Situation
Joanne is the production manager for a company that manufactures plastic containers. She has three production lines dedicated to one product, and she care-fully monitors the productivity of these lines. As part of a seminar on analyzing the variation in data, she put the productivity data for each line on a process behavior chart. Much to her surprise, Joanne discovered that the productivity of each line was predictable, although she was not impressed with the amount of variation, nor the overall average.
When she returned to work, she asked Brian, the quality manager, to set up charts on productivity so she could review them each day. After Brian set up the charts, Joanne was amazed to see that productivity for each line was unpredictable. In fact, the charts looked entirely different from the ones she had made.
Available data
In a meeting with Brian, Joanne wanted to get to the bottom of the discrepancy with the charts. Her charts showed predictability with a large variation, while Brian's charts showed each line operating unpredictably. The data they used is in the table, "Productivity Data for Lines 42, 43 and 44."
Questions
1. How can these data be organized to make a process behavior chart that shows a predictable process?
2. How can these data be organized to make a process behavior chart that shows an unpredictable process?
3. How can a process be both predictable and unpredictable?
4. If Joanne is interested in process improvement, which chart should she use?
Answers to November Brain Teaser
Q: Is Nelson correct in that the weights of parts from Line 2 are more variable than those from Line 1?
A: By comparing the histogram of Line 1 with that of Line 2, Nelson can show that the weights of parts from Line 2 are more variable than those from Line 1. The weights of the parts from both lines meet the specifications.
Q: Because Nelson is running both lines, what could cause one line to be more variable than the other?
A: The average and range charts for the two lines show that both of them have essentially the same amount of routine or common causes of variation. The average range for Line 1 is 2.9 while the average range for Line 2 is 3.1. However, the real story is in the charts for the averages for each line. Line 1 has only routine variation because all the points are inside the control limits and there are no nonrandom patterns. Line 2, on the other hand, has lots of points outside the limits signifying that a source of variation from some assignable cause is present. Nelson is new to Line 2, and so he makes a list of possible assignable causes of variation. Some of the possibilities include: different raw materials supplied to Line 2, the maintenance schedule for Line 2 and a set up problem on Line 2. Nelson believes that he runs both lines the same, and at the moment does not consider operator differences a likely cause.
Q: After continued questioning, the supervisor tells Nelson that Line 2 has an automatic controlling device that adjusts the amount of powdered metal that is metered into the cavity. What impact could this have on the behavior of the process?
A: If the automatic controlling device is not correctly installed or programmed, it could cause overadjusting of the process. This would result in the increase in variation shown on the process behavior chart.
Q: How might Nelson test the impact of the automatic controlling device to see if it is causing additional variation in the part weights?
A: With his supervisor's support, Nelson could turn off the automatic controlling device to test the possibility that it is the assignable cause shown on the process behavior chart. He would plot the data just as before and see if the points are now inside the control limits. If so, this would show that he had found the assignable cause of variation.
