Trending “Bad Practice” Recommended as “Best Practice”
I was just looking at a “Best Practice” from a nuclear utility about
I saw lot’s of “bad practices” listed as best practices.
What happens if you adopt a bad practice as a good practice? You get bad result! And when those bad results are related to trending it means that you will waste effort responding to trends that don’t exist and miss trends that do exist.
I won’t say which nuclear utility it was, but you need to be careful when accepting advice about trending – I’ve seen lot’s of bad practices out there.
Let’s talk about just a few of the “bad practices” that were recommended in this industry “best practice” …
1) They mentioned Pareto Charts but didn’t mention the 80/20 rule (Pareto Principle) that it is based on and how it controls the use of Pareto Charts for choosing which targets to attack first. This could lead to choosing items to improve that really are NOT that significant.
2) They recommend pie charts and matrixes to analyze data. I would never recommend using these as the appropriate Pareto Chart would be much better (and you only have to learn a single method for analysis).
3) For trending over time they recommended a mixture of techniques including making “trend lines” with linear, second order, third order, and fourth order polynomial approximations. This will lead to false trends and management knee-jerk reactions. (Just what you are trying to avoid.)
4) They then made a graph that looked like an XmR Chart or Process Behavior Chart but they didn’t provide the proper mathematical methods for setting the Upper Control Limit (UCL) which we call an Upper Process Limit (UPL).
“The UCL for each trending category and subcategory is set by mutual agreement between the trend group and the line organization responsible for the program, process, or issue that category or subcategory represents. Organizations typically started with initial UCL calculated on the basis of the mean over a specified time frame (usually 18 months) plus two standard deviations above the mean …”
Ahhh! This is exactly what Dr. Deming said NOT to do. Management arbitrarily setting and changing limits.
The 3 sigma limits were proven by extensive testing by Dr. Walter Shewhart back in the 1930’s. This has been accepted by quality experts around the world. Why would the nuclear industry “best practice” choose a different basis (and not explain how they chose to derive it). All this new standard will do is cause more “false alarms” and more knee-jerk reactions.
5) They didn’t show any appropriate techniques for trending infrequent data. This can lead to missing serious trends and management believing that they can’t detect trends in infrequently occurring data. (And thus even more knee-jerk reactions.)
Why is this bad practice that is represented as a good practice so troubling? Because we have been teaching best practice trending techniques based on a foundation of science and accepted math for over a decade. Everyone in the nuclear industry should now have someone at their plant that understands these advanced trending techniques. Yet no one has challenged this false “best practices.” Some are probably thinking about adopting it!
Where can you learn the advanced trending techniques that can help you understand and improve your facility’s performance? At this pre-Summit course:
Don’t miss this course that is only offered once a year.
Also, please don’t think that this course is ONLT for nuclear industry root cause analysis trending. It will work in any industry. We’ve had attendees appy it at:
- Oil Refineries
- Oil Platforms
- Manufacturing Facilities
- Chemical Plants
- Utilities (fossil and transmission & distribution)
- Government Agencies
Class size is limited. Sign up today!