What is the difference between incomplete fusion and incomplete joint penetration?

Incomplete fusion (LOF) means the weld metal failed to fuse to the base metal sidewall or to a previous weld pass — it is a planar discontinuity lying between the weld and the groove wall. Incomplete joint penetration (ICP) means the weld did not penetrate the full depth of the groove; the root area is unfilled. Both are planar defects, but LOF is generally the more dangerous because it can be hidden in mid-joint thickness where UT beam angles may miss it.

Does AWS D1.1 accept any incomplete fusion in a CJP groove weld?

No. AWS D1.1:2025 requires complete fusion in CJP groove welds — any lack of fusion with the base metal sidewall or between passes is rejectable. The visual and UT acceptance criteria both treat LOF as a defect with zero tolerance in complete joint penetration applications.

Can ultrasonic testing reliably detect incomplete fusion?

UT is the primary volumetric method for detecting LOF in structural groove welds, but its reliability depends on beam angle, probe placement, and operator technique. LOF oriented parallel to the weld axis and perpendicular to the fusion face is the most difficult to detect because the planar discontinuity may not reflect the UT beam back to the transducer. PHASED ARRAY UT with angled beams improves detection probability. RT (radiography) misses LOF reliably when the defect is parallel to the beam.

Which WPS essential variables most directly affect the risk of incomplete fusion?

Travel speed, heat input, electrode angle, and current/voltage range are the most direct controls. Too high a travel speed reduces the time the molten pool has to wet the sidewall. Too low an amperage produces insufficient heat to melt the groove wall. Electrode or wire angle toward the groove wall affects where the arc energy is directed. All four are either essential or supplementary essential variables or are bounded by the WPS operating ranges.

Incomplete Fusion in Structural Welds: Causes and WPS Controls

Lack of fusion is one of the most serious weld discontinuities in structural work. Learn why it happens, how AWS D1.1:2025 limits it, and which WPS variables control it.

Incomplete fusion — also called lack of fusion (LOF) — is a planar discontinuity that occurs when weld metal fails to bond to the base metal or to a previously deposited weld pass. Unlike porosity (a volumetric defect) or undercut (a surface-accessible notch), LOF is a subsurface planar flaw that concentrates stress like a crack. In cyclically loaded or tension-zone structural welds, it is among the most dangerous discontinuities that can slip through a fabrication process.

Understanding what drives LOF, how AWS D1.1:2025 addresses it, and which WPS parameters control it is essential for any CWI responsible for procedure review or production inspection.

What incomplete fusion is and where it forms

An LOF discontinuity is a cold lap between weld metal and the adjacent fusion boundary — either the groove sidewall, the root face, or the surface of a previous pass. The weld metal solidified in physical contact with the adjacent metal but without the atomic bonding that constitutes fusion.

LOF forms in three characteristic locations:

Sidewall LOF. The weld bead does not penetrate to the groove wall and instead bridges across the gap, leaving an unfused plane running parallel to the groove face. This is especially common in narrow-gap groove welds, where the arc energy is concentrated in the center of the joint and the sidewall does not get enough heat input to melt.

Inter-pass LOF. The heat of the current pass is insufficient to remelt the surface of the previous pass, particularly at the toe of a convex bead. The incoming weld metal wets the convex crown but leaves the toes in cold-lap condition.

Root LOF. At the root of the joint, insufficient penetration combines with the geometry of the joint to leave the root face unfused. Root LOF is a first-pass problem: once the root pass is completed, the defect is buried and can only be found by volumetric NDE.

AWS D1.1:2025 acceptance criteria

AWS D1.1:2025 does not set a size threshold for incomplete fusion in complete joint penetration (CJP) groove welds — any LOF in a CJP weld is rejectable. This reflects the structural function of a CJP weld: it must carry the full cross-sectional stress without reduction, and any unfused plane reduces the effective throat and creates a stress concentration.

For partial joint penetration (PJP) groove welds, LOF in the deposited weld metal is also rejectable, though the effective throat calculation already accounts for the unfilled root — the issue is LOF within the deposited weld volume itself, not the designed-for root condition.

Fillet welds must be free from LOF with the base metal. The AWS D1.1:2025 visual acceptance criteria for fillet welds prohibit cracks and incomplete fusion, though the detection of subsurface LOF in a fillet weld requires NDE methods.

For visual inspection, LOF that breaks the surface (typically at the toe of a bead) is detectable with MT or PT. Subsurface LOF requires UT or RT, and UT is generally preferred for planar defects. For more on UT acceptance limits, see UT acceptance criteria for AWS D1.1 structural welds.

Root causes of incomplete fusion

Inadequate heat input

The arc must transfer enough energy to melt both the deposited wire or electrode and a layer of the adjacent base metal. If the arc energy is too low — either because amperage is below the qualified range, voltage is low causing a short arc, or travel speed is too fast — the groove wall does not reach its melting point before the weld pool passes.

