Attenuation (Ultrasonic)

Attenuation is the loss of ultrasonic energy as it propagates through a material due to absorption, scattering, and beam spread, expressed in decibels per unit path length and material-dependent.

Definition

Attenuation describes how rapidly an ultrasonic signal loses amplitude with distance. It combines absorption (energy converted to heat) and scattering (energy redirected by grain boundaries, porosity, or other heterogeneities). Coarse-grained austenitic stainless steel attenuates much more than fine-grained carbon steel.

Technical Context

Attenuation is corrected for in DAC and TCG calibration.

The principle

The progressive loss of signal amplitude as a wave travels through material. In ultrasonics it comes from absorption, where energy converts to heat, and from scattering at grain boundaries and inclusions; both rise with frequency, and scattering rises sharply once grain size becomes comparable to wavelength.

How it shows up in practice

It governs the practical choice of frequency: higher frequency gives better resolution and worse penetration, and the balance is decided by the material. Transfer correction exists because the calibration block and the component rarely attenuate identically, and ignoring the difference biases every amplitude-based assessment.

Where the codes address it

ASME Section V Article 4 requires transfer correction be addressed; ISO 17640 and DGS/AVG sizing approaches build attenuation compensation into the evaluation.

Where practitioners get caught

Calibrating on a fine-grained block and examining a coarse-grained casting or an austenitic weld at the same sensitivity. Real attenuation difference can exceed the acceptance margin, so flaws are under-called with complete confidence.

Where Attenuation fits in an inspection programme

A term is only useful when it connects to a decision. Attenuation appears in written procedures, in technique sheets, and in the records an owner or accreditation body reviews afterwards — which means the way it is defined in your documentation has to match the way it is applied on site. Where the two drift apart, audits find it. Atlantis writes and reviews procedures against the governing codes, trains inspection personnel to apply them, and builds the record-keeping that makes the evidence retrievable years later. Procedure development and code consulting · NDT training and certification · Ask us about your programme.

Related terms

  • Decibel (dB) — A decibel is a logarithmic unit of relative amplitude ratio used throughout ultrasonic testing; 6 dB equals a factor of 2 in amplitude, and 20 dB equals a factor of 10.
  • Beam Spread — Beam spread is the angular divergence of an ultrasonic beam beyond the near-field distance, governed by transducer diameter and frequency; it sets the lateral resolution of the inspection.
  • Distance-Amplitude Correction (DAC) — DAC is an ultrasonic calibration curve plotted on the A-scan display that shows expected echo amplitude from a reference reflector at varying depths, used to evaluate flaw size while compensating for beam spread and attenuation.

More physics terms

Back-Wall Echo · Near-Field · Far-Field · Dead Zone · Shear Wave · Longitudinal Wave · Surface Wave · Lamb Wave · Mode Conversion · Geometric Unsharpness

Where this comes up in practice

Terms like this one appear in three places that matter commercially: the written practice that governs how your personnel are qualified, the procedures and technique sheets that define how an examination is actually performed, and the evidence an auditor or client asks for when they want to know why an inspection was accepted. Getting the terminology right is the easy part; being able to produce the qualification record, the calibration traceability and the procedure revision that applied on the day of the inspection is the part that decides audits.

Atlantis NDT provides NDT training and certification against ASNT SNT-TC-1A and ISO 9712, ASNT Level III consulting for written practices and procedure approval, inspection management software that holds qualification, calibration and procedure-revision evidence in recoverable form, and an asset integrity platform that binds inspection results to the asset they describe. Browse the full NDT glossary or ask a Level III directly.

Attenuation splits into absorption and three distinct scattering regimes, and which one dominates depends on grain size relative to wavelength. Rayleigh scattering, where grains are far smaller than the wavelength, rises with the fourth power of frequency; stochastic scattering with the square; the diffusive regime is frequency-independent. That relationship, not a material name, decides the inspection frequency.

The practical consequence is that two carbon steels with the same specification attenuate differently after different heat treatment, and a centrifugally cast austenitic component defeats a technique that works on a forging of identical chemistry. ISO 16811:2014 puts transfer correction into the sensitivity-setting procedure for exactly this reason: the value is measured by comparing two-probe response over a fixed separation on the calibration block and on the component and taking the decibel difference. ASME Section V, Article 4 requires the difference between block and component to be addressed. Raising gain does not solve high attenuation, because grain noise rises with the signal and signal-to-noise ratio is what limits detection. For austenitic and dissimilar-metal welds the working answer is a lower frequency and a transmit-receive longitudinal probe, which trades resolution for penetration and avoids the mode conversion that makes shear-wave examination of those welds unreliable.

