Situation
Situation When Sergio was promoted to manager of continual improvement for his company that makes plastic food containers, his first action was to work with customer service to address ongoing key customer issues. He wanted to target those processes associated with the customer issues to remove the causes of poor quality and poor performance. The number one customer issue based on complaints was container breakage, and several customers have already made comments about taking their business elsewhere. Sergio conferred with a team of process engineers and operators to prioritize what process measures were associated with container breakage and select the critical ones for process monitoring and control.
The team selected the temperature of the batches of raw material as the critical measure with the highest priority. They decided to record the temperature of two consecutive batches of material every two hours. After several days of data collection they would analyze the data and determine how the process behaved.
Answers to September Brain Teaser
Coating thickness on paper is a critical characteristic for performance of the paper and a cost in production. As process engineer for improvement projects, Eva helped a team set up a control plan for monitoring the coating thickness over time. Data were collected in subgroups of five once a shift. When the team met to review the process behavior chart they saw a pattern in the range chart, but they were not certain what the cause was.
Q: What is the pattern in the ranges on the average and range chart for these data?A: The pattern is a zigzag that continues for the entire time period. See the average and range chart, “Coating Thickness.” The existence of a regular pattern is an indication of an exceptional cause of variation in this process.
Q: What factor is associated with this pattern?A: The zigzag is associated with the different shifts; higher ranges are associated with day shift and lower ranges are associated with swing shift.
Q: What possible specific causes could explain the pattern?A: In considering possible causes for this type of pattern in the ranges, the question is what can make the five values in a subgroup different. If analyzing the day shift and night shift values separately, the average range is 11.7 for day shift and 2.5 for night shift.
So the more focused question is: What makes the day shift variation more than four times larger than the night shift? Typically, differences in shifts due to changing settings in equipment will impact the average result, but not the variation. The most likely cause of the increase in variation is the timing of the data collection. Eva would need to observe the timing of the data collection that the two shifts use to determine what causes this large a difference in routine variation.
Q: What is the impact of this pattern on the monitoring plan?A: All of the calculations of capability of the coating thickness in this process are suspect. When using the average range of 7.19 from the initial average and range chart, one will get capability results that are not appropriate for either shift. When using the average for each shift, one will get Cpk values of 1.3 for day shift and 6.1 for swing shift. These capability analyses are shown on the charts, “Capability Analysis for Day Shift” and “Capability Analysis for Swing Shift.” Neither value is useful until the timing of the data collection is verified and the causes of routine variation identified.
When Eva did observe the timing of the data collection, she found that day shift was taking the five readings spread out over the entire shift and swing shift was taking all five readings at one time during the shift. The process behavior chart clearly shows that there is more than four times as much variation over a shift than at one time during the shift.
The critical question now is what timing is appropriate to use to subgroup the data to understanding the routine variation that this process can hold over time?
