Penetration Welding Definition: CJP vs PJP Explained

A fabricator cuts identical V-grooves in two 12 mm plates, runs the same wire at the same amperage, and one job comes back from radiography rejected while the other passes. The welder swears the beads look identical. They aren’t: one weld fused completely through the root, and the other left a sliver of unmelted base metal behind. That gap between “looks like a weld” and “performs like a full-strength weld” is exactly what penetration welding describes.

What Is Penetration Welding? A Clear Definition

Two plates get the same V-groove cut, the same wire, the same amperage and the same travel speed. One job comes back from the radiographer rejected. The other passes. The welder swears the beads look identical, and from the outside they do. But one weld fused all the way through the root and achieved complete root fusion, while the other left a sliver of unmelted base metal trapped in the bottom of the groove. That gap between a weld that looks finished and a weld that actually behaves like a full-strength connection is exactly what penetration welding describes.

Here is the direct answer. Penetration welding is not a process name, a special machine or a different heat source. It describes the result: a joint in which the weld metal fuses completely through the thickness of the base metal at the root, with full root fusion. North American codes prefer the term complete joint penetration (CJP) for that outcome. British and European practice calls the same thing a full penetration weld. The opposite case has its own name in both traditions: partial joint penetration (PJP), where fusion depth is deliberately less than full. TWI defines a full penetration weld as a welded joint where the weld metal fully penetrates the joint with complete root fusion, and notes that the preferred US term is complete joint penetration in AWS D1.1. The same source defines partial penetration as a weld in which the fusion penetration is intentionally less than full penetration, with the preferred US term partial joint penetration.

That wording matters. A penetration weld and a partial penetration weld can look nearly identical from ten feet away. One is the design intent, the other is a defect. Naming the result gives inspectors, engineers and welders a shared vocabulary for telling them apart.

Term	What it means	Where you will meet it
Full penetration weld	Weld metal fully penetrates the joint with complete root fusion	British and European practice, plus many ISO-based specifications
Complete joint penetration (CJP)	The same condition, using the preferred US term	AWS D1.1 and most North American codes and drawings
Partial penetration weld / partial joint penetration (PJP)	Fusion penetration is intentionally less than full penetration	Where only part of the joint thickness is meant to be joined, with the depth specified

Now the part most people get wrong. Penetration is not something a process hands you automatically. Almost every arc process can produce a penetration weld, because the outcome depends on joint design, procedure and execution. Shielded metal arc welding, gas metal arc welding, flux-cored arc welding, gas tungsten arc welding, submerged arc welding, electroslag welding, laser welding and electron beam welding can all be taken to full penetration. Two practical references show how much the heat source and filler metal change while the fusion requirement stays the same: the practical guide to aluminum MIG welding and aerospace TIG welding standards and technique. The idea runs the other way too, which is why why SMAW still wins in so many shops has little to do with penetration being a stick-only property.

It is worth pausing on what welding actually is, because the distinction from brazing and soldering is the foundation everything else rests on. TWI defines welding as a fabrication process whereby two or more parts are fused together by means of heat, pressure or both, forming a join as the parts cool. What separates welding from brazing and soldering is that welding is a high-heat process that melts the base material, while brazing and soldering use a filler metal that flows between close-fitting parts without melting the parent metal. A completed welded assembly is called a weldment.

Braze a fitting onto a tube and the difference is obvious at the joint: the tube itself never became liquid. In a fusion weld, it does. That is why penetration depth only makes sense in fusion welding, and why a joint made by torch brazing and soldering can never be upgraded into a CJP weld by adjusting the flame.

One more point worth having in your head early: penetration does not automatically make a joint stronger. It makes the joint behave like the material it joins, and that behaviour can exceed the original plate. TWI notes that a weld produced by a suitable process can end up stronger than the parent metal, because the weld metal is a fresh casting with its own chemistry and microstructure. A properly executed full penetration weld in a fatigue-critical structure is routinely as strong as, or stronger than, the plate it joins.

All of this sits inside a serious standards framework. The American Welding Society has authored more than 350 standards covering welding practices and procedures, and those documents are where CJP and PJP get their precise definitions. Browse the current list at the AWS codes and standards page.

With the vocabulary straight, it is time to look at the anatomy of the weld itself, because penetration, root fusion and reinforcement are three different things that get confused every day.

Penetration, Root Fusion, and Reinforcement: Untangling the Terms

Most weld disputes are vocabulary disputes. One inspector calls the joint full penetration, the welder points at a bead that looks full penetration, and the engineer checks the drawing, which never said. Name the parts of a joint precisely and most of the confusion disappears.

The basic vocabulary

Parent metal is the base material being joined. It is not consumed by the weld; it melts back a short distance and mixes with the filler. Filler metal is what you add: stick electrode, continuous wire, or flux-cored powder. Weld metal is the result after filler and melted parent metal mix and solidify. That mixture matters, because the weld is a new alloy rather than simply added material.

While the arc is live everything is liquid. That liquid is the weld pool, where filler melts, base metal melts, and chemistry homogenises. When it solidifies it becomes weld metal, and the lowest point of the pool, where it melted deepest into the joint, is the weld root.

Root fusion describes that bottom edge: did the weld metal fuse completely with the parent metal across the full width of the joint, with no cold laps or unfused slivers? Penetration depth is how far the pool reached into base metal at the root. Fused metal, heat-affected metal and untouched parent metal together form the weld zone.

Reinforcement, and why the word is a problem

Most groove welds carry slight reinforcement: a little extra weld metal sitting proud of the parent surface at the crown. It looks tidy and it is normal. But TWI points out the term is inappropriate, because metal above the parent surface does not make the joint stronger. Sizing a weld for design means working to the design throat thickness, which measures the weld metal plus the parent metal fused into it, and excludes the cap on top. A fat bead is not a strong weld; a modest bead with perfect fusion is. Ground off or left proud, reinforcement does not change design strength.

