How ASTM E164 Is Actually Applied to Weld Ultrasonic Testing

Scope: What E164 Covers and the Boundaries People Ignore

E164 is a Practice for contact ultrasonic testing of weldments, and each of those words narrows it. Contact means a hand-held or mechanised probe coupled directly to the surface, not immersion. Ultrasonic testing here means pulse-echo A-scan with straight and angle beam probes. Weldments means welds and their heat affected zones in wrought ferrous and aluminium alloy products, in the thickness band from about 1/4 in. to 8 in.

The exclusions are where jobs go wrong. It does not cover castings, whose grain structure and geometry defeat the technique. It does not cover austenitic or dissimilar metal welds, where beam skewing and attenuation make a carbon-steel-derived distance amplitude curve meaningless. It does not cover phased array or time of flight diffraction, which have their own practices in E2700 and E2373. And it contains no acceptance criteria at all.

Two boundary errors recur. The first is thin wall: applying E164 to 4 or 5 mm pipe and claiming full volumetric coverage when the near-surface dead zone consumes a meaningful fraction of it. The second is material: running a carbon steel technique on 316L weld metal and reporting a clean result that means nothing. Both are procedure problems rather than technician problems, and both are best caught during procedure review by ASNT Level III consulting rather than during a client audit.

The Base Metal Straight Beam Scan Comes First

Before any angle beam work, E164 requires the scanning surface base metal to be examined with a straight beam probe. The purpose is not to find rolling defects for their own sake. It is to discover whether anything in the plate will intercept the angle beam on its way to the weld, because a lamination parallel to the surface is an almost perfect reflector at normal incidence and a total obstruction to a shear wave trying to pass through it.

The consequence of skipping this step is uniquely nasty, because it produces a clean screen. The technician scans, sees nothing, and reports full coverage, when in reality a portion of the weld volume was shadowed and never insonified at all. There is no signature on the instrument that distinguishes an undisturbed volume from an unreachable one. That is why this step is mandatory rather than advisory.

It also drives the technique. A lamination discovered in the scanning surface means the affected area must be examined from another surface or another side, or the coverage limitation must be recorded explicitly on the report. Recording a limitation is an entirely acceptable outcome; silently omitting the volume is not. Reviewers assessing an E164 package check for the straight beam record first, because its absence invalidates everything that follows.

Calibration, the DAC, and the Transfer Correction Nobody Records

Angle beam calibration under E164 is built on side-drilled holes at defined depths in a reference block of material representative of the part in composition, thickness range, product form, surface condition and heat treatment. Responses from those holes at successive sound paths build the distance amplitude curve, which is the reference against which every indication is later sized in decibels. Notches provide the corner-reflector response used for surface-connected geometry.

Representative is the word that gets ignored. A block from a different heat treatment condition, a different product form, or a substantially different thickness produces a DAC that does not describe the part being examined. It is the block that happened to be in the van, and the resulting reference level can be several decibels away from correct in either direction. Sensitivity established on the wrong block is not sensitivity, it is an assumption.

Transfer correction is the second half of the problem and the more frequently missed. Surface roughness, curvature, coating, grain structure and temperature all change how much energy actually enters the part compared with the block. E164 requires that difference to be evaluated and applied. It is usually a few decibels, occasionally more, and it almost always runs in the direction that makes the part look cleaner than it is. Its absence from the report is the single most common finding on contact weld UT packages, and it is one that competent NDT training to SNT-TC-1A addresses directly at Level II.

Coverage: Half Node, Full Node, Both Sides, Both Surfaces

The weld volume plus the heat affected zone has to be swept, and geometry decides how. The half node, or first leg, covers the far portion of the weld directly; the full node, after one reflection from the opposite surface, reaches the near portion. Which parts of the cross-section each leg reaches depends on the refracted angle, the wall thickness, the groove geometry and the width of the cap.

Two consequences follow. First, more than one angle is usually required, because a single 45 degree beam will strike a narrow-groove sidewall at a poor incidence and return little. Selecting angles so the beam approaches the expected fusion faces as near normal as possible is the whole art of the technique. Second, scanning from both sides of the weld is normally necessary, and on thick sections from both surfaces, because a planar flaw favourably oriented to one direction can be effectively invisible from the other.

Scan mechanics matter as much as angles. Index increment must overlap by at least ten percent of the effective transducer dimension so that no stripe of the volume is skipped, and scanning speed must stay slow enough for a real indication to register on the display and on the operator. A technique sheet that specifies angles but not overlap and speed has left the two variables most likely to produce a false clean result to the technician's discretion.

Acceptance Criteria Belong to the Referencing Code

E164 will tell you how to find and characterise an indication. It will not tell you whether to reject it. For structural steel that decision belongs to AWS D1.1 Clause 8, which uses an indication rating in decibels derived from its own calibration and read against a table that varies with thickness and sound path. For pressure vessels it is ASME Section VIII Division 1, with Mandatory Appendix 12 for ultrasonic examination of welds. For process piping it is ASME B31.3 Table 341.3.2, and for pipeline girth welds API 1104 Section 9.

