![]() The funnel experiment demonstrates that well-intended efforts to improve a stable process, without knowledge of cause and effect, can make things much worse. The balls come to rest farther and farther away from the target. Under this rule, we see somewhat of a random walk. In an application with dyes or paints, an example would be color-matching based on the last swatch or batch. One of the most common examples of this rule is on-the-job training. The strategy for this rule is consistency. Under Rule 4, the funnel is moved over the resting position for the previous drop. Now the process is moving farther and farther away from the target with each drop, on either side of the target. This strategy is often used in resource planning, inventory management, and managing budgets, lead times and production levels. The goal of this rule is to average out to the target level. Under Rule 3, overcompensation, the funnel is moved to the opposite position from the target. Many of the points fall close to either the upper or lower control limit. Also notice the pattern in the individuals chart. Notice that the moving range chart shows more variability, with more points beyond the upper control limit. In fact, if we apply this rule long term, the process will be 40% more variable. The process has become more variable than if we had left the process alone. Another example is when an operator overreacts to common cause variation and makes unnecessary adjustments. A common example in a machining operation is applying an offset adjustment or adjusting to zero. This rule is based on the idea of making adjustments to meet a target on the next production run. So if the ball's final resting position was 3 inches north of the target, the funnel is moved 3 inches south of this position. ![]() Under Rule 2, exact compensation, the funnel is moved to the exact opposite position of the last drop. We apply the other adjustment rules to see if we can improve the process. This is a process with only common cause variation. The points appear to be randomly scattered in the individuals chart, with no points outside the control limits. Most of the points are close to the target. The final position of each ball is marked, and the distance from the center of the target is plotted on the I and MR chart. Fifty successive balls are dropped, without changing the position of the funnel. The graph on the left shows the view of the target from above, marked by a circle. This script is available in the Sample Scripts Directory under Help, Sample Data. Which adjustment strategy do you think is the best? We'll conduct our own funnel experiment using the demoFunnelExperiment script. After the ball lands a certain distance from the target, aim the funnel right over the spot where it came to rest. After the ball lands a certain distance from the target, move the funnel an equal and opposite distance from the target. After the ball lands a certain distance from the target, move the funnel an equal and opposite distance from its last position. ![]() Leave the funnel fixed and aimed at the target. The volunteer was given different adjustment rules, or strategies, to try to improve the performance of the process. The critical measurement for the process was the distance from the target to the ball's final resting position. The goal was to get the ball to come to rest on the target. The ball's final resting position was marked, and this was repeated 50 times. A volunteer was selected from the audience and was asked to drop a ball through the funnel. The demonstration involved a funnel, a stand for the funnel (to lift it off the ground), a ball or a marble, and a target. To illustrate the effects of tampering to managers, Deming used a demonstration called the Funnel Experiment. These efforts to improve a stable process, without knowledge of cause and effect, are known as tampering. If a process is in control and on target, trying to improve the performance of the process by 'tweaking' machines or work methods generally makes things worse. As you have learned, a stable, or in control, process is one in which only common causes of variation are present. A special cause, for example, might be due to an equipment issue, to a part wearing out, or to a power failure. Only a small proportion of the problems are the result of unpredictable influences on the system, which are driven by special causes of variation. These are the result of predictable sources of variation, or common causes. According to Deming, most opportunities for improvement are driven by the system. A pioneer in the field of quality, and a peer of Walter Shewhart, was W. In a previous video, you learned about common and special causes of variation.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |