This section discusses the vital concepts of variation and accuracy, and how they relate to a manufacturing process. This excerpt shows how the book uses extensive figures to teach these concepts, and prepares students to understand and use control charts in a manufacturing operation.

Two factors define a manufacturing process' ability to meet specifications. These are (1) variation, and (2) accuracy. Precision is the opposite of variation

A manufacturing operation's goal is to make product that meets the customer's needs. Specifications reflect the customer's needs. Product that is in specification is good, and we can sell it to the customer. Product that is out of specification is bad, and the customer will not accept it. We must rework it, or scrap it. Two factors affect the process' performance. These are (1) variation, and (2) accuracy. It is easy to explain these by treating the specification as a target for a gun. Shots that hit the target are good, while those that miss are out of specification. We have a good idea of what accuracy is. If we are shooting at a target, we want to be aiming at its center. However, even if we are aiming at the center, there is some unavoidable random variation in where the shots hit. Precision is the opposite of variation. A precise process has very little variation. A musket is a smoothbore gun that was common in the 15th through 18th centuries. Soldiers used them in the American Revolution and Napoleonic Wars. A musket has no rifling to make the ball (bullet) spin as it goes down the barrel, so there is a lot of variation in where it hits. It is hard to hit anything with a musket at more than 100 to 150 yards, no matter how good the shooter is. Figure 3-1 shows a simulation of 100 musket shots at a target.

Figure 3-1 Simulation of Musket (high variation tool)

The musket is accurate, because its point of aim is exactly in the center of the target. However, some shots are off the target because of the tool's high variation. (We will later see that variation is relative to the target's size, or specification width.) The musket is not capable of better performance. We cannot improve the performance of a non-capable tool by (1) adjusting it, or (2) looking for a better operator. To improve performance, we must get a better tool, or improve the one we have.

How can we improve this process? Here are three ideas.

1. Adjust the musket's sights to improve its aim.
2. Hire a better marksman.
3. Replace the musket with a rifle.
We've already said that the musket's point of aim is the exact center of the target. Any adjustment will make it worse, not better. A common mistake in industry is to try to improve a process by adjusting it when it does not need adjustment. Terms for this mistake include overadjustment and tampering. [The chapter also shows the "funnel experiment," in which the computer automatically adjusts the aiming point after every shot.]

Hiring a better marksman won't help, either. Putting Daniel Boone or Annie Oakley behind this musket would not improve the results. The computer simulation assumes that a bench rest is holding the musket in place, so there is no operator-induced variation. We also said that the musket is not capable of better performance. Another common mistake is to blame the tool operator for random variation from a non-capable tool.

A rifle barrel causes the bullet to spin as it goes down the barrel. This makes the bullet go in a straight line. If you've seen a quarterback throw a good pass, the football spins and goes in a straight line. If a defender hits the quarterback as he throws the pass, or forces him to hurry, the football may tumble end over end. It usually doesn't go where the quarterback wanted it to go. A rifle bullet in flight acts like a good football pass, and a musket ball acts like a bad one. A spinning projectile has much less variation than a tumbling one. While a musket's extreme range is 150 yards or so, a modern rifle is effective at up to 750 or 1000 yards. Figure 3-2 simulates a rifle, and uses the same target that the musket used.

Figure 3-2 Simulation of Rifle (low variation tool)

The terms "capable" and "not capable" have special meanings in industrial statistics. A tool's process capability index measures its ability to meet specifications. The capability index uses the ratio of the specification width to the variation.

The following figure compares capability to control.

Figure 3-10 Capability and Control

The chapter later places the target figures side by side with the corresponding histograms and control charts. Students learn that the targets, histograms, and control charts all have the same meaning but, in a manufacturing operation, all they can see are the control charts. They must infer the process' condition from the charts.
• "When the point on the x-bar chart is above the upper control limit, the rifle's sights are out of alignment and the aiming point is above the bullseye."
• "When the point on the range chart is above the upper control limit, my rifle has been replaced by a musket: a high variation tool."
Here is the figure that explains why we should not adjust the process after every sample.

Figure 3-12 Overadjustment or Tampering


New (8/5/98)Animated Target Simulator for demonstrating in-control and out of control processes. Animated simulation of 300 shots at a target: an in-control process with Cp=1.33 (capable) is shown side by side with out of control processes of the user's choice. ActiveX, requires Internet Explorer 4.0.  Download via Internet Explorer. (Do not use the Exit button in IE 4.0, use the back key or close the browser. The Exit button goes with a Visual Basic version of the program.)

For more information, or to order the book, click on the following icon. Or contact ASQC Quality Press, 800-248-1946, and ask for item H0937.