Failed Welding Test? Causes, Retest Rules, and How to Fix It

The coupon looks good while you’re welding it. Then the inspector cuts the bend specimens, takes them to the saw, and one of them snaps open right across the root. Twenty minutes later you’re standing outside the fab shop wondering whether you just lost the job. Here is the part most welders never hear: a failed welding test is one of the most recoverable setbacks in this trade. Nearly every failure traces back to a handful of specific, teachable causes, and both AWS D1.1 and ASME Section IX are written to give you a second chance.

Why Welding Tests Fail — and Why It Feels Worse Than It Is

The coupon looks good while you’re laying it down. Then the inspector cuts the bend specimens, takes them over to the break, and one of them snaps open right across the root. Suddenly you’re wondering whether you just lost your qualification over a few inches of weld.

Here is the reassuring part: a failed welding test is recoverable in almost every case. It is a setback, not the end of a career. What matters now is knowing exactly what kind of failure you are dealing with and how to approach the fix.

What Does a “Failed Welding Test” Actually Mean?

In code terms, a failed welding test is not simply an ugly weld. It is a performance qualification test coupon that does not satisfy the acceptance criteria of the governing code, which means the welder is not qualified for that specific process, position, or thickness range at that time. Depending on the job, that governing code might be AWS D1.1, ASME Section IX, API 1104, or ASME B31.3.

Before jumping to retest rules, it helps to know that not every “bad result” is the same thing. These four scenarios are very different, yet they often get lumped into one category:

• A failed performance qualification test (WQT/WPQ): The coupon itself failed the required examination against the code’s acceptance criteria. This is the failure that limits qualification for that process, position, or thickness range.
• An invalidated specimen: The failure came from a testing artifact, such as poor machining, a specimen cut outside the required zone, or a mislabelled coupon. In these cases the specimen is replaced and does not count against you.
• A rejected production weld: A weld already placed in service or scheduled for service that fails production inspection. This is an entirely different matter from a qualification coupon.
• An expired or lapsed qualification: Under ASME Section IX QW-322, a welder who has not used a qualified process in production for more than six months must requalify. That is a continuity requirement, not a penalty for a bad weld.

The distinction matters because the path forward for each one is different. An invalidated specimen will not consume your retest allowances. A true qualification failure will.

The Stakes and Why Tests Are Harder

Any failed test carries real consequences. It can limit a welder’s qualifications, affect an employer’s ability to ship code work, and cost money for testing. But the biggest hurdle for most welders is the emotional hit, and that pressure is exactly what causes so many people to rush their next attempt before they understand the root cause.

There is a reason test coupons trip up welders who run solid production welds every day. Qualification tests are deliberately harder. They are often in restrictive positions, offer little to no allowance for comfort, are performed with the coupon restrained, allow no “second pass to fix everything” mentality, and sometimes do not allow the same preheat assistance available on a production job. These factors stack the odds against you, which is why failure happens more often than it should feel like it should.

That said, the codes are written to give qualified pathways back. Both AWS D1.1 (clause 6.25) and ASME Section IX allow retesting, with ASME Section IX QW-321 providing for an immediate retest of two consecutive test coupons for each failed position, both of which must pass. AWS D1.1 offers two retest pathways depending on whether additional training or practice was provided. These are procedural options, not permanent disqualifications.

The Two-Part Framework For Recovery

With the right mindset, every failed test can be used as data. For the rest of this article, we will use one simple framework: (1) what the code says failed and (2) why the metal did that. The code tells you which rule was broken and what retest path is available. The metal tells you which technique needs correcting.

To diagnose either side properly, you need a baseline of the process. Before changing anything, review fundamentals such as aluminum cleaner for welding, proper test coupon preparation, and how to read a WPS. That way any changes you make are to your technique, not an unintentional violation of your qualified procedure.

Ultimately, a failed welding test is a diagnosis, not a verdict on your worth as a welder. Get the diagnosis right, follow your code’s retest path, fix the cause, and your next coupon has every chance of passing.

The Four Ways a Welding Test Actually Fails

Nearly every rejection lands in one of four buckets: visual inspection, a bend test, a macro etch or fillet break test, or a volumetric examination. Each bucket points at a different family of causes, so “just weld it better” is useless advice. Find the bucket and you know what to practise.

1. Visual inspection (VT) rejection

Visual inspection is where most coupons die, and the code is least forgiving here. Under the AWS D1.1 visual inspection criteria, any crack is unacceptable regardless of size or location, and incomplete fusion is prohibited outright. Undercut deeper than the limit for your plate thickness, joint type and loading condition is a reject, as is a profile, convexity, reinforcement height or overlap outside the allowable range.

One rule has no exceptions: an arc strike outside the weld zone is an automatic rejection. It does not matter how small the crater looks or how neatly you ground it flush. Striking an arc on the base metal to start a pass disqualifies the coupon.

2. Mechanical (guided bend) test failure

This is the test most welders dread, because it comes after the coupon looked perfect. The specimen is bent until it fractures and the broken face is examined under magnification. ASME Section IX QW-163 is blunt: there shall be no open defects greater than 1/8 in in the weld or heat-affected zone on the convex side of the bend specimen, and open discontinuities at specimen corners are disregarded unless there is definite evidence of lack of fusion, slag inclusions or other internal discontinuities.