Toe, face, fusion line and the heat-affected zone

The weld face is the weld metal surface exposed to the joint. The weld toe is where weld metal meets the parent surface. TWI describes toes as points of high stress concentration and common initiation points for fatigue and cold cracks, which is why they must blend smoothly into the parent metal rather than end in a sharp notch.

The fusion line is the boundary between weld metal and the heat-affected zone. TWI notes it is a non-standard term for weld junction, the traditional name for the same boundary.

The heat-affected zone is the band of parent metal heated enough to change its microstructure without melting. For conventional steels TWI breaks it into four zones: grain-coarsened, grain-refined, partially transformed (intercritically heated), and tempered. Each behaves differently in service, so the HAZ often governs joint life even when the weld metal is sound.

Reading the cross-section

 weld metal (cap / reinforcement)
        ______________________________
  parent | | parent
  metal  |        ^  fusion line        | metal
         |   HAZ  |      |   HAZ         |
         |________|______|______________|
              root (fusion here = CJP)

Everything above the root line is weld metal, the narrow bands beside it are HAZ, and the parent metal beyond that is unchanged. Where the pool melted down into the base metal and joined it, that is where penetration actually lives.

Heat tint, sensitisation and what colour tells you

On stainless steel, oxide colour in the HAZ maps to peak temperature. TWI’s scale for AISI 304 in open air runs from light yellow at 290 degrees C (550 F), through blue at 540 degrees C (1000 F), to dark blue at 600 degrees C (1110 F). It is a practical field read on how hot the HAZ actually got.

That matters because of sensitisation, also called weld decay. In stainless HAZs, chromium carbides precipitate around grain boundaries and local chromium content can drop below 10.5 percent, so the steel can no longer form its passive film and intergranular corrosion attacks the boundaries. It is a metallurgical warning written in oxide colour, and the reason stainless heat input and interpass temperature are tightly controlled. For a shop welding stainless or aluminium, cleaning before you strike an arc is metallurgical control, not housekeeping: see our guide to cleaning aluminium for welding, then pair it with the practical notes in aluminium MIG welding, where the HAZ behaves very differently from steel.

Runs, passes and layers

A run or pass is a single length of weld made in one operation; a layer is weld metal built up from one or more passes. A multi-pass CJP weld is a stack of layers, and each pass must remelt the previous one to fuse with it. That remelting is also why multi-pass welds achieve deeper, more consistent penetration than a single pass: the pass two pool washes into the partially fused zone of pass one.

Complete Joint Penetration vs Partial Joint Penetration

Two terms decide what a weld has to accomplish before anyone lights a torch. TWI defines a complete joint penetration (CJP) weld as one in which the weld metal fuses completely through the thickness of the base metal at the root, with complete root fusion. A partial joint penetration (PJP) weld, by the same source, is one in which fusion penetration is intentionally less than full penetration. That single word, intentionally, is the entire argument, and it is why the penetration welding definition on a job drawing deserves a careful read.

Aspect	Complete Joint Penetration (CJP)	Partial Joint Penetration (PJP)
Definition	Weld metal fuses completely through the thickness of the base metal at the root, with complete root fusion	Fusion penetration is intentionally less than full penetration, to a depth specified by the designer
Purpose	Creates a continuous load path through the full thickness of the joint	Transfers load through a deliberately limited fusion area, sized by the throat
Where used	Pressure piping and vessels, primary structural members, moment connections, fatigue-loaded joints	Non-critical attachments, enclosure seams, box sections, pads, and secondary framing
How verified	Usually by ultrasonic testing, with radiography where UT is impractical	Often by visual inspection; UT or MT where the design requires proof of a stated fusion depth
Common processes	GMAW, FCAW, SMAW, GTAW, SAW, submerged arc, laser and electron beam	Any arc process capable of a controlled, repeatable fusion depth, including fillet welds
Risk if misread	Incomplete penetration becomes an acceptance rejection	A weld intended as PJP gets inspected and rejected as if it were CJP

When Partial Penetration Is the Right Answer

A PJP weld is not a failed CJP weld with a kinder name. It is a specified condition with a stated fusion depth, and it is entirely legal in many structures. Common cases include:

• Brackets, stiffeners and reinforcing pads attached where only part of the plate thickness needs to be joined and no full-strength load path is required.
• Box sections and enclosures where the seam only needs to keep panels aligned and slow the spread of a fire or leak, not carry primary load.
• Flange and channel connections in secondary framing, where the joint is sized by the weld throat rather than by continuity of the parent metal.
• Lap-plus-flange details that are not load-carrying, such as cosmetic trim, wear strips or non-structural stiffeners.

The point is that a PJP weld is a deliberate, drawn, specified fusion depth. The designer says how deep it needs to go, the procedure is qualified to deliver that depth, and the inspector measures it. Nothing is being cut short. For the process side of setting a controlled, repeatable bead on thin aluminum, see our guide to MIG welding aluminum.

What a CJP Weld Actually Requires

A CJP weld has to deliver three things at once: fusion through the full thickness of the joint, complete root fusion at the back of the groove, and a procedure qualified for that joint. Miss any one and you do not have a CJP weld, however good the bead looks.

The joint geometry is what makes that achievable. Common groove shapes include the square groove (flat, no bevel, ideal for thin material and submerged arc), the single-V (the workhorse for plate and pipe), the double-V (two single-Vs back to back, doubling welding speed on thick sections), a bevel, the U and J grooves (where a single pass must deposit a large volume of metal), and the flare-bevel, where the groove is formed by flaring the tube or pipe wall rather than cutting it. Backing matters too. A steel or ceramic backing bar behind the root can force the molten pool to fuse into the back of the joint, a legitimate way to guarantee root fusion on a square groove or a thin wall.

