In-Depth Interpretation of the Seven QC Tools · Control Chart

Introduction

"Every process has variation—the key is whether you can distinguish: is this normal variation, or is there a problem?"

The Control Chart is the most valuable tool for "process monitoring" among the Seven QC Tools.

In the first five issues, we learned about tools for analyzing problems, while the Control Chart is a tool for real-time process monitoring. It doesn’t wait for a problem to occur before analyzing; instead, it alerts you when the process is about to go wrong.

This is the core value of the Control Chart: shifting from "post-event analysis" to "pre-event prevention."

Chapter One: The Essence of Control Charts

1.1 What is a Control Chart?

Control Chart (Control Chart) is a graphical tool that plots process data in time sequence and uses control limits to determine whether the process is in a statistically controlled state.

Core Logic:
  Every process has variation, but there are two types of variation:

  ① Common Cause Variation (Common Cause / Ordinary Cause)
     → Inherent random variation in the process, unavoidable
     → Examples: minor variations in raw materials, slight environmental changes

  ② Special Cause Variation (Special Cause / Specific Cause)
     → Non-random variation with identifiable causes
     → Examples: tool breakage, operator error, material batch change

  The purpose of the Control Chart: to distinguish between these two types of variation.

1.2 Three Key Elements of Control Charts

A Control Chart consists of three lines:

  ── UCL (Upper Control Limit) Control Upper Limit
     μ + 3σ, the upper boundary of normal variation

  ── CL (Center Line) Center Line
     Process average μ

  ── LCL (Lower Control Limit) Control Lower Limit
     μ - 3σ, the lower boundary of normal variation

  If data points exceed UCL or LCL → the process has a special cause → intervention is needed

1.3 Control Limits vs. Specification Limits

Many people confuse control limits with specification limits; they are different concepts:

Key Understanding:
  Process in control ≠ Product conforming
  Product conforming ≠ Process in control

  The Control Chart focuses on "whether the process itself is stable"
  Specification limits focus on "whether the product meets requirements"

1.4 Three Major Functions of Control Charts

Chapter Two: Classification and Selection of Control Charts

2.1 Classification of Control Charts

Control Chart
  ├── Variable Control Chart (Continuous data)
  │   ├── X̄-R Chart (Mean-Range Chart): Most commonly used
  │   ├── X̄-s Chart (Mean-Standard Deviation Chart): For larger sample sizes
  │   └── Individual-Moving Range Chart: For sample size = 1
  │
  └── Attribute Control Chart (Discrete data)
      ├── p Chart (Defect Rate Chart): For varying sample sizes
      ├── np Chart (Defect Number Chart): For fixed sample sizes
      ├── c Chart (Defect Number Chart): For fixed units
      └── u Chart (Defects per Unit Chart): For varying units

2.2 How to Choose a Control Chart

Steps to choose a control chart:

Step 1: What type of data do you have?
  ├── Continuous data (length, weight, temperature, pressure)
  │   → Variable Control Chart
  │   → Next question: What is the subgroup size?
  │     ├── n=1: Individual-Moving Range Chart (X-MR Chart)
  │     ├── 2≤n≤10: X̄-R Chart (Mean-Range Chart) ← Most commonly used
  │     └── n>10: X̄-s Chart (Mean-Standard Deviation Chart)
  │
  └── Discrete data (conforming/nonconforming, defect count)
      → Attribute Control Chart
      → Next question: Is the data "counted by pieces" or "counted by points"?
        ├── Counted by pieces (nonconforming product number/rate)
        │   ├── Fixed sample size → np Chart (Defect Number Chart)
        │   └── Varying sample size → p Chart (Defect Rate Chart)
        │
        └── Counted by points (defect number/rate)
            ├── Fixed unit → c Chart (Defect Number Chart)
            └── Varying unit → u Chart (Defects per Unit Chart)

2.3 Stability Criteria (Judging Process Control)

Criteria for judging process control (four criteria):

  Criterion 1: No points exceed control limits
  Criterion 2: Points show no obvious trend or periodicity
  Criterion 3: Points are randomly distributed on both sides of the center line
  Criterion 4: No special patterns that violate the out-of-control criteria

