Angle Beam Testing in UT: Refraction Angles & Applications

Angle beam testing is a contact ultrasonic technique that sends a shear wave into steel at a set refracted angle, usually 45°, 60° or 70°, to find planar flaws tilted to the surface. For a weld fusion face, the working rule is probe angle ≈ 90° minus the bevel angle, so a 30° bevel calls for a 60° probe.

This guide is about technique: how to pick the angle, how to work out skip distance and the scan zone, how to scan a weld, how to verify the probe on an IIW block, how to set DAC or TCG, how AWS D1.1 and ASME use angle beams, and how rail testing uses them. The wave physics (mode conversion, Snell's law and the critical angles) is covered in full in our guide to shear wave ultrasonic testing and is only summarized here.

Contents: Physics in brief · Choosing the angle · Skip distance · Locating an indication · Scan patterns · IIW checks · DAC and TCG · D1.1 and ASME · Rail · Limits · FAQ

What physics does angle beam testing rely on?

A plastic wedge refracts a longitudinal wave at the steel surface so that only a shear wave enters the part, at an angle set by Snell's law. That is the whole of the physics an operator needs at the probe.

Snell's law is:

sin θ₁ / V₁ = sin θ₂ / V₂

where θ₁ is the incident angle in the wedge, θ₂ the refracted angle in the part, V₁ the longitudinal velocity of the wedge and V₂ the shear velocity of the part. With an acrylic wedge (about 2,730 m/s) on carbon steel (shear about 3,250 m/s), the numbers that matter are:

These values come from Evident's published velocities and are worked through step by step, with the critical-angle derivation, in our shear wave UT guide. The practical consequence is that a wedge marked 70° gives 70° only in steel with the design velocity. In other metals, and as the wedge wears, the true angle moves, which is why it is measured on a block before every job.

Key takeaway: the wedge sets the angle; the block tells you what the angle actually is.

How do you choose the angle for angle beam testing?

Choose the angle that meets the expected flaw face as close to 90° as possible, then check that the beam can cover the full weld volume within a practical sound path for that thickness. The written procedure or code often fixes the angle, but the reasoning behind it is the same.

Rule 1: match the bevel (90° − bevel angle)

Sidewall lack of fusion lies on the bevel face. Probe angles and bevel angles are both measured from the normal to the plate surface, so a beam refracted at (90° − bevel) strikes the fusion face square on. EN ISO 17640 asks for one probe angle to examine the fusion faces at, or as near as possible to, normal incidence.

A 70° probe on a 30° bevel is 10° off the face normal, and a smooth planar flaw hit 10° off can return a much weaker echo than its size suggests. Our article on lack of fusion in welds works this through with an AWS D1.1 rating.

Rule 2: keep the sound path reasonable for the thickness

A steeper angle travels farther to cover the same thickness. The leg (sound path from the scanning surface to the back wall) is T ÷ cos θ, and the full skip distance along the surface is 2 × T × tan θ.

On thin plate, a 45° beam returns to the surface within 0.75 in. on 3/8 in. material, so much of its useful travel falls under the wedge and the cap; a steeper angle spreads coverage over a scan band the probe can actually work. On thick plate, 70° runs nearly 9 in. per leg on 3 in. material. Under the AWS D1.1 attenuation convention of 2 dB per inch beyond the first inch, a reflector at the end of that first 70° leg already carries an attenuation factor of about 2 × (8.77 − 1) = 15.5 dB, against 6.5 dB at the end of the 45° leg. Holloway's 2017 modeling also found the 2 dB/in. convention furthest from reality at 60° and 70° on long paths.

That is why a common pattern is 70° on thinner structural butt welds, adding 60° or 45° as thickness grows, and two or three angles on thick or complex joints. There is no universal thickness cutoff; the procedure governs.

Rule 3: add angles for other flaw orientations

One angle tuned to the bevel is not enough. Inter-run lack of fusion, root cracks and transverse cracks sit at other angles, so procedures add a second angle and a scan along the weld axis (see the scan patterns below).

Key takeaway: start from 90° minus the bevel, check the leg length in the table, and add a second angle for flaws the first can't face.

How do you calculate skip distance, leg and surface distance?

For refracted angle θ and thickness T: leg = T ÷ cos θ, half skip = T × tan θ, full skip = 2 × T × tan θ, and for any sound path SP the depth is SP × cos θ on the first leg and 2T − SP × cos θ on the second. These are the same formulas set out in our ultrasonic testing formulas guide; this section applies them to scan planning.