Which Code Governs Which Decision

AWS D1.1/D1.1M, the Structural Welding Code for Steel, is described by AWS as one of the most-referenced codes in the world. It governs structural detailing, the allowable design strength of a CJP groove weld versus a PJP groove weld, and the inspection criteria for the finished product. AWS currently lists the edition as AWS D1.1/D1.1M:2025-AMD1.

ASME’s Boiler and Pressure Vessel Code plays a different role. ASME states that its BPVC standards provide the single largest source of technical data used in the manufacturing, construction and operation of boilers and pressure vessels, with a new edition released every two years; the 2025 edition is the current one. Within that code, Section IX is the qualification standard: it governs how welders and welding procedures are qualified, not the acceptance criteria for a finished weld. AWS standards, by contrast, are developed by committees of experts and approved by the American National Standards Institute (ANSI).

If you carry responsibility for these joints, certification matters as much as the weld itself. AWS offers CWI and SCWI endorsement pathways covering Structural Steel (D1.1), Bridge Welding (D1.5), Pipeline (API 1104) and ASME BPVC Section IX, alongside the B31.1 and B31.3 piping codes. Those endorsements are the formal recognition that someone can read a drawing, qualify a procedure and judge a weld against the code rather than by eye. Details on the specific inspection methods sit further down in this guide, and the certification side of the trade is where that proof starts.

How Weld Penetration Actually Happens: Groove, Root Gap and Heat Input

Penetration is not a machine setting but the result of a molten pool forming, melting into base metal and freezing into a bond metallurgically continuous with the parent metal. The same wire, amperage and joint can produce one passing weld and one that returns from the radiographer as a rejection.

The Pool, the Arc and the Root

An arc does not sit in the metal. It melts the consumable and part of the base metal around it; the two liquids mix into the molten weld pool, and the arc’s job is to supply the heat.

As the torch travels, the pool is replenished at the front and freezes behind it. That frozen metal must be strong enough to carry the load, ductile enough to absorb strain and compatible with the base metal it joins.

How far the pool melted into the base metal at the root is the penetration depth. Melt the root face completely and you have complete root fusion and complete joint penetration. Stop short and you have incomplete penetration, a defect in a CJP weld however good the bead looks.

The Variables a Welder Actually Controls

Penetration depth is not a single knob but the combined result of a dozen inputs, all managed at once.

• Groove angle and face angle. A wider groove lets the arc reach base metal and raises penetration, but costs filler and heat. A narrow groove traps the arc, risking lack of sidewall fusion.
• Root opening. The gap between the joint edges at the root. Too narrow and the electrode cannot reach the bottom, giving lack of root fusion; too wide brings excess heat and burn-through. Procedures specify a root opening and root face.
• Electrode or wire diameter. A fat electrode dumps more heat but reaches the root poorly; a thin one reaches the bottom but cannot supply enough melting capacity. Oversizing causes incomplete penetration.
• Current and voltage. Current sets melting rate and penetration depth; voltage, tied to arc length, sets bead width and pool shape. More voltage means a longer arc and more total heat.
• Travel speed. The biggest lever on heat input. Slow travel concentrates the same energy over less length; faster travel gives the pool less time to melt the base metal.
• Thickness and preheat. Cold thick sections chill the pool until it freezes before wetting the root. Preheat and interpass temperature manage this; a cold start on a hot joint is the worst of both worlds.
• Bead placement. Stringers give the cleanest, most repeatable fusion. Weaving spreads heat wider, helping on thick sections and hurting a narrow, deeply penetrated root pass.
• Polarity. DCEP puts more heat into the workpiece than AC and, at the same current, typically gives greater penetration than DCEN. The procedure must state which to use.

Heat Input: Energy per Unit Length

Heat input describes the energy deposited per millimetre of travel: Heat input (J/mm) = (Voltage x Current x 60) / Travel speed (mm/min). The 60 converts minutes to seconds. It ignores arc efficiency, but it gives a consistent basis for comparing welds, setting procedure limits and predicting the heat-affected zone.

TWI on the heat-affected zone puts the consequence plainly: low heat input cools faster and gives a smaller HAZ, while higher heat input and slower welding speed enlarge it in the same material. TWI’s overview of welding makes the same point in process terms.

Process Comparison

Each process delivers heat differently, and that difference decides how easily you get deep penetration.

Process	How it heats, and what that means for penetration
SMAW (stick)	Per the overview of welding processes, the flux coating shields the weld by producing carbon dioxide gas, and the electrode core is the filler, so no separate rod is needed. Voltage follows the welder’s arc length.
GMAW (MIG)	Continuous wire feed means no electrode changes, faster deposition and travel than stick, plus good pool control. See our MIG welding guide and flux core guide.
FCAW	Tubular wire with a powder core, allowing higher welding speed and greater penetration than GMAW. Self-shielded flux-core wire debuted in 1957, reaching pipelines and structural steel.
GTAW (TIG)	Non-consumable tungsten and an inert gas shield give precise pool control, which is why it dominates thin stainless and aerospace work. See TIG welding standards and techniques.
SAW and ESW	SAW blankets the arc in flux, ESW uses a conductive slag. Both give deep single-pass capability on thick sections, hence their hold on heavy plate and large vessels.
Laser and electron beam	Electron beam welding, introduced in 1958, made deep, narrow welds possible with a concentrated heat source; laser welding works the same way. See laser welding systems.

Why the Electric Arc Changed Everything

Before the arc, most structural and pressure welding used oxy-fuel gas, whose flame runs about 3,100 degrees C (5,610 F) near the tip and spreads heat widely, so welds cool slowly and carry greater residual stress and distortion. The arc gave fabricators a compact source that could be aimed down a groove; GMAW followed in 1948 and GTAW was perfected in 1941.