2.4 Out-of-Control Criteria (Eight Out-of-Control Rules)

Most commonly used out-of-control rules (Western Electric Rules):

  ? Rule 1: 1 point exceeds UCL or LCL
    → A clear anomaly has occurred, immediate investigation is needed

  ? Rule 2: 2 out of 3 consecutive points fall in Zone A (2σ-3σ)
    → The process mean is starting to shift

  ? Rule 3: 4 out of 5 consecutive points fall in Zone B or above (1σ-3σ)
    → The process mean shift is significant

  ? Rule 4: 8 consecutive points on the same side of the center line
    → The process mean has changed

  ? Rule 5: 6 consecutive points show an upward or downward trend
    → There is a trend factor (e.g., tool wear)

  ? Rule 6: 14 consecutive points alternate up and down
    → There may be two different processes alternating

  ? Rule 7: 15 consecutive points in Zone C (±1σ)
    → Data may have been modified or the process is abnormally stable

  ? Rule 8: 8 consecutive points on both sides of the center line but not in Zone C
    → There may be stratified mixing

Chapter Three: Practical Cases of Control Charts

Case 1: X̄-R Chart — Monitoring Injection Molding Product Weight

Background: An injection molding workshop produces plastic parts, measuring the weight of 5 pieces daily
Objective is to monitor process stability

X̄-R Chart Analysis:
  → Point 15, R Chart exceeds UCL (sudden increase in range)
  → Investigation reveals: Mold 15 began to wear
  → Replacing the mold restored normalcy

Value: Identified the issue before a batch of nonconforming products occurred

Case 2: Individual-Moving Range Chart — Monitoring Small Batch Production

Background: A precision machining workshop produces only 1-2 pieces daily
Cannot use X̄-R Chart (subgroup too small)

Using X-MR Chart (Individual-Moving Range Chart):
  → MR Chart: 3 consecutive points show an upward trend
  → Investigation reveals: Tool is gradually wearing
  → Replaced the tool early to prevent batch nonconformities

Value: Effective process monitoring even in small batch production

Case 3: p Chart — Monitoring Welding Defect Rate

Background: An SMT workshop monitors daily welding defect rates
Daily production varies (sample size changes)

Using p Chart (Defect Rate Chart):
  → Point 8 exceeds UCL
  → Control limits change with sample size
  → Investigation reveals: A new employee started work that day
  → Enhanced training restored normalcy

Value: Accurately judges process status even with varying sample sizes

Case 4: c Chart — Monitoring Surface Defects on Products

Background: An electroplating workshop monitors the number of surface defects on products
Each product is a unit

Using c Chart (Defect Number Chart):
  → 5 consecutive points above the center line
  → Out-of-control rule triggered (consecutive points on the same side)
  → Investigation reveals: The concentration of the electroplating solution changed
  → Adjusting the solution ratio restored normalcy

Value: Early detection of mean shift

Chapter Four: Common Misunderstandings of Control Charts

Misunderstanding 1: Confusing Control Limits with Specification Limits

× Incorrect Understanding:
  "Data points within control limits → Product is conforming"

✓ Correct Understanding:
  Control limits only indicate "process stability"
  Whether the product is conforming depends on specification limits
  Process is stable but nonconforming → Need to adjust the process center or reduce variation

Misunderstanding 2: Focusing Only on Points Exceeding Limits, Ignoring Patterns

× Incorrect Practice:
  Only check if any points exceed control limits
  No points exceed → "Process is fine"

✓ Correct Practice:
  Even if no points exceed, check for patterns
  Continuous rise, continuous on the same side, periodicity → All are signals of anomalies

Misunderstanding 3: Using Control Charts Without Understanding the Process

× Incorrect Practice:
  Start using control charts immediately without understanding the process

✓ Correct Practice:
  Step 1: Use a histogram to understand the distribution
  Step 2: Use stratification to understand if the data is stratified
  Step 3: Establish control limits after the process stabilizes
  Step 4: Continuously monitor the process with control charts

Misunderstanding 4: Not Updating Control Limits

× Incorrect Practice:
  Control limits calculated the first time are never changed

