May 7, 2009 | Mark Paradies

Trouble Trending in the Nuclear Industry


I recently read 11 nuclear plant trend reports. Not one had a mathematically valid way to trend. (One did come close.)

These 11 nuclear industry trend reports are not unusual … They are typical of trend reports from various industries.

This makes one ask: “Why do so many companies have trouble trending?”

Let’s look at specific problems in the 11 reports.

First, many “trend” reports showed no trends. Three of the eleven reports didn’t even have a single graph. They were just discussions of incidents and audits in a particular quarter with discussions of “trends” with no data.

Most of the reports that had graphs used simple bar and line graphs. Bar graphs with no timeline don’t really show “trends.” They are best used to look for the Pareto Principle and help to identify the biggest problems to attack first.

One report tried to use a mathematically based approach. They used a c-Chart with limits based on 3 times the square root of the mean of the data. But c-Charts require independent data with a constant area of opportunity – which is unlikely for incident data.

The most common error made in trending reports (and by management) is looking at the number of problems in one quarter and comparing them to the next quarter. If the number of problems goes down – that’s a good trend. If the number of problems goes up – that’s a bad trend. They tend to use straight-line approximations to “project the future” (as in the example below).


What’s wrong with this approach? Management starts reacting to noise rather than signals.

Some might say that there’s nothing wrong with trying to “fix things when they aren’t broke.” But they are wrong.

They don’t understand the cost of reacting to non-problems.

They don’t understand how employees get tired of the new improvement flavor of the week.

They haven’t thought about the negative cycle of blame, cover-up, and surprise that is all too common when major accidents occur.

What should people do? Where can they find out about the best practices in trending? First, they can read Chapter 5 of TapRooT® – Changing the Way the World Solves Problems. The new TapRooT® Book explains the problems with trending and the ways to implement best practices in performance measures and trending.200905031637.jpg

Would you like to go beyond reading? Then you should plan to attend this year’s Advanced Trending Techniques Course on October 5-6 in Nashville, Tennessee (just before the Summit).

What will you learn? First, you will learn why trending is needed and the basics of trending (including the proper way to apply Pareto Charts).

Second, you will learn the math behind Process Behavior Charts (the only chart you need to view trends over time). This math only requires addition, subtraction, multiplication, & division (no different¬ial equations).

Third, You will also learn special ways to use the charts to trend infrequently occurring accident data.

Finally, you will learn how to use your own trend data including how to improve the data and how to present it to management. Get more information about the course by clicking on the “Courses” button above.

Show Comments

Leave a Reply

Your email address will not be published. Required fields are marked *