Why You Cannot Dial Penetration In Directly

Which Joints Give You Penetration? Groove, Fillet, Plug and Slot Welds

Here is the thing most people get wrong: penetration is not a joint type, it is a result that only some joint types can deliver. You can weld almost any configuration you like, but only certain geometries let the weld metal reach all the way through the base metal at the root. So before you pick a process, you need to pick a joint, and the joint dictates what kind of weld you are actually trying to make.

According to TWI’s welding definitions, joint types are sorted by the angle between the two parts being joined:

Joint	Angle between the parts	Typical weld
Butt joint	135 to 180 degrees, inclusive	Groove weld (CJP or PJP)
T joint	More than 5 up to and including 90 degrees	Fillet weld, or groove on prepared edge
Corner joint	More than 30 but less than 135 degrees	Fillet weld
Edge joint	0 to 30 degrees	Usually a groove or edge weld
Lap joint	0 to 5 degrees	Fillet, plug or slot weld

Add the cruciform joint, which is simply two T joints meeting at 90 degrees, and you have covered the practical range. TWI groups these along with weld features such as the weld root, toe, face, run and layer to give shops four ways to classify what they are building: joint configuration, penetration, accessibility, and features of the completed weld.

Groove welds: where penetration happens

A groove weld is a butt-type joint in which the edges of two parts are prepared with a bevel, groove or flange so the surfaces are separated enough for the welding process to reach the root. That preparation is what creates the room for the arc to reach down and melt through the thickness. A groove weld can be a complete joint penetration weld or a partial joint penetration weld, and it can be a butt joint, a T joint with a prepared edge, or an edge joint.

For a butt joint between two flat plates, the classic setups are square groove, single V, double V, bevel, U and J. Double V and flare bevel grooves get deep penetration with less deposited metal, and a backing bar or run-on plate can support the root while it melts.

Fillet welds: strength without penetration

A fillet weld is a triangular joint between two surfaces that are approximately at right angles to each other. It carries load by shear across its throat, the shortest distance from the root to the face of the weld. It does not rely on base metal penetration, and that is a perfectly good thing. A fillet is the workhorse of structural and sheet fabrication because it is fast, it tolerates a range of fit-up, and it is straightforward to inspect by eye and by fillet gauge.

The important catch: if a fillet is specified as the load path, the designer has sized it for shear across the throat, and the base metal continuity you would get from a CJP groove weld simply is not there. Do not casually substitute one for the other.

Both are unusual in heavy fabrication and common in sheet metal and plate work. The hole has to be sized so the metal actually melts into it and bonds to the surface below. Too small and you get a cold, unmelted plug sitting on top. Too large and you get a hole full of loose slag. Also worth knowing: AWS treats most plug and slot welds as detail welds that are not considered to carry primary load, so you cannot lean on them structurally.

Worked examples: joint choice drives process choice

• 6-inch Schedule 40 pipe butt joint in pressure service. This is a groove weld, normally CJP with a matching or +1/2 inch gap, welded by a qualified procedure. Root fusion is mandatory, and the weld gets radiography or ultrasonic testing. On stainless or nickel alloys, GTAW root passes with TIG welding of stainless and aluminium are common so you can control the root and keep the backing clean.
• Structural moment connection, beam flange to column. This is a groove weld detail, and it usually calls for CJP on one side with a reinforcing detail at the other. Fillet welds alone will not develop the flange stress. This is standard AWS D1.1 territory, and the joint geometry plus the process go together: you might use stick welding with its forgiving heat control in the field or flux core for speed.
• Shelf bracket welded to a tube. Often a simple fillet, sized to the load and to the tube wall, with the option of a plug weld where a gusset sits inside. No penetration needed, no deep groove to prepare.
• Tube to baseplate where the designer wants the full tube wall engaged. This can be called out as a PJP groove weld, with a specified depth of penetration into the tube wall and often a matching insert or backing ring. The drawing has to say PJP and state the required fusion depth, because an unqualified CJP would be a defect and a plain fillet would not meet the design intent.

Once the joint is fixed, the process falls into place. A thin stainless or aluminium sheet bracket calls for MIG welding on aluminium with the right gun and wire, a thin-gauge decorative detail might go to TIG for control, and a thick structural joint on a job site goes to stick or flux core. None of those choices is about penetration on its own. Penetration is what the joint design and the qualified procedure are supposed to deliver, and the joint tells you which NDE method you will need later.

Note: Groove, fillet, plug and slot welds are all named features of a completed weld under AWS terminology. The name tells you the geometry, not the strength. Only the specification tells you whether penetration is required.

Why Penetration Matters: Strength, Weight and the End of Rivets

Here is the engineering argument in one sentence: a complete joint penetration weld turns two pieces of steel into one continuous piece of steel. No rivet head to tear off, no shear plane to slip along, no interface for a crack to hide in. Load enters the plate on one side, crosses the fused metal and leaves on the other as if the joint were never there.

That is fundamentally different from a fillet weld or a partial joint penetration weld. Those load only a limited area, so the designer sizes the joint by its throat rather than by the continuity of the parent metal. Get that difference clear and most of the reasoning behind a weld drawing falls into place.

Joint	How load crosses the joint	How the designer sizes it
CJP groove weld	Continuously, through the full joint thickness	Like parent metal; effective throat based on joint thickness and matching filler
PJP groove weld	Through a specified depth only, then stops	Specified depth of fusion plus any reinforcement
Fillet weld	Shear across the throat of the triangular deposit	Throat dimension, roughly leg size times 0.707

A CJP weld also saves weight, because doubler plates, angles and closure plates are no longer needed to build up the load path. A 12 mm plate welded full through can do the job of a 12 mm plate plus a 6 mm doubler, in less steel and less space. On anything that moves through air or water that saving compounds, which is one reason aluminum boat welding leans so heavily on full-penetration seams: a welded aluminum hull carries its loads with far less structure than a riveted one.

