Managing with robust metrics
Point-of-use delivery should be measured using statistical process control charting methods such as this "fraction conforming" or "p" chart.
By Dr. Thomas L. Landers -- Modern Materials Handling, 10/1/1998
There are plenty of horror stories about abusing performance metrics. One involves a manufacturing manager who went to a quality seminar and learned the valuable lesson that processes have natural random variation and unnatural disruptions due to identifiable root causes.The manager came home resolved to control process variation at a world-class level: no more than three product defects out of every million opportunities.
As worthy as that goal might seem, the problem came in implementation. The plant made large-scale electronics equipment and produced only a few units per year. It was not easy to define performance in terms of millions of opportunities.
To implement the metric and achieve an acceptable performance level, quality engineers became a little creative in defining "opportunities" for defects. Runs of wire and cable were arbitrarily divided into artificial segments treated like parts. Unfortunately, management had fostered target fixation.
There is a better way. In the August 1998 issue of Modern Materials Handling, I suggested that good metrics should be balanced, manageable, robust, and relevant.
In this sequel, I want to expand on the idea of robust metrics that measure both the level and the variability of performance. Statistical process control (SPC) charts provide this information, by showing actual variation about average performance (the centerline). Does SPC apply in materials handling? A real-world example is shown in the accompanying sidebar. Materials handling is vital to success in point-of-use material pull systems.
A single measure of average performance tells us very little about how to manage. Because of uncontrollable factors, there will tend to be period-to-period random variation. If we measure the variation, we can manage by exception and we can answer questions like, "How often is our performance X% better (worse) than average?"
The upper and lower control limits of an SPC chart indicate the natural variation of the process. With rare exceptions, we can expect random variation to stay within the control limits. When performance goes outside the control limits chances are there is an identifiable root cause. To improve the average level (centerline) or the variation (control limits) we have to make some investment of resources, such as training, process equipment and management information systems.
Use SPC charts to manage your materials handling systems. Watch for significant events. Then emulate successes and eliminate failures.
Talkback
Related Content
Related Content
Sponsored Links
View All Blogs
