it on taste. You may still report it if the procedure requires recording all relevant indications.

Defect means the discontinuity failed the acceptance standard. It is not a synonym for crack, and it is not a synonym for indication.

When to reinspect with a different field

Reinspection is an evaluation technique, not an admission of failure.

Change field direction (about 90°) when:

• The suspected discontinuity would run parallel to the first field (longitudinal seam after only a coil shot; transverse crack after only a head shot).
• The first pattern is weak, broken, or one-sided and may be orientation-limited.
• The procedure requires two directions on every area and the second direction is still due.

Change current type when:

• An AC-yoke pass is clean but the concern is near-surface (underbead, shallow burst). A DC or HWDC yoke, prods, or a bench head shot may be the procedure's next step.
• A DC residual pattern is cluttered. An AC field can both examine the surface and help shake off residual writing.

Demagnetize and reinspect when:

• You suspect magnetic writing or residual leakage from a previous shot.
• Particles cling everywhere and you cannot tell geometry from discontinuity.

Do not keep dusting more dry powder onto a maybe. Over-application creates false lines and buries relevant ones.

Recording essential variables

If it is not recorded, it was not a controlled examination. The procedure lists essential variables; typical MT records include:

• Procedure identification and revision
• Part identity, material, and area examined
• Equipment identity (yoke, bench, power pack) and magnetization method
• Current type (AC, HWDC, DC) and amperage or pole spacing / ampere-turns
• Particle system (dry visible, wet visible, wet fluorescent), color or designation, and bath batch if wet
• Lighting measurements (UV-A μW/cm² and ambient lux, or white-light lux)
• Process-control status for that shift (yoke lift, bath centrifuge, system-performance piece)
• Indication data: relevant / nonrelevant / false; linear / rounded; length; location; disposition
• Technician identification and date

ASME Section V, Article 7 and ASTM E1444 both expect the examination to be reproducible from the record. A report that says only "MT OK" fails topic 9 even if the weld is perfect.

The procedure is the law

Evaluation has a hard stop. You may characterize. You may measure. You may recommend a grind-and-reinspect sequence if the procedure allows blending of shallow relevant indications. You may not:

• Accept a crack-like relevant indication because "it will be painted" or "it always looks like that."
• Reject a geometric collector because a supervisor "does not like powder on threads."
• Swap in a different acceptance table from another job.
• Use particle brightness or "how bad it looks" as a severity scale that the standard does not define.

If the governing specification is silent, the Level II does not invent a limit. The report describes the indication and states that no acceptance criterion was provided — and the work stops until the Level III or the customer names a standard.

Realistic exam scenarios

Particle buildup follows a bolt thread root for 360° and stops at the thread. No crack-like extension runs into the shank. The procedure treats geometric collectors as nonrelevant. Classification: nonrelevant. Disposition: accept for that indication. If a radial line leaves the root into the shank, that extension is relevant and goes to the length table.

A 3/16 in (5 mm) sharp line sits at a fillet-weld toe. The procedure rejects linear relevant indications longer than 1/8 in (3 mm). Classification: relevant, linear. Disposition: reject. The Level II does not call it undercut by eye and walk away.

A longitudinal coil shot on a bar is clean. The drawing shows a history of seams. Evaluation is incomplete until a circular field is applied. A clean single-direction exam is not a clean part.

A fluorescent line appears, then vanishes after a gentle rinse and does not return on a controlled reapply. That is a false drain or dirt line, not a crack that "healed."

Topic 9 rewards slow language: classify, describe, measure, compare, record. Speed without a standard is how Level II reports get torn up.