AWS D1.1 applies the same logic in slightly different numbers. An open discontinuity larger than 1/8 in [3 mm] on the convex (bent) surface, or exceeding the aggregate length allowance, is a failure. A crack longer than 1/4 in [6 mm] is disregarded only when clearly attributable to a slag inclusion or fusion-type defect, then the tighter 1/8 in [3 mm] limit applies. Corner cracks over 1/4 in [6 mm] are ignored when the examiner sees no slag or lack of fusion.

3. Macro etch and fillet break test failure

Macro etch means cutting the coupon, polishing the cut face, etching it and reading what the chemistry reveals. Acceptance requires complete root fusion, no cracks, and no visible slag exceeding 1/4 in [6 mm] total accumulated length. Slag left in a root pass shows up here even when a bend test barely squeaked by.

Fillet welds use the break test instead. The coupon is loaded until the weld fractures, and the fracture face must show complete fusion into both members and a specified fraction of the weld cross-section fractured. A fillet that tore off the base metal without fusing is a failed specimen, not a strong weld.

4. Radiographic or ultrasonic rejection

Volumetric methods look inside the weld, and cracks are prohibited outright. Rounded indications, the signature of porosity, are limited relative to weld size: the maximum is commonly S/3, not exceeding 1/4 in [6 mm], with up to 3/8 in [10 mm] permitted where the weld size S exceeds 2 in [50 mm]. UT sees the same flaws from a different angle, so a coupon that passes RT can still fail UT on sidewall lack of fusion.

How many specimens, and why one is enough to sink you

A single CJP groove plate coupon commonly yields four bend specimens: two face bends and two root bends. All four must pass. One crack in one specimen fails the entire qualification, even if the other three look textbook, which is why a coupon that “felt good” can still cost you the test. Position matters too: the same root geometry behaves differently vertical or overhead, which our all position welding rod guide covers in detail.

Failure type	What the examiner sees	What it says about your technique
Visual (VT)	Crack, incomplete fusion, undercut past limit, profile or convexity out of tolerance, arc strike outside the weld zone	Travel speed or filler deposition out of balance, weak arc control, stray arc start
Bend test	Open discontinuity over 1/8 in [3 mm] on the bent face, or a crack through the root	Lack of root fusion, slag between passes, tungsten or wire contamination
Macro etch / fillet break	Missing root fusion, etched-face cracks, slag pockets, partial cross-section fracture	Incomplete penetration, dirty or rust-covered joint face, no slag removal
RT / UT	Cracks, rounded indications above the size limit, lack of sidewall fusion	Porosity from moisture or wind, low gas flow, contaminated consumables

Read your report against this table before you touch the torch; the right-hand column is your practice plan.

The Numbers That Decide Pass or Fail: Acceptance Criteria

Most welders don’t fail their qualification test by a mile, they fail it by a hair. A bend specimen that looked solid on the table can snap right open over a discontinuity that’s only a fraction of an inch over the limit. With that being said, knowing the exact numbers that the inspector is grading against can be the difference between a pass, and having to weld your coupon all over again.

Undercut: Easy To Get, Easy To Fail

Undercut is one of the most common near-misses for welders, and proper travel control plays a massive role in keeping it in check. For statically loaded nontubular connections, the limits are broken down by thickness. If the material is under 1 in [25 mm] thick, undercut shall not exceed 1/32 in [1 mm]. However, there is a very narrow exception allowing up to 1/16 in [2 mm] for an accumulated length of 2 in [50 mm] in any 12 in [300 mm] of weld. For material equal to or greater than 1 in [25 mm] thick, undercut shall not exceed 1/16 in [2 mm] for any length of weld.

These limits tighten drastically under more demanding loading. For cyclically loaded primary members with undercut transverse to tensile stress, it is limited to just 0.01 in [0.25 mm] deep. For tubular connections, undercut is limited to 1/32 in [1 mm] for wall thickness up to 5/8 in [16 mm], and 1/16 in [2 mm] for thicker walls.

Porosity, Reinforcement, and Profile

Not all porosity is treated the same. For CJP groove welds in butt joints transverse to the direction of computed tensile stress, there shall be no visible piping porosity whatsoever. For other static groove and fillet welds, the sum of visible piping porosity 1/32 in [1 mm] or larger is limited to 3/8 in [10 mm] in any linear inch of weld, and 3/4 in [20 mm] in any 12 in [300 mm]. For cyclic welds, the frequency is limited to one pore in any 4 in [100 mm] with a maximum diameter of 3/32 in [2.5 mm].

With the 2025 edition of AWS D1.1, welds shorter than 12 in have been clarified as well. For these shorter welds, the sum of porosity is limited to the weld length multiplied by 0.06, and accumulated undercut deeper than 1/16 in is limited to the weld length multiplied by 0.16.

Profile is just as strict. Reinforcement height is commonly limited to up to 1/8 in [3 mm] for groove welds (with allowances varying by application), whilst fillet weld convexity is limited to 1/16 in for face width up to 5/16 in, 1/8 in for width under 1 in, and 3/16 in for width of 1 in or more. Most importantly, overlap (cold lap) is absolutely prohibited, no matter how small it may appear to the naked eye. For those working with more exotic alloys such as those covered under Alloy 20 applications, always confirm the profile requirements with the project specific code as well.

Bend Tests: The Breaking Point

For guided bend tests, the numbers are brutally specific. The maximum permitted discontinuity size is 1/8 in [3 mm] on any side of the bending surface. In addition to that, the code also caps the total of the highest sizes of all discontinuities, commonly cited as 3/8 in [10 mm] where the largest indications are over 1/32 in [1 mm], but not more than 1/8 in [3 mm].