✓ Correct Practice:
  Regularly recalculate control limits
  After process improvements, control limits should narrow
  Control limits reflect the current true capability of the process

Misunderstanding 5: Overusing Control Charts

× Incorrect Practice:
  Use control charts for any data
  Insufficient data (less than 20 subgroups)
  Data is not in time sequence order

✓ Correct Practice:
  Control charts require data to be arranged in time sequence
  At least 20-25 subgroups are needed to establish control limits
  Data must come from the same process

Chapter Five: Combining Control Charts with Other Tools

5.1 Control Chart + Histogram

Sequence for combined use:

  ① Collect data
  ② Draw a histogram → Understand the distribution shape (is it normal?)
  ③ Draw a control chart → Determine if the process is in control
  ④ Calculate Cpk → Assess process capability
  ⑤ Continuously monitor with control charts

5.2 Control Chart + Stratification

When the control chart detects an anomaly:

  Step 1: The control chart identifies an abnormal point
  Step 2: Use stratification to categorize data by different dimensions
  Step 3: Identify which dimension the anomaly occurs in
  Step 4: Implement targeted improvements

→ The control chart tells you "there is a problem"
→ Stratification tells you "where the problem is"

5.3 Control Chart + Pareto Chart

Combined use:

  Step 1: Use a Pareto chart to identify the "vital few" problem types
  Step 2: Establish control charts for key problem types
  Step 3: Real-time monitoring of the incidence rate of key problems
  Step 4: Compare control chart changes after implementing improvement measures

5.4 Control Chart + Scatter Plot

When the control chart detects a process anomaly:

  Step 1: The control chart identifies an abnormal point or trend
  Step 2: Use a scatter plot to analyze the relationship between suspicious variables and process output
  Step 3: Identify the variable causing the anomaly
  Step 4: Adjust the variable to restore process stability

Chapter Six: Advanced Usage of Control Charts

6.1 Before and After Improvement Control Chart Comparison

Pre-improvement control chart:
  ── Points exceed control limits
  ── Trends or periodicity exist
  ── Process is out of control

Post-improvement control chart:
  ── All points within control limits
  ── Points are randomly distributed
  ── Control limits are significantly narrower

→ Comparing the two charts side by side → Intuitively demonstrates the improvement effect

6.2 Short-Cycle Control Charts

Traditional control charts require a lot of data
Short-cycle control charts are suitable for small batch, multi-variety production environments

Methods:
  ── Standardize data (convert data from different products to Z-values)
  ── Target control chart (use deviation values to draw control charts)
  ── Pre-control chart (can judge with a small amount of data)

6.3 Control Charts in the Digital Age

Limitations of traditional control charts:
  ── Manual plotting, lagging
  ── Manual judgment, dependent on experience
  ── Independent existence, not integrated with other systems

Digital control charts:
  ── Automatic data collection: MES/SCADA systems automatically collect process data
  ── Automatic plotting: Data is displayed in real-time on the control chart
  ── Automatic out-of-control detection: The system automatically detects the eight out-of-control rules
  ── Automatic alerts: Notifications are sent automatically when anomalies occur
  ── Automatic analysis: Fishbone diagrams and scatter plots are linked to assist in root cause analysis

Chapter Seven: Evaluation Standards for Control Charts

Summary: The "Art" and "Science" of Control Charts

Science (How to use):
  ── Choose the right type of control chart
  ── Correctly calculate control limits
  ── Plot points in time sequence
  ── Use out-of-control rules for judgment

Art (Why to use):
  ── Not just to draw a "chart with control lines"
  ── To shift from "post-event analysis" to "pre-event prevention"
  ── To detect signals before the process goes wrong

The greatest value of the control chart is not "monitoring," but "predicting."

By the time you discover a batch of nonconforming products, the damage has already been done. The control chart alerts you before the first nonconforming product is produced.

This is the shift from "firefighting" to "fire prevention."

Document Version: v1.0
Generated Date: 2026-05-03
Author: Excellence Quality Think Tank

Issue 7: Control Chart (Control Chart / 管制图)