But wouldn’t it be easier just to use an effective cycle time? Let’s look at effective cycle time vs. EPEI.
The cycle time should be the lowest repeatable time to produce a good unit. This should be confirmed with a time study for manual processes and is often the rated cycle time for machines. With an effective cycle time, the cycle time is increased to account for downtime, yield loss, and rework in order to correctly reflect rate at which the process is actually able to produce over a longer period of time.
Now we come to the awkward issue of changeovers. Of course from a lean perspective, changeovers should be reduced as much as possible; however, this is not completely possible in all cases.
To calculate an effective cycle time where there are changeovers, the changeover time would have to be factored into the cycle time for each unit. In order to do so, a period of time must be selected and the number of changeovers during that time must be estimated. From there, the rest of the math is easy. The total number of changeovers X time per changeover / total units gives an amount of time that the effective cycle time must be increased to accommodate changeovers.
First challenge: Over what period do we calculate the average number of changeovers? A day, a week, a month, longer? Certainly over a longer period, the average is going to be closer, but we don’t actually run our operation over a longer time; we run by hour, by shift, or by day. However, if we choose a shorter time, we’ll commonly find that the number of changeovers is very different from the average.
Second challenge: When we are running in production, we’ll find that we are actually producing faster than our effective cycle time. And then, when we stop for a changeover, our effective cycle time is zero. So, you can’t take a time measurement against an effective cycle time on an individual unit basis but only over some average.
Third challenge: The dynamic nature of high mix/low volume value streams means that we cannot simply use an expected average number of changeovers reliably. While the total number of units may be equal, the mix of those units may have far more changeovers or far less changeovers in any period. And since not all changeovers are necessarily equal, the mix may even further skew our effective cycle time.
EPEI provides a more dynamic solution that resolves each of these challenges. First, the time period to pick for changeovers is itself the length of the calculated interval. Second, by not including changeovers in an effective cycle time, we can track actual performance vs. effective cycle time more precisely.
Third, EPEI can be calculated for a specific mix and volume of products so setting an average number of changeovers is not necessary.
Next time, we’ll look at how EPEI sets both our lot size and our supermarket level.