Cracks are treated with zero leniency in most cases. The maximum size of any crack on a bend specimen is 1/4 in [6 mm], unless it is established that the crack is due to slag inclusion or another fusion-type defect, in which case the limit drops down to 1/8 in [3 mm]. Keep in mind that one failed specimen is enough to fail the entire test coupon, so it only takes one bend to ruin your whole attempt.

Retest Rules: AWS D1.1 Compared with ASME Section IX

A failed welding test doesn’t have one universal fix. The path back to qualification depends entirely on the code governing the work. For most welders, that means choosing between AWS D1.1 for structural steel and ASME Section IX for welding and brazing qualifications. Getting that framework wrong can waste both time and a perfectly good coupon.

AWS D1.1: Two Paths Forward

For a welder performance qualification under D1.1, Clause 6.25 provides two options to regain qualified status for the failed type and position. The decision comes down to whether or not further training is taken.

• Immediate retest (no additional training): Two consecutive test coupons must be welded for each failed type and position. Both must meet every visual and mechanical or NDT requirement with no exceptions.
• Retest after further training or practice: Only one coupon is required for each failed type and position, provided it is welded and inspected to all of D1.1’s criteria.

ASME Section IX: QW-321 in Practice

For boilers, pressure vessels, and much of ASME B31 piping work, the retest requirements fall under QW-321. Unlike a blanket rule, these are broken down by the examination method that caused the failure.

• QW-321.1 (visual failure): Two consecutive coupons are required for each failed position, and all must pass visual examination.
• QW-321.2 (mechanical test failure): Two consecutive coupons must be tested by mechanical means, with both passing.
• QW-321.3 (volumetric failure): Two consecutive coupons must be examined by the same volumetric method. For pipe, the production weld alternative allows examination of twice the original weld length, or twice the number of pipe circumferences.
• QW-321.4 (after further training or practice): Only a new test per failed position is required.

Several ASME interpretations are vital here. IX-83-13 confirms the doubled coupon requirement does not apply after further training. IX-92-73 allows a welder to elect either the volumetric alternative or mechanical testing under QW-304 after training. Most notably, IX-86-07 extends the immediate retest provision to renewal test coupons under QW-322, meaning a lapsed qualification can be re-earned under the same immediate retest path.

Most importantly, the retest must use the exact examination method that failed. A failed bend test cannot be substituted with radiography to try and pass.

Side by Side Comparison

Code	Immediate Retest	After Training or Practice	Coupon Reuse	Primary Application
AWS D1.1	Two consecutive coupons per failed type and position, all criteria met.	One coupon per failed type and position.	Not permitted. Fresh plate required.	Structural steel, bridges, and construction qualifications.
ASME Section IX	Two consecutive coupons per failed position, tested by the method that failed.	One new test per failed position (QW-321.4).	Not permitted.	Boilers, pressure vessels, and ASME B31 piping.
API 1104	Governed by the API standard and its operator qualification rules.	Varies by operator qualification programme.	Not permitted.	Pipeline welding, including automatic and semiautomatic processes.
ASME B31.1 / B31.3	Generally follows Section IX qualification logic.	Varies by project specification.	Not permitted.	Power and process piping, often with stricter project additions.

What Sits On Top of the Code

The cited code is rarely the last word. Project specifications, state building departments, and employer quality systems can all add conditions. For example, Los Alamos National Laboratory GWS 1-05 calls for an immediate retest to be performed within five working days of failure, with up to eight hours of training or practice allowed beforehand. This timing is not universal, but it is a practical discipline worth following anywhere.

Before booking a retest, verify four things: the governing code, the exact clause and edition cited by the project, any project specification additions, and your employer’s retest procedure. For a free cross-check, TWI’s summary of Section IX retesting and retraining can be helpful, along with resources such as this overview of immediate retest rules.

Finally, only the failed type and position are affected. All other processes and positions remain valid subject to continuity requirements. A failed 6G does not invalidate flat or horizontal qualifications, so focus your efforts where they count.

Root Causes: Why Welder Qualification Tests Fail

Codes tell you when a coupon is invalid. They rarely tell you why the metal cracked. That part is physics, and it repeats in a short, learnable list: incomplete root fusion, trapped slag, gas in the puddle, an open crater, and a travel speed that outran deposition.

Start with what a guided bend test actually does. It takes a strip of weld metal and tries to break it at the root, squeezing stress into a narrow band at the fusion line, the hardest-working metal in the joint. Anything that never fused, or a pocket of slag or gas sitting on that line, becomes the crack that splits the sample. That is why the bend test punishes fusion defects harder than anything else, and why Los Alamos procedure guidance on welding practice treats root fusion as the make-or-break variable.

Root and interpass fusion problems

The AWS D1.1/D1.1M:2025 requirement is unambiguous: complete fusion shall exist between adjacent layers of weld metal and between weld metal and base metal, with no exceptions across statically loaded, cyclically loaded, or tubular connections. Failing that is usually mechanical, not mysterious.

