Measurement and indicators

Cycle time

In French: Temps de cycle

Cycle time is the duration between two successive outputs of the same station or the same process. It describes a rate, is recorded with a stopwatch, and is never summed up by its mean.

What exactly is measured

Cycle time is counted from one end to the next end: the moment one unit is finished, then the moment the next one is. That choice of boundaries looks trivial and is not — measuring from the start of a unit to its own end gives a different quantity, which ignores what happens between two units.

It therefore includes everything the station actually does to produce one more unit: handling, picking up again, waiting for the machine, the entry that follows. Those are the components the record brings out, and they often weigh more than the operation itself.

It is recorded at the station, by observing, never by asking. A time given from memory is systematically the time of a good cycle with no incident, that is, the best case and not the usual case.

What the mean hides

A cycle time is observed over a series of consecutive measurements, not over three readings. The spread obtained is at least as informative as the central value: two stations with the same mean, one consistent and the other erratic, do not pose the same problem at all and do not call for the same actions.

An unstable station disrupts those that follow it, even when its mean is good, because downstream has to protect itself against the long cycles. Consistency therefore has a value of its own, independent of speed.

High values deserve to be noted with their cause rather than discarded as abnormal. A cycle that takes twice as long because the part was badly presented is not a measurement aberration: it is the record of an event that will happen again, and its frequency is a datum of the problem.

What it is used for once recorded

It serves first to place each station in relation to the pace demand imposes, which immediately points to the stations in difficulty and those with margin. That comparison is the basis of any line rebalancing.

It then serves to establish real capacity: the number of units a station can produce over a period is derived from the observed cycle time, not from the theoretical time written in the routing. The gap between the two is frequent and explains many schedules that do not hold.

It finally serves as a benchmark before and after a change. It is the measurement that makes it possible to say whether a change of layout or tooling produced an effect, and by how much — provided the initial situation was recorded, which can never be made up afterwards.

The common mistakes

Recording three cycles and concluding. Three measurements show no spread and almost always land on favourable cycles, because the presence of the observer is enough to keep the ordinary disruptions away for a few minutes.

Timing the person instead of the station. Apart from the climate it creates, the measurement becomes unusable: what is sought is the content of the work and its irregularities, not the speed of the person carrying it out.

Retaining only the central value. It is enough for a rough capacity calculation and says nothing about the problem to treat, which almost always lies in the spread of the values.

On the shop floor — Medical device manufacturing

At an assembly station, the routing said 52 seconds. Thirty consecutive cycles gave values from 49 to 96 seconds, the longest all corresponding to the moment when the operator had to open a new pack of components. The subject was neither the gesture nor the person, but the packaging supplied from upstream.

The other tools illustrated by this trade: healthcare and life sciences.

Frequently asked questions

How many cycles should be recorded?

Enough for the spread to appear, which means several dozen consecutive cycles on a repetitive station, and more when the work varies from one unit to the next. The practical criterion is the stability of what is observed: as long as a new reading still brings out unseen durations, the series is not long enough.

Does cycle time include breaks and breakdowns?

No, it describes the station running: planned stoppages and breakdowns are handled separately, in the availability indicators. Mixing them produces an artificially long cycle time that corresponds to no observable reality, and that then prevents knowing whether a deviation comes from the work content or from the unavailability of the equipment.

How do you measure a cycle time on non-repetitive work?

By relating it to a comparable unit of output rather than to a gesture — a file processed, an order entered, a call closed — and by accepting that the spread will be wide. The measurement remains useful for sizing a capacity, provided it is segmented by type of request: mixing simple and complex cases produces a mean that describes no real case.

Should the aim be to reduce the cycle time of every station?

No, only that of the station that limits throughput changes anything in the overall result. Improving a station that already has margin produces only additional waiting downstream, and often work in progress. It is the most frequent mistake after a complete record, because the temptation is to treat every deviation observed.