00151 žac5d664405d180127a449f8d3c8023

152

Simpson & Keats

values of the ARLs. Table 15 shows the design and performance statistics grouped by size of shift. In this case, 64 runs from the designed experiments were used to calculate each measure. The results indicate that, using this

Table 13. Shewhart and CUSUM Control Chart Sensitiyity Analysis for Smali and Large Shifts (LV example)

Response	R^2		Input Variables
Expected Cost per Time	X	CUSUM	X E C0	C, a	b	Y	6
Smali Shift	0.94	0.91	x“ć X c	X c			X c
Large Shift	0.91	0.90	X C X c	X c			X c
CUSUM Reference Value
Smali Shift	-	0.88					c
Large Shift	-	0.80					c
Decision lnterval
Smali Shift	0.85	0.67				c	X c
Large Shift	0.85	0.60			X c	X	X c
Sampling lnterval
Smali Shift	0.81	0.64	o o	X c	X c		X
Large Shift	0.76	0.76	X c	X c	X c		c
Sample Size
Smali Shift	0.87	0.67	X x c	X		X	X c
Large Shift^1	0.88	-	X X	X		X	X

1

CUSUM model optimal sample size = 1 for 57 of the 65 runs, not enough yariability for a model.