• Root gap and root face. Too little root opening, or a root face left beveled when the WPS calls for square, gives the puddle nowhere to fuse into. A root opening too wide on thin plate is just as bad, because the arc cannot stretch across the gap.
• Electrode angle. Standing the electrode too steep and pushing the arc toward the root produces the classic lack of fusion: arc force drives the molten pool downward and backward instead of wetting the sidewall. The technique details in our aerospace TIG welding technique guide explain the same principle, because TIG and SMAW both live or die on how the arc is directed.
• Travel speed on the root pass. Dragging the root along faster than you can fill melts the leading edge and leaves it behind.
• Slag left between passes. A skipped chip and brush puts slag inside the joint, and the next pass melts over it.
• Contamination. A dirty tungsten or contaminated electrode tip puts oxides right at the fusion line.
• Preheat and interpass temperature. On heavy section, welding cold to thick plate raises hydrogen levels and leaves the HAZ soft and crack-sensitive.

Porosity

Porosity comes from gas, and gas comes from four places: moisture on the electrode or in the flux, wind disrupting the shielding, insufficient shielding gas flow, and contamination of the joint face by paint, rust, or mill scale. A dirty or damaged gas cup adds a fifth. Bend a porous coupon and it opens on bright, shiny, rounded voids.

The rule that catches people out: on CJP groove welds in butt joints transverse to the direction of computed tensile stress, no visible piping porosity is permitted. Zero tolerance means one pore can fail an otherwise excellent coupon, so a clean joint face matters as much as a clean electrode.

Shape and profile defects

Undercut is a groove melted into the base metal adjacent to the weld toe that is left unfilled by weld metal. It cuts base metal thickness and leaves a sharp stress concentration, and it is a straight rejection. It usually comes from too much travel speed or too little filler deposition. The same category collects excessive reinforcement and convexity, slag trapped in the toe, spatter, and arc strikes outside the weld zone.

Crater cracks are the quiet killer. Under AWS rules, all craters shall be filled to provide the specified weld size, except at the ends of intermittent fillet welds outside their effective length. Leave a crater open and the bend test finds it every time.

Human and situational factors

A 6G or 2G coupon is a different skill from flat plate. Position changes your heat, filler access, and sight line, and practice material matters too. Cold weather stiffens your hands and slows wire feeding, poor fit-up puts the root where you did not plan it, and the wrong electrode size or F-number for the qualified WPS changes bead shape and heat input. Choosing the right rod for the position, as covered in our all-position welding rod guide, is part of the test, not a detail outside it.

Procedure and documentation problems

Some failures are not welder failures. Welding outside the qualified WPS variables, welding the wrong base metal, or misunderstanding what the coupon requires will fail you regardless of skill. A common self-inflicted failure is welding inside the WPS variable range but with an electrode or process setup you have never run consistently in the test position. The numbers were legal and the practice was not. Match the consumable to the WPS and to the position, as in our alloy steel electrode selection guide, and rehearse that exact setup before the coupon goes on the table.

Diagnostic Table: Symptom, Cause, Drill

Symptom	Likely cause	Drill that fixes it
Open bend, smooth grey fracture	Root lack of fusion	Root gap and angle drill on scrap plate, checking fusion on each pass
Bright round voids in the fracture	Porosity	Clean-stub drill with a flow check and a wind shield
Grey lamination in the fracture	Trapped slag	Interpass cleaning drill: chip, brush, inspect, then weld
Crack at specimen end	Open crater, run-in or run-out	Crater fill drill: end every bead with a clean fill and a deliberate tail
Undercut rejection	Travel speed outran deposition	Match travel speed to deposition rate on flat coupons and measure cap height
Same rejection on every specimen	Procedure or setup problem	WPS and fit-up review with your supervisor before more practice beads

Diagnose Your Failure the Way an Inspector Does

A failed coupon is not scrap. It is the only physical evidence you will ever get about what your arc was really doing, and once it is cut up, it is gone. Good welders read it like an X-ray: find the problem, name the cause, then pick up the torch.

Step 1: Read the report before you read the coupon

Identify which test failed: visual inspection (VT), guided bend, macro etch or fillet break, tensile, or radiographic/ultrasonic examination (RT/UT). Note the position and type, because “failed in 6G on a fillet weld” is a different conversation from “failed flat on a groove weld.”

Write both down. Most self-inflicted errors happen when a welder assumes the visual passed and blames the bend test, when the report shows a visual failure on undercut and the bend test was never reached. If procedure documents feel intimidating, brush up on how the procedure documents drive the test.

Step 2: Classify the failure into one of four families

• Weld metal problem — the deposit itself is defective: porosity, slag, lack of fusion, cracking.
• Base metal problem — the parent metal failed: cracks, laminations, or weak plate properties.
• Heat-affected-zone problem — the hardened or cracked zone beside the weld, where hydrogen cracking in restrained, hardenable steel shows up.
• Test artifact — the coupon failed because of how it was machined, cut, or handled, not the weld. Many failures hide here.

Step 3: Locate the defect along the coupon

• Start of the weld — suspect an arc strike, poor run-in, or a dirty start area.
• Repeating at regular intervals — a defect every pass length means inconsistent electrode angle or travel speed, often with too much weave.
• End of the weld — crater cracking or a bad run-out, from not filling the crater or jerking the electrode out.
• Root, mid-height, or cap — root points to root fusion and cleanup, mid-height to interpass cleaning and slag, cap to surface tension and crater control.

Note which specimens failed. If both face bends snapped in the same place but the root bends were clean, the problem is in the cap or upper passes. If every specimen failed in the same spot, the cause is systematic.

