Inspecting a Braze Joint: Shear, Dye Penetrant and Visual Criteria

A braze joint fails in the field for reasons that were visible at the bench, or would have been visible to a cross-section. Braze joint inspection exists to catch those conditions before parts are packed, and it is worth understanding what each method can and cannot establish, because the cheapest checks find most of the defects and the expensive ones are there to answer a question the cheap ones cannot.

What a Good Joint Looks Like

A complete fillet, sound bond line, no overheating.

A sound brazed joint shows an even filler fillet along the insert, a continuous bond line under section, and no discoloration or distortion that suggests the cycle ran hot.

Three features define acceptance visually. The first is fillet formation: the filler should appear as a continuous, slightly concave fillet where the insert meets the pocket, with no gaps, skips or unfilled corners. The second is evidence of wetting across the whole interface, which on a sectioned sample appears as a bond line of consistent thickness with no dark voids between the filler and either surface. The third is the condition of the surrounding materials: discoloration of the steel beyond the joint, distortion of a thin section, or a blue or grey tint near the insert all point to a cycle that ran outside the intended window.

Two features are hard to judge by eye and are commonly over-read. A joint can look complete on the outside and contain internal voids that reduce the load-bearing area, and a joint can have a perfect fillet while the carbide beside it has lost hardness to an overheating cycle. Inspection criteria should therefore state which conditions are judged visually and which require a destructive or sectioned check, rather than leaving the distinction to the inspector.

Defect Classes and Causes

Joint defects fall into a small number of classes, and each one has a characteristic cause in the process rather than in the material. Recognising the class is what makes a corrective action possible.

Voids and incomplete fusion are the most consequential. They occur when filler does not flow into part of the interface, usually because of surface contamination, an oxide layer, insufficient filler, or a joint gap that the filler cannot bridge. The part then carries load over a fraction of its intended area, and failure appears as an insert that sheds cleanly under load.

Overheating is the second class, and it damages the carbide rather than the bond. Excessive temperature or time at temperature promotes grain growth and can degrade the binder, so the insert meets its hardness specification on paper and behaves as a softer grade in service. Discoloration, a widened heat-affected zone and distortion are the visible signs, and hardness or structural checks are needed to confirm them.

Distortion and residual stress form the third class. A thin section heated unevenly will move, and the joint may then be sound but the assembly out of tolerance. Even without visible distortion, a cooling profile that quenches the joint leaves stress that shows up as delayed cracking. The metallurgy behind these mechanisms is described in the references published by ASM International, and the process fundamentals, including the conditions that produce incomplete fusion, are set out in the TWI brazing explainer.

Destructive and Non-Destructive Checks

The inspection methods divide into those that destroy the sample and those that do not, and the choice between them is a question of what has to be proved rather than of which is better.

The shear test is the direct destructive method: a sample assembly is loaded until the bond fails, and the failure load is compared with the joint area the design assumed. It answers whether the filler and the joint geometry carry the intended load, and because it destroys the part it is applied to samples made on the same cycle. Sectioning is the other destructive method, and it is the only routine way to see voids, bond line thickness and the condition of the carbide beside the joint.

Non-destructive methods cover production. Visual inspection with defined criteria catches fillet defects, missing filler and obvious overheating. Dye penetrant reveals surface-breaking discontinuities that reach the surface, which is useful on an exposed fillet but cannot detect an internal void. Volumetric methods such as radiography or ultrasonic inspection can reveal internal conditions, and are normally used where a joint is safety-critical or where a defect class has appeared and needs to be characterised; the applicable test standards are published by bodies such as ASTM and the welding society guidance from the American Welding Society. Filler metal data for the alloys involved is published by suppliers such as Lucas Milhaupt.

Sampling Frequency and Records

Sampling turns an inspection method into a control. A shear test on one sample proves that the cycle can produce a good joint; a sampling plan proves that it still is.

A practical scheme has three layers. In-process checks run continuously and are cheap: filler placement, part positioning, and the cycle record itself. Batch-level visual inspection applies defined criteria to production parts, with any non-conforming item quarantined rather than corrected. Periodic destructive verification — a shear test and a section — is performed on samples from the same cycle, at an interval agreed with the customer or set by the manufacturer’s own control plan.

The record is what gives the scheme its value. A batch identification that links parts to their cycle data and inspection result means a field failure can be traced to a specific production window and compared with other batches from it. The traceability arrangement used at SENTHAI is described on the quality control page, and the batch-level approach applied to blades is set out in the article on blade quality control and batch traceability. Where a customer wants a specific sampling rate, that is agreed as part of the order and stated in the written quotation or order documents.

What Inspection Cannot Prove

It cannot predict life or certify a whole batch.

Inspection establishes that the joints examined met the criteria at the time of examination; it does not establish service life, and a sample cannot certify parts it did not include.

The distinction matters in three ways. First, shear testing measures the strength of a sample joint, not the load that a specific insert will see on a machine, which depends on geometry, fitment and operator behaviour. Second, a defect that is not surface-breaking may go undetected by a visual or penetrant check, so an inspected batch can still contain an internal void; only sectioning or a volumetric method can rule that out, and only on the parts examined. Third, an inspection that passes at dispatch says nothing about what a later repair, a field weld or an impact will do to the joint.

The honest use of inspection is therefore comparative and diagnostic. It separates good batches from bad ones, it identifies when a process has drifted, and it produces the record that allows a failure to be investigated. It cannot replace a duty-matched grade, a correct geometry or a fitment that holds the insert in place, and a procurement specification that treats inspection as a substitute for those will not be improved by adding test methods.

Cost and Time of Each Method

The methods differ by orders of magnitude in cost and time, which is why a sensible programme uses all three layers rather than choosing one.

The practical sequence is to prove the process with destructive checks, verify it continuously with the cheapest checks, and reserve volumetric methods for a specific question. That is also the sequence in which the costs make sense: a programme that leads with the most expensive method spends time characterising a process it has not yet proved. The measurement principles behind these tests are the subject of the work of the national measurement bodies such as NIST, and published research on brazed joint integrity, such as the peer-reviewed work indexed at MDPI, provides independent context for why void content matters.

Inspection Criteria to Write Into a Contract

Inspection only protects a buyer if the criteria are agreed before production. Written generically, they become a matter of opinion at delivery.

A workable clause names five things:

• The acceptance criteria, stated in measurable terms: fillet continuity, permitted discoloration, bond line condition on a section, dimensional tolerance for the assembly.
• The method for each criterion, so it is clear which conditions are judged visually and which require a destructive or volumetric check.
• The sampling plan, including the interval for shear testing and sectioning and whether samples come from the same batch as the delivered parts.
• The record set, naming what accompanies the delivery and how it links to a batch identification.
• The non-conformance route, stating what happens to parts that fail, and the evidence needed for a claim.

Two commercial points sit alongside the technical ones: the cost of destructive sampling is normally reflected in the agreed terms, and any additional test report beyond the standard set is agreed at order stage. The applicable terms are stated in the written quotation or order documents. Buyers who write these criteria once and reuse them across suppliers get a comparable basis for deciding where to place work.