The Riveted Era and What It Cost

For most of industrial history, joints were riveted, and a riveted joint is a stack of discrete fasteners, each one a separate load path and a potential failure point. Boilers and ship hulls built that way failed with terrible regularity, and the failures were not subtle. Wikipedia records that the Sultana boiler failure killed more than a thousand passengers, and that Titanic sank in part because of failures in its riveted hull, where rivets failed and seams opened. In each case the weakness was not the steel, it was the joint.

The response was welding. Once joints could be fused rather than fastened, the design rule changed from “how many fasteners do I need” to “how do I make the joint as strong as the plate it joins.” That shift is why pressure vessels, offshore platforms and aircraft frames are now specified by penetration rather than rivet count.

The Payoff Numbers

The M/S Carolinian, launched in 1930, was the first all-welded merchant vessel, and Time magazine reported it was built by a smaller team of just nine workmen, saving 75 percent of the cost and 77,000 pounds (35,000 kg) of steel rivets. No rivets means no riveting crew, no holes in the plate and no seam preparation beyond the weld itself.

Land structures followed. The Maurzyce Bridge in Poland, erected in 1928, was the first welded road bridge in the world, and the Sydney Opera House’s shells sit on a stud-welded steel frame where every primary connection behaves as continuous steel rather than a series of pins.

Pressure equipment followed the same logic. The strength and uniformity of welded steel allowed entirely new vessel types, most notably the LNG carrier, whose tanks carry cryogenic liquid in welded, stress-relieved seams that could not have been riveted. It is worth remembering that ASME’s Boiler and Pressure Vessel Code was developed in direct response to deadly boiler failures, and it exists precisely to make welded pressure boundaries predictable. In aerospace, where every gram counts and inspection is unforgiving, aerospace welding standards push the same logic further, demanding full penetration and rigorous verification on primary structure.

The Counterweight: Not Everything Should Be CJP

None of this means every joint should be fully penetrated. Penetration is a specification, and specifying it well means knowing where it does not pay. TWI notes that some materials are considered effectively unweldable, and where a filler metal of very different composition has to be used, as with brittle cast irons, you get a heterogeneous weld whose properties differ from the parent metal on both sides. Such joints can be perfectly adequate, but they are not continuous steel, and they are sized by the weaker material in the joint.

So specifying penetration is a trade-off between the load the joint must carry, the code minimum, the cost and time of the weld, and the weldability of the materials involved. Sometimes a fillet is genuinely right and calling for CJP would be wasteful. Sometimes a PJP detail is deliberate and cheaper. But when the joint is the weak link in a structure and a failure would be catastrophic, the whole argument for penetration welds comes down to a single idea: make the joint as strong as the metal, so the design no longer has to worry about the connection at all.

How Penetration Is Verified: Visual, Radiographic and Ultrasonic Testing

A bead looks the same whether it fused through the plate or not. Root fusion happens inside the joint, where only a testing method can reach, and every code requiring complete joint penetration pairs that requirement with an inspection method.

Four families of inspection cover it: visual, radiographic, ultrasonic, and the surface-break methods, liquid penetrant and magnetic particle. Only two can confirm root fusion, and neither is visual.

Visual Inspection: The First and Cheapest Check

Visual inspection (VT) is where every job starts, and in many shops the only inspection. It costs almost nothing and catches a surprising share of real problems: wrong weld size, poor contour, excessive reinforcement, undercut, visible cracks, overlap and surface lack of fusion. It also confirms correct process and filler, interpass cleaning, and backing.

What VT cannot do is prove the weld fused through the thickness. A bead can look clean while a sliver of unmelted base metal sits at the root, exactly what failed the plate that came back from radiography in the story that opened this article.

Ultrasonic Testing: The Standard for CJP Groove Welds

Ultrasonic testing (UT) is the reference method for verifying CJP in groove welds. A probe sends high-frequency sound into the metal and the instrument reads the echoes. Sound travels through soundly fused weld metal and the display shows continuous entry and exit signals. Where fusion is incomplete, sound bounces off the unbonded interface and returns early, indicating the flaw at a known depth.

That geometry is why UT is the standard. It senses reflectors along the fused volume and confirms depth of fusion and lack of fusion, which a radiographic image cannot. It suits thick sections, runs once the weld has cooled, and leaves a reviewable data record.

The catch is that UT is operator-driven: it needs a skilled technician, a calibrated procedure, the right probe and couplant, and a written scan plan for complex joints. Stainless and aluminium attenuate differently, so the setup changes again. Aerospace shops treat NDE as a discipline of their own, as the aerospace welding processes, standards and careers guide explains, and thin-wall stainless needs its own UT approach, as covered in aerospace TIG welding standards and techniques.

Radiographic Testing: Strong on Volumetric Defects

Radiographic testing (RT) places the weld between an X-ray or sealed gamma source and a film or digital detector. Density differences become image contrast, so porosity, slag inclusions, cracks and lack of fusion appear as dark shapes.

Its weakness is superposition and orientation. In a double-sided weld, two layers of image land on one detector, so a root defect on the far side can hide behind sound weld metal on the near side. Defects lying nearly parallel to the detector can also be missed. RT needs access on both sides and a qualified interpreter.

Surface-Breaking Defects: Penetrant and Magnetic Particle

For surface-breaking cracks, liquid penetrant testing (PT) and magnetic particle testing (MT) are the tools. PT uses a capillary dye that bleeds out of a surface-breaking flaw and needs a clean, non-porous surface; MT uses magnetic particles that gather at discontinuities in ferromagnetic material. Neither says anything about penetration, because both only find flaws that reach the surface.

