Part sorting after cutting: when it becomes the cell's main constraint ART-147

Laser output is not yet a completed set

The controller reports that the program has finished. For production, this is only an intermediate event. The remaining work is to:

1. gain safe access to the sheet; 2. separate acceptable parts from the skeleton and small scrap; 3. avoid damaging the surface or geometry; 4. identify the item and revision; 5. count the actual quantity; 6. isolate parts that require inspection; 7. assemble the order set or the set for the next operation; 8. label the container or pallet; 9. transfer the routing status; 10. remove the skeleton and clear the area.

If planning treats every part as ready when the NC program ends, the completion time looks better than reality. A more accurate final event is the confirmed transfer of an identified set.

Four types of constraint

Physical. There are too few people, tables, containers, lifting aids or square metres. Large or interlocked parts are difficult to separate, small parts are lost, and sensitive surfaces are damaged.

Informational. The operator cannot see which geometry corresponds to an item, where to place the set or which revision is current. Similar parts become mixed.

Organizational. Sorting starts only after several sheets accumulate, priorities change, and responsibility between the laser, warehouse and next operation is unclear.

Quality-related. There is no simple rule for what to accept, what to hold for inspection and how to record recutting. The employee either delays the whole set or passes a doubtful part onward.

Automation can primarily help with physical movement and repeatable stacking. It cannot correct a wrong drawing revision, a missing route or contradictory quality criteria.

Metrics that demonstrate the constraint

Measure the time and material flow, not merely whether “people are always busy.”

| Metric | Definition | What it shows | |---|---|---| | time from program completion to a free pallet | minutes between the machine event and readiness for the next cycle | direct machine blocking | | time to a complete set | from completion of the first or last sheet to set confirmation | actual elapsed time after cutting | | work in progress at separation | sheets, nests or standard hours waiting | queue accumulation | | labor time per sheet or order | actual separation labor | labor intensity | | share of mixed or unidentified parts | cases where traceability is lost | information defect | | damage after cutting | defects created during removal or stacking | part-handling quality | | recutting caused by loss or confusion | confirmed replacements | hidden cost | | waiting at the next operation | time when bending or welding waits for a set | system-wide impact |

MTConnect distinguishes states for waiting on material and part unloading. This is not a complete method for every laser, but it confirms an important semantic difference: program execution and completed unloading are separate events.

A simple two-week test

Do not begin by purchasing a system. For 10 working days, record the following for every nest:

• order and nest IDs;
• program completion time;
• start and finish of access and separation;
• pallet-release time;
• number of people involved;
• complexity category;
• set-confirmation time;
• number of exceptions: stuck, lost, damaged or doubtful parts;
• delay reason;
• destination of the set.

Separate “pallet blocked” from “set incomplete.” With exchange tables, the laser may continue cutting while the previous sheet is separated. Machine availability does not fall immediately, but work in progress grows and the problem appears after several cycles or at the next operation.

Calculate not only the mean but also the median, 80th and 90th percentiles and a list of the worst cases. One large sheet can distort the average. Group results by format, thickness, part count, minimum size, share of small items, protected surface and number of orders per nest.

An example flow map without a universal threshold

Assume one nest passes four control events. At 10:00 the controller confirms program completion. At 10:12 the previous sheet is removed from the pallet area and the machine can accept the next job. At 10:38 physical separation is complete. At 11:05, after counting and finding two similar items, the set is transferred to bending.

The data provides three intervals rather than one:

• 12 minutes from program completion to pallet release — direct impact on the next machine cycle;
• 38 minutes until physical separation is complete — labor required for the post-cut operation;
• 65 minutes until the set is complete and accepted by the next operation — actual time until the next process can proceed.

If only the first 12 minutes are recorded, the cell appears fast even though bending waited for more than an hour. If only 65 minutes are recorded, it may appear that the laser was blocked throughout. The map separates machine, manual and inter-operation constraints.

Add a reason and action owner to every interval. In the example, the first 12 minutes may be a normal pallet exchange; the delay until 10:38 may result from many small parts; the remainder may be an information search for similar items. The conclusion then becomes specific: “for mixed nests of this type, identification during set assembly delays the next operation,” rather than “sorting is slow.”

The figures are illustrative, not standards. Determine the main constraint by recurrence and system impact. After 10 working days, build a table:

| Nest group | Observations | Median pallet-release time | Median set-completion time | Main confirmed exception | Impact on next operation | |---|---:|---:|---:|---|---| | typical series | actual | actual | actual | from the log | queue or no queue | | many small parts | actual | actual | actual | from the log | queue or no queue | | mixed orders | actual | actual | actual | from the log | queue or no queue | | large or sensitive parts | actual | actual | actual | from the log | queue or no queue |

First assess sample quality: are difficult sheets missing from the log, are events defined consistently, and did staffing change? Then find a group in which the delay repeats and creates a queue. Only after that should you compare options: change nesting rules, prepare containers, add a picking map, reallocate resources or assess the technical feasibility of automatic picking.

An automation decision cannot be derived from one percentile. It requires a portfolio of geometries, the exception share, labor time, required area, integration and the effect on completed sets. ART-147 ends with evidence of the constraint; system selection belongs to ART-148.

Why mixed nests make sorting harder

Combining different orders may improve material use but raises the information burden after cutting. An employee must recognize similar geometries, place them in different containers and confirm quantities. Metal savings can be smaller than the additional labor, delay and recutting risk.

