Total Productive Maintenance (TPS) has become the industry standard for maintaining equipment in the best possible shape, increasingly important as automation increases. The favorite key performance indicator of equipment health is OEE.
This is measured by multiplying three factors:
OEE = Availability x Performance x Quality
Availability = run time/total time
Performance = Total count of parts/target count (based on a standard)
Quality = Good count/Total count
Rob van Stekelenborg finds “Four Reasons to Dislike OEE.” In sum:
- OEE Hampers Learning and Improvement. It is a composite measure making it difficult to understand the “voice of the process.” You must dig deeper to really understand what it is telling you looking at the components. And a stable OEE can mean change in one factor is cancelling out change in another factor.
- OEE Scares Employees Away. The idea of autonomous maintenance is to give team members control and responsibility for their equipment and they simply find it hard to understand OEE. They may even game the system such as running larger batches to reduce changeovers and make OEE look better.
- OEE is a Local Measure. Each OEE metric measures a single piece of equipment rather than how that equipment contributes to the objectives of the value stream.
- “World Class OEE” is a Fallacy. Companies that set an OEE target to be world class, like 85%, will then encourage game playing as in 2 above. There are a variety of ways to game the system to make OEE look good while actually degrading overall performance.
And I would also note that OEE performance is relative to a base line for a given piece of equipment therefore it is specific to that equipment and not comparable across departments or plants.
These are all excellent points, but to me the biggest problem is that OEE is the product of three different things which almost guarantees confusion and makes it difficult to establish targets for improvement. While Gary Convis and I were writing The Toyota Way to Lean Leadership he became the CEO of Dana (truck chassis parts supplier). He quickly learned that the company was operating without KPIs that he could see from his level to the work group level and worked with a team to develop and pilot KPIs by region. One that survived was OEE. I was surprised and asked Gary about it mentioning I have never seen that measured in Toyota. He said it was an accepted measure within the Dana culture and he wanted to support what they have used and believe works.
Within one year I found he had eliminated OEE as a measure and replaced it with equipment uptime, or what some in Toyota calls "operational availability."
Operational Availability = Time equipment actually able to run / Time when equipment is needed to run
This is very easy to understand throughout the shop floor. And it is directly related to the global value stream measure of takt—how much does it need to run to meet the customer demand? It turned out that when Gary visited a Dana plant he was hard pressed to find anyone who actually understood what OEE meant and was using it for improvement.
The best argument I have heard for OEE is that at a macro-level in a plant it provides a high level picture of how your equipment is functioning. You can multiply it across say 5 linked machines in series and get a picture of the percentage of time you can expect to get a good part out with the frequency you need. Then you can use it as a signal detector to know where to focus improvement activity. But this is at the plant level usually led by engineering. For kaizen at the work group level and as a KPI for operations I believe it is confusing because it aggregates so much and is hard to act upon.
I recall visiting Toyota’s engine and transmission plant in Buffalo, West Virginia. In the transmission plant they explained that the way everybody, salaried and hourly, earn raises and bonuses each year is based on production of transmissions. That is what pays for the plant. The key factors in production are equipment running when it is supposed to be running and quality. Everybody understands “operational availability” (OA). They have a counter showing how many transmission parts each machine should be producing through the day based on the takt (customer demand rate) and what they are producing. They measure OA by shift. They also measure defects/100 in a separate quality measure, both externally shipped to the customer, and more important internal defects that are caught in inspection and repaired or scrapped.
The way OA is used is not for comparing performance or any type of individual reward or punishment system. It is used for improvement. The overall performance board for the plant shows OA by shift. This then is tied to department boards and even drills down to hourly OA. Every hour OA is examined and comments are written about the reason for any downtime for a specific machine. Most of the problems are unique, one of a kind and lead to immediate countermeasures by maintenance. Recurring items go onto a tracking sheet and follow a more rigorous problem solving process to get to the root cause.
In Toyota standards for operational availability, set in Japan, are very high—in the range of 97 to 98 percent. There is little inventory and often machines are set up sequentially so one machine going down quickly shuts down a whole line of machines. It is rare walking around in a Toyota plant to see any piece of equipment stopped for more then a few minutes. There equipment works so well, not because of elaborate measurement systems, but because of rapid response to breakdowns and rigorous problem solving when a recurring cause is detected. Continuous improvement is at the center of equipment wellness and continuous improvement can only be done by capable, motivated, thinking people.