Method	Shows well	Proves root fusion?
Visual (VT)	Contour, size, undercut, surface cracks	No
Ultrasonic (UT)	Depth of fusion, lack of fusion, cracks, slag	Yes, with a qualified procedure
Radiographic (RT)	Porosity, slag, inclusions	Partly, subject to masking
Penetrant / Magnetic Particle	Fine surface-breaking cracks	No

Regulation, Standards and Who Signs Off

OSHA treats inspection of critical welds as a safety matter. The welding standard lists industrial applications including transmission pipelines and mechanical piping systems, both relying on X-ray inspection of critical joints. 1910.252(d)(1)(vii) requires X-ray and radioactive isotope inspection of welded pipeline joints to conform to ANSI Z54.1, the American National Standard Safety Standard for Non-Medical X-ray and Sealed Gamma-Ray Sources, and 1910.252(d)(2)(ii) applies that same ANSI Z54.1 requirement to X-ray inspection of welded mechanical piping systems. See OSHA 1910.252.

On the people side, AWS offers the Certified Welding Inspector (CWI), Senior CWI, Certified Radiographic Interpreter (CRI) and a Penetrant Testing endorsement, plus ASME BPVC Section IX endorsements. It describes the CWI as a credential that advances careers and ensures quality, delivered in person, online instructor-led and self-paced, and states it has authored more than 350 standards for welding practices and procedures, developed by expert committees and approved by ANSI, in its codes and standards catalogue.

Keep the distinction clean: qualification and acceptance are separate acts. Passing a performance test proves the welder can weld. Proving a specific joint fused through its thickness is a testing question, and for CJP groove welds in a pressure boundary the answer usually comes back from UT.

Common Penetration Problems and How to Fix Them

Most penetration problems are not exotic. They are the same handful of defects showing up again and again, and each announces itself through a symptom you can read on the bench: a bead that looks fine but radiographs as a dark streak.

Incomplete Penetration and Lack of Fusion

Partial penetration means the weld metal did not fuse through the full thickness of the joint. Lack of fusion is the broader cousin: the weld metal never melted into the base metal, so the boundary is not metallurgically bonded even where the bead looks solid.

Usual causes:

• Heat input too low for the thickness, or travel speed too fast.
• Electrode or wire diameter too large for the root opening.
• Groove angle too wide, letting the arc spread.
• Incorrect work angle, blowing the arc off the root.
• Dirty, oxidised, painted or rust-covered surfaces.
• Insufficient preheat on thick or cold sections.
• Wrong polarity for the process.
• Excessive root opening, starving the root of filler metal.

The fixes follow directly: raise current or slow travel, step down to a smaller electrode, tighten the groove angle, correct the work angle, and clean the joint before striking the arc. On aluminium, prep is not optional; see our guide to cleaning aluminium correctly before welding.

Slag Inclusion

Slag inclusion is trapped flux or oxide inside the weld metal. On radiography it shows as a bright, sharply defined irregularity, and it is a serious defect in pressure and structural work because it breaks the load path.

Stick and flux-cored welding produce slag that must be removed between passes. Shielded metal arc welding is slower partly for this reason: electrodes are replaced constantly and slag chipped away every pass.

Causes are slag left in place before the next pass, a groove too narrow for the electrode to reach the root, weaving so wide that slag folds into the puddle, and poor interpass cleaning. Fixes are unglamorous: chip and brush every pass, keep the groove wide enough to clean and fill, narrow the weave, and if inclusions persist, re-gouge and reweld instead of welding over them.

Excessive Penetration and Burn-Through

Excessive penetration means the weld metal ran past the root of the joint; on thin material it means a hole. Causes are too much heat, slow travel, a root gap set too wide, and excess reinforcement at the root of the cap pass.

Backing is the standard defence: a sacrificial bar or strip under the root. Beyond that, control dilution and keep the root gap inside the procedure, because an oversized gap sends extra filler to the root and the penetration follows the filler rather than the base metal. On thin sections, cut amperage and speed up rather than adding a second layer.

Porosity

Porosity is gas trapped in the weld metal, visible on radiography as round voids. Causes are contamination, shielding gas problems in GMAW (wrong flow rate, leaks in the hose, wind disrupting the gas curtain), wet or rusty surfaces, and oil or cutting fluid on the plate.

The repair is to find and remove the source, not bury it: clean back to bright metal, check gas flow and hose integrity, and keep moisture and oils away from the weld area. Surface preparation for aluminium is covered in our aluminium MIG welding guide.

Distortion and Residual Stress

Uneven heat input compresses the weld while it is hot and pulls it into tension as the surrounding metal contracts, producing angular distortion, bowing and residual stress. On restrained joints it can open a crack while the arc is still running.

Counter it by distributing heat: alternate the side of the joint you weld, sequence welds to balance shrinkage, use smaller and more numerous passes instead of one large one, and keep fit-up tight. Plan for distortion you cannot avoid into the assembly rather than fighting it later.

Heat-Affected Zone Problems

The heat-affected zone is where base metal changed properties without melting, and it is where many cracks start. TWI notes that weld toes are frequent initiation points for fatigue and cold cracks, because they are points of high stress concentration.

In aluminium the HAZ can be softer and weaker than the parent metal, while in other materials it can be harder and stronger, and both create design challenges per TWI. Machining is the most effective way to remove an entire HAZ, but it cuts material yield and adds labour and machine time, so it is reserved for critical components. Post-weld heat treatment can reduce HAZ effects, but TWI cautions it is costly, time-consuming, limited by part size and not always a complete solution.