Worked example: scan zone for a 3/4 in. single-V butt weld

Inputs. 0.750 in. carbon steel plate; single-V groove with 30° bevels and a 1/8 in. root opening; 60° probe (measured 60.0° on the block). Cap width = 0.125 + 2 × 0.750 × tan 30° = 0.125 + 0.866 = 0.99 in., so each cap toe is about 0.50 in. from the weld centerline.

1. Leg: 0.750 ÷ cos 60° = 0.750 ÷ 0.500 = 1.500 in.
2. Half skip: 0.750 × tan 60° = 0.750 × 1.732 = 1.30 in. With the index point 1.30 in. from the centerline, the first leg reaches the root.
3. Full skip: 2 × 1.30 = 2.60 in.
4. Outer edge of the scan zone: to bring the second leg up to the top of the near fusion face, at the near toe, the index point must sit one full skip beyond that toe: 0.50 + 2.60 = 3.10 in. from the centerline.
5. Inner edge: as close to the near toe as the wedge front allows. The first leg then covers the lower part of the far fusion face and the root.

So the 60° probe is moved from the toe out to about 3.1 in. from the weld centerline on each side, and the surface must be clean and couplable across that whole band. A 70° probe on the same weld would need 0.50 + 4.12 = 4.62 in.

Key takeaway: the scan zone width is one full skip plus half the cap width, measured from the weld centerline, for each side scanned.

Worked example: locating an indication on the second leg

An indication's sound path tells you which leg it is on, and only then can you work out its depth. Comparing the sound path with the leg length first prevents the most common plotting error.

Inputs. Same 0.750 in. weld and 60° probe. The index point is 2.40 in. from the weld centerline. An indication peaks at a sound path of 2.40 in.

1. Which leg? The leg is 1.500 in., and 2.40 in. lies between 1.500 and 3.000 in., so the reflector is on the second leg.
2. Depth: 2 × 0.750 − 2.40 × cos 60° = 1.500 − 1.200 = 0.300 in. below the scanning surface.
3. Surface distance: 2.40 × sin 60° = 2.40 × 0.866 = 2.08 in. ahead of the index point, so 2.40 − 2.08 = 0.32 in. short of the centerline, on the probe's side.
4. Plot against the joint. At 0.300 in. deep (0.450 in. above the root), the near fusion face sits 0.0625 + 0.450 × tan 30° = 0.0625 + 0.260 = 0.32 in. from the centerline. The reflector lies on the near fusion face.

That position is consistent with near-side sidewall lack of fusion, which a 60° beam meets square on as it rises on the second leg. The operator would confirm by scanning from the other side and with a second angle before reporting it. Had the operator taken 2.40 × cos 60° = 1.20 in. as the depth, the reflector would have landed below a 0.750 in. plate.

A first-leg check on the same plate with a 70° probe: an indication at 1.60 in. sound path (less than the 2.19 in. leg) is 1.60 × 0.342 = 0.55 in. deep and 1.60 × 0.940 = 1.50 in. ahead of the index point.

Which scan patterns cover a weld?

A full weld examination combines a raster scan across the weld for flaws running along it, a scan along the weld axis for transverse flaws, and scanning from both sides of the weld wherever access allows. Evident's weld tutorial describes the basic motion:

"Moving the probe back and forth causes the sound beam to sweep across the full height of a weld." (Evident, Angle Beam Inspection tutorial)

The naming confuses people: a scan "for longitudinal discontinuities" points the beam across the weld, and a scan "for transverse discontinuities" points it along the weld.

Why scan from both sides

On a single-V weld scanned from one side, the near fusion face is reached only by the second leg, and the far face's upper part sits under the cap where the first leg can't reach. Scanning from the second side reverses both, so each fusion face gets a direct look. A secondary copy of AWS D1.1:2020 Clause 8 states that testing "from both sides of the weld axis shall be made wherever mechanically possible", and that its patterns are symmetrical about the weld axis except the one run directly over it. Confirm the wording in your edition. Where only one side is accessible (a backing bar, a stiffener, a T-joint), the procedure should record the restriction and add an angle, use both legs, or call in another method.

Overlap and speed

Raster lines must overlap. ASME-based procedures commonly specify a minimum overlap of 10% of the transducer dimension perpendicular to the scan direction and a maximum scan speed of 6 in./s unless the calibration is verified at the faster speed. These figures are taken from published procedures rather than the code text, so confirm them in ASME Section V, Article 4, for your edition.

Key takeaway: one angle from one side in one direction misses flaws; the scan plan should show both sides and both beam directions.

How are the beam index point and angle verified on the IIW block?

