Introduction: Why Do We Need Quality Assurance?
Welcome to the study guide on Control Charts and Quality Assurance! Imagine you buy a bag of crisps labelled \(35\text{ g}\). When you open it, do you expect it to weigh exactly \(35.000\text{ g}\)? Probably not! Some bags might weigh \(35.2\text{ g}\), while others weigh \(34.8\text{ g}\). A tiny bit of variation is completely normal in manufacturing.
However, what happens if the machine breaks down and starts packing bags with only \(20\text{ g}\) of crisps, or overfilling them to \(50\text{ g}\)? The company would either upset customers or lose a lot of money! This is where Quality Assurance and Control Charts come to the rescue. In this chapter, you will learn how statisticians use sample data over time to decide if a manufacturing process is running smoothly or if it needs to be stopped and fixed.
Don't worry if this seems tricky at first—we will break down every line, limit, and graph step-by-step!
1. Key Concepts in Quality Assurance
Before we draw or read graphs, let's understand why we monitor processes in the first place.
Why Not Test Every Single Item?
You might wonder: Why don't factories just check every single item they make?
There are two main reasons for this:
• Cost and Time: Checking millions of items one by one would take far too long and cost too much money.
• Destructive Testing: Some tests destroy the product! For example, to test the lifespan of a matchstick or a firework, you have to light it. If you test every match, you have nothing left to sell!
Instead, factories take regular, small samples over time (for example, choosing \(5\) bottles every hour) and calculate summary statistics like the sample mean \(\bar{x}\) or the sample range \(R\).
Types of Variation
In any production line, variation occurs for two main reasons:
• Natural (Common Cause) Variation: Small, random differences that are unavoidable, such as tiny fluctuations in room temperature or slight vibrations in a machine. The process is still working correctly.
• Special (Assignable Cause) Variation: A specific problem that causes a noticeable change, such as a blunt cutting blade, a worn-out gear, or incorrect settings entered by an operator. This needs to be investigated and corrected!
Quick Takeaway: Quality Assurance uses regular samples to spot special problems early without needing to test every single item produced.
2. The Anatomy of a Control Chart
A Control Chart (or process control chart) is simply a time-series graph used to plot sample statistics (usually the sample mean \(\bar{x}\)) over time. It has a horizontal axis showing the sample number or time, and a vertical axis showing the measurement.
A standard control chart contains five horizontal reference lines:
1. Target Line (Process Mean)
This is the middle line, showing the desired or expected value (the ideal average measurement, \(\mu\)). For example, if bottles should contain \(500\text{ ml}\), the target line is drawn at \(500\text{ ml}\).
2. Warning Limits (Inner Limits)
These two lines are drawn above and below the target line:
• Upper Warning Limit (UWL)
• Lower Warning Limit (LWL)
These limits are typically set at \(2\) standard errors from the target line. Under normal conditions, about \(95\%\) of sample means should fall inside the warning limits. Only \(5\%\) of samples (or \(1\) in \(20\)) are expected to fall outside them purely by chance.
3. Action Limits (Outer Limits)
These two lines are drawn further out, at the very top and bottom of the chart:
• Upper Action Limit (UAL)
• Lower Action Limit (LAL)
These limits are typically set at \(3\) standard errors from the target line. Under normal conditions, about \(99.8\%\) of sample means should fall inside the action limits. It is extremely rare (only about \(0.2\%\) or \(1\) in \(500\) samples) for a point to fall outside the action limits by pure chance.
Memory Aid: The Traffic Light Analogy
To remember how these limits work, think of a traffic light:
• Between Warning Limits: GREEN — Everything is fine. Keep the process running.
• Between Warning and Action Limits: AMBER — Be cautious! Take another sample immediately to double-check.
• Outside Action Limits: RED — Stop! The process is out of control. Halt production and fix the problem.
Quick Takeaway: A control chart has a Target line in the centre, two inner Warning Limits (around \(95\%\) confidence), and two outer Action Limits (around \(99.8\%\) confidence).
3. Interpreting Control Charts: Making Decisions
In your exam, you will often be given a completed control chart and asked whether the process is in control or out of control, and what action the manager should take.
When is a Process "In Control"?
A process is considered in control (operating normally) when:
• All sample points lie comfortably between the Upper and Lower Warning Limits.
• The points scatter randomly above and below the target line with no clear pattern.
