How to laser-mark tube parts without losing traceability ART-200

Traceability is more than an inscription

A text number, DataMatrix or another symbol is only a carrier. It transfers an identifier but does not contain the whole history by itself. A reliable system preserves the relationship between that identifier and the material batch, supplier record, program revision, order, part number, operation, event time, machine and configuration, inspection and assembly.

The GS1 traceability standard provides a general model for critical tracking events and key data elements. It is a useful foundation, but it does not mean that every internal metal part must carry a GS1 identifier. A company may use its own immutable identifier if uniqueness, record lookup and history-retention rules are clearly defined.

Human-readable information must be distinguished from a machine-readable carrier. A person needs a short clue about the part or heat, while a scanner needs a robust code. One method may back up the other, but both must point to the same object.

What object is being identified

First define the level: part type, individual item, batch, kit, stock length or production order. If all identical parts share one number, this is not item-level traceability. If a serial number is assigned before nesting, the process must guarantee that it remains attached to the correct position in the cutting program.

Critical products may require an identifier for every item. A batch number may be sufficient for high-volume, non-critical parts. For a frame kit, a kit identifier together with component positions can be useful. The decision depends on the quality plan, customer requirements, regulatory context and cost of an error.

Do not encode changing business logic directly into an immutable identifier. Warehouse location, status, the current person responsible for an order or a web address are better stored as attributes. The identifier should remain stable while the part moves between operations.

The ID assignment event

An identifier may be assigned before nesting, while the machine program is created, at the moment of cutting or after physical separation. Each option needs a control point. The essential requirement is to link the future mark unambiguously and at the same time to the material stock and the specific program position.

If the programming system creates serial numbers, verify that they survive nesting changes, copying, reruns and new revisions. When an operator reruns one part, the system must not silently print a duplicate item identifier. Scrap and recutting require separate events, not manual reassignment of the old code.

ART-188 covers the overall route of program data. ART-200 focuses on the identification operation itself and the physical durability of the mark.

Where to place the mark on a tube

The position must remain accessible to the scanner and stay outside a weld zone, a bending contact area, a fixture clamp, a decorative face or an area that will later be machined. Angular orientation matters on a round tube; on a rectangular section, choose the face and the distance from the corner radius.

Consider visibility after assembly. A mark that is easy to read on a separate part may end up inside a closed frame. This may be acceptable if scanning is required only before welding, but the event must be recorded before access is lost.

Define no-mark zones in the design and manufacturing system or in a controlled table. Do not rely on the operator's memory. For each part family, define placement, orientation and minimum free-area rules that match the selected carrier and inspection plan.

Effect of curvature and orientation

A machine-readable code consists of elements of a defined size. On a small tube, curvature changes the visible projection for the camera and the focal position across the width of the mark. If the code is too wide, its outer modules are viewed at a different angle and may decode inconsistently.

In practice, assess more than contrast. Verify decoding with the real scanner geometry, including distance, angle, lighting, surface condition, oil and scale, and the condition after subsequent operations. A single microscope image does not prove shop-floor readability.

For a difficult carrier, it may be possible to reduce the physical width, change the placement or use another identification method. Any change to symbol size, error-correction level or data content must comply with the selected system specification and be validated.

Laser-marking method

Depending on the material and system, laser marking may produce an annealed, engraved, ablated or other type of mark. TRUMPF confirms the industrial use of laser marking, but there is no universal setting for every tube. The result depends on wavelength, optics, focus, coating, alloy, surface and required depth.

When the tube-cutting head creates the mark, distinguish low-power scribing from a dedicated marking process. When a separate marker is used, the identity of the part must be transferred between two machines and its orientation in the fixture must be controlled. Both routes can work, but their failure scenarios differ.

Set parameters using the manufacturer's documentation and a process trial. Do not specify a depth that may unacceptably weaken a thin wall, create a stress concentrator or damage a coating without engineering approval.

Mark quality and readability

For human-readable text, inspect character completeness, height, contrast, orientation and lack of ambiguity. A machine-readable carrier requires a validated verifier or at least a controlled scanner test under a defined procedure. A normal phone reading the fresh code once is not sufficient evidence of process capability.

Define acceptance criteria before and after every significant operation. A mark may be readable immediately after the laser but become unclear after tumbling, blasting, bending, welding heat, painting or corrosion exposure. The test must reproduce the actual production route.

Do not confuse readability with data correctness. The scanner must return an identifier, and the system must show the expected part, revision and status. These two levels are verified separately.

Code content

The most robust model uses a short opaque identifier that retrieves a record from a controlled registry. Encoding a long set of fields makes the symbol larger, while every format change complicates compatibility. At the same time, complete dependence on a network connection may be a risk for an autonomous work area.

A compromise may combine an identifier with a minimal version or checksum. The exact design depends on the ERP/MES and quality-system requirements. Do not use a direct web address as the only identifier: the domain and route can change. A web address may lead to the record, but a stable identifier must exist separately.

When GS1 DataMatrix is used, its content and application identifiers must comply with GS1 rules. An internal arbitrary DataMatrix should not be called GS1 DataMatrix merely because it looks similar.

Linking the mark to the material batch

When a stock length is loaded, the operator or automatic system must confirm the source-material identifier. The stock record is linked to the supplier certificate or batch, but the certificate does not have to be duplicated in every mark. A reliable relationship in the data repository is sufficient.

When a nesting plan uses a remnant from an earlier stock length, the identity of that remnant must be preserved. If a short remnant is returned to storage without a label or recorded event, traceability has already been lost before the next part is cut.

ART-194 covers nesting optimization and end remnants. The key point here is that optimization must not break the material-origin history.

