3.5 Geometric Discontinuities: Undercut, Overlap, and Profile
Key Takeaways
- Undercut limits (AWS D1.1): 0.010 in. for cyclically loaded, ~1/32 in. for statically loaded members
- Overlap is surface lack of fusion — a stress-concentrating notch, generally not acceptable
- Groove-weld reinforcement limits: 1/8 in. for material up to 1 in. thick, 3/16 in. over 1 in.
- Arc strikes must be ground smooth and visually inspected for cracks; spatter must be removed
- Theoretical fillet throat = 0.707 × leg size; most geometric discontinuities are found by VT
Surface and Profile Discontinuities
Geometric discontinuities are conditions of the weld's shape and surface — they change the cross-section or surface profile and act as notches. Most are surface-breaking, so visual testing (VT) is the primary detection method, making them the bread-and-butter of CWI Part B (Practical) gauge work. They are generally less severe than cracks or lack of fusion, but several are routinely rejectable, and undercut and overlap are among the most frequently tested topics on the whole exam.
Undercut
Per AWS A3.0, undercut is "a groove melted into the base metal adjacent to the weld toe or weld root and left unfilled by weld metal." It removes load-carrying section and leaves a sharp notch that initiates fatigue cracks — especially dangerous in cyclically loaded structures.
- Toe undercut — at the weld face/base-metal junction.
- Root undercut — on the root side of the joint.
Causes: excessive current (amperage too high), excessive arc length/voltage, travel speed too fast, improper electrode (work) angle, and excessive weave that overheats the toe.
AWS D1.1 acceptance (visual):
| Loading | Undercut limit |
|---|---|
| Cyclically loaded, member transverse to tensile stress | ≤ 0.010 in. (0.25 mm) deep |
| Statically loaded | ≤ 1/32 in. (1 mm) for any length, OR ≤ 1/16 in. (1.6 mm) for cumulative lengths up to a code-specified amount |
The tighter cyclic limit exists because fatigue cracks readily nucleate from the undercut notch.
Overlap (Cold Lap)
Per AWS A3.0, overlap is "the protrusion of weld metal beyond the weld toe or weld root" — weld metal that has rolled over onto the base-metal surface without fusing to it. It is essentially surface lack of fusion and creates a sharp re-entrant notch at the toe. Overlap is generally not acceptable because it both forms a stress-concentrating notch and signals probable lack of fusion underneath. Causes: travel speed too slow, amperage too low (the puddle runs ahead of the arc), improper electrode angle, and oversized electrode for the joint.
Reinforcement, Underfill, and Convexity
Weld reinforcement is weld metal in excess of that required to fill the joint. Some face reinforcement is normal; excessive reinforcement creates an abrupt toe transition (stress concentration) and wastes filler. AWS D1.1 limits groove-weld face reinforcement to 1/8 in. (3 mm) for material up to 1 in. thick and 3/16 in. (5 mm) for thicker material; the weld must also blend smoothly to the base metal.
Underfill is the opposite — a groove-weld face (or root) below the base-metal surface, leaving the joint under-thickness and rejectable. Excessive convexity (fillet face too bulged) creates a sharp toe; excessive concavity reduces the effective throat and undersizes the weld.
Fillet-Weld Throat and Size
For an equal-leg fillet weld, the theoretical throat = 0.707 × leg size (cos 45° = 1/√2 ≈ 0.707). A 1/2-in. fillet therefore has a 0.354-in. theoretical throat. Undersized legs or a throat below the specified minimum (often from excessive concavity) make the weld rejectable.
Mismatch (High-Low), Arc Strikes, Spatter, and More
| Discontinuity | Description | Significance / treatment |
|---|---|---|
| Mismatch / high-low | Misalignment of the two members' surfaces at a butt joint | Creates a stress concentration and eccentric load path; limited by code (e.g., D1.1 caps offset of abutting parts) |
| Arc strike | Localized remelted/heat-affected spot from striking the arc outside the joint | Leaves a hard, brittle (martensitic) HAZ and possible micro-cracks; D1.1 requires it be ground smooth and VT'd for cracks, especially on cyclically loaded members |
| Spatter | Metal droplets thrown onto the surface | Usually a workmanship/cosmetic issue; must be removed because it can mask other discontinuities and start corrosion |
| Excessive convexity | Bulged fillet face | Sharp toe — stress riser |
| Excessive concavity / insufficient throat | Sunken fillet face | Throat below minimum = undersized weld |
| Melt-through / excessive root reinforcement | Excess root-side metal | Limited by code; abrupt root profile concentrates stress |
Discontinuity → Cause → Detection Quick Reference
| Discontinuity | Typical cause | Primary detection |
|---|---|---|
| Undercut | High current, fast travel, long arc, bad angle | VT (and undercut gauge) |
| Overlap | Slow travel, low amperage, bad angle | VT |
| Excessive reinforcement / convexity | Too much fill, low travel speed | VT (fillet/weld gauge) |
| Underfill / insufficient throat | Too little fill, excessive concavity | VT (gauge) |
| Mismatch (high-low) | Poor fit-up/alignment | VT (Hi-Lo gauge) |
| Arc strike | Striking arc off the joint | VT, then MT/PT for cracks |
| Porosity (surface) | Gas/moisture/wind | VT; subsurface by RT |
| Slag inclusion | Poor interpass cleaning | RT / UT |
| Incomplete fusion / penetration | Low heat, bad angle, joint geometry | UT (RT for IJP) |
| Cracks | Hydrogen, restraint, S/P films | VT/MT/PT (surface), UT (subsurface) |
Exam essentials: Memorize the D1.1 undercut numbers — 0.010 in. cyclic, 1/32 in. static — and reinforcement limits 1/8 in. (≤1 in. thick), 3/16 in. (>1 in.). Overlap is generally not acceptable (surface lack of fusion). Arc strikes must be ground and re-inspected for cracking. Fillet throat = 0.707 × leg.
Per AWS D1.1, what is the maximum allowable undercut depth for a cyclically loaded member transverse to the applied tensile stress?
Why is overlap (cold lap) generally treated as unacceptable?
What does AWS D1.1 require for an arc strike found on a cyclically loaded member?
The theoretical throat of an equal-leg fillet weld equals: