Control charts can save you time and money by helping you to focus on only those changes in your process that may indicate serious stability problems.
By default, Minitab uses all of the data in a control chart to calculate the control limits. These lines mark the boundaries between an in-control process and an out-of-control process.
However, sometimes, using all of the data isn’t the best option.
When should I set my own control limits?
Extreme values or periods of great instability can have undue influence over the control limits, which results in a chart that may not identify an out-of-control process. Here are three cases when it makes sense to set the control limits yourself.
Established, historical values
With some longstanding processes, you may already know the parameters of a process when it’s in control. Likewise, you may have arrived at these values through an earlier control chart from a period of stability. If you know these parameters, it may make sense to use them rather than the parameters that are generated from the data that are being charted.
Measurements taken throughout the years have shown the stable mean width of lumber produced in a milling process to be 3.5" with a standard deviation of 0.04", so these values are used to set the control limits. The default limits (based on the sample data) would have missed the out-of-control point in this chart.
Known periods of instability
If you know certain subgroups reflect temporary abnormalities in the process, you may want to leave them out when you calculate control limits. Including them can reduce the sensitivity of your control chart. In other words, while the control chart may still be able to detect extremes, it may overlook subtler cases of instability.
Points four and five on this chart represent weight measurements taken on a breakfast cereal filling line when a stuck control resulted in overfilled boxes.
While included in the chart, these extreme values weren’t considered when placing the control limits, so the limits reflect the smaller range of values of the other points.
Stability is reached
Once a process is stable, you can set the control limits so that they don’t change as you incorporate new data into your chart. This will alert you sooner to gradual process shifts.
After you monitor the defects from a casting process for some time, you’re confident that the process is stable. You decide to freeze the control limits to ensure that they don’t drift further apart as you enter new data.
Note that control limits are not the same as specification limits. Specification limits indicate how you’d like your process to perform. Control limits are based on the actual data and indicate out-of-control points.
Remember also that a process can be stable and yet not perform within your specification requirements. Ensure process stability first, and then tackle other process improvements.
Setting control limits in Minitab
There are two simple methods for setting control limits: parameters and estimation. The best choice depends on your reason for setting control limits.
To base control limits on historical values, simply specify the parameters that are used to calculate the limits (generally the mean and standard deviation).
Suppose that you monitor the fill weights on a bottling line with an Xbar-R Chart. The mean and standard deviation of this process are well established, so you use these values.
- From the control chart dialog box, choose Xbar-R Options > Parameters.
- In Mean, type 0.92.
- In Standard deviation, type 0.01385.
- Click OK in each dialog box.
Freezing control limits after a certain point or excluding outlier data from the calculation of control limits are both accomplished in the same way: simply include or exclude the desired subgroups (or individual observations, in the case of individuals charts) from the calculation.
Suppose that you monitor the rate of successful sales calls at a call center with a P Chart. Because this process is new, you want to discount the observations in the first three subgroups when you calculate the control limits. The operators were unfamiliar with the new call scripts, and you knew that the process was unstable.
- From the control chart dialog box, choose P Chart Options > Estimate.
- From the dropdown list, choose Omit the following subgroups when estimating parameters.
- Type 1:3.
- Click OK in each dialog box.
Putting control chart optimization to use
When you investigate the stability of a new process, Minitab’s default method for establishing control limits in a control chart is a great place to start. But monitoring a process is not a one-time event—it’s an ongoing process itself. As you become more familiar with your process, setting your own control limits can be an important part of maintaining process stability.