Step 4: Read the fracture face

Five minutes with a hand lens pays for itself:

• Smooth grey face, no penetration past the root — lack of fusion. The metal never melted into the base.
• Bright, shiny face with rounded gas pockets — porosity. Gas expands after the metal solidifies.
• Flat face with grey inclusions, sometimes laminated — slag trapped between passes.
• Rough, crystalline, or faceted face — a solidification or hot crack, a more serious problem.

For a tensile-type break in base metal, the key criterion is whether the breaking load met the minimum strength requirement. If it did, the specimen may be acceptable even though it broke outside the weld metal. The inspector or lab makes that call.

Step 5: Ask the procedure question

Ask plainly: skill issue or WPS and setup issue? More practice helps only when the cause was your hands. If preheat was not achieved, the electrode was the wrong F-number for the qualified procedure, the WPS was wrong for the joint, or fit-up and root gap were bad, you have a procedure problem. The fix is a revised WPS and a re-qualified PQR.

LANL GWS 1-05 requires the Welding Test Supervisor to review failed specimens to find the conditions contributing to the failure and allows up to 8 hours of documented training or practice before a retest. AWS D1.1 permits two retest specimens for a particular test specimen type cut from the same qualification material, and both must meet the test requirements. The metal you already welded may still work in your favour.

Step 6: Document everything before you cut or discard anything

Before the coupon goes in the bin, write the file:

1. The failed test report, with the fracture description.
2. A root-cause statement naming the defect and cause.
3. The corrective action applied, with practice hours.
4. The retest results, including replacement specimens.

A file showing only passing results is a credibility problem the first time somebody audits it.

Aluminum and stainless work follows the same logic, but surface prep matters more. If an aluminum coupon fails by lack of fusion, look hard at cleaning and oxide removal, as our guides on preparing aluminum for welding and aluminum flux core welding show.

Do these six steps properly and the next coupon is not a gamble but an experiment with one variable changed.

Seven Mistakes That Turn a Failed Test Into a Longer Problem

Failing a welding qualification test is not the hard part. What happens in the week after is. These seven mistakes turn a two-week setback into a two-month one.

1. Taking the retest immediately because an appointment happens to be open

Under AWS D1.1, an immediate retest with no additional training requires two consecutive passing coupons for each failed type and position, all meeting the visual and mechanical requirements. The unprepared welder takes two shots, not one, and a second failure lands on the record, not the first.

ASME Section IX arithmetic is similar: QW-321 requires two consecutive qualifying coupons after an immediate retest, while QW-321.4 allows a single new test per failed position after further training or practice. One nuance from QW-321.1: you may weld more than one test coupon, and the examiner may pick either that passed visual for the mechanical test. Both still have to pass.

2. Changing the procedure for the retest

Nothing on a coupon should differ from the original test except the welder’s skill. If the WPS variables were not followed on the failed coupon, it was never a valid qualification attempt, so the retest must use the correct WPS: the right filler metal, F-number, preheat and interpass temperatures, travel speed and voltage range, and cleaning method.

Welding “better” with a different process or electrode produces a coupon for a procedure nobody qualified, not a qualifying coupon. If the shop’s setup was the problem, fix and document it first, which may mean a revised WPS and re-qualified PQR.

3. Not documenting the failure

Keep the failed test report, write a root-cause note in your own words, and record the corrective action you took. That turns a failed coupon into usable data instead of scrap steel on a rack.

It also protects you later. When an inspector asks why a coupon was replaced, a documented invalidation supports the decision; an undocumented one raises doubt about everything else you welded.

4. Calling a failure an “invalidation” because the result was inconvenient

Invalidation means something specific: the specimen was machined incorrectly, cut in the wrong place, contaminated, mislabelled, or otherwise compromised. It is a determination made and documented by the test organisation, usually when the examiner finds a problem with the specimen itself, not with the weld.

Relabelling a real failure as an invalid specimen does not remove it from the record, it creates a credibility problem. If the claim fails, the welder has a failed test and a failed test report dispute.

5. Practising the wrong thing

A welder who failed in 6G and spends a week running flat-plate beads has not addressed the cause. The skill that failed was welding in that position, on that thickness, with that joint configuration, under that restraint.

Work through the specific defect instead: root fusion drills with consistent root gap and electrode angle, slag control and interpass cleaning between passes, crater fill technique, travel speed matched to deposition rate to control undercut, and a wind shield for porosity.

6. Assuming the whole qualification is gone

Only the failed type and position are affected. A failed 6G plate coupon does not cancel your flat and horizontal qualifications, nor your qualification under another process.

Continuity is the piece to watch. ASME IX QW-322 allows requalification by welding a single test coupon where there is reason to question a welder’s ability to deposit sound welds, and it governs welders who have not welded with a qualified process for a while. Check your continuity dates; a lapse in one process can become a second failure.

7. Chasing the cheapest certificate mill

A coupon welded under an unapproved procedure, inspected by someone who is not the authorised inspector, or reported on a form nobody can audit will be rejected by the client, inspector or code official. It also costs more: you pay for the cheap test, then again for the real one, plus delay and a reputation hit with the shop that sent you.

Ask before you pay: who is the authorised inspector or welding test supervisor, will the lab issue a report signed by the inspector, can they offer both mechanical and volumetric options, what is the turnaround, and what happens to the coupon if the retest fails. And never sign for a coupon you did not weld; it ends careers fastest.

