Changeover time is the time from the last good unit of the old product to the first good unit of the new product, at full rate. Start a clock at the first, stop it at the second, and record the phases in between. Do it for many changeovers, not one.
What counts as changeover time?
Many plants time only the stop. That is too short. A line that restarts but makes scrap for 20 minutes has not finished the changeover.
The most useful definition, and the one used in this guide, has two markers:
- Start: the last good unit of product A.
- End: the first good unit of product B, running at the normal rate.
Everything between is changeover. It includes clearing, cleaning, tool change, adjustment, trial pieces and the wait for approval.
Pick this rule once and write it on the standard sheet. If two crews use two rules, their numbers cannot be compared.
Where it sits in OEE
Overall Equipment Effectiveness (R. C. Hansen, 2001) calls a product changeover a planned stop that is part of normal operation (stop time, operational). It stays inside scheduled time. It is not excluded.
The same book says to calibrate reported changeover times against actual clock time. A time written from memory is not a measurement.
Nakajima's six big losses call this loss setup and adjustment. See the six big losses.
How to time a changeover
- 1Write the two markers. Last good unit of A, first good unit of B at rate. Put them on the line sheet.
- 2Choose the phases. For example: stop and clear, remove old tooling, fit new tooling, adjust and trial, wait for first good unit.
- 3Time it from a clock, not memory. Use a stopwatch, a shared wall clock or a timestamp from the machine. Write the time when each phase ends.
- 4Film the full changeover. Hansen recommends video of the whole changeover, including preparation and follow-up. It shows waiting and walking that nobody mentions.
- 5Record the context. Product from, product to, crew, date, and anything unusual such as a missing tool.
- 6Repeat. Time at least 5 to 10 changeovers, with different crews if you can.
- 7Look at the spread. The gap between the fastest and slowest is the first thing to explain.
The film is the most useful tool here. The video method is described in Overall Equipment Effectiveness (R. C. Hansen, 2001). A team watches it together and puts each step and its time on a timeline.
What to write on the changeover sheet
Keep the sheet to one page. It should take under a minute to fill in.
- Date, line and crew.
- Product from and product to, with the format or size.
- Clock time of the last good unit of product A.
- End clock time of each phase.
- Clock time of the first good unit of product B at rate.
- Notes: missing tools, waiting for people, parts that did not fit.
If the machine can time-stamp stops and counts, take the two markers from it and add only the phases by hand. For the wider picture of logging stops, see how to track downtime.
Worked example
Example: a packing line has six changeovers in two weeks. Times are taken from the last good unit to the first good unit.
| Changeover | Minutes |
|---|---|
| 1 | 52 |
| 2 | 61 |
| 3 | 47 |
| 4 | 74 |
| 5 | 55 |
| 6 | 49 |
| Total | 338 |
Mean = 338 / 6 = 56.3 minutes. Median = (52 + 55) / 2 = 53.5 minutes. Fastest 47, slowest 74.
The spread is 27 minutes. The same job takes 57% longer on a bad day (74 / 47 = 1.57). That gap is a better target than the average, because the best crew has already shown it can be done.
Split it into phases
A single number tells you how long. Phases tell you why. Example: a typical 55-minute changeover from the film.
8 + 12 + 15 + 14 + 6 = 55 minutes. The shares add up to 100%.
The tool swap (remove plus fit) is the obvious half, at 27 minutes. But adjusting and trial pieces take 14 minutes, a quarter of the time.
That points to set-point boards and fixtures, not faster hands.
Next, mark each step as internal (only possible while stopped) or external (can be done while the line runs). That sorting is the heart of SMED. The method is in SMED changeover reduction.
Check reported time against real time
Example: the shift sheet says the changeover took 40 minutes. The machine timestamps show the last good unit at 10:05 and the first good unit at 11:01.
That is 56 minutes. The sheet missed 16 minutes of trial pieces and waiting.
This is why Overall Equipment Effectiveness (R. C. Hansen, 2001) says to reconcile reported times with clock time. A baseline built on remembered numbers will make every later improvement look smaller than it is.
What to do with the number
- Cost it. Multiply minutes by changeovers per week, then by your cost per minute. The changeover cost calculator does this.
- Set a baseline. Use the mean and the best result together.
- Sort by product pair. Some pairs are much slower than others. Tell the scheduler which sequences are short.
- Report it every week. One chart, one line: mean and best changeover time.
Example: 5 changeovers a week at 56.3 minutes is 281.5 minutes, or 4.7 hours of scheduled time. Cutting the mean by 20% saves about 56 minutes a week.
Cutting a changeover also needs a good restart. Any slow running or scrap after the end marker belongs to startup losses.
Common mistakes
- Timing the stop only. The line looks fast and the scrap after restart is blamed on production.
- One measurement. One changeover is a story. Ten changeovers are data.
- Mixed rules. Cleaning counted by one crew and not by another.
- Timing the best crew. The best case is a target, not the baseline.
- Using a code called "other". Give changeover its own reason code so it never lands in a general bucket.
Changeover numbers drift when they depend on memory. Fabrico takes the stop and restart from the machine signal, and the operator tags the cause at the line, so each changeover has a clock time. See how Fabrico logs every stop with its cause.
What people ask next
Should changeover time include the first-piece approval?
Yes, if the line cannot run without it. The changeover ends at the first good unit at rate, so the wait for approval is part of it.
What is the difference between changeover and setup time?
In practice they are used for the same thing. Changeover is the full switch from one product to the next. Setup is often used for just the preparation of the new product.
How many changeovers should I time?
At least five to ten for a baseline, spread over different crews and product pairs. A single measurement can mislead.
Is changeover planned or unplanned downtime?
Planned, because it is a normal part of running different orders. It still costs capacity and you can reduce it. See planned vs unplanned downtime.
How much can I cut it?
Overall Equipment Effectiveness (R. C. Hansen, 2001) expects 20 to 50% the first time a team works on a changeover, then about 10% for each repeat once the new method has been stable for 6 to 12 months. Your result depends on how much waiting there is today.
Related guides: Tool Change Downtime: Measure It and Cut It · Adjustment Losses After a Changeover Explained · Cycle Time Variation: Causes and How to Measure · What Is a Good MTTR? Set It From Your Own Data · What Is a Good MTTR? Set It From Your Own Data