OEE - Overall Equipment Effectiveness
We have found that there are some misconceptions regarding the calculation of OEE that can lead organizations to Mass Production behaviors.
Introduction
In our travels to many companies we always stress the need to develop new measures that lead to lean behaviors. What we mean by a Lean Behavior is that we want everyone in the organization to be measured and rewarded for taking action to reduce process waste. If an organization retains their old style productivity measures as they try to become lean, they will find that they have a difficult time getting their staff at every level and in all functions to change. Why would I change my behaviors if I'm measured and rewarded via mass production-type measures? In a previous paper on measurements, we have discussed which measures support lean, and which drive the organization in the opposite direction.
One of the measures commonly used by organizations on the road to lean is Overall Equipment Effectiveness or OEE. This measure is commonly used to assess success in a total productive maintenance program. The common definition of OEE is: a percentage calculation obtained by multiplying the availability rate, the performance efficiency performance rate, and the quality rate.
While working with various companies as advisors to their lean efforts we have found that there are some misconceptions regarding the calculation of OEE that can lead organizations to Mass Production behaviors. This short paper will address some of our observations.
Performance Efficiency
This is probably the most contentious issue that we have seen with OEE. For those of you familiar with value stream mapping, you will understand that we want to set the pace of production in our value stream at the pacemaker process. At our pacemaker we issue a schedule in small chunks of time called a pitch. The pitch is based on the rate of customer demand or takt time. If we produce faster that takt time, we are over producing - creating all kinds of waste. We also need more people to do the work than if we slowed down to takt time. If we work slower than takt time, we will not meet our rate of customer demand leading to overtime, missed shipments, and unsatisfied customers.
Just because a machine can produce 3600 parts/hour doesn't mean that we should be doing so. We need to be working at our takt time, and providing a visual image of how we are doing to takt time to the people on the work floor. If our takt requires that we produce at 1200 pieces/hour, then that is what we should be doing. Measuring the performance efficiency of a machine does not mean that the machine should work at the performance specification of the machine. In some cases, this is much faster than takt time, and will lead to overproduction. That's why many lean senseis prefer smaller, more flexible machines that can be run at the right rate rather than big Monuments that will overproduce.
So beware that measuring performance efficiency may lead your people to run the machine faster than takt to meet their required measures. That's a behavior that you want to avoid! What we really want is for the machine to be tuned-up enough to run at the performance specification if we needed it to in order to achieve our takt time.
Availability Rate
This is a less contentious issue than performance efficiency, but we need to understand exactly the behavior we want to achieve here.
Again, we want to be producing no faster, or slower than takt time. Waste is produced either way. When we say that we want a machine to be available, this means that it must be available when we want it to run. It doesn't matter to our customers if the machine isn't available when we don't need it to run.
What we are trying to measure here is the reliability of the machine to run on demand - not whether the machine is used all of the time. There is a major problem in a few organizations. The finance group bought the machine out of the organization's capital, and to meet their return on investment they naturally want to keep that machine running. But this attitude can lead to production faster than takt time.
So availability should be measured by understanding if the machine was available only when an operator requires it in order to produce to takt time. Having a machine available on a non-operating shift doesn't matter unless the machine is needed because it is a shared resource that is a bottleneck - like a heat-treat oven. So we shouldn't be measuring its availability then.
Quality Rate
We need machines to be producing quality parts. An integral component of Lean is that we never pass a defect to the next process. If we identify a part coming off of a machine as a defect, we should also understand that there might be more than 1 defect on that part. If we have 10/100 defect parts produced by the machine and each of the incorrect parts has 3 defects on it, then what is our quality rate? What if there is a varying amount of defects on each of the 10 incorrect parts?
How do we define a defect on the part? How many opportunities are there to produce a defect on a part? Do 2 different operators define defects the same? How do we measure to make sure? Are the measurement tools accurate? There are many questions to be answered before we know what quality means at the end of any process. First Time Through quality doesn't always give you enough information to be able to problem solve properly. Before measuring quality rate of a machine, you need to define what a defect is, and how to measure it. Realize that the quality rate of a machine varies over time, and that statistical process control is required to understand whether the machine quality is in control or not.
The more that machines and processes move towards one-piece flow, the easier it is to understand quality, and implement mistake proofing devices to eliminate the root causes of defects. This is largely due to the fact that one-piece flow requires a greater level of inspection at source, and fast feedback. Measuring a quality rate is necessary in every value stream. You must ensure that defects are not passed onwards and that you are working to resolve the causes.
Aggregating The Numbers
This is one more potential problem in using OEE for a measurement. Aggregating the three factors, efficiency, availability, and quality, and using the result as a measurement can lead you to some wrong conclusions, and potentially increase costs in your value stream. We'll illustrate by example:
Say we take a measurement and obtain the following results:
Measurement 1: Efficiency = 99%, Availability = 80%, Quality = 95%
OEE = 99% x 80% x 95% = 75%
A day later we look at another measurement:
Measurement 2: Efficiency = 98%, Availability = 95%, Quality = 89%, OEE = 98% x 95% x 89% = 83%
The analyst might draw the conclusion that things have gotten better. This may be wrong for several reasons:
The change may be due to the natural variation that happens in every process. (Which we will never know unless we use SPC to review the numbers and determine whether there is a special cause of the variation). This is harder to do when the numbers are aggregated.
Quality has gotten much worse, which may mean dramatically increased operating costs to rectify defects. Yet the OEE number looks better. So why aggregate the numbers if they might allow you to draw the wrong conclusions?
Conclusion
OEE taken in isolation, and without understanding the critical elements which go into making up this number, may cause you to react or not react properly. The result being that you and your colleagues are driving behaviors which are detrimental to your Lean progress. Getting a good OEE score should not be about running a machine faster or harder - rather, the machine has to run at the speed that you need it to, be available (only) when you need it, and be capable of producing parts to specification. Creating an aggregate of the efficiency, availability, and quality rates can only serve to mask the unwanted variability in any of these sub-components. In addition, beware that your OEE measures do not overtake the importance of operator balancing in a cell. Remember that a critical part of Lean is the separation of Man and Machine - not a focus on optimizing the machine.