Source: ISO 16811:2014 — Non-destructive testing — Ultrasonic testing — Sensitivity and range setting; ASME BPVC Section V, Article 4 (2025 Edition)

Technically reviewed by Anoop Rayavarapu — ASNT NDT Level III (UT, RT, MT, PT, VT, ET) · API 653 · ISO 9001:2015 Lead Auditor
Attenuation mechanisms and which one is governing your setup
Loss mechanismFrequency dependenceCondition that makes it dominantHow it shows on the A-scanCompensation
AbsorptionRises with frequencyPolymers, composites, elastomer-lined and coated componentsUniform amplitude loss against a clean baselineLower frequency; raise reference gain with the noise floor checked
Rayleigh scatteringRises with the fourth power of frequencyGrain diameter far below wavelength — fine-grained steels examined at high frequencyAmplitude loss with a rising grass-like baselineHalving frequency cuts scattering loss sixteenfold
Stochastic scatteringRises with the square of frequencyGrain diameter comparable to wavelength — coarse castings, austenitic weld metalGrain noise climbing toward the evaluation levelLow frequency, focused or TRL probes, or a technique change
Diffusive scatteringFrequency-independentGrain diameter far above wavelengthBack-wall echo lost entirely, no usable penetrationUltrasonics is the wrong method here; radiography or an alternative technique
Beam divergenceFalls as frequency and element size riseSmall crystals over long sound pathsAmplitude falling with distance for reasons unrelated to the materialDAC or TCG curve; DGS sizing already accounts for it
Interface and couplant lossWeakRough, scaled, painted or sharply curved surfacesUnstable back-wall echo, amplitude changing with probe pressureSurface preparation, then transfer correction measured on the actual surface
Anisotropy and beam skewWeak, but strongly geometry-dependentAustenitic and dissimilar-metal welds with columnar grain growthBeam arrives where it was not aimed; back-wall echo displaced sidewaysTRL longitudinal probes, modelled scan plan, mock-up demonstration
Attenuation is measured, not looked up: it is the decibel difference between calibration block and component taken with two probes at fixed separation. A transfer loss below 2 dB is customarily recorded and ignored; above that it is added to the reference level before scanning.

How do I choose inspection frequency for a coarse-grained material?

Compare grain size to wavelength. When average grain diameter sits far below the wavelength, scattering rises with the fourth power of frequency, so halving frequency cuts scattering loss sixteenfold. Set frequency low enough that back-wall echo and grain noise separate by the signal-to-noise margin the procedure demands, then confirm the resulting wavelength still resolves the smallest flaw that must be found.

Why does raising the gain not fix a noisy austenitic weld?

Gain amplifies grain noise and flaw signal together, so signal-to-noise ratio is unchanged. Detection depends on that ratio, not on absolute amplitude. What improves it is lowering frequency, using a transmit-receive longitudinal probe that suppresses near-surface interface noise, focusing at the depth of interest, or moving to an imaging technique whose processing averages the random noise.

What probe works on dissimilar-metal and austenitic welds?

A transmit-receive longitudinal (TRL) probe at low frequency, commonly 1 to 2.25 MHz. Longitudinal waves attenuate and skew less than shear waves in anisotropic weld metal, and separating transmitter from receiver kills the interface noise that swamps a single-crystal probe. The trade is coarser resolution and a fixed focal depth, so the setup targets a defined depth band.

Does attenuation affect TOFD sizing the same way it affects amplitude-based sizing?

It affects both, differently. Amplitude-based sizing compares a signal to a reference level, so any attenuation difference between block and component biases the result directly. TOFD sizes from the arrival time of diffracted tip signals, which attenuation does not shift; what attenuation costs TOFD is the visibility of those weak tip signals, so a flaw is missed rather than mis-sized.

Is high attenuation itself a rejectable condition?

No code rejects material for attenuation. What high attenuation does is invalidate the examination: when grain noise rises to the evaluation level, or the back-wall echo cannot be maintained at the required amplitude, the technique has no demonstrated detection capability and the report cannot claim coverage. The correct outcome is a technique change recorded in the report, not an acceptance.

How much attenuation difference between block and part is too much?

No universal threshold exists; the procedure sets one. Common practice records a transfer loss below 2 dB and ignores it, and adds anything larger to the reference level. Once the measured difference exceeds the margin between the recording level and the noise floor, the block no longer represents the part and a representative block has to be made.