This does not mean mixing is prohibited. Include downstream cost in the nesting rule:

• maximum number of orders per sheet;
• no mixing of critically similar parts;
• separate sheet zones;
• a printed or digital picking map;
• in-process marking where technology and requirements permit it;
• containers prepared before the cycle ends.

ART-141 examines nesting economics. The boundary for ART-147 is that high material utilization does not guarantee fast set completion.

A standardized work area

Separation is accelerated by preparation, not by telling people to “work faster”:

• a safe access area consistent with equipment instructions;
• containers bearing the order ID and next operation;
• a visible holding place for doubtful parts;
• separate routes for skeletons and small scrap;
• suitable lifting or vacuum aids used within their documentation;
• the current revision of the nest map;
• space for quick inspection and counting;
• an unobstructed route to the next point;
• a completion rule identifying who confirms the set.

An article cannot prescribe a safe lifting method for a specific part. Mass, sharp edges, temperature, centre of gravity, gripping area and equipment require a local risk assessment and the manufacturer’s instructions.

Use a picking map instead of searching by eye

The map should show:

• nest ID and version;
• sheet outline and orientation;
• item, quantity and order;
• picking sequence or zone;
• destination container or pallet;
• parts requiring separate inspection;
• remnant that must be registered;
• any change made after the map was printed.

If the program is regenerated, the old map must become invalid. This is the same version discipline as the cutting-condition library in ART-144. A printout without a version is a risk.

How to organize set assembly

Choose the flow unit. It may be an order, route set, assembly or batch for the next operation. All parts for one customer order do not necessarily need to stay together if some go to bending and others to welding, but the rule must be explicit.

Every container should carry:

• a stable ID;
• order or assembly;
• items and expected quantity;
• actual quantity;
• inspection status;
• next operation;
• date, time and responsible person;
• reference to an exception or recut.

Barcode scanning can reduce manual entry, but a code does not compensate for a faulty data structure. Define the entities and events first.

Visual queue management

Four physical or digital states are useful:

1. awaiting separation; 2. being separated; 3. set incomplete or under inspection; 4. ready for transfer.

A sheet must not disappear from the list when separation begins. Completion is recorded only after quantity and route confirmation. Show “pallet released,” a machine event, separately from “set complete,” a production event.

When sorting really is the main problem

Sorting is a constraint when several conditions are present:

• the queue grows consistently during typical shifts;
• the machine or exchange table waits to be released;
• downstream operations wait for parts although cutting is complete;
• overtime or additional people are needed specifically for separation;
• set assembly accounts for the largest part of elapsed time after the program;
• sorting errors cause recutting;
• increasing laser speed expands the queue rather than output of complete sets.

If a queue appears once because of an unusual batch, capital automation may not be justified. Use a load profile rather than the worst single episode.

What to change before automation

1. Standardize IDs for orders, nests, items and containers. 2. Prepare containers before cutting ends. 3. Introduce a revision-controlled picking map. 4. Separate physical removal, inspection and set assembly. 5. Establish a rule for similar parts. 6. Reduce mixed orders where economically justified. 7. Plan separation as a resource, not as “support work.” 8. Measure transfer to the next operation. 9. Eliminate repeated entry of the same data. 10. Identify exceptions that will remain manual.

Salvagnini offers both manual and automatic sorting systems; according to its description, the manual TN device allows parts to be separated while cutting continues. This confirms that intermediate organizational and mechanized solutions exist between fully manual work and robotic picking.

When to assess technical feasibility

After stabilizing the data, prepare a representative set of nests:

• the most frequent;
• the worst by time;
• small parts;
• large or heavy parts;
• long narrow parts;
• similar items;
• sensitive surfaces;
• different thicknesses;
• parts that do not separate cleanly from the skeleton;
• mixed orders.

The supplier should demonstrate which parts the system can lift, recognize, stack and transfer, and which will remain manual. Do not extend a promise based on one demonstration sheet to the whole product range.

Common mistakes

• measuring only the time until the next machine start;
• ignoring search and set assembly;
• optimizing material use without downstream cost;
• mixing map revisions;
• stacking parts without a container ID;
• recording a defect only after recutting;
• treating busy people as proof of a constraint;
• purchasing sorting without analyzing actual geometries;
• ignoring skeletons, small scrap and remnants;
• failing to plan space for completed stacks.

Checklist

• [ ] Program completion, pallet release and set completion events are defined.
• [ ] Two weeks of nest-level data has been collected.
• [ ] The queue is measured as work, not merely as sheet count.
• [ ] The geometries that create the greatest labor demand are known.
• [ ] The picking map has an ID and version.
• [ ] Containers are prepared and labeled.
• [ ] Doubtful parts are physically separated.
• [ ] The next operation confirms acceptance.
• [ ] Handling damage is recorded separately.
• [ ] The mixed-nest rule accounts for sorting.
• [ ] Safe movement methods are defined locally.
• [ ] A representative set is ready for the automation study.

Conclusion

Sorting is a constraint not when it is merely unpleasant, but when it delays the next cycle or transfer of an acceptable set. The best first step is to make three time points, the queue and exceptions visible. Identification, prepared containers, a picking map and mixed-nest rules can often remove part of the loss. What remains forms an honest technical brief for automation.

Safe boundaries

• The article gives no instructions for manually lifting specific parts.
• It does not claim automation is suitable for the whole product range.
• It gives no universal return-on-investment or productivity figures.
• ART-148 remains a separate topic on automatic unloading.

Need help selecting equipment?