Welding Safety and Health: What Penetration Work Actually Exposes You To

Every rule in this section exists because somebody got hurt. Deep penetration welding is one of the hardest jobs in a metal shop: more heat, more time at temperature, more fume, more radiant energy and more hot work than light fabrication. OSHA treats it accordingly. The agency’s own welding, cutting and brazing topic page pulls together the general industry rules in 1910.252 and the construction rules in 1926.351, and those are enforceable requirements, not suggestions. General industry and construction shops run under different rule sets, so know which covers your site before hot work starts.

Hot Work Permits and Fire Prevention

OSHA 1910.119 defines hot work as work involving electric or gas welding, cutting, brazing, or similar flame or spark-producing operations, and requires a written hot work permit whenever that work happens on or near a covered process. A covered process means any flammable liquid, flammable gas, combustible dust or hydrogen-producing process. The permit is where you record the fire watch, the gas tests and the emergency contacts before the arc strikes.

The fire prevention requirements in 1910.252(a) incorporate NFPA Standard 51B, the Standard for Fire Prevention in Use of Cutting and Welding Processes. Three numbers matter on every job. Under 1910.252(a)(2)(vii), where practicable, all combustibles must be relocated at least 35 feet (10.7 m) from the work site. Under 1910.252(a)(2)(v), a 35-foot radius of floor must be swept clean of scrap, debris and combustibles. And a fire watch must be maintained for at least a half hour after welding or cutting is finished, so a smouldering slag under a plate has time to show itself. On a penetration weld with a wide, hot bead, that half hour is not negotiable. Disciplined housekeeping and slag removal, covered in advantages of stick welding, pay off here.

Ventilation, Hoods and Fume Control

Deep penetration means more base metal melted into the fume. Under 1910.252(c)(2), mechanical ventilation is required at a minimum of 2,000 cubic feet (57 m³) per minute per welder when welding happens in spaces of less than 10,000 cubic feet per welder, in rooms with ceilings under 16 feet, in confined spaces, or where partitions obstruct cross ventilation. Under 1910.252(c)(3)(i), freely movable hoods must maintain a control velocity of 100 linear feet (30 m) per minute in the welding zone, and that has to hold at the most remote distance from the arc, not just at the hood face. Position the hood a few inches from the work, not across the room.

Manganese and the Health Picture

According to the NIOSH welding fumes overview, welding fumes are composed of metals, and most contain a small percentage of manganese. Prolonged exposure above 1 mg/m³ may lead to manganism, a Parkinsonian syndrome with symptoms that resemble Parkinson’s disease. The limits stack up uncomfortably, because each one is written against a different fraction of the same substance.

Limit	Value
NIOSH recommended exposure limit (TWA)	1 mg/m³
NIOSH short-term exposure limit (STEL)	3 mg/m³
OSHA permissible exposure limit	5 mg/m³ ceiling
ACGIH threshold limit value, respirable manganese	0.02 mg/m³

NIOSH also notes that welding in confined spaces significantly increases exposure, because the fume has nowhere to go. If you cannot move the work, move the extraction. If you truly cannot ventilate the space, you need supplied air and a qualified respiratory protection program.

Eye Protection, Cylinders and Electrics

Eye protection is graded by shade number, and the OSHA table for shielded metal arc welding starts at shade No. 10 for work up to 30 amps and climbs to No. 15 in the 301 to 400 amp range. Auto-darkening helmets should be set well inside those numbers, since a filter that is too light lets you squint into the arc and one that is too dark leaves you blind when you need to read the puddle.

Gas welding adds its own rules. Under 1910.253(a)(2), acetylene must never be generated, piped or used at a pressure above 15 psig, and acetylene cylinders must be stored valve end up. Oxygen cylinders must be kept 20 feet from combustible material and separated from fuel gas cylinders. If your shop still runs an oxy-fuel torch for repair work, our air acetylene welding safety guide covers torch and cylinder handling in more detail.

Electrically, 1910.254(b)(3) sets voltage limits of 80 volts for manual AC arc welding and 100 volts for manual DC, and 1910.254(d)(8) states that cables with splices within 10 feet of the electrode holder must not be used. Ground return matters too: under 1926.351(c)(2), pipelines containing gases or flammable liquids must never be used as a ground return for arc welding. A stray arc through a gas line is a rupture waiting for a spark.

Practical Takeaways and Conclusion

Take one idea away: penetration is not a setting on a machine. It is an outcome that comes from three things working together. The joint design has to permit it, the qualified procedure has to produce it, and the welder has to execute it. Only then does a third party verify it. Penetration is proven, not assumed.

That changes how a shop runs a job. Instead of asking “will this weld penetrate?”, ask what the designer intended, what procedure was qualified to deliver it, and what proof the inspector accepts. A joint that answers all three passes. One that answers two comes back.

