Guided bend tests for PQR qualification under AWS D1.1

The guided bend test is the primary ductility check in a welding procedure qualification record under AWS D1.1. Before a WPS can be issued based on a test weld, that coupon must pass both tensile strength testing and guided bend testing — together they confirm that the weld meets strength and soundness requirements. For CWI and QC managers running or witnessing PQR tests, understanding which specimen types apply to a given coupon, what the jig geometry does, and what the acceptance limits mean is daily operational knowledge.

What the guided bend test measures

Tensile testing tells you whether the completed weld meets the minimum strength requirement for the base metal group. Guided bend testing tells you something different: whether the weld and heat-affected zone (HAZ) are ductile enough to deform plastically without cracking or opening up discontinuities under controlled bending stress.

A weld that is strong but brittle can pass a tensile test and still fracture in service under loading that induces bending. The guided bend test applies this bending stress to specific surfaces of the weld cross-section — the face (as-deposited surface), the root (back side), or the through-thickness cross-section (side) — to expose any fusion defects, porosity, or cracking that would otherwise pass visual examination undetected.

This is why the guided bend test cannot be waived based on radiographic or ultrasonic testing results. NDE detects indication size and location; it does not measure ductility or reveal the consequence of those indications under bending load. The bend test is a separate and complementary qualification requirement.

Face, root, and side bend specimens

AWS D1.1 Clause 6 specifies which specimen types apply to groove weld PQR tests based on base metal thickness:

For base metal under 3/8 in (10 mm) thick: the test coupon yields two face bend specimens and two root bend specimens — four total. Face bends are cut so the as-welded surface goes into tension on the convex side of the bend. Root bends are oriented the opposite way, putting the root pass region into tension. Each specimen samples only the surface it is oriented toward.

For base metal 3/8 in (10 mm) and thicker: four side bend specimens are used in lieu of face and root bends. Side bends are cut transverse to the weld axis through the full thickness of the joint, then bent so the full weld cross-section is the convex surface. This gives a complete picture of the fusion zone and HAZ through the entire plate thickness in one specimen — far more representative on thick material than a face or root bend that samples only the surface layer.

The distinction matters in practice. A fab shop qualifying a WPS for 1/4 in plate will machine face and root specimens; the same shop qualifying for 3/4 in plate will machine side bends. Getting this wrong — side bending thin material, or running face and root specimens on thick plate — is a PQR deficiency that will surface during an audit or third-party review.

The guided bend jig

The guided bend is a three-point bend using a jig with a plunger (male die) and either fixed shoulders or rollers. The plunger pushes the specimen down through a slot, bending it around the plunger radius to a controlled bend angle.

The jig dimensions — specifically the plunger diameter — control the outer fiber strain applied to the specimen convex surface, which determines how severe the test is relative to material ductility. AWS D1.1 specifies plunger diameters based on base metal yield strength. For lower-strength carbon steels (Fy ≤ 50 ksi [345 MPa]), a smaller plunger diameter is used; for higher-strength steels, the plunger diameter increases to avoid fracturing a specimen in the HAZ that would otherwise be acceptable. Using the wrong jig geometry — either too small a radius or a worn, out-of-spec jig — can cause false failures or miss real ones.

When witnessing PQR tests at a test laboratory, verify that the jig conforms to the dimensional requirements in the current edition of D1.1 before any specimens are bent. Ask the lab for their jig calibration or dimensional verification records. Dimensional nonconformance of the test jig is grounds for rejecting test results, which means remachining specimens and retesting — a costly delay if caught after the fact.

Acceptance criteria

After bending to the required angle (typically 180° for structural carbon steel), examine the convex surface of each specimen under good lighting:

• No single open discontinuity may exceed 3/16 in (5 mm) in any direction. This includes porosity cracking open at the surface, lack of fusion opening up, slag tears, or any other open discontinuity visible on the convex side of the bent specimen.
• Corner cracks up to 3/8 in (10 mm) that appear to originate at the machined edge of the specimen — not from a weld or HAZ discontinuity — do not constitute failure. These typically result from edge preparation condition or base metal surface condition rather than weld quality.
• Transverse cracks passing across the weld or HAZ that exceed the limit are a clear failure. These indicate brittleness in the fusion zone or HAZ and point to a procedure issue — insufficient preheat, excessive heat input, inappropriate filler metal, or base metal incompatibility.