This is where the most expensive procurement trap on the page appears. A specification that says examine per ASTM E164 and accept per AWS D1.1 has combined two calibration philosophies. The D1.1 indication rating is arithmetic performed on a reference level that D1.1 itself defines; substitute a DAC built to E164 on a different reflector and the resulting number no longer corresponds to the acceptance table. The report will look complete and the dispositions will be unsupported.

The resolution is straightforward and has to happen before scanning: calibrate to the document whose acceptance criteria you will apply, and use the other as supporting technique guidance. Writing that decision into the procedure, with the reference reflector and reference level stated explicitly, removes the ambiguity permanently. Where legacy specifications already carry the conflict, an independent read of the procedure and the resulting reports against the code actually contracted is what independent report validation is for.

E164 Against ASME Section V Article 4

For Code work, ASME Section V Article 4 is the governing ultrasonic document, invoked by the referencing Code Section. It sets essential variables for the procedure, calibration block requirements, the temperature relationship between block and part, scanning speed limits, couplant restrictions on austenitic and nickel alloys, and the demonstration expectations. It is enforced by the Authorised Inspector, and it is not optional.

E164 is invoked instead by ASTM material specifications, by owner and EPC specifications, and by fabrication contracts outside ASME jurisdiction. The two documents describe the same physics and reach broadly compatible techniques, but they are not interchangeable in a procedure. A vessel fabricated to Section VIII Division 1 needs Section V Article 4 with the applicable appendix; writing that procedure to E164 because the technician is familiar with it will not survive the Authorised Inspector's review.

In practice many organisations maintain one technical technique and two procedure covers, each mapping the same scanning plan onto the essential variables of the document that governs the contract. That works, provided the mapping is explicit and the calibration requirements of the governing document are the ones actually followed on the job. Keeping procedure revisions, essential variable changes and the resulting requalifications traceable per job is the kind of records burden an inspection management system removes.

Reporting: What Has to Be on the Sheet

An E164 report has to let a competent reader reconstruct the examination without being present for it. That means the weld or component identification with the datum and scanning directions defined, the material, product form and thickness, the surface condition and preparation, the instrument and probe with serial numbers, frequency, element size and nominal angle, the wedge, the couplant, and the reference block identification.

It then has to record the calibration: the reference level, the DAC construction, the measured transfer correction in decibels, the time of calibration and the verification checks performed during and at the end of the shift. Indications are recorded by surface distance from the datum, depth, sound path, amplitude relative to the reference, length and the angle from which they were detected, together with the scanning limitations encountered.

Limitations are the section that separates a professional report from a decorative one. Where a nozzle geometry, a support, a coating or a lamination prevented coverage of part of the volume, the report says so and identifies which part. An owner reading a report years later during a fitness-for-service assessment needs to know what was actually examined, and a report that quietly implies complete coverage misleads the person who most needs the truth.

The Findings That Recur on E164 Jobs

Reviewers who audit contact weld ultrasonic packages find the same handful of issues repeatedly. Transfer correction absent from both the procedure and the report. No straight beam base metal record. A calibration block whose product form or heat treatment does not match the part. Scanning overlap unspecified. Couplant used for the examination different from the couplant used at calibration. Certification expired for the technician who signed the sheet.

A second cluster concerns the boundary between documents. Procedures citing E164 for acceptance criteria it does not contain. Procedures citing E164 while the crew runs phased array. Procedures written for carbon steel and applied to austenitic weld metal without any demonstration on representative material. Each of these can be closed in an afternoon before the work starts, and each is expensive after the joints are welded, painted and installed.

The underlying pattern is that the failures are documentary before they are technical. The physics is well understood and the technicians are usually competent; what is missing is a procedure that fixes every variable the practice leaves to judgement, and a report that proves the procedure was followed. Getting that pairing right, on a specific contract and against the specific code that governs it, is the substance of ASNT Level III consulting work on ultrasonic scope.

E164 against phased array: what the older standard does not attempt to cover

E164 was written around conventional single-element ultrasonic technique, and its calibration and transfer correction methodology reflects that — a manual DAC curve built on a fixed reference block, applied by a technician making point-by-point angle and gain decisions. Phased array and TOFD introduce sectorial scanning, encoded data recording, and TCG built across a swept angular range, none of which E164's calibration procedure was designed to control, and a shop running phased array under an E164-only procedure has a real gap between what the document specifies and what the technique actually requires.

ASTM E2700 exists specifically to close that gap for phased array weld examination, with its own calibration, TCG-at-every-angle and encoded-data requirements. The practical implication for a procedure library is that E164 and E2700 are not interchangeable and a shop offering both conventional and phased array UT needs separate procedures citing the correct governing document for each — not one procedure with phased array details appended to an E164 framework that was never built to carry them.

This distinction is worth stating on a proposal or a procedure review specifically, because the two techniques are sometimes treated as a hardware choice within one procedure rather than as two examinations governed by two different standards with different acceptance mechanics underneath. None of this substitutes for reading the referencing construction code directly, since E164 itself is explicit that the acceptance decision is never its own to make.

Request a consultation