What should a folding carton plant track as downtime?

Every stop on the presses, die-cutters, and folder-gluers, with duration taken from the machine and cause taken from the operator at the time of the stop. Planned makeready and PMs should be recorded separately from unplanned stops so availability numbers mean something. Micro-stops below a set threshold deserve their own category because they behave differently from breakdowns.

Why do micro-stops matter more than they look?

A folder-gluer that stops for ninety seconds ten or twelve times per shift loses a meaningful fraction of an hour, and none of those stops are large enough for anyone to write down. Because manual logs are biased toward long events, the improvement effort aims at breakdowns while the short stops continue unopposed. Automatic capture is the only practical way to see them.

Do we need to replace our machine controls to capture downtime?

Usually not. Most presses, die-cutters, and folder-gluers already expose run state, faults, and counts at the PLC, and those can be read over OPC UA or the machine's native protocol without touching control logic. Older equipment sometimes needs a sensor or a small retrofit, which is normal scope rather than a reason to stop.

How is downtime tracking different from OEE?

Downtime tracking is the input; OEE is one way of summarizing it alongside performance and quality. Plenty of plants compute an OEE number from unreliable stop data and get a figure that looks precise and is not. Fix the capture first, then the summary metric becomes worth arguing about.

Who should assign the downtime reason code?

The operator at the machine, in the moment, from a short list of codes written in the plant's own language. Software can suggest a likely cause based on the fault signature and recent history, but the person confirms it. Automatic classification without confirmation tends to produce clean-looking data that the floor knows is wrong.

Downtime Tracking in Folding Carton Plants

Most carton plants track the four-hour breakdowns and miss the two-minute stops, which is where the shift actually goes.

Downtime tracking in a folding carton plant is the practice of capturing every stop on the presses, die-cutters, and folder-gluers with a cause attached, and it only works when the capture comes off the machine rather than out of an operator's memory at the end of a twelve-hour shift. Most carton plants already track downtime in some form. What they usually have is a paper log or a spreadsheet that records the long, obvious events: the folder-gluer that went down for three hours waiting on a maintenance tech, the press that lost a blanket mid-run. Those events are real, and they are also the minority of lost time. The stops that quietly consume the shift are short, frequent, and unlogged, and no downtime program that depends on hand entry will ever see them.

Where the time actually goes on a carton line

A folding carton plant is a sequence of machines with very different failure signatures, and lumping them into one downtime bucket hides the pattern. The sheetfed press loses time to makeready, plate changes, color matching, and washup. The die-cutter loses it to makeready and setup, stripping problems, and nicks that were cut too light or too heavy. The folder-gluer, which is usually the real constraint in a carton plant, loses time to feeder jams, blanks that fan out of square, glue pot temperature and glue line faults, camera reject clusters, and endless small adjustments on a job that is fighting the fold. Then there is board itself: warp, moisture pickup from a humid warehouse, caliper variation across a pallet, all of which show up downstream as jams that get logged, if at all, as “feeder issue.”

The scale problem is arithmetic. A folder-gluer running at 20,000 pieces an hour that stops eleven times in a shift for ninety seconds each has lost roughly seventeen minutes. Nobody wrote down a single one of those stops, because each one felt like nothing. Across three shifts, five days, and several machines, that pattern is worth real capacity, and it is invisible in every report the plant currently produces. This is the core reason downtime tracking on folding carton lines tends to be directionally wrong rather than slightly wrong: the recorded data is biased toward long events, so improvement work aims at the long events, and the short ones keep running unopposed.

Why manual downtime tracking in folding carton plants drifts

Recall is not measurement. A reason code written at end of shift is a summary of a shift, not a record of events. Durations get rounded, small stops get dropped, and the cause recorded is whatever the operator concluded, not what the machine did.
Reason codes get compressed. When there are twenty codes on the sheet and one shift to fill it out, everything lands in the two or three broadest ones. A plant that finds 60 percent of its downtime coded “mechanical” or “other” does not have a downtime problem, it has a coding problem hiding the downtime problem.
Makeready and downtime blur. Setup on a die-cutter is planned time, but a makeready that ran ninety minutes past standard is partly unplanned, and manual logs almost never split the two. The overrun is where the improvement is.
Nobody trusts a number they entered under pressure. Operators know the log is soft, supervisors know it is soft, and so the weekly review argues about the data instead of acting on it.

What good downtime tracking looks like on carton equipment

The workable pattern separates two jobs that manual logging tries to do at once. The machine reports that it stopped and for how long; the operator supplies why, on a screen at the machine, choosing from codes that match how the plant actually talks. Duration and frequency come from the equipment and are not up for debate. Cause comes from the person standing there, in the moment, which is the only time the answer is reliable.

Getting the machine signal is usually less exotic than it sounds. Presses, die-cutters, and folder-gluers on a modern floor have a PLC, and run state, cycle counts, faults, and speed are typically available there. Harmony connects at the PLC layer, Allen-Bradley and Rockwell, Siemens, Omron, Mitsubishi, over OPC UA or whatever the machine speaks, which is how folding carton plants get machine-level stop data without replacing controls. Older equipment sometimes needs a sensor or a simple retrofit instead, and that is a normal part of the work rather than a blocker. The broader architecture, capture at the machine, reason at the operator, one record for the plant, is covered in our production tracking guide.

The distinctions worth getting right

Planned versus unplanned. Scheduled makeready, breaks, and PMs are not downtime in the same sense as a jam. Keep them separate or your availability number will be meaningless.
Micro-stops versus breakdowns. Set an explicit threshold, often thirty or sixty seconds, and count what falls below it as its own category. Micro-stops are a maintenance and setup signal, not a repair queue item.
Machine cause versus material cause. A folder-gluer jam caused by warped board is a purchasing and storage problem wearing a maintenance costume. If board lot travels with the run record, that pattern becomes visible; if it does not, the folder-gluer takes the blame forever.
Constraint versus non-constraint. An hour lost on a press that is not the bottleneck may cost nothing. An hour lost on the gluer that gates every order costs an hour of plant output. Weight the Pareto accordingly.

What plants do with the data once it is real

The first month of honest data usually reorders the improvement list. Plants commonly find that a single job or a single SKU family accounts for a disproportionate share of gluer stops, which points at a die or a score line rather than at the machine. They find that one shift's stop profile differs from another's on identical work, which is a training conversation, not a discipline conversation. They find that makeready overruns cluster on particular job transitions, which is a scheduling and sequencing fix. None of these conclusions are available from a log that records only the big events. We would be careful about promising a specific number here, because the honest answer is that it depends on how much unrecorded time the plant is currently carrying, and that varies widely between floors.

Starting small and keeping it honest

The way this tends to go wrong is instrumenting the whole plant at once and drowning the floor in reason-code prompts. The better start is one constraint machine, usually a folder-gluer, with automatic stop capture, a short and specific code list, and a weekly review that acts on the top two causes and nothing else. Harmony's published pilot is $15–20K one-time over 4–6 weeks with engineers on-site and working software by week three, and the approach is software and hardware agnostic, so the existing ERP and controls stay in place. Where the system suggests a cause or a corrective action, it proposes and a person approves, because an operator who is overruled by a model stops trusting the screen, and a downtime system nobody trusts is just a slower version of the paper log.