When is a Process "Out of Control"? (The Decision Rules)
A process is considered out of control if any of the following happen:
Rule 1: A single point falls outside the Action Limits
What it means: It is almost impossible for this to happen by random chance.
Action required: Stop the process immediately, investigate the cause, and reset/repair the machine before restarting.
Rule 2: Two consecutive points fall outside the Warning Limits
What it means: While one point between the warning and action limit might just be an unlucky \(1\text{-in-}20\) chance, two in a row strongly suggests the machine has drifted.
Action required: When one point crosses a warning limit, take another sample immediately. If the second sample is also outside the warning limit, stop the process and investigate.
Rule 3: A clear trend is visible (e.g., \(5\) or more points steadily increasing or decreasing)
What it means: The measurements are drifting in one direction over time. This often happens when a machine tool is slowly wearing out or a fluid is steadily leaking.
Action required: Inspect the machine and perform maintenance before points cross the action limits.
Rule 4: A sudden shift or run (several points in a row on one side of the target line)
What it means: If \(7\) or \(8\) consecutive points are all above the target line, the average has shifted upwards, even if none have crossed the warning line yet.
Action required: Check machine calibration to bring the average back to the target value.
Rule 5: Cyclic or repeating patterns
What it means: If the points rise and fall in a predictable wave pattern, an external factor may be influencing production (e.g., temperatures rising during the afternoon and cooling at night).
Quick Takeaway: Stop the machine if a point goes beyond the Action Limit, or if two successive points lie between the Warning and Action Limits. Look out for trends and shifts!
4. Step-by-Step Worked Example
A factory fills cartons with apple juice. The target volume is \(200\text{ ml}\). Every hour, a sample of \(4\) cartons is measured, and the sample mean volume is plotted on a control chart.
The chart has limits set as follows:
• Upper Action Limit: \(206\text{ ml}\)
• Upper Warning Limit: \(204\text{ ml}\)
• Target Line: \(200\text{ ml}\)
• Lower Warning Limit: \(196\text{ ml}\)
• Lower Action Limit: \(194\text{ ml}\)
Scenario A:
At \(10\text{:00}\), the sample mean is \(201\text{ ml}\). At \(11\text{:00}\), the sample mean is \(205\text{ ml}\).
Question: What should the quality manager do at \(11\text{:00}\)?
Answer: The point \(205\text{ ml}\) lies between the Upper Warning Limit (\(204\text{ ml}\)) and Upper Action Limit (\(206\text{ ml}\)). The manager should take an immediate second sample to check if the process has truly drifted. Do not stop the production line just yet unless the next sample also exceeds the warning limit.
Scenario B:
The follow-up sample at \(11\text{:15}\) gives a mean of \(205.5\text{ ml}\).
Question: What action should now be taken?
Answer: Since two consecutive sample means have fallen outside the Upper Warning Limit, the process is out of control. Stop the production line immediately and recalibrate the filling machine.
Scenario C:
At \(14\text{:00}\), a sample mean is recorded at \(193\text{ ml}\).
Question: What action is required?
Answer: The value \(193\text{ ml}\) is below the Lower Action Limit (\(194\text{ ml}\)). The manager must stop the machine immediately, find the fault, and inspect cartons produced recently.
5. Common Mistakes to Avoid in Exams
• Mixing up Warning and Action Limits: Remember that Warning Limits are inner (closer to the target, \(95\%\)), while Action Limits are outer (further from the target, \(99.8\%\)).
• Stopping the machine too quickly: A single point between the Warning Limit and the Action Limit is an alert to take another sample, not to shut everything down immediately.
• Ignoring trends: Even if all points are inside the Warning Limits, a continuous upward or downward slope indicates that the process is drifting and needs attention.
• Confusing sample mean with individual items: Control charts usually plot the mean of a sample of items, not individual single items.
6. Summary Checklist for Quality Assurance
Before sitting your exam, make sure you can:
• State why sampling is used instead of inspecting every item.
• Identify the Target Line, Warning Limits (\(95\%\)), and Action Limits (\(99.8\%\)) on a chart.
• Explain the difference between natural random variation and assignable causes.
• State clearly what action to take when points lie inside warning limits, between warning and action limits, or beyond action limits.
• Identify out-of-control patterns such as trends, runs, and points beyond action limits.