One more thing about the after-training retest

After further training or practice, the new test counts as a fresh qualification, not a repeat of the failed attempt. Section IX does not require proof of training or that an inspector witness it. ASME interpretation IX-18-01 leaves that to the welder’s supervisor and the inspector. What matters is that your practice addressed the defect that failed.

I’ll rewrite this section to fit the560–840 word range while keeping the heading, facts, and all links.

A Retest Plan That Actually Works

A failed coupon is only useful if you do something with it. Most welders treat the second attempt as a rerun of the first: same setup, same habits, same result. The welders who pass on the second try spend a week attacking the specific defect the report named, not the trade in general.

Here is a realistic 10-to-14-day cycle. Adjust the dates to your schedule, but keep the order. The order matters more than the calendar.

Days 1-2: Read the Report and Name the Failure

Sit down with the failed test report and your supervisor, instructor, or lead welder. Write one sentence: “I failed because of ____.” No hedging. “Lack of fusion on the root pass because I ran a hot, fast bead with too much travel speed on the second pass.”

That sentence is the plan. Everything after it is practice aimed at one thing. If you cannot write it, you do not understand the failure yet, and practice will not fix it. Reading the report next to the acceptance criteria in your governing code, such as the AWS D1.1 inspection requirements, turns a vague feeling of failure into a measurable defect with a numeric limit attached.

Days 2-6: Drill the Specific Defect

Work on scrap plate, one problem at a time. Each drill targets a defect that shows up on real failure reports:

• Root fusion. Run consistent root gaps and electrode angles. Too steep drives the arc toward the root and you get lack of fusion on the face of the root pass. Too flat gives penetration but poor tie-in. Practice until your hand stops drifting.
• Slag control and interpass cleaning. Chip and brush every pass. No exceptions, no “it looked clean.”
• Porosity control. Clean stubs, dry storage, and a wind shield. Drafts undo an otherwise good weld.
• Crater fill. Every crater gets filled and the electrode lifted deliberately, not snuffed.
• Undercut control. Match travel speed to deposition rate. Dragging the puddle behind a fat pool cuts into the base metal.

Days 5-9: Position-Specific Practice

Practice in the position you failed, on plate of the same thickness as the test coupon, using the same joint configuration. A 6G vertical-up test is a different skill from a flat bead, and 1 in [25 mm] plate behaves nothing like 1/4 in [6 mm]. Have a partner or instructor watch and call out what they see: arc angle, travel speed, bead length, and where you paused.

Days 8-11: Full-Length Practice Coupons

Weld full coupons at the correct size, mark them like the real thing, and, if your test lab or in-house shop will do it, have them cut and bend tested. This is where you find out whether the training is landing.

Days 12-13: Preparation

Review the WPS variables you will be graded on: preheat and interpass temperatures, filler metal, travel speed, cleaning requirements. Confirm electrode storage and conditioning. Set up the coupon the night before so you are not scrambling on test day. Fatigue and last-minute rushing cause more failures than lack of skill.

Day 14: The Retest

Verify the position marking on the coupon before you strike an arc. Have the WPS in hand. Follow the test supervisor’s instructions exactly and weld the coupon as though it were a production weld, because under the code it is one. Remember that ASME Section IX retest rules mean a failed coupon is welded fresh, never repaired and resubmitted.

Do Not Neglect Safety While You Practice

Repetition means more fume exposure, not less. Under 29 CFR 1910.253, mechanical ventilation is required at a minimum of 2,000 cubic feet per minute per welder wherever general ventilation is required, and local exhaust hoods are designed to hold a control velocity of 100 linear feet per minute in the welding zone. Those general ventilation rules apply in spaces under 10,000 cubic feet per welder, in rooms with ceilings under 16 feet, and in confined spaces or wherever partitions significantly obstruct cross ventilation. See the OSHA Part 1910 Subpart Q welding, cutting and brazing rules and the OSHA welding and cutting safety fact sheet for the full requirements.

Hexavalent chromium is the hazard that gets most welders and stainless and chrome-plated work. OSHA sets its permissible exposure limit at 5 µg/m³ as an 8-hour time-weighted average under 29 CFR 1910.1026 and 1926.1126. For welding fumes, NIOSH has not set a recommended exposure limit because fume composition varies by process; the recommendation is to reduce exposure to all welding emissions to the lowest feasible concentration. NIOSH lists metal fume fever, interstitial pneumonitis, and lung cancer among the health effects associated with welding fume exposure.

Use local exhaust hoods close to the arc, keep your position and electrode selection deliberate rather than improvised, and wear the right respirator and protective gear for the filler metal you are running. Practising in a poorly ventilated space for two weeks to prepare for a retest is a poor trade.

Two weeks of focused practice followed by one clean coupon beats four weeks of general welding followed by a hopeful retest. If you drilled the right defect, in the right position, on the right thickness, the pass usually arrives because the cause was removed rather than because you got lucky.

Choosing the Retest Route: Test Labs, Costs, and the Gear That Helps

You know the defect and you have drilled the right thing. Now the unglamorous half: who welds your retest coupon, and who inspects it? Three routes exist, and the best depends on your employer, geography and how often your shop welds.

Route One: Go Back to the Organisation That Issued the Failure

Usually the fastest path. The lab already holds your original report, knows the type and position you failed, has the forms loaded and an examiner who knows your work. Nothing gets re-entered, so paperwork rejection is unlikely.

One rule matters. Under ASME Section IX QW-321, the two-coupon immediate retest must use the same test method that failed. A coupon that failed guided bend testing gets retested by guided bend testing, not swapped for radiography. The TWI summary of ASME retest requirements puts it plainly: the method that produced the failure is the method that has to be repeated.

Route Two: A Different Third-Party Test House

Useful when the original lab is weeks out or far away, or you want a second opinion. A new lab will want your previous records to confirm continuity. Two cautions: confirm the client or employer accepts tests from that laboratory, since some specifications list approved houses by name, and note that the standard does not soften. The acceptance criteria in AWS D1.1 or ASME Section IX do not care whose turnstile you walk through.

Route Three: In-House Qualification

Companies that weld constantly often qualify welders internally with an authorised inspector or a Certified Welding Inspector on staff. Marginal cost per test drops once the inspector is salaried, and scheduling gets easier. The catch is independence: the inspector must be authorised to run the test and sign the report, and your client’s quality system may still require an external AI for verification.

Route	Who administers it	Main strength	Watch out for
Original organisation	The lab that issued the failure	Records and examiner already in place	Queues, travel, waiting for a slot
Different third-party house	Independent test house	Fresh examiner, sometimes more capacity	Client approval, record transfer, no change in standard
In-house	Your own authorised inspector	Lowest marginal cost for high-volume shops	Inspector authorisation and independence

What to Verify Before You Sign Up

• Who signs. Ask for the name, certification and authorisation of the AI or Welding Test Supervisor. GWS 1-05 requires that supervisor to verify the coupon against the WQT and the Welding Technique Sheet, including position, fit-up, electrode selection and preheat. If that check is missing, ask whether the lab even supplies the technique sheet. GWS 1-05 has the detail.
• A properly completed report. Request a sample. It should record test method, position, type, each result, and the inspector’s signature and stamp.
• Volumetric capability. If the retest needs radiography or ultrasonic testing, confirm the lab does it on the same turnaround. Small labs often handle bend testing only and farm out the NDT, which adds days.
• Coupon rules in writing. How many coupons, how many specimens off each, what happens if one fails. Agree it before you weld.
• Record retention. Coupons (pass or fail) and radiographs are kept until viewed by the welder or 30 days from qualification, whichever is earlier. The file records why the test failed, and failed bend coupons and radiographs are retained under the standard’s records provisions.

What Retesting Costs

Nobody publishes a clean price list, because the total depends on coupon size and thickness, the process and the test method. Volumetric testing costs considerably more than guided bends, and third-party labs add administrative fees plus a possible retake charge on a failed coupon. Expect a retest to cost a meaningful fraction of a full first qualification, with the price driven by the test method rather than by the failure.

Who pays varies too. Employer-funded qualification is common for code work on company product, since the company needs qualified welders to ship. A personal qualification, taken for your own certification card or to work elsewhere, is usually self-funded, and many shops cover the first attempt only. If you pay out of pocket, ask about a package covering both coupons if one fails, because that is where the cost lands.

The Gear That Actually Shortens the Learning Curve

You do not need a new welding machine. You need a small set of things that remove the excuses you used to give for the defect.

• Electrode holder with a clean ground clamp. Poor ground causes erratic arc behaviour and unstable fusion. This all-position welding rod guide covers consumables for the position you failed.
• Electrode storage. Damp electrodes cause porosity and failed coupons. A rod oven or heated box is cheap insurance.
• Grinder, flap discs, chipping hammer, wire brush. Interpass cleaning is not optional: slag at the toe of a previous pass becomes an inclusion and a crack in the bend specimen. This welding equipment guide covers the supporting gear worth owning.
• Fume extractor with the hood near the arc. Effective extraction protects you on long practice sessions. OSHA FS-3647 is worth reading before you set up a station.
• Tape measure and square. Consistent root gap and leg length are free. Most fit-up defects that cause lack of fusion are measurement errors, not skill errors.
• Plate coupons at test thickness. Practising on 6 mm and being graded on 20 mm is a different problem. Ask your shop or training provider for offcuts of the right thickness and joint configuration.

Disclosure: Some shop and supplier links in this section are affiliate links. We may earn a commission if you buy through them. It does not change our advice, and every tool above is described with its limits, because a good electrode holder will not fix a bad root gap.

Choose the route that gets you a properly signed report from a properly authorised inspector, on the code your project cites. Buy the gear that removes your specific defect, not the gear that looks impressive. Then go back to the drill that fixes the metal.

Conclusion: Turn the Failure Into a Plan

A failed welding test is not a verdict, it’s a diagnosis. Regardless of how discouraging it may be to see a bend specimen snap in front of an inspector, nearly every failure can be corrected with the right approach. The difference between welders that bounce back, and those that remain stuck in a cycle of rejections, isn’t talent, it’s their ability to treat their failure as data instead of defeat.

