Failed PQR Mechanical Test: AWS D1.1 Retesting Provisions Explained

When a PQR Test Fails: What Happens Next

A Procedure Qualification Record failure is one of the more stressful events in a fab shop's QC program. Hours of welding time, laboratory fees, and schedule pressure ride on a few bend or tensile specimens. When a specimen comes back failed, the immediate question is whether the qualification run is over or whether recovery is still possible.

AWS D1.1 provides retesting provisions for exactly this situation. Using them correctly requires understanding the difference between a failed specimen, an invalid specimen, and a failed qualification — distinctions that matter both technically and commercially.

Test Requirements for a Tested WPS Under AWS D1.1

Before discussing retesting, it helps to understand what the standard requires. A tested WPS under AWS D1.1 Clause 6 requires the welder to produce one or more test plates, from which specific specimens are cut. For groove welds, the typical test requires:

• Reduced-section tensile specimens to verify minimum base metal or weld metal strength
• Guided bend specimens (face, root, and/or side bends, depending on base metal thickness) to verify fusion and soundness
• Supplemental Charpy V-notch (CVN) specimens when toughness requirements apply, governed by AWS D1.1:2025 Table 6.8

For fillet welds, a separate fillet weld procedure test uses nick-break or macro specimens. The number and type of specimens depend on the weld joint configuration, base metal thickness, and whether supplemental qualification is required.

Rule library based on AWS D1.1:2025; verify against your governing edition.

What Constitutes a Failure

A specimen fails when it does not meet the acceptance criteria for its test type.

Tensile specimen failure: The specimen breaks below the required minimum tensile strength. For most structural steel, the WPS must demonstrate tensile strength at or above the base metal specification minimum. If the specimen breaks in the weld metal, the weld metal tensile strength applies; if it breaks in the base metal heat-affected zone, the base metal specification minimum governs.

Guided bend specimen failure: Any open discontinuity exceeding 1/8 inch [3 mm] in any direction on the convex surface after bending constitutes a failure. Cracks originating at specimen corners during bending are generally acceptable unless they exceed this threshold and do not appear to originate from weld discontinuities.

Failures typically indicate problems with the welding parameters — preheat, heat input, travel speed — filler metal selection, joint fit-up, or operator technique on the test plate.

The Retesting Provision

When a specimen fails, AWS D1.1 permits the welder or qualifying organization to invoke retesting. The provision requires cutting two additional specimens from the same test plate at locations adjacent to the failed specimen. Both replacement specimens must pass the test. If either fails, the procedure is disqualified.

Two requirements govern this provision:

Same test plate: The additional specimens must come from the original test weld, not a new weld made after observing the failure. This eliminates any opportunity to adjust parameters between the failure and the retest.

Both must pass: There is no further retesting opportunity once the two replacement specimens are tested. A single failure among the two stops the qualification.

If the original test plate does not contain sufficient remaining material — because the plate was cut too small or too many specimens were already taken — a new test plate must be welded and the full complement of required specimens must be tested. In this scenario, the welder cannot selectively retest only the failed specimen type; all required specimens must be cut from the new plate.

Invalid Specimens vs. Failed Specimens

Not every test that produces a poor result is a failure for retesting purposes. AWS D1.1 distinguishes between a failed specimen and an invalid one.

An invalid specimen is one that did not produce a valid test result due to a cause unrelated to weld quality:

• A machining defect created a stress concentration that caused fracture in the wrong location
• The specimen was cut from a location outside the required zone
• The specimen dimensions do not meet the standard's requirements
• A tensile fracture occurs entirely in the base metal at a location away from the weld, and the breaking load still meets or exceeds the minimum tensile requirement

When a specimen is invalid, a replacement is cut and tested — this is not the retesting provision and does not consume the retest opportunity. The lab report and QC documentation should clearly distinguish invalid-replaced specimens from failed-retested specimens.

Disagreements about whether a specimen is invalid or failed tend to arise on tensile breaks in the base metal. The key criterion is whether the breaking load met the minimum strength requirement. If it did, the specimen may be acceptable; if not, it is a failure regardless of fracture location.

Investigating Root Cause Before Retesting

The retesting provision is not a second chance to repeat the same mistake. Before cutting additional specimens, a competent QC program should investigate why the original specimen failed.

For bend failures, common causes include:

• Incomplete root fusion to the base metal on the first pass
• A weld profile that created a sharp notch at the toe, initiating cracking during bending
• Inadequate preheat maintained throughout the qualification run
• Incorrect bead sequence or insufficient interpass cleaning between passes

For tensile failures, common causes include:

• A filler metal not matching strength for the base metal grade
• Heat input outside the range needed to achieve full fusion
• Base metal that is not actually certified to the grade specified on the WPS
• Actual welding parameters (current, voltage, travel speed) that deviated from the WPS ranges on the test plate

If the investigation reveals a welding parameter error or a fit-up problem on the test plate, correcting it for the retest is appropriate — the corrected parameters become the qualified range in the final PQR. If the root cause is the WPS itself, such as a fundamentally mismatched filler metal, the WPS must be revised and a full new test plate welded.

Documenting the Failure and Retest

The PQR record must include the original test results and any retests. Do not conceal or exclude failed data from the record. A complete PQR that shows an initial failure, a root cause investigation, and a successful retest is defensible under a third-party audit. A PQR that shows only the passing results — with no record of earlier testing — becomes a credibility problem if the laboratory retains records of the failed tests.

Minimum documentation for a failed-and-retested PQR:

• Original test report showing failed specimen data and fracture description
• Written statement of root cause investigation
• Description of any corrective action applied to the test plate setup or procedure before retesting
• Retest results from both replacement specimens, with laboratory certification

The final PQR package presented to the Engineer of Record or third-party CWI for acceptance should include all of the above. Submitting only the passing retests without the failed originals is a documentation integrity issue that experienced reviewers will question.

The Distinction From WPS Requalification

A failed PQR test is not the same as a WPS requalification trigger. Requalification occurs when an essential variable on an already-qualified WPS is changed — for example, a change in filler metal classification, base metal group, or preheat reduction beyond the limits in AWS D1.1:2025 Table 6.6. The retesting provisions apply to a WPS that was never successfully qualified in the first place.

See AWS D1.1 Table 6.6 Essential Variables Explained for the complete list of changes that require a new PQR rather than a revision to the existing document.

Cost and Schedule Implications

A PQR test run at a qualified laboratory costs $400 to $1,500 depending on specimen count, CVN testing, and shipping. A failed test and retest adds laboratory turnaround time — typically one to three business days — plus the cost of additional specimen machining.

The faster approach is to build test plates correctly the first time: verify and document preheat before and during welding, ensure the test plate dimensions allow for all required specimens plus the two retesting specimens if needed, and have a CWI witness the run. If your fab shop generates PQRs regularly for new procedures, software that pre-loads required specimen counts and layout from the WPS parameters reduces costly test plate sizing errors before you ever strike an arc.

For related guidance, see PQR Tensile and Bend Test Requirements and Reading a PQR Test Report. To manage your full WPS/PQR library in a structured, audit-ready system, visit our pricing page.