Chapter 13 โ Think like a quality analyst. Answer like a pro.
Attribute data ยท Binary (pass/fail) ยท Tracks % defective units
pฬ = total defectives / (n ร k)
UCL = pฬ + 3โ[ pฬ(1โpฬ) / n ]
LCL = pฬ โ 3โ[ pฬ(1โpฬ) / n ] (โฅ 0)
Ex: % wrong orders per day, % defective chips per batch
Attribute data ยท Count ยท Tracks # defects per unit/time
cฬ = total defects / k
UCL = cฬ + 3โ(cฬ)
LCL = cฬ โ 3โ(cฬ) (โฅ 0)
Ex: scratches per windshield, broken chips per bag, errors per report
Variable (continuous) data ยท Subgroups of n โ 2โ10
Xฬ Chart: UCL = Xฬฟ + AโRฬ LCL = Xฬฟ โ AโRฬ
R Chart: UCL = DโRฬ LCL = DโRฬ
Exhibit 13.7 (n=5): Aโ=0.58, Dโ=0, Dโ=2.11
(n=4): Aโ=0.73, Dโ=0, Dโ=2.28
Ex: bag weight (g), fill volume (ml), bolt length (mm)
Measures if process fits within spec limits ยท Must be in control first!
Cpk = min[ (Xฬฟ โ LSL) / 3ฯ , (USL โ Xฬฟ) / 3ฯ ]
Cpk โฅ 1.0 โ process is capable
Cpk โฅ 1.33 โ good standard
Cpk < 1.0 โ process NOT capable
Key: process must be in statistical control BEFORE assessing Cpk
| n | Aโ | Dโ | Dโ |
|---|