The Weld Fab Welding & Fabrication Planning Desk

SHEET TWF·G-06 REV 2026-08-20

Weld bead geometry and fatigue life: what the profile tells you

Fatigue cracks start where geometry concentrates stress. This sheet names the bead-profile features, explains why they matter and states what visual inspection cannot prove.

7 min read 2 sources evidence-led · source register below

Macro view of a polished weld-bead toe beside profile and radius gauges.
G-06 · generated illustration created for this site — visual reference, not inspection evidence

The profile vocabulary

Talk about weld quality eventually becomes talk about shape, and shape needs names. The cross-section of a fillet weld — the corner weld that joins two surfaces at an angle — carries most of the vocabulary:

FIG. 1 · Fillet weld cross-section, named

Cross-section of a fillet weld with its features labelled A vertical plate stands on a horizontal plate, joined by a fillet weld in the corner between them. Labels point to the two toes where the weld face meets each plate, the slightly convex weld face, the two legs measured along each plate from the corner, the root at the corner itself, and the throat measured from the root to the face. A band of heat-affected zone is shaded along the fusion boundaries in straw amber. Toe (upper) Toe (lower) Weld face Throat Root Leg (vertical) Leg (horizontal) where face meets parent metal — the classic fatigue-crack start root-to-face distance, dashed blue straw shading: heat-affected zone
FIG. 1 — fillet weld cross-section: toes, face, legs, root and throat, with the heat-affected zone shaded straw. Schematic proportions; a real profile is set by process, position and procedure.
  • Legs — the distances from the root out along each parent surface to the toe. The weld volume calculator’s equal-leg fillet is this measurement idealised into arithmetic.
  • Throat — the shortest distance from root to face; the dimension design calculations usually stand on. The design throat assumes an idealised triangle; the actual throat of a convex or deeply penetrated weld differs, which is one reason profile is worth caring about.
  • Toes — the lines where the weld face meets the parent metal. Remember these; the fatigue story is mostly about them.
  • Face, root, reinforcement — the visible surface, the corner (or, in a butt weld, the far side of the joint), and the weld metal standing proud of a butt joint’s surface. “Reinforcement” is the standard word and a small lie: the excess metal does not strengthen the joint, and its edges create exactly the sharp transitions this sheet is about.
  • Undercut — a groove melted into the parent metal at a toe and left unfilled: a built-in notch at the worst possible address.

Why profile matters for fatigue

Fatigue is failure by repetition: a crack that starts small at a point of concentrated stress and grows a little with every load cycle. Where stress concentrates is a matter of geometry — load flowing through a section crowds around every sharp change of shape, the way water crowds around a rock. A welded joint is a collection of exactly such changes, and the sharpest of them are usually at the weld toes, where the face meets the parent plate at an angle, sometimes with a micro-notch or undercut thrown in[S2].

That is why two welds with identical strength on paper can behave very differently under cyclic load. A flat or gently blended face flowing smoothly into the plate gives the stress somewhere to go; a sharply convex bead, a ropey cold lap, an undercut toe or a proud, hard-edged cap concentrates it. None of this changes the static arithmetic of throat sizes — it changes how many cycles the joint survives before a crack finds the notch. Fatigue design standards exist precisely to classify welded details and their expected lives; applying them is structural engineering, not welding folklore, and this sheet stops at the vocabulary.

What a visual check can see — and what it cannot promise

A careful eye and a set of gauges genuinely can verify a useful list:

  • leg lengths and apparent throat against the specified size, with a fillet gauge;
  • face shape — flat, convex, concave — and how the toes blend;
  • undercut, surface-breaking porosity, cold lap, spatter, arc strikes;
  • misalignment and distortion of the parts around the weld.

The boundary matters: a visual check cannot promise fatigue life. What the eye cannot reach decides too much — internal flaws, lack of fusion, penetration depth, residual stress, parent material condition, and above all the real load spectrum the joint will see in service. Procedure qualification testing exists because surface appearance alone is not evidence of mechanical performance: the qualification record behind a procedure is built on non-destructive examination and mechanical tests of an actual test piece[S1]. A good-looking bead is necessary evidence. It is never sufficient.

Profile comes from procedure

A bead profile is not chosen at the moment of welding; it is the output of everything a procedure fixes — process, consumable, parameters, position, preparation, technique. That is why the qualification system treats changes to its essential variables as invalidating: alter the material, process or thickness range and the evidence for what the weld — including its profile — will be is gone, and a new test is required[S1]. If the profile you are getting is not the profile the job needs, the conversation belongs upstream, with the procedure documents — not downstream with a grinder and hope.

Improvement methods and their limits

TWI lists weld-toe grinding and peening among methods used to improve the fatigue strength of welded joints[S2]. Selecting a method, executing it and taking credit for it in a fatigue calculation must follow the design standard for the structure and the competent engineer responsible for the detail. Use the TWI source as a technical introduction, then work to the governing standard under that engineer’s direction. This desk quotes no improvement factors and issues no verdicts on any weld — its contribution is the planning arithmetic, such as the theoretical volume and filler mass of the joint you are specifying, from figures you enter yourself.

Source register

  1. S1 Welding procedure — Job Knowledge 39 TWI Ltd Technical body accessed
  2. S2 Methods for improving the fatigue strength of welded joints TWI Ltd Technical body accessed