How a new section batch affects the accuracy of series cutouts ART-208

What should be treated as a separate batch?

A production-control batch is not always the same as one invoice or truck load. A useful batch groups material of traceable origin and sufficiently similar manufacturing conditions. Grade, standard, manufacturer, heat, forming line, seam type, coating, nominal section and wall thickness may match only partly.

If the supplier changes the production site, forming method, allowed range or packaging, that can represent a new condition for a sensitive part. Material after long storage, repeated transport, corrosion, mechanical damage or mixed bundles should also be identified separately.

An internal batch identifier should connect the certificate, goods receipt, bundle, machine program, cut parts and inspection results. Without this link, it is impossible to determine whether a trend arrived with the material, a machine adjustment or an operator change.

Which section characteristics become part geometry?

For round tube, critical characteristics may include actual diameter, ovality, wall thickness, straightness, local dents and seam position. Square or rectangular hollow sections add corner radii, convex or concave walls, lack of face parallelism and twist along the length. Open or special sections may present their own unstable locating surfaces.

These characteristics influence the part in different ways. An outside-size change can move the surface relative to the nominal CAD model. Twist rotates a local face so a hole with the correct nominal axial coordinate appears in the wrong place on the real wall. Bow changes the position along the length and loading on supports. A different internal radius changes beam access and the internal edge near a corner.

The seam matters for more than its external bead. Local thickness, hardness, reflectivity, internal geometry and behaviour during subsequent bending or welding can differ around it. This does not make the seam a defect by definition; it makes seam orientation a controlled input to the setup and inspection plan.

Nominal size does not describe the actual cross-section

A rectangular section designation defines a nominal family, not an identical real contour for every bar. If the program uses an idealised section while the machine neither measures nor compensates for the difference, part of that difference transfers into cutout coordinates.

Do not reduce the incoming specification to one average measurement. Two sections can have the same average width but different corners, twist or local waviness. The surface from which a functional cutout is located matters. A bolt hole, assembly lock and end for robotic welding have different sensitivities.

Build an error budget covering the material, location, clamping, machine motion, cutting process, measurement and assembly. This turns a discussion about the batch from blame finding into an assessment of each contribution.

Incoming inspection without excessive measurement

Incoming inspection should not become complete scanning of every stock length. First identify the characteristics that can actually move critical cutouts. One part may be sensitive to twist and seam position, another to ovality and bow, and a third to actual wall thickness and corner radius.

The sample should cover different bundles, positions within a bundle and lengths. On each selected bar, inspect several cross-sections near both ends and in the middle. Record actual values rather than only “accepted.” If the delivery contains several heat numbers or manufacturing labels, do not hide them in one average.

Use a capable method: calibrated hand tools for simple dimensions, an indicator arrangement for straightness, a profile template, or optical and coordinate measurement for a complex contour. The method needs sufficient repeatability. An unsuitable measurement system can create the appearance of an unstable batch.

Use a control sample after a batch change

The best first check is not an arbitrary finished part but a short control sample containing sensitive elements. It should reproduce critical angular positions, lengths, areas near the seam, transitions across a corner and required ends. Include repeats so that variation is visible rather than relying on one result.

Place features near the beginning, middle and end of the working length. If the process includes regripping, include one feature before and another after it. On long profiles, inspection only at the loading end does not reveal how bow interacts with supports.

Measure the sample from the same datums that matter in assembly. Where practical, perform a trial fit with the mating part or fixture. Do not let the fixture force the section into a shape that hides twist or another deviation that will create stress in the assembly.

Separate material variation from machine condition

Before blaming the material, check a stable master artifact or known control bar. If the reference result has also moved, investigate the system: origins, chucks, supports, nozzle, optics, calibration and the software baseline. If the reference remains stable and the problem returns with the new tube, the batch becomes a stronger hypothesis.

A paired test is useful: cut old and new material with one unchanged setup in a short time window. Balance the order so warm-up or contamination does not coincide only with the second group. Preserve actual parameters, program version, jaw and support positions, and seam orientation.

Do not change compensation, cutting settings and clamping simultaneously. If a result improves after three adjustments, the cause remains unknown and the next batch may need the opposite response.

Automatic correction is useful but not universal

Some tube lasers can measure section deformation or form error and adjust the program. Sensors on a specific machine can reduce the influence of material variation, but the mere presence of such a feature does not prove accuracy for every part and section.

Verify which deviations are measured, over what range, with which configuration, and whether measurement occurs before every cutout or only at selected points. Correction of the external surface may not account for an internal radius, wall-thickness change or seam behaviour. It also does not replace acceptance against the specific drawing.

After a batch change, retain raw sensor data where available and compare it with an independent measurement. If the function is optional, record its exact name and confirm that it was active.

Old compensation can become a new error

Compensation developed for a previous batch contains that batch’s geometric signature. If the sign of bow, amount of twist or seam position changes, the old correction can double the error. Every compensation file therefore needs a revision, a defined material range and a validation date.

Separate machine correction from a part-specific adjustment. The first restores the geometric baseline of the equipment. The second accounts for confirmed behaviour of a particular material and part family. Do not hide a system problem with a local part offset.

For a new batch, first verify the baseline without transferring arbitrary local corrections. If a shift is repeatable and its cause is understood, introduce a controlled correction with a new revision and repeat control sample. Do not copy values across diameters, wall thicknesses or suppliers without evidence.

