Sizing, Uncertainty and Fitness for Decision

Learning outcomes

By the end of this chapter, the learner should be able to compare amplitude, boundary, tip and time-based sizing methods; distinguish measured response size from physical flaw size; quantify repeatability, bias and uncertainty; select methods suited to the reflector and decision; and report limitations without false precision.

14.1 What is being sized?

Length along a scan, through-wall height, depth, ligament and remaining wall are different measurands. The Level 3 must define which is required and which ultrasonic feature represents it. Apparent response extent is not automatically physical extent.

14.2 Amplitude-drop methods

Fixed-amplitude-drop techniques move the probe from maximum response until amplitude falls by a specified amount. The drop may correspond to a fraction of peak amplitude: 6 dB is approximately one-half amplitude and 20 dB is one-tenth. The method's validity depends on reflector size, shape and directivity relative to the beam. It can estimate beam-defined edges for larger reflectors but may be poor for small or irregular cracks.

Maximum-amplitude techniques track the probe position of peak response along the reflector. Last-significant-echo approaches use a stated threshold to define response endpoints. Results depend on gain, noise, smoothing and threshold definition.

14.3 Tip diffraction and time-based sizing

Crack tips can produce diffracted responses that support depth or height measurement using arrival time. TOFD and pulse-echo tip techniques can be less dependent on specular amplitude, but tip identification, phase, dead zones, velocity and geometry remain critical. A missing tip is not proof that the crack ends at the visible response.

14.4 Equivalent reflector sizing

DAC comparison or DGS methods express response relative to reference reflectors. They support evaluation or equivalent-size estimates under controlled assumptions. They do not directly describe crack height or area. Report terminology must prevent an equivalent flat-bottom-hole or disc result from being read as a physical flaw dimension.

14.5 Measurement uncertainty

Uncertainty contributors include time-base accuracy, velocity, probe index, angle, encoder position, surface position, thickness, beam width, response selection, analyst repeatability, algorithm and reference-standard tolerance. Some create random variation; others create systematic bias.

Repeatability trials, known-reflector comparisons and independent analyses can estimate performance. When the acceptance boundary is close to the measured value, the applicable decision rule must state how uncertainty is handled. The examiner should not silently round a marginal result toward acceptance.

14.6 Reporting precision

Report only the precision supported by the method. A value of 3.47 mm is not more credible than 3.5 mm if the demonstrated error is plus or minus 2 mm. Include sizing method, calibration basis, scan direction, result, uncertainty or demonstrated error where required, and known limitations.

Failure modes and decision control

Weak practice commonly includes reporting response length as crack length without qualification; choosing a drop method because it gives a preferred answer; missing one tip and assuming zero ligament; treating DGS equivalent size as physical area; quoting excessive decimal places; and ignoring uncertainty at an acceptance boundary.