In our exploration of high-mix, low-volume (HMLV) value streams, we see that changeovers between items occur all the time when we have multiple items being produced in the same value stream. Our goal is always to eliminate changeover all together since it is non-value add but that sometimes isn’t possible. Changeovers are not relevant if our take time exceeds cycle time + changeover time. In that case, we can afford a changeover on every unit and we can run production in any sequence.
In my work with HMLV value streams, I’ve found that there are a few general kinds of changeovers. Understanding the kind of changeovers that you are faced with can provide some insights into the strategy for how best to deal with them and how to incorporate them appropriately into your EPEI calculation.
First kind of changeover is what I’d think of a normal changeover, the time between the production of the last unit of one item and the first good unit of the next item. Often times in a HMLV operation, the changeover time is the same between any two parts and so single changeover time is sufficient for all. In other cases, the changeover time is different depending on the part being run next regardless of what was run before it. In this case, a single changeover time by item is sufficient. In other cases however, the sequence of items in production may make a difference in the changeover times requiring a matrix of from/to items with specific changeover times depending what was run previous and what comes next.
The second kind of changeover is what I think of as a major changeover that occurs between groups of items. In other words, this changeover is the time between the last unit of one item in one group of items and the first good unit of the next item in a new group. A classic example is a painting process where there may be no changeover between multiple items of the same color but a changeover is needed for a new color. This also can occur with heat treat processes for temperature changes, changing machine settings for gauges of metal in a mill or rolling operation. Combining the major changeover with the idea of sequencing and a matrix of changeover times support many common manufacturing processes where the lowest overall changeover is going from light to dark, from low temperature to high temperature, from wide to narrow, or from thick to thin.
The third kind of changeover is what I call a useful life changeover. This captures situations where intermittent activities are required that stop operations after a period of time or after a certain number of units. Often times this has to do with tool changes, recalibration, or lubrication. For example, a slitting process may require blades to be changed every 4 hours or every 1200 meters of cutting. While this may not be technically a changeover, it is an intermittent interruption in operation that can be accurately planned in advance based on the mix and volume of products required.
These three different kinds of changeovers in combination with sequencing and a matrix of changeover times has been able to cover all of the HMLV value streams I’ve worked with.
If it sounds like this can get complicated, you’re right. It can. But the more precisely you can calculate the EPEI, the less you will have to provide a larger buffer than is absolutely essential. And therefore your value stream will run with its lowest level of inventory and shortest lead time.
We’ll ease up on the complexity next time and talk about how to use EPEI as a measure of capacity.