Weld Examination and Geometric Responses

Learning outcomes

By the end of this chapter, the learner should be able to connect weld preparation and process with likely discontinuities; plan beam directions for fusion faces, root and heat-affected zones; distinguish common geometric and mode-converted responses; address backing, counterbores and restricted access; and define supplementary verification.

11.1 Build the weld model first

The Level 3 needs thickness, bevel angles, root face, root gap, cap and root profile, mismatch, backing, counterbore, branch geometry, material zones and accessible surfaces. As-built information is preferable to nominal details. Weld-process knowledge helps predict lack of fusion, incomplete penetration, slag, porosity, cracking and profile conditions.

The scan plan should direct sound toward relevant fusion faces and likely crack planes. More angles do not automatically mean more coverage; each beam should have a stated purpose and usable signal-to-noise ratio.

11.2 Root and backing responses

The root region can produce strong responses from penetration profile, mismatch, suck-back, counterbore and backing. A backing strip may trap and reverberate energy, producing delayed signals beyond the nominal wall. A counterbore can return or mode-convert energy near the expected crack zone.

Discrimination uses probe movement, response dynamics, calculated position, opposite-side or opposite-direction scans, alternate angles, straight-beam checks and knowledge of the actual geometry. A response should not be declared non-relevant merely because it occurs where geometry is expected; flaws also enjoy convenient hiding places.

11.3 Surface and access limitations

Spatter, scale, reinforcement and curvature can vary coupling and wedge contact. Restricted access may prevent full skip distance or transverse scanning. High-low alignment changes beam path. If scanning from one side only, the technique must demonstrate the affected fusion faces and root zones or declare exclusions.

11.4 Austenitic and dissimilar-metal welds

Columnar grain structure can redirect, split and attenuate beams. Longitudinal-wave techniques, lower frequency, specialised wedges, dual-matrix arrays or transmit-receive arrangements may improve capability. Qualification should use representative parent, weld and buttering materials with realistic geometry.

11.5 Interpretation matrix

A response that repeats at the predicted sound path along an entire known feature strongly supports a geometric source, unless a continuous discontinuity is physically plausible. Sharp amplitude change with small probe rotation supports directional or planar behaviour, although some geometry can also be directional. Detection from opposing directions at a consistent point strengthens location confidence.

A response that follows a counterbore shown on the drawing is likely to include a geometric contribution, but that evidence does not exclude cracking beside the feature. Tip-like satellite responses may support crack extremities only after their paths, repeatability and relationship to the main response have been confirmed. No single observation is decisive; interpretation relies on the combined pattern.

No single observation is conclusive. Interpretation should combine independent features.

Failure modes and decision control

Weak practice commonly includes designing from nominal weld symbols without as-built checks; using one angle for every fusion face; assuming all root-zone responses are geometry; ignoring beam spread into adjacent structures; applying carbon-steel models to austenitic weld metal; and concealing one-sided access limitations.