The Fallacy of OEE (Overall Equipment Effectiveness) in System Design

OEE (Overall Equipment Effectiveness) and its components are excellent tools when analyzing an existing system to identify where improvements can be made.

While the origin of OEE comes from losses due to availability, productivity, and quality calculations often expresses as OEE = Availability * Performance * Quality. The formula below consolidates these equations into an easier way to look at OEE.

Where OEE falls short is during the design process

OEE targets will often muddy the waters during the design process since the equipment design and system layout are still in flux. Designers need to consider many factors including the cost of labor and equipment; the value of system buffering, and opportunity cost of floor space. All these factors impact the Ideal Cycle Time portion of the equation shown above.

At the core, production facilities are looking to make a profit. Cost of production is the true measure of how successful a system is for a manufacturer. Once a design is set in terms of capital, layout, labor, floor space, and buffering.

The OEE factors give us insight on where we can make the system better. Focusing on a specific OEE target during design can lead to a higher cost per part.

Almost every system design starts with demand. How many of these things do we need to make per hour? Per day? Per week? Once that is known, we have choices to make. Let’s say we want to make 63 parts per hour. We could design a system that runs at an ideal rate of 70 PPH that has an OEE of 90% and it would yield 63 PPH production. Alternatively, we could design a system that runs at an ideal rate of 75 PPH with an OEE of 84% and we would make the same 63 PPH. The question is which one has a lower cost per part? The answer is that we do not know. However, if there is an arbitrary target OEE for our systems of 90%, we would never even consider the 75 PPH concept.

Option 1: 70 PPH @ 90% OEE = 63 PPH produced
Option 2: 75 PPH @ 84% OEE = 63 PPH produced

Why would we ever want to make a faster, less efficient system? Let’s describe them in more detail, and outline examples using FlexSim software.

Designer 1

The designer is told that it takes 24 minutes of assembly time to put together a part. The demand for this part is 63 PPH and the OEE target is 90%. Doing some quick math, the designer sees that a rate of 70 PPH will allow the system to net 63 PPH at 90%. 70 PPH gives the designer a little over 51 seconds of time for each part. Traditionally, these parts are manually assembled on an assembly line with parts indexing between stations on a conveyor.

So how many stations are needed on this line?

24 minutes equals 1440 seconds. Using approximately 51 seconds between parts, we’ll need 29 operators working on the line to assembly this part. The designer thinks that is a really long line. We can cut down on the size of the system if we put an operator on each side of the line. The designer decides on a 15 station line with 30 operators. With 30 operators, each one has 48 seconds of work out of the 51 second target. This seems reasonable so they move ahead to process the line so each worker gets an average of 48 seconds but no more than 51 seconds of content.

“…We would not have though about a second design if an OEE of 90% was a requirement for the design…”

Designer 2

This designer is given the exact same specifications as designer 1 except no OEE expectation is set. We need 63 PPH. The designer knows that manual assembly is inherently inefficient. Instead of targeting 90% OEE, the designer decides to run a bit faster and live with the inefficiency caused by manual assembly. By setting the target cycle time at 75 PPH, the designer knows they only need 84% efficiency to reach the 63 PPH target.

Designer 2 takes the same linear approach to the line with operators on each side. Since the line is running faster (48 seconds to make 75 PPH), he’ll need 32 operators instead of 30. So what about buffering? The simulation study shows that the 16 station, 32 operator line can make 84% without any buffers. The line is 16 stations long instead of the 22 stations needed for Designer 1’s concept. It will require two more operators, but it requires less capital and less floor space.

Summary

So which one is better? From a cost per piece standpoint, it is hard to say. We would have to know the value of floor space, the cost of the conveyor and the labor cost to compare them accurately. There is a more conceptual argument. By eliminating buffers in implementing a “leaner” system, designer 2 exposes the system to traditional improvement techniques we see in lean manufacturing. Every over cycle condition will be noticed more. The production team will strive to eliminate that variation because its impact is obvious.

Over time, the second system has more room to improve. As the process becomes more refined, it may be possible to eliminate the extra two workers and still make production. In the end, the second design promotes lean thinking and will probably end up with a better system in the long run. The key concept is that we would not have thought about the second design if an OEE of 90% was a requirement for the design.