Half of a good time study happens before the stopwatch starts. A study that was not prepared produces a number nobody accepts, and the second attempt is always harder than the first.
Prepare before you measure
Pick one station with a method that is stable and documented. Timing an unstable method measures the chaos, not the work.
Fix the purpose in writing. A planning time and a time used for pay are not the same promise, and a time taken for one and later used for the other ends the cooperation you need. Say which one it is before you measure, and then keep that answer for the whole study.
Inform the people. Operators and the supervisor must know what is measured and why, and that they see the record first. Where the rules of your site require the employee representation to be involved, that is settled by the responsible function before the invitation goes out, not at the station on the morning of the study.
Break the operation into elements
An element has a start and an end point you can see or hear, and runs roughly five to sixty seconds. Write the elements on the sheet before the first cycle and do not invent new ones while timing.
Four rules keep a breakdown usable. One activity per element, and if you cannot say in four words what happens, the element is too big. Separate manual work from machine running time. Give an element that occurs only every tenth unit its own line and a frequency. And keep the elements in the order of the work, so the sheet can be read at the station.
At station S3 in area M2, assembling a hydraulic control block came out as six elements: take the housing from the bin, insert six cartridges with the torque tool, fit seals and plugs, fasten the cover with four screws, scan and label, and set the block in the test carrier.
How many cycles
The number follows from the cycle length and then from the spread you actually find, not from the time available. Below half a minute, plan twenty to twenty-five cycles, because the reading error is large compared with the cycle. Between half a minute and two minutes, fifteen to twenty. Above two minutes, five to fifteen, and time rare elements separately with their frequency.
After timing, check the precision. Two standard deviations divided by the mean times the square root of the count gives the half width of the 95 percent interval as a share of the mean. Above five percent, add cycles. Four times the cycles halve that figure.
The cycle that went wrong
Fifteen cycles were timed at S3. Cycle eleven took 196 seconds: a cartridge jammed and the torque tool had to be re-set.
| Measure | Value |
|---|---|
| Mean of all fifteen cycles | 153.5 s |
| Mean of the fourteen cycles used | 150.4 s |
| Standard deviation of those fourteen | 2.6 s |
| Relative precision reached | 0.9 % |
Three rules decide whether that exclusion is honest. A cycle is excluded only with a written reason that names an event, never because the value looks high. The cycle is never deleted: it stays on the sheet, marked as excluded, so an audit can see it. And the exclusions are counted, because three interruptions in fifteen cycles are not an outlier problem, they are a process problem.
Read the spread, not just the mean
The widest element at S3 was inserting the cartridges, at 1.8 seconds of spread. The cause was found at the station within ten minutes: two cartridge types with almost identical heads. Color marking removed most of that spread in the next study.
A wide element points at a missing method detail. That finding is often worth more than the time itself.