In prequalified WPS procedures, the heat input range is implied by the amperage and voltage ranges recorded in the WPS. In qualified-by-test procedures, the PQR records the actual parameters, and the WPS allows ± tolerances around the qualified values. Dropping outside the lower bound without amending the WPS is a nonconformance; running at parameters inside the range but at the low end still carries LOF risk if technique is poor.

Electrode or wire angle

For SMAW, FCAW, and GMAW, the electrode or wire angle relative to the groove face directly controls where the arc energy is directed. When welding a groove weld in the vertical or overhead position with a beveled sidewall, the welder must angle the electrode toward the fusion face on each side alternately to ensure both walls receive arc energy. Maintaining a steep work angle (perpendicular to the joint axis) without oscillating toward the walls produces a bead that bridges the groove without bonding to the sidewalls.

WPS documents typically specify travel angle (the drag or push angle in the direction of travel) but less often specify work angle (the angle toward the groove face). For wide grooves or multi-pass work, the WPS should call out oscillation technique or weave pattern — specifically that the arc should be directed to the groove wall momentarily at the end of each weave to ensure sidewall fusion.

Root face geometry and fit-up

At the root, the gap between the two groove faces and the land thickness (root face height) determine how much the arc can penetrate before the weld pool freezes. A root face that is too tall prevents the arc from reaching the bottom of the joint and produces root LOF. A root gap that is too tight produces the same result by limiting penetration depth.

AWS D1.1:2025 prequalified joint details in Clause 5 specify root openings and root face tolerances for each joint type. Field fit-up must be confirmed to these tolerances before welding starts. Tight joints that don't meet the minimum root opening should be rejected and re-fit — a CWI who accepts a tight joint to avoid fit-up rework and then encounters root LOF in production NDE is facing a repair problem that is both more expensive and more complicated than the fit-up correction would have been.

For fit-up tolerances and their effect on the procedure, see groove weld fit-up tolerances under AWS D1.1.

Cold starts and stop-and-restart locations

When an electrode is restarted after a stop, the arc strikes cold metal before the crater of the previous weld is remelted. If the welder does not pause long enough at the restart point to remelt the crater and establish fusion with both the previous bead and the sidewall, LOF occurs at the restart. AWS D1.1:2025 requires restarts in CJP groove welds to be made such that the previous pass is remelted and fusion is established before forward travel resumes.

High-restraint joints — thick flanges, box columns, heavy splices — are particularly susceptible to restart LOF because the adjacent base metal acts as a large heat sink that cools the stop crater rapidly.

WPS variables that control LOF risk

The following WPS parameters are the primary process controls for LOF:

Amperage (current) range. The minimum qualified amperage sets the lower bound on arc energy. Running below the WPS minimum is a nonconformance; it also reduces heat input, shortens the fusion boundary, and raises LOF probability. The WPS must record the amperage range, and the welder must stay within it.

Voltage range. Arc voltage controls arc length (in SMAW) and arc characteristics (in GMAW/FCAW). Low voltage produces a short, tight arc that concentrates heat in the center of the weld pool rather than washing it across the fusion face.

Travel speed range. Fast travel speed reduces heat input per unit length. Many WPS documents bound travel speed by heat input formula rather than explicitly listing travel speed — when using heat input limits, the welder must keep track of bead length per unit time.

Technique (stringer vs. weave). AWS D1.1:2025 Table 6.6 (essential variables) treats a change from stringer bead to weave bead technique as an essential variable that requires WPS revision. This is because weave technique directly affects how the weld metal deposits across the groove — a change in technique changes how the arc energy is distributed to the fusion faces. For more on stringer vs. weave bead documentation, see stringer vs. weave bead WPS documentation under AWS D1.1.

Position. Position affects gravity's influence on the weld pool. In vertical-up welding (3G), the pool tends to sag away from the fusion faces if travel speed is too high. The WPS must qualify for the position used in production, and the parameter ranges used during PQR testing in each position represent the limits for that position.

Shielding gas (GMAW/FCAW-G). Shielding gas mixture affects arc characteristics and metal transfer mode. A change in gas mixture is an essential variable. For the relationship between shielding gas and arc penetration profile, see shielding gas essential variables under AWS D1.1.

CWI inspection focus for LOF prevention

Pre-weld: verify joint fit-up meets the minimum root opening and maximum root face tolerances for the WPS-specified joint type. Confirm the electrode or wire diameter and classification match the WPS. Check that the groove surfaces are free of mill scale, rust, oil, and moisture — contaminated surfaces suppress fusion.

During welding: for the root pass especially, watch for cold bridges — where the arc skips across the root gap without penetrating to the root face. Listen for arc characteristics (a crisp, consistent crackling usually indicates good fusion; sputtering or irregular arc suggests contact or parameter problems). Verify the welder is directing the arc to the sidewall on each oscillation, not just running beads down the center of the groove.

Post-weld: require NDE appropriate to the joint and application. CJP groove welds in tension zones on cyclically loaded structures require UT per AWS D1.1:2025. Review NDE reports for LOF indications before releasing the weld for the next operation.

Rule library based on AWS D1.1:2025; verify against your governing edition. The authority having jurisdiction (AHJ) or contract may specify AWS D1.1:2020 or an earlier edition.