The Roadmap To Recovery

With that being said, recovery doesn’t come from luck or a last minute change of tactics. It comes from discipline. Before you ever strike an arc for your retest, commit these takeaways to memory:

1. Find the exact culprit. Don’t settle for “I failed”. Locate exactly which test failed (VT, bend, macro, or RT/UT), the position, and the defect that caused it.
2. Know the rulebook. Every code plays by its own game. Under AWS D1.1 clause 6.25, a retest after further training or practice only requires one coupon for each failed type and position, making documented, targeted practice the cheaper and more reliable route. Under ASME Section IX QW-321.4, a new test is made for each position on which the welder failed, allowing all of your other valid qualifications to remain in place.
3. Fix the metal, not the paperwork. Never try to label a legitimate failure an invalidation to dodge your retest allowance. If your variables were incorrect, the coupon itself is not valid to begin with.
4. Build a paper trail. Keep your failed test report, fracture descriptions, and any notes from your inspector. From there, write out your root cause and your planned corrective action. These records will be invaluable for any future audit.
5. Only qualify what’s broken. You do not need to requalify your entire book. Due to the position based requirements of ASME IX, as well as the specific “failed type and position” allowance within AWS D1.1, retest only what was rejected.
6. Practice with intent. There’s a massive difference between running beads in the shop and practicing for the exact coupon you’ll be graded on. Train in the failed position, with the same thickness, joint prep, and procedure that will be used for your retest.
7. Know when to escalate. If this is your second attempt and the same defect persists, stop guessing. A repeated failure is far more often a fit-up, preheat, WPS, or procedure issue, opposed to a lack of effort. At that point, escalate to a qualified trainer or welding engineer.

Mindset Is Your Greatest Tool

The habits that get you to pass your next test are the exact same habits that will save you thousands of dollars on the job. Welders that analyze their coupon fracture faces learn to catch lack of fusion before it ever makes it to an NDT booth. Those that document their fixes learn to build consistency between passes. Those that practice with intent learn to combat their weakest position until it becomes their most stable one.

Most importantly, recognize the scale of it all. A failed performance coupon is aggravating, however, a rejected, fully completed CJP production weld is exponentially more expensive to repair, document, and re-inspect. The ten minutes that you spend analyzing your failed specimen now will prevent those hours of rework later.

For the sake of credibility, always ensure that any acceptance criteria that you’re being graded against are pulled from an authoritative source. All of the numeric limits that are referenced throughout this article are derived from AWS D1.1/D1.1M, ASME Section IX, and Los Alamos National Laboratory’s GWS 1-05. With codes constantly updating between editions, always confirm against the exact edition and code that your project cites.

Your First Step Starts Now

You don’t need to solve every possible defect at once. You only need to solve the one that failed you.

Pull your failed test report off of your desk, read over it one last time, and write down a single sentence naming the defect that caused the rejection. That one sentence is where your recovery begins.

From there, build your plan off of it. Drill it, document it, and retest it with intent. If any of the common questions from this topic are still lingering in the back of your head, be sure to take a look at our FAQ answers for further clarification. You’ve been knocked down by this test once, but with this diagnosis in hand, you’re more than equipped to pass it the next time around. You’ve got this.

Frequently Asked Questions

Can you retake a failed welding test?

Yes. AWS D1.1 clause 6.25 provides two retest pathways, and ASME Section IX QW-321 permits an immediate retest in which you make two consecutive coupons for each failed position and both must pass. After documented further training, ASME IX QW-321.4 requires a new test for each failed position, and under D1.1 that path needs only one coupon per failed type and position.

How long do I have to retake a failed welding test?

ASME Section IX does not specify a waiting period, so it is up to the authorised inspector and your employer. Some site standards set a limit: the Los Alamos National Laboratory standard allows an immediate retest within 5 working days of the failure and permits up to 8 hours of training or practice beforehand. Practising deliberately for one to two weeks is the most common sensible path.

What does a bend test failure actually mean?

It means the specimen opened up under bending at a location that indicates a real weld defect. Under AWS D1.1 the bending surface may not show any discontinuity larger than 1/8 in [3 mm], and under ASME IX QW-163 no open defect greater than 1/8 in is permitted on the convex side. Corner cracks are usually disregarded unless there is clear evidence of lack of fusion or slag inclusion.

Is a failed test different from an invalidated test?

Yes, and the difference can save you a retest. A failed specimen does not meet the acceptance criteria. An invalid specimen produced an unusable result for reasons unrelated to weld quality, such as a machining defect or a specimen cut from the wrong location, and it is simply replaced. Only failures count against the retest allowance.

If I fail one position, do I lose my other qualifications?

No. Both AWS D1.1 clause 6.25 and ASME IX QW-321.4 limit the retest to the type and position that failed, so passing positions and other processes remain valid, subject to the normal continuity rules such as the six-month requalification requirement in ASME IX QW-322.

Can I use radiography instead of a bend test on my retest?

It depends on the pathway. An immediate retest must repeat the test method that failed, so a bend-test failure has to be retested by mechanical testing. After further training, the new test is treated as a fresh qualification, so a welder may elect the volumetric alternative or mechanical testing permitted by QW-304, as confirmed by ASME interpretations IX-83-13 and IX-92-73.

The failure was caused by the procedure, not me. What do I do?

Say so in writing and treat it as a procedure investigation, not a welder re-test. If the coupon was welded outside the WPS, or the WPS itself was wrong, correct the procedure, requalify the PQR if required, and weld the retest coupon under the corrected WPS. A welder cannot qualify to a procedure that was not followed on the coupon.

Does a failed welding test mean I can never weld again?

Not remotely. Failing a qualification test is common, and both major codes build in a recovery route. What does end a career is refusing to analyse the failure, repeating the same test without changing anything, or letting a lapsed qualification go undocumented.

What records should I keep from the failure?

Keep the original failed test report with the fracture description, a written root-cause investigation, the corrective action you applied, and the retest results. Site standards such as the Los Alamos National Laboratory welding standard require the reason for failure to be recorded and failed coupons and radiographs to be retained. A file that shows only passing results becomes a problem in any audit.