Peak the echo from the block's 100 mm radius to mark the beam index point on the wedge, then peak a side-drilled hole and read the true refracted angle from the block's engraved scale. Evident's angle-beam guidance says test procedures generally require verifying the beam index point and refracted angle on an IIW block or similar reference before testing.

1. Beam index point (BIP). Place the probe on the IIW Type 1 block facing the 100 mm radius. Move it back and forth until the radius echo peaks, then mark the wedge where it lines up with the radius center mark. That mark is where the beam leaves the wedge, and every surface distance is measured from it.
2. Range. Set the screen range and zero from known radius echoes (or from a DSC block), and confirm with a second known path.
3. Refracted angle. Peak the echo from the side-drilled hole the block's scale is built around and read the angle where the BIP mark meets the engraved scale. Use the measured angle, not the stamped one, in the instrument.
4. Record and recheck. Note the measured BIP and angle, then recheck at the interval the procedure sets and after any wedge change or visible wear.

AWS D1.1-based procedures commonly require the beam to be within ±2° of its nominal 70°, 60° or 45°. A wedge that has worn unevenly can move the angle by a degree or more and shift the BIP, which moves every plotted indication.

Worked example: checking the angle by calculation

Where no engraved scale is available, the angle can be computed from any side-drilled hole at a known depth. Peak the echo from a hole whose center is 1.00 in. below the scanning surface, then measure the surface distance from the BIP to a point directly above the hole center: 1.70 in.

• θ = arctan (surface distance ÷ depth) = arctan (1.70 ÷ 1.00) = 59.5°
• For a 60° probe, that is 0.5° low, inside a ±2° tolerance.
• Cross-check: the sound path to the hole center should be 1.00 ÷ cos 59.5° = 1.97 in.

If the screen reads a very different sound path, the range or zero is wrong, not the angle.

How is DAC or TCG set for an angle beam?

Build the curve from the same probe, at the same angle and from the same surface you will scan, by peaking equal-sized side-drilled holes at increasing sound paths on both legs. A distance-amplitude correction (DAC) curve joins those peaks on the screen; time-corrected gain (TCG) raises gain with time so the same holes all peak at the same height. Our DAC curve guide explains the method in general; the angle-beam specifics are below.

Worked example: DAC points for a 60° probe on a 1 in. ASME-style block

An ASME Section V basic calibration block has three side-drilled holes at 1/4 T, 1/2 T and 3/4 T (vendor listings citing T-434.2.1; confirm the figure in your edition). For a 1.000 in. block and a 60° probe, the leg is 1.000 ÷ cos 60° = 2.000 in.

Set the highest peak to the reference height the procedure names (80% of full screen height is common), mark each hole's peak and join the marks. The curve now covers sound paths from 0.50 to 3.50 in., which spans both legs of a 1 in. weld. Use the curve only within that range; most procedures don't allow evaluation beyond the last measured point.

Transfer correction and scanning gain

The block and the part rarely couple the same way. A transfer correction (the dB difference in a reference signal on block and part) is added when the part is rougher, coated, curved or more attenuative. Scanning is then done above the reference level; ASME-based procedures commonly require at least twice the reference (+6 dB); confirm the figure in Article 4. Indications are evaluated at the reference gain, not the scanning gain.

Key takeaway: an angle-beam DAC must be built on both legs with the scanning probe, or second-leg indications are judged against a curve that was never measured.

How do AWS D1.1 and ASME use angle beams?

AWS D1.1 rates each angle-beam indication against a single reference hole with the formula d = a − b − c, while ASME work compares indications with a DAC built on a basic calibration block and takes acceptance from the construction code. The probe work is similar; the sensitivity method and acceptance route differ.

Two points trip up inspectors who move between the codes. D1.1 does not by itself require UT: the contract or a building code such as AISC 360 does. And a D1.1 rating class is never a fixed dB band; it depends on thickness, angle and loading. Our guides to AWS D1.1 weld inspection criteria and ASME weld ultrasonic inspection cover the acceptance side, and UT acceptance criteria standards compares them.

ASTM E164-24 is the general practice behind both: contact testing of weldments 0.250 to 8 in. thick with angle beams, straight beams or both. When UT is the basis of acceptance, it leaves the reference standards and limits to agreement between the parties.

How is angle beam testing used on rail?

Rail flaw detection relies on a 70° shear beam, aimed forward and backward along the rail, to find transverse defects in the head, plus lower-angle beams of roughly 37° to 45° for the web and bolt holes. Probe sets differ between systems, so treat the angles below as typical rather than fixed.

One U.S. patent for a real-time rail test system (US 5,777,891) describes its 70° beam this way:

"It is primarily used to detect transversely oriented defects such as TD's and for defective welds." (US 5,777,891)

The defects that count as service failures under 49 CFR 213.237 are transverse fissures, compound fissures, detail fractures and vertical split heads, and the same section requires search equipment to detect defects between joint bars. The angled beams are what reach the rail ends and bolt holes under the bars.

The same angle-beam arithmetic applies on rail. A hand 70° probe used to verify a car's suspect call, with an echo at 1.50 in. sound path, puts the reflector 1.50 × cos 70° = 0.51 in. below the running surface and 1.50 × sin 70° = 1.41 in. along the rail from the index point. Shelling, head checks and a ground or corrugated surface can break coupling or mask the head, so a stretch without a valid search should be recorded rather than assumed clean. Defect-specific detail is in our guides to the transverse fissure in rail and bolt hole cracks.

What are the limits of angle beam testing?

Angle beam testing misses flaws parallel to the surface, depends on hitting a planar flaw near square on, and gives position only by arithmetic on a conventional A-scan. A scan plan should name the method that covers each gap.

Amplitude is not size. A tight, planar flaw at a poor angle can rate lower than a harmless rounded reflector facing the beam, which is why sizing and characterization use beam movement, a second angle and, on critical work, TOFD or PAUT.

How EMA Industries applies this

EMA Industries uses angle beam testing in AWS D1.1 structural weld UT, in rail weld inspection and in ultrasonic rail flaw testing for transit agencies, short lines and industrial track owners, and provides NDT Level III services, including procedure and scan-plan review. We're based in Southern California and mobilize nationwide. On a weld job, the scan plan should record the bevel, the angles chosen for it, the scan zone width for each side and the measured BIP and angle. See our ultrasonic testing services or contact EMA Industries to scope an inspection or a procedure review.

Related reading: UT calibration for rail inspection RFPs · A-scan, B-scan and C-scan · How ultrasonic rail flaw detection works

Frequently asked questions

Which probe angle should be used for a 3/4 in. butt weld?

Start from the bevel: 90° minus the bevel angle, so 60° for a 30° bevel. On 3/4 in. plate a 70° probe is also common for the root and near-surface zone. The code or written procedure sets the final angles.

How do you calculate skip distance in angle beam testing?

Full skip = 2 × thickness × tan θ. A 60° beam in 0.750 in. plate skips every 2.60 in., and the first leg reaches the back wall at half that, 1.30 in. from the index point.

Why do inspectors scan a weld from both sides?

From one side, the near fusion face is reached only by the second leg and part of the far face sits under the cap. Scanning from the other side gives each fusion face a direct look, and AWS D1.1 asks for it wherever access allows.

What is the beam index point, and how is it found?

The beam index point is where the beam leaves the wedge, and every surface distance is measured from it. It is found by peaking the echo from the IIW block's 100 mm radius and marking the wedge at the radius center mark.

Does AWS D1.1 use a DAC curve for angle beam testing?

No, not in the usual sense. D1.1 sets a reference level on a 0.060 in. side-drilled hole and corrects each indication with an attenuation factor of 2 dB per inch beyond the first inch. ASME work uses a DAC or TCG built on a basic calibration block.

What angle is used to find transverse defects in rail?

A 70° shear-wave beam, aimed forward and backward along the rail head, is the standard choice for transverse fissures and detail fractures. Lower angles of about 37° to 45° are typically used for the web and bolt holes.

Sources

• Angle Beam Inspection, Evident (Olympus IMS) flaw detection tutorial, accessed October 2026
• An Introduction to Angle Beam Assemblies, Evident white paper (wedge design velocity, IIW verification of beam index point and refracted angle)
• AWS Weld Inspection Tutorial, Evident (D1.1 indication rating and attenuation factor)
• AWS Announces Release of D1.1/D1.1M:2025, Structural Welding Code – Steel, American Welding Society, April 17, 2025
• BPVC Section V, Nondestructive Examination, ASME, 2025 edition
• ASTM E164-24, Standard Practice for Contact Ultrasonic Testing of Weldments, ASTM International
• 49 CFR 213.237, Inspection of rail, eCFR; text checked on Cornell LII
• US 5,777,891, Method for real-time ultrasonic testing system, 1998
• US 5,419,196, Ultrasonic side-looker for rail head flaw detection, 1995
• Holloway (2017), modeling of the AWS D1.1 attenuation factor at 45°, 60° and 70°, CINDE Journal