OEE Helps Measure, Analyze Performance Changes

When Measuring Dependent Processes, It's Important to Analyze the Interactions, Relationships to Identify, Understand the Root Cause of Changes in the Metrics

By Jack Chopper

Many years ago, during a Lean Manufacturing training session, I collected one of my favorite axioms: "If you change something before measuring it, you won't be able to determine if the change was an improvement."

The ideal process is to measure, make a single change, document, measure again. The key word is "single." But how many of us have the luxury of following that process for implementing changes? In production environments, I suspect very few.

As a controls guy, I like to measure and analyze. I also read everything I can get my hands on regarding best practices and overall equipment effectiveness (OEE) measurements and implementations.

Consider this question: Which road will get you to the 60 mile mark faster: Traveling 30 mph on a road with no traffic lights, or traveling 50 mph on a road with traffic lights every 4.75 miles on average, with each red light lasting an average 2.5 min?

To decide, many focus on the minimization problem and the rudimentary arithmetic, but what's described in this little exercise are the "ideal" conditions. We don't know the incidences of backups, the synchronization of the traffic lights, etc.

Ditto that for dynamic plant conditions, where few processes are truly independent. Properly measuring the performance of one process requires isolating it from the others. When measuring dependent processes, it's important to analyze the interactions and relationships to identify and understand the root cause of changes in the metrics.

In the road case above, a simple change to the duration of a particular traffic light might benefit the overall throughput of traffic. Or changing such timings only during specific hours might be more beneficial. Quickly, we begin to envision the design of experiments to determine the best strategy for the automatic signals.

We need to arm ourselves with information that lets us make good decisions in a timely manner to reap the benefits. Example: The sooner we know that the widget machine is struggling, the sooner we can take action. This action helps us either to mitigate that particular issue, or better use labor and materials on all of the downstream equipment.

We know the widget machine is struggling because we measured it — whether it's output, fluctuations in power consumption, or whatever. We have a history of prior measurements with which to compare that data.

Back to the road example: Assuming there are alternative routes to use when traffic grinds to a standstill, many drivers begin a mental exercise to choose to stay on their current route, or to invest the time necessary to navigate to an alternate. With GPS devices supplying real-time traffic information, browser-enabled smartphones, and traffic boards that announce distance and estimated travel times, the exercise now involves much more data than before. Not so long ago, many of us relied solely on traffic reports from local radio stations.

The takeaway is that the "better road" will change from one to the other — and back — as conditions change. The time of day that the changes occur is crucial if we're to react appropriately.

Although far more complex than a simple road example, the same thought process applies to plant equipment and its performance data. Many of us are engaged in supplying solutions to help enterprises become more effective with the time available; we want more throughputs, with fewer errors and rejects, with far more precision than ever before. More importantly, we want to know when we're failing to meet those goals, but we want to know as soon as possible so that we can mitigate.

To reach those goals, we must make good decisions. Properly implemented, OEE is one tool that will help us make those decisions, since we will understand the interrelationships of the various processes much more thoroughly. With some OEE history, we will benefit from a predictive component, which is among the most valuable data.

By itself, OEE isn't enough. We need to create the culture where changes are embraced. At the end of the day, how fast the production lines ran is far less important than the throughput minus the rejects. OEE will give you the numbers, but the changes necessary to improve those numbers comes from — you guessed it — you.