HKSM

A control chart is a time-ordered graph of process data with a central line and statistically calculated upper and lower control limits used to monitor stability. It helps distinguish routine variation from special-cause signals so the team can respond appropriately.

Key Points

• Shows process performance over time against a central line, upper control limit (UCL), and lower control limit (LCL).
• Control limits are calculated from process data; they are not the same as customer specification limits.
• Signals such as points beyond limits or nonrandom patterns indicate special causes that need investigation.
• Choose the chart type based on data: variable charts (X-bar/R, X-MR) or attribute charts (p, np, c, u).
• Limits are often set at approximately ±3 standard deviations unless the chosen method dictates otherwise.
• Use control charts to stabilize the process before assessing capability or making major adjustments.

What the Diagram Shows

A control chart plots sequential measurements and highlights whether the process is stable (in control) or showing special-cause variation. It provides a visual boundary of expected variation and flags unusual behavior.

• Data points ordered by time.
• Central line representing the process average or proportion.
• Upper and lower control limits derived from the data and chart type.
• Optional specification limits to compare against customer requirements.
• Annotations for out-of-control points or patterns.

How to Construct

• Define the measure, sampling frequency, and subgroup size or rational subgrouping approach.
• Collect an initial dataset under typical operating conditions.
• Compute the central line (mean, median, or proportion) appropriate to the chart type.
• Calculate control limits using the correct formulas for the selected chart (e.g., ±3 sigma or published factors).
• Plot data in time order, draw the central line and control limits, and add optional specification limits.
• Apply standard rules to flag signals, investigate special causes, and document findings.
• Periodically recalculate limits if the process changes and stabilizes at a new level.

Inputs Needed

• Operational definitions of the measure and how it is recorded.
• Historical or baseline data collected with consistent methods.
• Sampling plan, including frequency and subgroup size or moving range parameters.
• Chosen chart type and calculation method.
• Measurement system validation or calibration evidence.
• Optional specification or tolerance limits for reference.

Outputs Produced

• Control chart with central line and control limits.
• List of points or patterns signaling potential special causes.
• Recorded investigations and corrective or preventive actions.
• Updated process understanding and stability status.
• Input to performance reports and continuous improvement plans.

Interpretation Tips

• A single point outside the UCL or LCL is a likely special cause signal.
• Runs of several points on one side of the center line suggest a shift in the process.
• Trends of consecutive increases or decreases may indicate a systematic change.
• Tightly clustered points near a limit or wide swings may suggest stratification or measurement issues.
• Do not adjust the process for random fluctuation; investigate only when rules indicate a signal.
• Distinguish control limits (process behavior) from specification limits (customer needs) when making decisions.

Example

A team tracks daily cycle time (minutes) for a repetitive task using an X-MR chart. From 25 baseline days, the average is 10.2 minutes with calculated limits of LCL = 7.8 and UCL = 12.6.

• Day 12 shows 13.1 minutes, which exceeds the UCL, triggering investigation.
• The team discovers a temporary system slowdown and documents a corrective action.
• After fixing the issue, subsequent points return within limits and cluster around the mean, indicating restored stability.

Pitfalls

• Confusing control limits with specification limits and taking the wrong action.
• Tampering with the process based on random variation, increasing instability.
• Using the wrong chart type for the data, leading to misleading signals.
• Collecting inconsistent or infrequent samples that hide true variation.
• Ignoring measurement system errors that distort limits and signals.
• Failing to recalculate limits after a process improvement changes the mean or variation.

PMP Example Question

While monitoring a stable process with a control chart, you notice one data point beyond the upper control limit and the next five points within limits around the mean. What should you do first?

1. Adjust the process setpoints to bring the average down.
2. Ignore the point because the process returned to normal.
3. Investigate the out-of-control point for a special cause and document actions.
4. Tighten specification limits to prevent future deviations.

Correct Answer: C — Investigate the out-of-control point for a special cause and document actions.

Explanation: A single point beyond a control limit signals a likely special cause. Investigate and address it rather than tampering with the process or changing specifications.