Working Mechanics of D.O.E

<<<<<< Page 4 / 5 >>>>>>

 

Mean Vs. Variability

Analysis of mean is an excellent way of accumulating process knowledge, at the same time a poor means of process control if performed without considering variability. In many cases, however, analysis of mean may be sufficient to yield improvements. This is because by the time one gets to a stage of deciding between mean or variability analysis, much of the benefits from structured experimentation have already been realised. Mean analysis is based on the average value of the repetitions of a single run. Variability analysis is based on the variations between the repetitions of a run.

Run	CONTROL FACTOR				RESULTS Weight (g)
No		X1	X2	e	r1    r2   r3	Mean	Stdev
1		1	1	1	4     5    6	5	1
2		1	2	2	4     6    8	6	2
3		2	1	2	6     6    9	7	1.73
4		2	2	1	7     8    9	8	1

Standard deviation in the above Table is only an example of variability. In D.O.E Signal-to-Noise Ratio (S/N) is preferred. It is an index of variability (developed by Genichi Taguchi). It is a logarithmic statistical measure of performance used in evaluating the quality of the product in relation to the effect of noise. The larger the S/N ratio the smaller the variability. There are different S/N ratio for each of the types of characteristics, i.e., Nominal-The-Best, Larger-The-Better, and Smaller-The-Better