100-Metodos-de-Qualidade-Total

ANALYTICAL METHODS 105
Method 42
Parameter design
Purpose
To determine which factors are important in the manufacturing process
and to find the optimum set of working conditions.
When to use
When you want to reduce the variability of a process by changing the
variability-control factors, while maintaining the required average performance
through appropriate adjustments to the target-control factors.
How to use
Suppose that we have three components in our process and that we decide
to look at two stages of each component. We then use all eight possible
combinations and make four runs at each of the eight (23 = 8) combinations.
We can analyse the data in two ways. First, we take as the response
the mean of the sets of observations. In the second analysis, we take as the
performance characteristic the estimated variance at each of the eight
combinations.
Benefits
It can provide the means both to reduce costs and improve quality. By
making effective use of experimental design and statistical techniques, one
can identify the settings of easy-to-control product or process parameters.
Example
Consider an electrical power circuit where the characteristic of interest is
the output voltage with a target value of Yo. Assume that the voltage is
largely determined by the gain of a transistor in the circuit whose nominal
value (x) can be controlled. Suppose that the effect of the transistor gain on
the output voltage is non-linear. A transistor with a gain of Xo would
produce the required output voltage of Yo. The effect of a variation about
the nominal value Xo on the resulting variation about Yo is indicated by
bands straddling the nominal values. However, if the circuit designer
chooses a nominal gain of x, then, owing to the non-linearity of response, it
can be seen that the variation about the corresponding voltage, y, is much