The 3/16 in (5 mm) limit in AWS D1.1 is less restrictive than the 1/8 in (3 mm) limit in ASME Section IX. If your shop qualifies procedures under both codes, ensure test laboratory personnel know which acceptance criterion applies to each coupon. A result that passes D1.1 may not qualify under ASME IX.

Rule library based on AWS D1.1:2025; verify against your governing edition, as the AHJ or contract documents may specify the 2020 edition or earlier.

Specimen preparation: where failures come from

PQR bend specimens are machined by the test laboratory from the test coupon. Proper preparation is critical:

Reinforcement removal: On groove weld coupons, face reinforcement and root reinforcement must be ground flush with the adjacent base metal surface before bending. Leaving reinforcement creates a stress concentration at the weld toe that will initiate cracking even on a sound weld — the specimen will fail the bend test for a reason unrelated to weld quality.

Saw marks and machining defects: Saw cuts or grinding marks that run parallel to the weld axis on the long edges of the specimen become stress risers on the convex surface during bending. If edge cracks appear and appear to originate from a machined edge defect rather than a weld discontinuity, evaluate whether a replacement specimen is warranted.

Specimen dimensions: AWS D1.1 specifies the specimen width and thickness dimensions for each bend type. Face and root specimens are typically 3/8 in (10 mm) wide × material thickness. Side bend specimens are typically 3/8 in (10 mm) in the through-thickness dimension × 1-1/2 in (38 mm) in the weld-depth dimension. Specimens machined out of tolerance for width or thickness change the effective test severity and can invalidate the results.

Retest provisions

Not every failed specimen requires a complete new test weld. AWS D1.1 provides two retest paths:

Preparation defect retest: If a specimen fails and there is clear, documented evidence the failure originated from a preparation error — saw mark, machined edge crack, incorrect reinforcement removal — a replacement specimen machined from the original test coupon may be substituted. The replacement must pass. This provision requires good judgment: if there is any doubt about whether the failure is preparation-related or weld-related, treat it as weld-related.

Weld-related failure retest: If the failure is attributable to the weld itself — lack of fusion, porosity cluster, HAZ cracking — you may retest using two additional specimens of the same type taken from either the original coupon (if material remains) or from an additional weld coupon produced under the same procedure. Both retests must pass. One pass and one fail is not acceptable.

These retest provisions are frequently misapplied. The determination of whether a failure is preparation-related or weld-related should be made by the CWI or engineer of record and documented. Once a retest is initiated, the original specimen result and the retest rationale must both appear in the PQR package.

What the PQR records for bend tests

The PQR document must record each bend specimen with its type (face/root/side), orientation, acceptance result, and — if any discontinuity was found — its measured dimension. A record showing "NR" (no rejectable indications) or a measured value of zero for all specimens is the most common result for a well-qualified procedure.

If any specimen required replacement under the preparation-defect provision, that substitution must be noted with the rationale. If a retest was performed under the weld-defect provision, both the original failure and the retest results must appear.

For guidance on reading and interpreting the full PQR test report, see reading a PQR test report and what is a PQR. For the broader context of how PQR bend tests relate to tensile testing requirements, see PQR tensile and bend test requirements.

When the PQR supports multiple WPS documents — common in shops that generate several joint-type procedures from a single qualification weld — the bend test coverage applies to all WPS documents derived from that PQR. See using multiple PQRs to support a single WPS for how that relationship works in practice.

Practical notes for CWIs witnessing PQR tests

Be present for specimen extraction and labeling. Chain of custody from test weld to machined specimen is a recurring audit finding. Each specimen should be traceable to its position in the coupon (which face, which root, which side), the welding position used, and the process and filler metal — before it goes to the machine shop for preparation.

Photograph the convex surface after bending. Post-bend photographs are the permanent record of what the specimen showed. If the lab has a macro lens, closer shots of any indications — even acceptable ones — document the as-found condition regardless of what the written report states.

Record the bend test date on the PQR. The test date, not the document issue date, establishes when the qualification was performed. This matters for evaluating which edition of D1.1 governed the testing and whether the PQR remains current after an edition transition.

Managing PQR records, bend test documentation, and the WPS packages derived from them across multiple projects and processes is easier with a purpose-built platform. See WPS software options for structural fab shops for how qualification records can be maintained with essential-variable tracking and qualification status monitoring.