Changeover Tracker

What this tool does

CM8-289 is the ongoing log of every changeover you run. One row per changeover: the machine, the products either side of it, the standard time allowed, the minutes it actually took, who set it, whether the first-off passed first time, and what held it up. From that the tool works out the trend against your target, the average by machine, the minutes lost to each delay reason, and the spread the average hides.

Everything runs inside this single file — no account, no upload and no network request of any kind, so machine, product and setter names stay on the computer you are using.

Why logging every changeover matters

A SMED study answers the question "what is possible?" It takes one changeover apart task by task, separates the work that genuinely needs the machine stopped from the work that does not, and projects a much shorter changeover. It is the right way to design a new method, and the SMED Changeover Analysis tool in this range does exactly that job.

This log answers a different question: what actually happens? The gap between the two answers is where nearly all the remaining improvement lives. A study proves a press can be changed in twelve minutes; the log shows a mean of twenty-six, because the trolley is staged on earlies and not on nights, because one setter follows the new sequence and two never saw it, because the jig is borrowed by the rework bench every second week. None of that appears in a study, and all of it is fixable once it is visible.

A log is also the only way to prove an improvement happened. Without one, "changeovers are quicker now" is an opinion; with twelve dated rows and a falling monthly mean it is a fact.

What counts as a changeover, and where the clock starts

Measure from the last good part of the old job to the first good part of the new one — not from when the setter arrives, not to when the tool is clamped. That captures everything the machine did not produce: the trial parts, the adjustments, the wait for a first-off signature. Anything narrower hides the part of the changeover most worth improving.

Log every changeover, including the ones that went badly, the ones on nights, and the ones abandoned and restarted. A log containing only tidy changeovers is not a measurement, it is a highlight reel.

Planned versus actual, and an honest standard time

The planned minutes are the standard time your schedule allows for that changeover on that machine — the number the planner uses when deciding whether a job fits in a shift. Set it honestly. A standard everybody beats is padded, and it hides lost capacity inside a number nobody questions. A standard nobody has ever achieved is equally useless: people stop reading the variance and the planner adds a private buffer on top. The workable standard is the time a competent setter achieves when the preparation has been done. Review it when the method changes — the sample log shows a press whose standard drops from 25 minutes to 20 once the tooling trolley became normal.

Variance is actual minus planned, so a positive variance is time lost. Efficiency turns that into a percentage: 100% means you hit the standard exactly, above it means you beat it.

Delay reasons — most changeover time is lost before the machine stops

Ask why a changeover took forty minutes instead of twenty and the answer is very rarely that the clamping was slow. It is that the tooling was still on another machine, the material had not been brought to the line, the program revision was not loaded, or the setter was finishing something else. Every one of those is external work — work that could have been done while the machine was still running the old job — sitting inside the stop because nobody owns the preparation.

That is why the delay reasons here are mostly preparation failures — and the cheapest problems in the building to fix: a staging trolley, a shadow board, a rule that the next job's pack is printed when the current run has thirty minutes left.

The tool counts a delay reason as costing time only where the changeover ran over its standard, and it refuses a reason on a changeover that came in at or under standard. Use "None — ran to plan" freely; a log where every row carries a reason is one where the reason field has stopped meaning anything.

First-off right first time — the hidden changeover cost

The tooling swap is visible and everybody watches it. The adjustment cycles afterwards are invisible and frequently cost more. Three attempts at a first-off — run a part, measure it, tweak, run another — can take longer than removing and refitting the tool, and it happens after everyone has stopped paying attention because the machine is technically "running".

So the log asks two questions. First-off passed first time is the flag; adjustment cycles counts the attempts before it did. The tool will not accept both, because they contradict each other.

A low right-first-time rate usually means settings are being found by trial rather than recalled — no recorded parameter set, no locating pins, no boundary sample at the machine. Each has a fix that removes the adjustment cycles rather than making the setter faster at them.

Read the spread, not the mean

The distribution chart matters more than the trend line, and it is the one most people skip. A mean of 22 minutes built from a cluster of twelves and a cluster of forties is not one process performing averagely — it is two processes, one of which works. The question stops being "how do we get faster?" and becomes "what is different about the twelves?" That usually has an answer somebody already knows: a particular shift, product pair or setter, or simply whether the trolley was staged.

Consistency is worth more to a planner than speed: a changeover that reliably takes twenty-five minutes is easier to schedule around than one averaging eighteen with a tail out to fifty.

Comparing machines, and what a stuck machine means

The by-machine table gives each machine its count, average, best, worst, over-target count, first-off rate and total minutes. Read the best column as a target: if a machine has ever been changed in seventeen minutes, the gap between its best and its average is improvement that needs no new equipment.

A machine whose average will not move — the sample log has one, stuck in the mid thirties — almost never has a speed problem. It has a recurring, unaddressed cause: tooling shared with another operation, a jig with no home, or a method never taught to the second and third shifts. Look at its delay reasons before you look at its people. In the sample, three of that line's four changeovers were held up by tooling that was not staged, and one came in with a clean first-off — the setter was fine, the preparation was not.

Machines with genuinely different changeover work are not directly comparable. Use the ranking to decide where to spend attention, not to rank people.

The formulas

The average changeover is the mean of the actual minutes of every timed changeover in the current filter. Minutes lost to a delay reason is the sum of positive variances on rows carrying that reason, so a changeover that beat its standard never contributes lost time. Where planned minutes are missing, variance and efficiency print a dash.

The spreadsheet workflow

If changeovers are already written on a shift sheet, you do not have to retype them here.

• Spreadsheet template in the toolbar saves a CSV whose headings are exactly this tool's column names, with a guidance row showing what each column expects — the date format, the accepted values for the shift, type and delay-reason lists, and the unit for each number.
• Open it in any spreadsheet program, fill in one row per changeover, and delete the guidance row before saving. Keep the file as CSV.
• Import spreadsheet reads it back. Columns are matched by heading, so their order does not matter and extra columns are ignored; numbers, dates and list values are converted for you. Rows missing a required column, or failing a validation check, are skipped and reported by row number.

The file is read by this page in your browser: nothing is uploaded and the original spreadsheet is not changed. Importing adds to what is already here rather than replacing it.

FAQ

Do we log a changeover that was abandoned halfway? Yes, with the full elapsed minutes and a note. Abandoned changeovers are among the most expensive events on a machine and they vanish from every log that records only completions.

How many changeovers before the numbers mean anything? Per machine, roughly ten before an average is worth quoting and twenty before a trend is. Below that, read the individual rows and the spread.

Should the target be the single-digit goal? No — keep them separate, which is why the settings hold both. The target is what you expect now and what the trend line is drawn against; the goal is where the method is heading.

Saving your work

Changeovers, settings and the report header are written to this browser's local storage as you type. That storage belongs to one browser on one computer: another browser, a private window, a second machine or a clean-up tool that clears site data will not have it.

Treat Export .json as the real save — one file containing everything, which Import .json restores anywhere. Export CSV gives you the log for spreadsheet work. Reset asks twice, then erases everything this tool has stored. There is no undo.

Accuracy & disclaimer

The arithmetic here is simple and the tool does it faithfully. What it cannot do is see the changeovers nobody logged — and those are exactly the ones worth seeing. Watch the count of changeovers logged alongside the average: if the count falls as the average improves, the improvement is in the recording.

Planned minutes are your own standard, not an independent measurement, so variance and efficiency are only as meaningful as that standard. This is a record-keeping aid, not engineering advice — any physical change to tooling, clamping or guarding must be approved by whoever is competent for that machine.