The seven-item penetration checklist

1. Read the governing code before you draw or cut the detail. Structural steel means AWS D1.1/D1.1M, pressure equipment means ASME BPVC Section VIII (Section IX governs procedure and welder qualification), pipelines mean API 1104, and bridges mean AWS D1.5. Each code sets the acceptance criteria and the required level of examination before a single mark lands on steel. The current structural steel edition is AWS D1.1/D1.1M:2025-AMD1, and AWS revises on its own publication cycle, so confirm the edition in force for your project.
2. Decide CJP or PJP explicitly, right on the drawing. Write “CJP” or “PJP” with the required depth, or call out to the detail. Ambiguity is where arguments start, and on a job with a radiographer, arguments are expensive.
3. Qualify the procedure and the welder before the joint is made. AWS offers welder performance qualification and welding procedure qualification endorsement pathways, and the industry works the same way: people and procedures qualify first, the finished weld gets inspected afterward. A qualified procedure run by a qualified welder is not something you improvise on the table.
4. Match the nondestructive examination to the joint. For CJP groove welds, ultrasonic testing is the standard method, because it senses reflectors through the fused volume and confirms depth of fusion and lack of fusion. Radiography handles volumetric defects like porosity and slag well, but on a double-sided weld one root defect can hide behind the other. Surface-breaking defects are a job for liquid penetrant and magnetic particle testing, and neither proves penetration.
5. Control the variables that move penetration depth. Groove angle, root opening, electrode or wire diameter, voltage and amperage, travel speed, work thickness, preheat and interpass temperature, bead placement, and work angle. Heat input is the shorthand: arc energy per unit length, calculated as voltage multiplied by current multiplied by 60, divided by travel speed. It decides whether you get full-thickness root fusion or a sliver of unmelted base metal.
6. Keep the joint clean and remove interpass slag. Oxide, rust, paint, moisture, oil, and trapped slag sit right at the fusion line. Clean before tacking, between passes, and at the root before the cap pass. Aluminum is less forgiving, and a dedicated guide to prepping aluminum for welding is worth a read if that is your base material.
7. Record the weld map: weld ID, welder stamp, procedure number, date, inspection results, repair history. When a joint is rejected months later, it tells you whether you have one bad weld or a systemic parameter issue.

Why a correctly matched weld can beat the parent metal

TWI notes that a completed welded joint may be referred to as a weldment, and that welding can produce a join stronger than the parent metal when the process and consumables are matched correctly. The heat-affected zone need not be the strongest zone in the assembly, but with the right procedure the joint itself can carry more than the plate it came from. A qualified procedure and a properly prepared joint are not paperwork; they are the difference between a weldment and a liability.

Back to the definition

Penetration welding is a join in which the weld metal fuses completely through the thickness of the base metal at the root, with complete root fusion. North American codes call that a complete joint penetration weld. Where less penetration is intended and specified, it is a partial joint penetration weld, and that is a legitimate design choice, not a defect. The difference between the two is not how good the weld looks, it is what was specified, what was qualified, and what was proven.

That is where the two plates from the opening story separate. The welder handed a tight groove angle, a hot root and a slightly quick travel speed got a weld that looked fine on the surface and failed the radiograph. The welder handed a qualified procedure and a joint design that let the arc reach the root got a weld that passed. Between the two, it is usually a few degrees of travel angle and a few percent of heat input. That is why definitions matter: write them down, qualify them, verify them, and the radiograph becomes a formality instead of a verdict.

Frequently Asked Questions

What is the definition of penetration welding?

Penetration welding is a join in which the weld metal fuses completely through the thickness of the base metal at the root, giving complete root fusion. It is not a separate heat source or process name; most arc processes can produce it. In North American codes the preferred term is complete joint penetration (CJP) weld, while British and European practice calls it a full penetration weld.

What is the difference between complete joint penetration and partial joint penetration?

A complete joint penetration (CJP) weld fuses through the full thickness of the joint with complete root fusion. A partial joint penetration (PJP) weld has fusion penetration that is intentionally less than full, with a specified minimum fusion depth. The critical point is intent: in a PJP weld, partial penetration is the design; in a CJP weld, the same condition is a defect called incomplete penetration.

Does a fillet weld have penetration?

No. A fillet weld is a triangular joint between two surfaces and carries load by shear across its throat rather than by fusing through the base metal. Penetration welds are groove welds in butt-type joints. That is why codes specify fillet weld size by throat thickness and groove weld strength by penetration and root fusion.

How do you prove a weld achieved complete joint penetration?

You cannot prove it by eye. Visual inspection confirms contour, size, undercut and surface cracks but cannot see root fusion. Ultrasonic testing is the standard method for verifying CJP groove welds because it senses reflectors throughout the fused volume, while radiography shows porosity and slag but can hide one root defect behind a double-sided weld.

What are the most common causes of incomplete penetration?

In practice the usual culprits are insufficient heat input, travel speed too fast, an electrode too large for the joint geometry, an excessive groove angle, a dirty or oxidised surface, and inadequate preheat on thick or cold sections. Fixing it usually means changing the procedure rather than simply adding more filler on top.

Is penetration welding the same thing as full penetration welding?

Effectively yes, but ‘penetration welding’ is informal industry phrasing. The formal terms are complete joint penetration (CJP) and partial joint penetration (PJP), which is how AWS D1.1 and ASME codes describe them. Using the formal terms on drawings and in inspection reports avoids ambiguity.

Can you get complete joint penetration with stick welding?

Yes. Shielded metal arc welding is routinely used for CJP groove welds in structural and pipe work, especially in the field and on heavy sections, because the flux coating provides shielding and the electrode core doubles as filler. The same CJP weld can also be produced by MIG, flux-cored, TIG, submerged arc, electroslag or laser processes depending on thickness and production volume.

When is partial joint penetration acceptable?

It is acceptable wherever the design does not require a continuous load path through the full thickness, such as non-load-bearing attachments, reinforcing pads, enclosure and box details, and some flange connections. The requirement is that the PJP intent is stated explicitly with a specified fusion depth, so that inspectors and later trades are not left guessing.

Does more weld reinforcement make the joint stronger?

No. Reinforcement is the excess weld metal sitting above the plane of the joint, and it does not increase the load-carrying capacity of the weld. Design uses the effective or throat thickness, which excludes reinforcement, because the strength of the joint comes from the fused cross-section rather than the extra metal piled on the surface.

Which codes govern penetration weld acceptance?

For structural steel, AWS D1.1 sets the detailing, inspection and acceptance criteria, with the current edition listed as AWS D1.1/D1.1M:2025-AMD1. For boilers and pressure vessels, ASME BPVC Section VIII governs construction while Section IX governs welder and procedure qualification. Pipelines typically fall under API 1104, and bridges under AWS D1.5.