A custom production line is rarely finished the way a standard machine is. It is assembled from processes that have never run together before: a handling robot feeding a bonding station, a vision system checking alignment, a conveyor moving parts between stations with different cycle times. When the line is delivered, the buyer has usually paid for equipment that only a handful of people on the project team have seen run at full speed. The data collected during testing is what turns that demonstration into something you can accept, and later use to argue for a fix.
If you are sourcing automated production line solutions, decide the data list before the build starts. Requirements agreed after commissioning tend to be negotiated down. The items below are what we ask buyers to capture on custom lines, and what we record ourselves when a line we built goes through factory acceptance.
1. Cycle time at every station, not just the average for the whole line
A single figure such as "1,200 units per hour" hides more than it reveals. Record the actual time each station takes to release a part, the variation of that time across consecutive cycles, and where the part waits. Station data shows you which process sets the pace and which one is idle.
Ask for this data as a timed observation over a defined number of cycles, not a single demonstration run. Note the mix of products being run, since a line tested only on the easiest variant will not hold its numbers when the harder geometry arrives.
2. Yield, first-pass yield, and what happened to the rejected parts
Yield on a custom line is usually reported as a single percentage. Ask for two numbers instead: the share of units that pass at the end of the line, and the share that pass every station without being reworked or re-presented. A line can reach an acceptable final yield while quietly reworking a large share of parts at one station, which is capacity and labour you pay for every day the line runs.
Record what happens to a part that fails. Whether it is diverted, retried automatically, or stopped at the station tells you how the line behaves when it stops being perfect, and that is most of its working life.
3. Downtime events with cause codes and recovery time
Log every stop during the test: what triggered it, who cleared it, how long the line was down, and how it was restarted. A stop of 90 seconds caused by a mispicked part and a stop of 20 minutes caused by a controller fault look similar in an availability percentage and completely different in a maintenance plan.
Insist that the codes are meaningful to your operators rather than the builder's internal shorthand. If your maintenance team cannot read the stop history, you cannot plan spares, and you cannot tell a design fault from an operating error when a warranty question comes up a year later.
4. Process window data for the critical step, measured at the process
Every line has one or two steps where the product is actually made: a bonding press, a coating head, a laser path, a cure oven. For those steps, record the controlled variables — temperature, pressure, vacuum, dwell time, alignment offset — as they were measured at the process over the whole test, not as the setpoint shown on the screen.
This is where custom lines most often fail quietly. A station can hold its setpoint while drifting against the tolerance the product needs. Where a tolerance is stated in the specification, verify it with an instrument that is independent of the machine's own control system, and keep the calibration record for that instrument with the acceptance file.
5. Changeover records for the products the line must actually run
Custom lines are usually sold with a product family in mind. Time a real changeover from one variant to another, including the first good part after the change. Record which steps need a technician, which are recipe-driven from the interface, and how many parts are consumed before the line returns to specification.
A changeover performed once by the engineer who designed the fixture is not the same event as a changeover performed by your shift team on a Tuesday night. Ask that the changeover be repeated by someone who has not worked on the build, with the written instructions as the only support.
6. Safety function results and interlock response
For each guard door, light curtain, emergency stop and safety interlock, record the test that was performed and the result: whether the hazardous motion stopped, and how the machine was reset. Keep this with the machine documentation, because it is the part an inspector or an insurer is most likely to ask about.
Also record what the machine does with a part that was in the process when the safety function tripped. Half-finished parts, held vacuum, retained heat and stored pressure are the situations where operators are injured and where lines sit idle while someone works out the correct recovery.
7. Utility consumption and facility interfaces under load
Record measured electrical load, compressed-air demand, water consumption and heat rejection while the line is running at speed, along with the supply conditions the line was tested under. Peak demand and inrush behaviour matter for the electrical design of your bay; average figures do not cover them.
Note anything the builder assumed about your facility: floor space, door width, lifting points, air dryer quality, water hardness, ambient temperature, network access. Assumptions that were not written down before the test become change orders when the line arrives.
8. Software, recipe and data-interface behaviour
Capture the software and firmware versions running during the test, the recipe structure, who has permission to change parameters, and what happens after a power loss. Record whether recipes can be locked, and whether the machine retains traceability records for parts that were in process when it stopped.
If your MES or historian needs to read production data, test that interface during acceptance and save a sample of the exported records. A field that your quality team needs later is much cheaper to add while the line is still in the builder's factory.
9. Traceability: can you follow one part through the line?
Pick a small number of parts before the test, follow them through, and check that the records you were promised actually exist for them: which station processed them, which process data applied, which operator or recipe was active. Then do the reverse exercise — take a record and confirm the machine data supports it.
Traceability that only exists as a screen display is not traceability. Confirm what is stored, where it is stored, for how long, and how you export it without vendor assistance.
10. Long continuous run data and the wear it exposes
A line that works for an hour can still fail at hour eight. Run the line continuously for a defined period, and record temperature rise at drives and bearings, consumable life, drift in the critical process variables, and the points where operators intervene manually to keep things running.
Manual intervention is the number to watch. Each place where an operator compensates for the machine is a hidden labour cost and a future reliability question. A manufacturing automation engineer who has built lines before will look for exactly these workarounds during a long run.
11. Documentation and open issues, captured during the test
Alongside the measurement data, collect the documents that make the data usable later: the test plan with pass and fail criteria, the as-built parts list and any substitutions, the electrical and pneumatic drawings as built, spare-parts recommendations with the wear items separated, the operator and maintenance training records, and the open issues list with owners and dates. All of it should be produced while the line is still on the builder's floor, because after the line ships, every correction becomes a trip.
An open issues list is the most valuable acceptance document you can hold. A line can be accepted with known, dated, owned items; a line accepted with nothing recorded simply moves the same problems into your production schedule.
How we verify before a line leaves the factory
For projects we take on, testing is not a single event at the end of the build. We run functional, performance, stability, safety and lifetime checks against a written plan, then output the records, problem list and improvement suggestions as a package the customer keeps — the same structure we use on our hardware testing services, whether the equipment is an automation cell or a cleaning machine. Where a line needs a purpose-built station rather than a catalogue machine, the same engineering sequence applies: requirement analysis, scheme design, mechanical and electrical design, prototype build, testing and verification, then whole-line delivery, as described on our customized solutions page.
If you are at the stage where a custom line is being specified, send us the process description, the product variants, the target output and the acceptance criteria you have in mind. We will tell you which of the groups above can be measured with the design as you have described it, and which need an instrument or a fixture agreed before the build starts. That conversation costs nothing; discovering the gap during factory acceptance costs a delivery date.