Establish a sampling plan for the series

After a successful control sample, define the inspection points for starting the production run. Useful checks commonly include the first part, several consecutive repetitions to assess short-term stability, the first part from each new stock length and periodic inspection through the batch. Frequency depends on defect consequences, process capability and detection speed.

Not every feature has the same risk. A critical assembly lock or datum hole may require frequent inspection, while a decorative contour may require less. Separate quantitative data measured as a number from attribute checks that result in pass or fail.

Reduce frequency only after a stable history, not after one good bar. Return to enhanced control after a new batch label, an adverse trend, an unusual value, nozzle replacement, service intervention, a stop after impact or a program change.

Read the deviation pattern

A consistent shift of every cutout in one direction usually points towards a datum, an origin or a systematic size effect. A change along the length may relate to bow, supports, thermal state or regripping. A deviation repeated at one angular position requires examination of angular geometry, the seam and interaction with the chuck.

If individual bars in the new batch show deviations with different signs, one fixed compensation is a weak solution. Better material selection, active sensing or a functionally wider acceptance range may be needed. If only cut quality deteriorates while position remains stable, investigate cutting conditions separately from geometry.

A sequence plot is more informative than an average. Label the batch, bar, longitudinal and angular position, program revision and measurement method. This reveals a step change at the new batch and identifies individual unstable bars.

When can the batch be released to production?

A batch is ready when it has been identified, critical incoming characteristics lie within the agreed range, the control sample passes dimensional and functional inspection, short-term variation is acceptable, and the reaction plan is available to the shift. The conclusion must apply to a defined part family and machine configuration.

Acceptance of one part does not prove that the profile is “perfect.” It confirms only the tested combination of material, machine, program and measurement method. Another feature or assembly may have different sensitivity.

If a result is near a limit, do not start a large batch in the hope that the average will improve. Isolate the material, increase control, clarify the cause, widen the sample or obtain an authorised design decision.

Agree requirements with the material supplier

The purchase order should state the standard, grade, nominal dimensions and any truly necessary additional characteristics. Request bundle traceability, a certificate and marking that survives unpacking. Avoid undefined requests for “laser quality” without measurable criteria.

Discuss seam position and condition, straightness, twist, radii, coating, transport protection and whether mixed heats are acceptable. Incoming requirements should be realistic for the selected standard and inspected by a capable method.

When a production problem is confirmed, send the supplier useful evidence: batch identifier, bar number, cross-section, actual values, photographs, measurement method and functional effect. A photograph of one failed part without datums cannot separate material from process.

Control changes and preserve knowledge

For an approved route, retain the acceptable material range, control sample, approved setup, program revision, inspection instruction, baseline charts and reasons for revalidation. The operator should be able to see whether the current profile may be combined with previous stock or requires another check.

Record the comparison result, not only the fact that a batch changed. After several deliveries, the company can see which characteristics really predict a problem and remove unnecessary measurements. Historical data does not, however, cancel verification after a significant change.

Test handover between shifts practically: another operator should be able to locate the batch, establish the orientation, cut the control sample, measure it and make the decision from the documentation. If verbal hints remain necessary, the process still depends on one person.

Practical checklist for a new batch

1. Match the marking, certificate, supplier, heat and number of bundles. 2. Do not mix material before assigning the internal batch identifier. 3. Select characteristics critical to the current part. 4. Measure representative bars at several cross-sections. 5. Check a reference to exclude an obvious machine-state change. 6. Cut a control sample in sensitive angular and longitudinal zones. 7. Record actual values and perform a trial fit. 8. Do not transfer old part compensation without confirmation. 9. Define series sampling and the reaction plan. 10. Preserve the material identity, configuration, data and decision as one evidence package.

This sequence is not a universal acceptance plan. Sample size, characteristics and tolerances are determined by the drawing, contract, risk and current quality system.

Minimum release record

The decision to start a new batch should be concise but reproducible. Record the part number and drawing revision, internal batch identifier, supplier and material designation, selected bar identifiers, machine and configuration, program revision, measurement method, actual values of control characteristics and responsible person. State the differences from the previous batch and why they are acceptable.

Three decisions are useful. “Accepted” allows work within the agreed range. “Accepted with enhanced control” requires a higher measurement frequency or limits on orientation, length or zone. “Hold” prevents mixing and series cutting until the situation is clarified. “Seems fine” gives the operator no actionable rule.

If a new compensation produced the result, retain both original and corrected geometry, the reason for the change and the control part. The compensation belongs to a defined material range, not to a general profile name. It does not automatically remain valid after a change in heat, supplier, wall thickness or support condition.

Retain a material sample or control part when a repeat order has high risk or the supplier is investigating a cause. The quality system defines retention time. A photograph without scale and batch identity does not replace a measurement record.

This release record turns a one-time check into knowledge for the next delivery. It also allows batch trends to be compared without confusing material variation with service work, setup changes or program revisions.

Close the batch after the order

At the end of the run, compare the initial control sample, intermediate measurements and final control part. If the trend remained within the agreed limits, record the actual material range and result, but do not narrow the incoming specification from one successful delivery. If drift appeared, separate unused bars, preserve their batch identity and open a cause analysis before mixing them with the next delivery.

The closing record should include scrap, recuts, manual corrections and stops related to material. This allows the supplier to be assessed through real route stability rather than the certificate alone. One isolated problem, however, does not prove systematic material nonconformity until the equipment, program and measurement have also been checked.

Safe limits

This article does not establish universal cutting settings, service calibration or authority to change a drawing. Work must follow the machine manufacturer’s documentation, an internal risk assessment and an approved inspection plan.

Need to manufacture a batch from new section stock?