15.3 General Evaluation, Sizing, and Acceptance Logic

The ASNT NDT Level II ultrasonic general exam lists General Evaluation as official UT topic 6 on the outline administered prior to 5 February 2027. Topics 4 and 5 asked what the echo is. Topic 6 asks what you measure, what you write down, and which printed sentence decides accept or reject. A bright, well-characterized crack is still only an indication until the written acceptance standard named in the procedure is applied.

After 5 February 2027 ASNT's revised UT general exam rolls from six topics to four domains: principles and techniques; materials and product forms; data, interpretation, and reporting; and training, certification, and safety. That revision adds explicit data-quality, scan-plan, and OJT language. Those ideas are already good Level II practice on the current outline. Do not wait for the new domains to use a scan plan, to reject bad data, or to work only within documented qualification. Do not replace this chapter's teaching baseline with the post-2027 outline either — the exam you sit before that date is still topics 1–6.

ASME Section V, AWS D1.1, API 1104, and customer aerospace specifications are industry/code documents, not ASNT-published exam secrets. When a stem prints a recording level or a reject rule, use the printed rule. When it does not, describe the logic — %DAC, length, location, character — and refuse to invent a number ASNT did not publish.

6 dB drop sizing — as commonly taught, and its limits

The 6 dB drop (half-amplitude drop) is the length-sizing method almost every Level II course still teaches. The motion is simple:

1. Scan to the maximum echo from the indication.
2. Note that peak amplitude (or put it at a convenient screen height).
3. Move the probe along the indication until the amplitude falls by 6 dB (half the voltage, which is half the screen height if you started at a linear 80% and stop at 40%).
4. Mark that probe position as one apparent edge.
5. Return to the peak and move the other way to the other −6 dB point.
6. The distance between marks, after the procedure's index-point correction, is the reported length.

A 20 dB drop (one-tenth amplitude) is the sibling method some procedures name for a more conservative envelope. Encoded imaging and tip-diffraction methods are different families. Use the method the procedure names.

Limits you must be able to state:

• The method assumes a reflector that is larger than the beam and bright enough that the −6 dB points sit on the edges, not on the beam's own skirt.
• On a reflector smaller than the beam, the −6 dB contour is essentially the beam width. You over-report length. That is why a tiny pore can "size" at 8 mm with a large crystal.
• A planar reflector whose amplitude varies along its length (a twisting crack, a partial-bond island) produces −6 dB points that are not the true ends. You may under-size a long, fading plane or split one indication into two.
• 6 dB drop is not through-wall height unless the procedure specifically uses an amplitude-drop in the depth direction and the geometry supports it. The common shop 6 dB walk along the weld is a length estimate.
• Orientation, coupling, and a geometric corner under the same gate all move the −6 dB points. A root-bead echo has a 6 dB "length" too. That number does not turn geometry into a crack.

Tip diffraction, time-of-flight diffraction (TOFD), and amplitude-drop-plus-beam-spread corrections exist because 6 dB is a teaching default, not a physics identity between probe travel and discontinuity edge. If the stem asks for the commonly taught method, 6 dB drop is the answer. If it asks for the limit, say beam-width oversizing of small reflectors and poor through-wall performance.

DAC percent amplitude evaluation

A distance-amplitude correction (DAC) curve is built from reference reflectors — commonly side-drilled holes at several depths, sometimes flat-bottom holes — so that amplitude is compared at the same sound path. TCG (time-corrected gain) is the instrument's way of flattening that same idea into a constant recording height. Either way, an indication is described as a percent of DAC (or a decibel relation to the reference) at its metal path.

What %DAC is:

• A repeatable amplitude number you can write beside length and location.
• The input to many recording levels and acceptance levels (for example, record at 20% DAC, evaluate at 50% DAC — examples of industry practice, only if the named code says so).
• A way to compare two reflectors at different depths without pretending a deep echo is "small" only because of spreading and attenuation.

What %DAC is not:

• Not through-wall height. An 80% DAC lack of fusion can be a thin plane. A 30% DAC crack can be deep if it is poorly oriented.
• Not an automatic reject. Codes mix amplitude, length, location (weld metal vs HAZ vs base metal, surface-connected vs buried), and character (crack-like vs rounded).
• Not valid if the DAC was built on the wrong material, the wrong reflector, the wrong angle, or after a frequency change you did not restandardize.

Worked logic the specific exam likes (the stem will print the numbers):

• Indication 15% DAC, recording level 20% DAC → do not record as an indication under that paragraph (still note anything the procedure says you must, such as a suspected crack regardless of amplitude).
• Indication 80% DAC, 12 mm long, reject rule "crack-like above 50% DAC and longer than 10 mm" → reject if you have characterized it as crack-like.
• Indication 80% DAC, 4 mm long, same rule → record and accept for that amplitude-length pair if character and location also pass.
• Indication 120% DAC geometric root with symmetric echodynamics and no extension → not a defect if the standard evaluates discontinuities, not geometric corners. Amplitude alone did not make the bead rejectable.

