This article describes a novel approach to calculating the financial aspect of overall equipment effectiveness (OEE), with the result referred to as $EE (as in monetary units). By using $EE, a management team readily can “SEE” their operation in financial terms. Employees are then better able to focus on underperforming operations to improve the bottom line.
|
ADVERTISEMENT |
The formula for calculating OEE is straightforward, and the product of factors (with units in per cent) is a very useful metric. In modern manufacturing, challenges arise when setting cycle time targets, measuring quality, or objectively classifying downtime. When these problems have been overcome, companies then face the issue of where to direct resources to improve the bottom line. The solution is to use $EE.
Background and practical limitations of OEE
In many businesses there is a communication breakdown between the front office and the shop floor which arises by coincidence. Because the three OEE production metrics (performance, yield, and availability) are not expressed in monetary units, daily efforts to improve processes using OEE alone do not always translate into significant bottom-line savings.
…
Comments
I have developed OEE metrics
I have developed OEE metrics for many years and find your concept of $EE very interesting.
However, I find your analysis of the data you present very disturbing. The use of so-called "trend lines" in your first two charts of OEE and $EE is one of the biggest abuses of "statistical analysis" that I see all the time in meetings when someone is trying to push an agenda. There is no trend in your data. All you have to do is plot your data on a properly constructed Run Chart and/or a Process Behavior Chart and you will learn that you have a stable and predictable process generating $EE and OEE numbers. The so-called "trend lines" are just least squares fits of the data. If you add the 95% confident interval for the line you will find that the "trend line" can have a negative slope, a positive slope, or a slope of zero. Therefore, there IS NO TREND!!! In addition, when I plotted your data, the r-squared of the regression is 0.069 - not significant for the degrees of freedom in your data. You will also find that the prediction confidence interval is massively wide at all values of $EE.
Correlation between OEE and $EE is not necessarily causation, even if it is intuitively evident. You show no evidence of causation. so you end up trying to explain common cause variation is terms of special causes in a system that is clearly showing a reasonable degree of statistical control.
As W. Edwards Deming would have said about your analysis: "Simple...obvious...and wrong."
I appreciate your thoughtful
I appreciate your thoughtful comments. The simulated data which is used in these examples contains scrap, downtime and slow cycle time events for several machines, which cause financial losses. Those losses relate directly to tangible material, labor and machine overhead added costs. Practically speaking, removing those events translates into greater profits.
Admittedly causation lies deeper in the data, but that is precisely the point; you have to dig deeper. Although the data might be in a state of statistical control, businesses typically do not tolerate operational losses. Therefore the complex system in this example is out of tolerance and improvements must be made.
Our only agenda is to relate OEE metrics directly in financial terms in order to direct resources accordingly. Again, thank you for your feedback. We will continue to refine our analytical approach.
"If at first the idea is not absurd, then there is no hope for it." - Albert Einstein
Very interesting article
Very interesting article!
Can you please decipher the term "GROSS Unshed D/T"
GROSS
From the article: "Note also that scrap and downtime are prefixed by gross, meaning they are being compared to targets of zero scrap and zero downtime. "
Unsched D/T = Unscheduled downtime
Add new comment