Control Chart

A control chart is a diagram used to see how processes change over time. Data is plotted in chronological order. The control chart always has an average centerline, upper control upper limit and lower limit management lower limit. These lines are based on historical data. By comparing the current data with these lines you can draw conclusions about whether the process change is consistent (within control) unpredictable (out of control, the effect of special variation).

The management of variable data is used in pairs. The top chart monitors the average position or center position of process data. The chart below monitors the range or width of the distribution. If your data is a shot at the target exercise, the average is where the shots are gathered and the extent is how firmly they are gathered. Control chart for using attribute data alone

Process change pattern analyzing special causes (non-traditional events) or common causes (embedded in the process)

Determine whether your quality improvement program should be designed to prevent fundamental changes to specific problems or processes

Please look for "unmanaged signal" on the control chart. After identifying, investigate by marking the cause on the chart. Record the method of investigation, knowledge, reasons, and remedy

One point outside the control range In Figure 1, the 16th point is higher than the UCL (control upper limit)

Two of the three consecutive points are on the same side of the centerline and are further away from 2σ. In Figure 1, the fourth point is sending the signal.

Four of the five consecutive points are on the same side of the centerline and the distance from it is greater than 1σ. In Figure 1, the eleventh point sends a signal.

Run eight times on the same side of the centerline. Or 11 out of 10, 14 out of 14, or 16 out of 20. In FIG. 1, the point 21 is continuous at the eighth point on the center line.

An obvious consistent or persistent pattern that indicates that your data or process has changed somewhat

Continue drawing data as generated. Every time a new data point is drawn, it checks the signal outside the new control

This process can be frustrating when creating a new control chart. If so, the control limit calculated from the first 20 points is a conditional limit. If the process is running under control, there are at least 20 consecutive points, recalculating the control limits.

From Nancy R. Tague "The Quality Toolbox", 2nd edition, ASQ Quality Press, 2005, pp. 155-158

Short-term control charts or control charts for short-term production runs plot observations of variables or attributes of multiple parts on the same chart. Short-term control charts have been developed to meet the requirements for collecting dozens of process measurements before calculating control limits. Meeting this requirement is often difficult for the task of manufacturing a limited number of specific components during manufacturing run. For example, a paper mill can produce only three or four (large) rolls of a particular type of paper (ie, parts) and then transfer the product to another paper. However, when monitoring variables such as paper thickness and attributes such as defects such as dozens of roll paper, it is possible to calculate the thickness of the conversion and the control limits of variables of interest (in short-term production) I will. value

Advantages of variable control charts Variable control charts are more confidential than attribute management diagrams (see Montgomery, 1985, p. 203). Therefore, variable control charts may be warned about quality problems before the actual "unacceptable" (detected in the property chart) occurs. Montgomery (1985) calls a variable control chart as the main indicator of failure and issues a warning before the amount of waste increases in the production process. You can create a variable control chart so that you do a single observation from the production line instead of the observation sample. This is sometimes necessary when testing multiple observed samples is too expensive, inconvenient or impossible. For example, customer complaints and number of product returns may only be available on a monthly basis, but we recommend that you plot these numbers to detect quality problems.