If DAC cannot be drawn — coarse grain ate the holes, or the procedure uses a side-drilled reference at one depth plus a transfer correction — follow that written sensitivity system. Do not invent a 6 dB-from-back-wall shop custom and call it ASME.

Length versus through-wall

These are different dimensions. Mixing them is a standard trap.

Fitness-for-service and many construction rules care more about through-wall and surface connection than about a long shallow slag line. A 40 mm long, 1 mm tall lack of fusion and an 8 mm long crack that is 80% of wall are not the same evaluation, even if both plot at 60% DAC.

If the procedure only asks for length and %DAC, report length and %DAC. Do not add a guessed height. If it asks for height, use a method the procedure qualifies. "I dropped 6 dB along the weld and called that the depth" is a failed evaluation.

Procedure acceptance levels

Acceptance is a document, not a feeling. The Level II's job is:

1. Confirm the procedure is the one specified for this product, code year, and technique (angle, frequency, probe size, scan surface).
2. Confirm calibration and sensitivity (DAC, TCG, or the named reference) are in tolerance now, not last Tuesday.
3. Characterize the indication (topic 5).
4. Measure what the procedure says to measure — usually %DAC or dB, length, location, sometimes height.
5. Apply the printed combination of those numbers. Typical industry patterns include: record above a %DAC threshold; reject crack-like indications above a higher threshold when length exceeds a fraction of thickness or a fixed millimetre value; reject any surface-connected crack regardless of amplitude; treat grouped indications by a spacing rule. Those patterns are examples. The stem's document wins.
6. If the document is silent on a point (how to measure length, whether geometry is excluded), do not invent a shop custom. Report the indication and the silence, and escalate to the Level III or the customer as the written practice requires.

"Looks bad," "I've never passed one that high," and "the welder said it is just the root" are not acceptance levels.

Documenting location

A number without an address cannot be repaired, re-examined, or defended. Record at least:

• Datum. Distance along the weld or part from a stated zero — a match mark, a stamped reference, a flange face, a drawing station. "In the weld" is not a datum.
• Offset from weld centerline (or from a bevel toe) if the procedure uses that grid.
• Depth / metal path / skip. Write the metal path and whether the peak is at half-skip, full skip, or a plotted depth. Convert to depth only with the angle and thickness the plot used.
• Surface. OD or ID, face A or B, which side of a double-V.
• Amplitude (%DAC or dB to reference) and length (and height if required).
• Probe identity: angle, frequency, size, and the procedure or technique sheet number.
• Character: planar / volumetric, geometric vs relevant, and the echodynamic facts you used.

If you cannot reconstruct the indication from the report without standing next to the person who wrote it, the report is incomplete. Encoded scans still need the same words: what datum the encoder used, what gate produced the image, and what acceptance paragraph was applied.

Do not call without the written standard

Call, on this exam, means accept or reject — a disposition. Characterization is allowed without a code table ("planar echo at half-skip, 18 mm long, 70% DAC, not explained by the root plot"). Disposition is not.

• No written acceptance standard in the stem and none named in the procedure → you cannot legally reject or accept. You can record and escalate.
• Two documents disagree → follow the order of precedence the contract or written practice states. Do not average them.
• A construction code and a customer spec both apply → the stricter rule often governs, but only if the procedure says so. Read the paragraph.
• A geometric echo that some shops "always reject to be safe" is still geometry if the standard evaluates discontinuities. Over-calling is not conservatism when it hides a real crack you stopped looking for.

This is also why data quality and scan plans belong in evaluation, not only in a 2027 domain name. A scan that skipped the cap toes, a DAC drawn on the wrong block, a gate that never covered the root, or a file with lost coupling is not evaluable. The correct Level II action is to repair the data (rescan, recalibrate, follow the scan plan) rather than to disposition noise.

On-the-job training (OJT) and the employer's written practice already decide which product forms and techniques you are allowed to evaluate. The February 2027 language makes that expectation louder. It does not create it. If you have not been qualified on coarse-grain austenitics, do not sign an evaluation that pretends the carbon-steel DAC still applies.

Realistic exam scenarios

You peak an indication at 64% full screen, walk both ways to 32%, and mark 22 mm between index points. The procedure uses 6 dB drop for length. Report 22 mm length. Do not also write "22 mm through-wall." If a later tip-to-tip measurement gives 4 mm height, those are two different numbers.

An indication is 80% DAC and 12 mm long. The procedure records above 20% DAC and rejects crack-like indications above 50% DAC when length exceeds 10 mm. You have already characterized it as planar and not geometry. Disposition: reject under that printed pair of rules. If the stem withholds the standard, you record and you do not invent the reject.

A report says "crack, reject" with no datum, no skip, and no %DAC. That record is incomplete even if the call would have been correct. Evaluation includes the address.

A candidate sitting the current general exam is still studying topics 4–6. A candidate who will sit after 5 February 2027 will see those same skills under data, interpretation, and reporting — plus an explicit ask for scan plans, data quality, and documented OJT. Both candidates should already work that way.

Topic 6 language is short: what you measured, how, where it is, and which printed sentence you applied. Anything else is storytelling.