The optimization of complementary parts dimensions has been an area of much experimentation. Consider the case of a car body and door dimensions. Many engineers have tried to measure the dimension of the body (
z ) and the dimension of the door ( x ) to calculate the difference
y = z - x
and then set y to some value, usually a tolerance level. Is this a good method
of optimization?
No! Other similar areas include positioning of components on an X - Y table or component.
Is there a better method?
How do we do it?
Our strategy is as follows:
Simply let
- Function z approach its nominal value.
- Function x approach its nominal value.
If z and x share the same variables then re-parameterize z and x into a single performance measure.
|