Unloading, sorting and short parts

The order of parts on a conveyor or in a container may differ from the machine-program order. A part may bounce, jam, fall out with scrap or be removed manually. If the identifier is read only after unloading, a controlled reconciliation event is required.

Placement is particularly difficult on short parts: the carrier must fit, avoid the cut and remain readable. A controlled batch label on a container until the next operation may be the correct solution, with direct part marking performed later. This does not reduce traceability if the container is controlled and mixing is prevented.

ART-205 covers the physical separation of finished parts and scrap. ART-200 defines the information rules that must survive this stage.

Recutting, scrap and repair

When a part is rejected, its identifier must not simply disappear from history. Record the rejection event, reason and decision. A recut part receives a new item identifier or inherits the order relationship under a controlled policy, but the process must never create two active physical objects with one serial number.

For repair, retain the original identifier, the repair event, instruction, result and release decision. If the mark is damaged, link the replacement mark to the old identity through the change log. Manual re-entry without a second-person check creates a high risk.

Test cancellation, machine-program reruns, power loss, unavailable networks and repeat scanning. One error-free scenario does not validate the system.

Scan plan along the production route

Define critical events: material issue, cutting completion, sorting and packing, bending, welding fit-up, inspection, coating, final assembly and shipment. Not every operation requires a scan; excessive events encourage workarounds. A scan should change a controlled status or confirm a relationship.

For each event define who scans, where, with which device and under what conditions; what data is expected; what happens on error; whether an offline buffer is permitted; and how repeats are handled. Store the event time and performer together with the identifier.

Walk the actual production route with the part and scanner. Check access, glove use, lighting, convenience and response time. An architecture that works only at an engineer's desk is not production-ready.

Data integrity and access rights

Identifier generation must guarantee uniqueness within the required scope. Changes in assignment, merging, splitting and reprinting must leave an auditable record. An operator must not be able to silently reassign a code to another batch unless the correction, reason and approval are recorded.

Separate master data from event data. A part definition may be updated with a new revision, but the historical record must show the version used for manufacture. Do not overwrite the past with the current name without retaining a state snapshot or a version relationship.

Backups, lookup availability and cybersecurity matter, but this article does not design an ERP. For a pilot, it is sufficient to validate the full identifier life cycle and record interface requirements.

Pilot set

Select several families: a large round tube, a small diameter with visible curvature, a rectangular section, a thin wall, an as-rolled surface and a part that will be painted. Add a short part, a rerun and a rejection scenario. Create codes with representative content.

Verify mark creation, physical placement, scanning immediately after marking, reconciliation at output, readability after each operation and the correctness of the registry response. Ask another operator to complete the route without verbal guidance.

Retain photographs, verification results, scan logs, event history, and the links to the material and final assembly. For every nonconformity, identify whether the cause is physical, software-related, process-related or organizational.

Acceptance criteria

Criteria should cover a unique identifier, correct assignment, a readable carrier in defined states, placement in an allowed zone, no unacceptable effect on the part, successful scanning in the real geometry, correct event registration, duplicate protection and recovery after an error.

Set an acceptable unreadable-code rate and response procedure separately. If a code cannot be read, the process must not encourage the operator to select a similar-looking part from a list. The part needs isolation or controlled manual verification.

Traceability is ready only when the physical mark and the data chain pass validation together. A positive result in one layer cannot compensate for an error in the other.

Common mistakes

The first mistake is encoding only a part number and calling it item-level traceability. The second is creating a serial number before a nesting change without revision control. The third is placing a mark in a weld or bending zone. The fourth is testing only the fresh surface with a phone. The fifth is losing the identity of a remnant.

The sixth is reusing an identifier after rejection. The seventh is having no offline procedure. The eighth is calling any square code GS1 DataMatrix. The ninth is storing only a web address without a stable identifier. The tenth is failing to record baseline marking parameters and scanner model.

Scaling and change control

After the pilot, create approved marking-parameter sets for material and surface families, placement rules, a code scheme, a scanning instruction and an exception procedure. Train operators to recognize not only an unreadable mark but also a correct code applied to the wrong part.

Revalidation is needed after changes to alloy, coating, diameter, curvature, marker optics, software, code content, scanner, lighting, deburring, painting or the ERP interface. Repeat the test after moving the mark to another face or closer to a seam.

The parameter-set version and data-scheme version must be included in the evidence package. Otherwise, it is impossible to demonstrate that serial production still matches the approved pilot.

Semantic-correctness test

The pilot should include deliberately similar parts: left and right versions, different revisions, the same part number from different batches and a recut item. The operator scans them in random order. The system must show the correct description, revision, material relationship and permitted next operation.

Add negative scenarios: an unknown identifier, a retired serial number, a code from another order, a duplicated active identifier and an isolated part. Correct behavior is a stop or a clear exception procedure, not selection of an approximately matching record.

For human-readable backup text, verify that characters cannot be easily confused and that high-risk manual input uses a check digit or second-person confirmation. The backup path must not be so weak that it permits uncontrolled substitution.

Owner of every transition

Traceability breaks at responsibility boundaries. Assign owners for identifier generation, material linking, mark application, scanner maintenance, assignment corrections, data retention and rejection decisions. Every interface requires expected inputs and outputs and an error signal.

If the tube-cutting area sends parts for external coating, agree on the transport container, packing list, scan before dispatch and reconciliation after return. Loss of a visible mark after coating is acceptable only when another controlled carrier or a recorded reassignment event exists.

When the ERP or web-based lookup changes, the stable identifier must not change. A migration table, additional labels and a change log preserve the link between old marks and the new system. Test this before switching off the former lookup method.

Safe boundaries

This article does not define marking depth, allowable wall weakening, mandatory regulatory marking or ERP architecture. These decisions require approval from the responsible engineering, quality, information-technology and safety specialists.

Need to assess a manufacturing order?