OEE – A way to Speak with Data, But It Only Works If We Use It

Speaking with data is an important part of the journey to becoming a World Class organization.  It is a way to make the invisible visible, to see things that we are otherwise likely to miss. It is also a way to get the attention of people who control the purse strings, when we need resources to solve a problem. I first discussed speaking with data in my post March 15, 2010. (http://bobduckles.blogspot.com/2010/05/speak-with-data.html)

OEE, which stands for Overall Equipment Effectiveness, can be a useful tool for managing and continuously improving.  There are three components to OEE:

1.     Availability (Planned uptime minus unplanned downtime)

2.     Performance  (Planned operating time minus minor stoppages and running slow. It is calculated by dividing the number of pieces produced by the pieces that would be produced at the rated or standard speed)

3.     Quality (All pieces produced minus defective pieces produced divided by all pieces produced)

Some companies have a mistaken view of the OEE of their equipment. I have been told that a certain machine has an OEE of .90 or .95, when I can see that it is running slowly, creating scrap, or is down fairly often. There is a convention that any OEE of .85 or above means that the equipment is working at a world-class level.  

Fairly high numbers in each of the three components can lead to an OEE that is still below level of .85. I have seen estimates that OEE in American manufacturing tends to run around .60.  I have not seen studies on which this claim is based, but it feels right to me as a ballpark. I have also been in plants where most of the equipment does not even reach that level.

Let’s consider a few examples:

OEE	=	Availability	x	Performance	x	Quality
	=	0.8	x	0.9	x	0.99
	=	0.71

The first example does not reach and OEE of .85.  One combination that does reach the world-class level is:

OEE	=	Availability	x	Performance	x	Quality
	=	0.9	x	0.95	x	0.99
	=	0.85

This combination still falls short:

OEE	=	Availability	x	Performance	x	Quality
	=	0.9	x	0.9	x	0.9
	=	0.73

It does not make much sense to make these calculations, unless we use the data. It is also important that we not use the data to beat people up, but to help us figure out where we need to focus our attention in order to improve.

One way to do this is to calculate the OEE for a given machine daily, at the machine. If the OEE is consistently .85 or better, there are probably other pieces of equipment that need our attention more than this one.    We still might want to ask whether this is an unusual level for this machine.  To know this, we would need to have a history.  I advocate not only manually calculating but also manually plotting OEE for each piece of equipment we are monitoring.   Here is a way to present the data. 

(Click on the image to enlarge it) 

On the graph we see that on Monday, the 5th, we suddenly had a drop to .15.  We should want to know what caused this drop.  Our first question: Was it a problem of availability, of performance, or of quality? A glance at the bottom of the chart, where the three components are recorded, we can tell immediately that or biggest problem was availability.  The machine was down.  We ought to find out right away what caused the machine to go down, so that we can take action to insure that this problem never happens again.

On Friday, the 9th,  we had another drop in OEE. A quick look at the data tells us that this time we had a problem with performance.  That is where we need to dig in to find out what happened and take corrective action.

Increasingly, calculations such as OEE are made by statistical packages into which data is, in some cases,collected automatically. Interesting high level graphs and reports can be generated. When I work with clients, I insist that the operator, at the machine, create this graph manually.  If we only have a report that comes out at the end of the week we miss an opportunity to act.  When OEE suddenly drops, we want to start asking what happened right away—the same day.  OEE charts on the machines that are being monitored can a part of visual control.

To begin to use OEE, I would not suggest that we start monitoring and charting every machine. Begin with some critical operations that we need to improve. Strive for consistency, then strive for improving OEE.  Some equipment may never need to be plotted because it is reliable and has a capacity far in excess of what we need.

World class OEE, .85 or better, is a factor to consider when ordering new equipment and calculating the needed capacity. 

I have had occasion to work with teams planning a new line and the equipment to be purchased for them. They planned as if the equipment would run at its stated capacity 100% or the time.  The improvements above .85 can be pretty challenging.  While part of achieving world class is demanding excellence of ourselves, plan with a .85 OEE in mind, nothing higher.

To get operators to record the data and make the calculations, we should explain that these data will help us solve problems around availability, performance, and quality.  Then, we need to earn credibility by using the data for that purpose.  It is discouraging to engage in an activity that does not add value to the product or the process.  Too often we start data collection and quickly forget why we are doing it. Start small.  Stay with it.  Show some results.  Spread the process from the early examples.

Notes: The Lean Thinker  has recently described how to calculate the three components of OEE. I chose to focus more on what to do with the OEE calculation, once you have it.  


While for reasons that I hope are clear, I object to capturing the data and doing the calculations in a centralized system.  I do not object to the operator having templates on a screen into which we can plug the data that give us the components and overall OEE that we then plot manually.